University of Southern California Dissertations and Theses

Development of carbon molecular-sieve membranes with tunable properties: modification of the pore size and surface affinity

(USC Thesis Other)

Development of carbon molecular-sieve membranes with tunable properties: modification of the pore size and surface affinity

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content DEVELOPMENT OF CARBON MOLECULAR-SIEVE MEMBRANES WITH
TUNABLE PROPERTIES: MODIFICATION OF THE PORE SIZE AND SURFACE
AFFINITY

by

Hui-Chun Jocelyn Lee

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(CHEMICAL ENGINEERING)

December 2010

Copyright 2010 Hui-Chun Jocelyn Lee

ii
Dedication
To my dearest parents Shih-Chuan Lee and Yi-Hsiou Tsai, for believing in me and giving
me unconditional support and encouragement to pursue my own path, and
to my brother Chin-Fu Lee for all his support and encouragement.

iii
Acknowledgements
I would like to convey my sincerest gratitude to my advisors Professors
Muhammad Sahimi and Theodore T. Tsotsis, for giving me the chance to work under
their invaluable supervision. I am especially grateful for their personal guidance,
insightful comments, constructive criticism and patient encouragement during the entire
phase of my studies. This dissertation work would not be in the current state without their
persistent help, support and patience.
I wish to express my deep and sincere thanks to Professor Massoud Pirbazari for
serving on my dissertation committee. I thank him for his support and guidance. I would
also like to thank Drs. Katherine Shing and Ed Goo for serving on my qualifying exam.
I wish to thank Mrs. Tina Silva for her love and support. I am also grateful to my
colleague and officemate for the last 5 years, Dr. Ryan Mourhatch. Special thanks are due
to the administrative staff of the Mork Family Department of Chemical Engineering and
Materials Science, Ms. Karen Woo, Ms. Heather Alexander, and Mr. Brendan Char for
all their help and support throughout my graduate studies, and to all the fellow graduate
students in our research group for their friendly help and fruitful discussions.
Last but not least, my deepest appreciation goes to my parents and my brother
who shared my joy and frustration during the last five years, and gave me their endless
love and support. Their incredible understanding and encouragement made it possible for
me to complete this study.

iv
Table of Contents
Dedication ii

Acknowledgements iii

List of Tables vi

List of Figures ix

Abstract xiii

Chapter I: Introduction 1
1.1. Introduction 1
1.1.1. General Introduction and Background 1
1.1.2. General Gas Separation Mechanisms 7
1.2. Scope of the Present Work 10

Chapter II: Carbon Molecular Sieve Membranes 14
2.1. Introduction 14
2.2. Experimental Techniques 25
2.2.1. Membrane Preparation 25
2.2.2. Transport Investigations 30
2.3. Results and Discussion 32
2.4. Conclusions 47

Chapter III: Pore Size Modification of Carbon Molecular Sieve
Membranes
49
3.1. Introduction 49
3.1.1. Steam Activation 52
3.1.2. Carbon Deposition 58
3.2. Experimental Techniques 63
3.2.1. Membrane Preparation and Modification 63
3.2.1.1. Steam Activation 64
3.2.1.2. Methane Activation 65
3.2.2. Membrane Testing 65
3.3. Results and Discussion 66
3.3.1. Pore Modification via Steam Activation 66
3.3.1.1. Permeation Tests 66
3.3.1.2. Gas Adsorption Analysis 72
3.3.2. Pore Modification via Methane Activation 81
3.4. Conclusions 85

v
Chapter IV: Surface Affinity Modification of Carbon Molecular Sieve
Membranes
87
4.1. Introduction 87
4.2. Experimental Approach 93
4.2.1. Membrane Preparation and Modification 93
4.2.1.1. Impregnation 93
4.2.1.2. Direct Incorporation 95
4.2.1.3. The Sandwiching Method 96
4.2.2. Membrane Testing 97
4.3. Results and Discussion 98
4.3.1. Preparing Ni-CMS Membranes via Impregnation 98
4.3.2. Preparing Metal-CMS Membranes via Direct
Incorporation
104
4.3.2.1. Preparing Metal-CMS Membranes via
Direct Incorporation
104
4.3.2.2. Preparing Pd-CMS Membranes via Direct
Incorporation
113
4.3.3. Preparing Ni-CMS Membranes via the Sandwiching
Method
115
4.4. Conclusions 117

Chapter V: Economic Analysis for Membrane Applications in Post-
Combustion CO
2
Capture
119
5.1. Introduction 119
5.2. Model Development 119
5.3. Results and Discussion 120
5.4. Conclusions 122

Bibliography 124

vi
List of Tables
Table 1.1. Gas separation and membrane applications 2

Table 2.1. Initial characterization of substrates of the first batch of
membranes (permeances in m
3
/m
2
.bar.h)
33

Table 2.2. Permeances and separation factors of the first batch of
membranes (permeances in m
3
/m
2
.bar.h)
34

Table 2.3. Initial characterization of the substrates from the first batch
after each regeneration step (permeances in m
3
/m
2
.bar.h)
38

Table 2.4. Permeances and separation factors for the first batch of re-
used membranes (permeances in m
3
/m
2
.bar.h)
39

Table 2.5. Initial characterization of the second bath substrates
(permeances in m
3
/m
2
.bar)
39

Table 2.6. Permeances and separation factors for the second batch of
membranes at 120 °C (permeances in m
3
/m
2
.bar.h)
40

Table 3.1. Classification of pore system 50

Table 3.2. H
2
, CO
2
, and CH
4
permeances and ideal separation factors
of four 3-layer base-CMS membranes prior to steam
treatment
69

Table 3.3. H
2
, CO
2
, and CH
4
permeances and ideal separation factor
of a 3-layer based-CMS membrane prior to steam
treatment
74

Table 3.4. Estimated total amount of CH
4
deposited within the
membrane as a function of activation period
82

Table 3.5. Permeances and separation factors for membranes
subjected to methane activation at 700 °C (permeances in
m
3
/m
2
.bar.h)
83

Table 3.6. Permeances and separation factors for base-CMS
membranes subjected to methane activation (permeances
in m
3
/m
2
.bar.h)
84

vii
Table 3.7. Permeances and separation factors for base-CMS
membranes subjected to methane activation (permeances
in m
3
/m
2
.bar.h)
85

Table 4.1. Permeances and separation factors measured at 120 °C for
Ni-impregnated membrane – first batch (I-B1)
(permeances in m
3
/m
2
.bar.h)
99

Table 4.2. Permeances and separation factors measured at 120 °C for
Ni-impregnated membrane – second batch (I-B2)
(permeances in m
3
/m
2
.bar.h)
99

Table 4.3. Permeances and separation factors measured at 120 °C for
Ni-impregnated membrane – third batch (I-B3)
(permeances in m
3
/m
2
.bar.h)
100

Table 4.4. Permeances and separation factors measured at 120 °C for
Ni-impregnated membrane – fourth batch (I-B4)
(permeances in m
3
/m
2
.bar.h)
101

Table 4.5. Permeances and separation factors measured at 120 °C of
Ni-CMS membranes prepared by the direct incorporation
method – first batch (D-B1) (permeances in m
3
/m
2
.bar.h)
106

Table 4.6. Permeances and separation factors measured at 120 °C of
Ni-CMS membranes prepared by the direct incorporation
method – second batch (D-B2) (permeances in
m
3
/m
2
.bar.h)
106

Table 4.7. Permeances and separation factors measured at 120 °C of
Ni-CMS membranes prepared by the direct incorporation
method – third batch (D-B3) (permeances in m
3
/m
2
.bar.h)
107

Table 4.8. Permeances and separation factors measured at 120 °C of
Ni-CMS membranes prepared by the direct incorporation
method – fourth batch (D-B4) (permeances in m
3
/m
2
.bar.h)
108

Table 4.9. Permeances and separation factors measured at 120 °C of
Ni-CMS membranes prepared by the direct incorporation
method – fifth batch (D-B5) (permeances in m
3
/m
2
.bar.h)
109

Table 4.10. Permeances and separation factors measured at 120 °C of
one-layer carbon membranes before additional Ni/PEI
coatings (permeances in m
3
/m
2
.bar.h)
112

viii
Table 4.11. Permeances and separation factors measured at 120 °C of
Pd-CMS membranes prepared by the direct incorporation
method – seventh batch (D-B7) (permeances in
m
3
/m
2
.bar.h)
114

Table 4.12. Permeances and separation factors measured at 120 °C of
Pd-CMS membranes prepared by the direct incorporation
method – eight batch (D-B8) (permeances in m
3
/m
2
.bar.h)
115

Table 4.13. Permeances and separation factors measured at 120 °C of
Ni-CMS membranes prepared by the conventional
sandwiching method (S-B1) (permeances in m
3
/m
2
.bar.h)
116

Table 4.14. Permeances and separation factors measured at 120 °C of
Ni-CMS membranes prepared by the conventional
sandwiching method using pyrolysis of metal solution
instead of drying (S-B2) (permeances in m
3
/m
2
.bar.h)
117

Table 5.1. CO
2
recovery per membrane area (kg/m
2
.day) as a function
of CO
2
permeance and separation factor of CO
2
/N
2

(permeances in m
3
/m
2
.bar.h)
121

Table 5.2. Amount of CO
2
recovered (kg/day) and its corresponding
membrane area (m
2
) as a function of CO
2
permeance and
separation factor of CO
2
/N
2
(permeances in m
3
/m
2
.bar.h)
122

ix
List of Figures
Figure 1.1. Classifications of inorganic membranes 3

Figure 1.2. Types of membranes based on morphology 4

Figure 1.3. Transport mechanisms of porous membranes: (a) Knudsen
diffusion, (b) surface diffusion, (c) capillary condensation,
(d) molecular sieving
9

Figure 2.1. Permeability-permselectivity trade-off curve for O
2
/N
2

separation properties of polymeric materials
14

Figure 2.2. Two-layer composite membrane form by coating a thin
layer of a selective polymer on a support
15

Figure 2.3. Idealized structure of a pore in a carbon material 17

Figure 2.4. General structure of PEI 27

Figure 2.5. Ultem
®
1000 resin 27

Figure 2.6. Schematic diagram of the furnace set-up 29

Figure 2.7. Model for membrane transport 30

Figure 2.8. Schematic diagram of the experimental apparatus 31

Figure 2.9. CO
2
and CH
4
permeances of the first batch of membranes
as a function of the number of carbonized layers
35

Figure 2.10. CO
2
/CH
4
separation factor of the first batch of membranes
as a function of the number of carbonized layers.
36

Figure 2.11. CO
2
and CH
4
permeances of the second batch of
membranes as a function of the number of carbonized
layers at 120 °C
41

Figure 2.12. CO
2
/CH
4
separation factors of the second batch of
membranes as a function of the number of carbonized
layers at 120 °C
41

Figure 2.13. CO
2
permeance of membrane 5 as a function of time 43

x
Figure 2.14. CH
4
permeance of membrane 5 as a function of time 43

Figure 2.15. CO
2
permeance of membrane 6 with two carbon layers as
a function of time
44

Figure 2.16. CH
4
permeance of membrane 6 with two carbon layers as
a function of time
45

Figure 2.17. CO
2
permeance of membrane 6 with two carbon layers as
a function of temperature
46

Figure 2.18. CH
4
permeance of membrane 6 with two carbon layers as
a function of temperature
46

Figure 3.1. Mechanisms of a carbon deposition in the pore system of
carbon membranes: (a) Homogeneous carbon deposition
on membrane pore walls; (b) in-layer carbon deposition
on membrane pore wall entrances; and (c) adlayer carbon
deposition outside the membrane pores
58

Figure 3.2. Schematic diagram for steam activation module 64

Figure 3.3. Schematic diagram of methane treatment module 65

Figure 3.4. He permeance of membranes as a function of time and
temperature in steam activation
68

Figure 3.5. Ar permeance of membranes as a function of time and
temperature in steam activation
68

Figure 3.6. H
2
, CO
2
, and CH
4
permeances of membranes as a function
of the duration of treatment
.
70

Figure 3.7. Separation factors of H
2
/CO
2
of membranes as a function
of steam activation period
71

Figure 3.8. Separation factors of H
2
/CH
4
of membranes as a function
of steam activation period
71

Figure 3.9. Pore size distribution of the support substrate 74

Figure 3.10. Nitrogen adsorption isotherms of the support substrate and
CMS membrane
75

xi
Figure 3.11. Nitrogen adsorption isotherms of steam-activated CMS
membrane at each activation time (0 h, 6 h, 12 h, 18 h, and
24 h)
76

Figure 3.12. Nitrogen adsorption isotherms at low relative pressure of
steam-activated CMS membrane at each activation time (0
h, 6 h, 12 h, 18 h, and 24 h)
77

Figure 3.13. The Horvath-Kawazoe cumulative pore volume plot of
steam-activated CMS membrane at each activation time
(from bottom of graph: 0 h, 6 h, 12 h, 18 h, and 24 h)
79

Figure 3.14. The HK median pore diameter of steam-activated CMS
membrane as function of activation time
79

Figure 3.15. The BJH adsorption cumulative pore volume of steam-
activated CMS membrane at each activation time (0 h, 6
h, 12 h, 18 h, and 24 h)
80

Figure 3.16. The BJH adsorption average pore diameter of steam-
activated CMS membrane as function of activation time
80

Figure 3.17. CH
4
permeances of membranes as a function of methane
activation period
82

Figure 4.1. Percentage change in permeances of H
2
and CO
2
as a
function of Ni loading for Ni-impregnated membranes (I-
B4)
103

Figure 4.2. Percentage change in separation factor of H
2
/CO
2
as a
function of Ni loading for Ni-impregnated membranes (I-
B4)
103

Figure 4.3. Permeances measured at 120 °C as a function of number
of layers for membranes prepared by direct incorporation
method – sixth batch (D-B6) (permeances in m
3
/m
2
.bar.h)
110

Figure 4.4. Separation factor measured at 120 °C as a function of
number of layers of membranes prepared by direct
incorporation method – sixth batch (D-B6)
111

xii
Figure 4.5. Permeances measured at 120 °C as a function of Ni
content in 2% PEI solution of membranes prepared by
direct incorporation method (permeances in m
3
/m
2
.bar.h)
112

Figure 4.6. Separation factors measured at 120 °C as a function of Ni
content in 2% PEI solution of membranes prepared by
direct incorporation method
113

xiii
Abstract
Carbon molecular-sieve (CMS) membranes have been studied in the past few
years as an alternative to both inorganic and polymeric membranes. They are known to
have considerable resistance to high temperatures and pressures for gas separation
applications, such as those involving mixtures that contain H
2
, CO
2
, and CH
4
.
In the previous studies as well the preliminary studies in this thesis, the CMS
membranes have been prepared by the carbonization of a polyetherimide precursor
placed on the top of a tubular substrate. The substrate consists of a ceramic support with a
thin layer of γ-alumina on the top. The membranes were prepared by dip-coating the
tubular substrates in polyetherimide solution with the appropriate concentration, followed
by controlled carbonization in an inert atmosphere. The single gas transport properties
and separation characteristics of the resulting CMS membranes were studied, using
various gases. In addition, the thermal stability of the membranes was studied at 120 °C.
A CO
2
/CH
4
separation factor of 25 was obtained at room temperature. The permeance of
CO
2
decreased and that of CH
4
increased as the temperature was raised. As a result, the
corresponding separation factor also diminished.
Developing the ability to tune the membrane’s structure in order to improve its
properties is desirable. So far, tuning of the CMS membrane structure has primarily been
attempted through the selection of the polymeric precursors, and of the appropriate
pyrolysis conditions. However, tuning based on such factors does not provide a generic
solution for the membrane development for a wide range of industrial applications. The
focus of the present work is to develop an alternative technique based on a post-treatment

xiv
step, in order to adjust the pore size distribution of the CMS membrane, which is
accomplished by activating the carbon surface using steam and depositing carbon using
methane. In this study, we investigated the influence of the various activation parameters,
including the temperature and duration of the treatment on the properties of the resulting
CMS membranes. Steam activation is a known technique used to prepare activated
carbons with various pore structures. Application of this technique to the CMS membrane
resulted in the gases permeances increasing significantly after 12 h steam activation at
600 °C, accompanied with a small enhancement of the H
2
/CH
4
separation selectivity. The
resulting CMS membrane was also evaluated using BET, a structural characterization
technique, in order to relate the transport properties to its structure. In addition, methane
activation at 700 °C decreases the gases permeances of the resulting CMS membranes. It
was shown that, relative to hydrogen, the methane treatment of the membrane impacts the
CO
2
permeance, resulting in a H
2
/CO
2
separation of 10. However, the molecular sieving
behavior was shown to diminish after the methane treatment.
We also studied the modification of the surface affinity of such membranes
through the incorporation of metal and solid oxide nanoparticles within the structure.
Three metal precursors were chosen for the preparation of metal-impregnated CMS
membranes, namely, nickel-formate and nickel-acetylacetonate (as Ni precursors), and
palladium acetylacetonate (as a Pd precursor). The choice of the metals was based on
their strong affinity towards H
2
, while the choice of the metal precursors was due to the
fact that they are known to self-decompose to nickel/palladium nanoparticles at elevated
temperatures. Several techniques were utilized to incorporate the metal particles within

xv
the CMS membranes. The transport and separation properties of the resulting CMS
membranes were characterized in terms of their permeability and selectivity using single
gases, such as H
2
, CO
2
, and CH
4
. Compared with the base-CMS membrane, the Ni-CMS
membrane, prepared by a depositing layer (using a 6% PEI solution), and 2 layers using
0.1%Ni/2%PEI solution, had higher gases permeances without showing inferior
separation characteristics.

1
Chapter I
Introduction

1.1. Introduction
In chemical and petroleum processing plants, traditional separation processes
(e.g., distillation) are among the most energy consuming units, and energy consumption,
as a result of using such systems, has increased significantly in recent years. The high
energy consumption related to such processes results not only in high operating costs, but
is also harmful to the environment. As a result, a number of alternative and more energy-
efficient technologies have been developed in recent years in order to replace the
traditional separation processes. Membrane-based separation is one such technology, the
use of which is becoming wider, with pertinent applications in various fields.
1.1.1. General Introduction and Background
There are six major membrane processes that are currently used in the industry,
namely, microfiltration, ultrafiltration, reverse osmosis, electrodialysis, gas separation
and pervaporation. A membrane is generally defined as a semi-permeable thin barrier
between two fluid mixtures with certain selectivity and permeability towards one or more
components of the mixtures. Typically, the passage of some components through the
membrane is easier than that of others, which provides a suitable means of separating the
components of a mixture.
The use of membrane processes, for the separation of gaseous mixtures, has
expanded dramatically during last two decades in industries, such as natural gas

2
processing, landfill gas recovery, air separation, etc. Table 1.1 lists examples of
commercial membrane applications for gas separation [Ismail et al., 2002]. Gas
separation processes require a membrane with both high permeability and selectivity. On
the other hand, typically, there is a trade-off between membrane selectivity and
permeability, and membranes with higher selectivity usually exhibit lower permeability,
and vice versa. This presents a ―grand challenge‖ in the membrane field, namely, how to
improve permeability without simultaneously sacrificing selectivity

Table 1.1 Gas separation and membrane applications

Membranes can be fabricated by using either organic or inorganic precursors, or
both – the latter are called hybrid membranes and consist typically of a thin organic layer
on top of an inorganic support. Organic and inorganic membranes typically operate under
different operating conditions, with inorganic membranes being more appropriate for
operation under high pressure and temperature conditions. The interaction between the
fluid to be separated and the membrane surface also plays an important role when
choosing the right membrane materials. This interaction usually affects the membrane’s

3
resistance to transport, which is encountered by the fluid during the separation. The
intensity of such interacting forces depends strongly on the chemical species involved,
and the nature of the membrane materials [Rios et al., 2002].
Inorganic membranes, which are of primary interest in this thesis, are classified
into two major groups based on their structure: porous inorganic membranes and dense
(non-porous) inorganic membranes [Hsieh, 1990]. The porous inorganic membranes are
further divided into two sub-groups: symmetric and asymmetric.

Figure 1.1 Classifications of inorganic membranes

In summary, and as shown in Figure 1.1, the membranes are classified by:
i. the material used to make the membrane (organic or inorganic)
ii. the membrane structure or morphology (porous or dense, symmetric or
asymmetric), which relates to the way transport occurs through the membrane
Porous membranes allow transport through the pores, whereas dense membranes
enable transport through the bulk of the material. Symmetric membranes generally

4
consist of a uniform structure along the cross section. Asymmetric membranes, on the
other hand, have several layers all with different characteristics. They generally consist of
a thin permselective layer, either dense or microporous, supported on top of a thicker
porous layer [Bhave, 1991; Dalmon, 1997; Dixon, 1999; Drioli & Giorno, 1996; Drioli,
2004; Mulder, 1991; Noble, 2002, 2003].

Figure 1.2 Types of membranes based on morphology

The separation properties of symmetric membranes are determined by their entire
structure. On the other hand, for asymmetric membranes, the separation normally takes
place at the thin surface layer. The advantage of this type of membrane is its good
mechanical strength resulting from the thick support layer, which is able to withstand the
large driving pressure gradients, up to several tens of bar, while simultaneously
maintaining reasonable selectivity.
Most of commercial membrane-based gas separation applications are using
nonporous polymeric membrane [Sedigh et al., 1998]. This type of membranes performs
well in separating molecules with similar size and shape. The separation properties of
nonporous polymeric membrane are dependent upon either the differences in solubility of
the molecules into the polymeric film, or the differences in the diffusivity of the
molecules into the solid-phase, or both. High separation factors for several mixtures of

5
gases have been obtained by utilizing nonporous polymeric membranes. Their low
permeabilities are, however, the main issue in industrial gas separation applications.
Moreover, their low resistance to elevated temperatures makes it impractical to use them
in high-temperature petrochemical applications. Although some glassy polymeric
membranes can withstand high temperatures, the majority of nonporous polymeric
membranes are restricted in their use in the range of low to moderate temperatures. Such
limitations on the use of polymeric membranes have inspired the development of dense
SiO
2
and metal membranes, as well as the development of porous inorganic membranes
as more reliable systems for use at higher temperatures.
Dense membranes, commercially made of palladium and its alloys with various
other metals (e.g., silver), have been used mostly in gas separations. Application of dense
metal membranes is primarily for the highly selective separation of hydrogen from
various gaseous mixtures. Hydrogen is able to transport through the metal lattice, while
other gases cannot. However, the generally low permeability, high cost, and relatively
poor mechanical stability [Sedigh et al., 1998] have limited their potential industrial use.
Recent developments in this area have resulted in the improvement in the properties of
such membranes. Dense SiO
2
membranes, prepared by chemical vapor deposition
(CVD), are similar to the metal membranes in that, they are able to transport hydrogen
while excluding other gases. The only disadvantage of such membranes is their
sensitivity to water vapor which causes irreversible compaction of the pore structure.
Porous membranes may be classified, based on their average pore sizes, into three
categories: macroporous (with pore diameters > 50 nm), mesoporous (with pore

6
diameters between 2 to 50 nm), and microporous (with pore diameters less than 2 nm)
[Karger & Ruthven, 1992]. The earlier porous inorganic membranes were mesoporous γ-
alumina membranes, prepared by sol-gel techniques [Leenaars et al., 1984; Keizer et al.,
1988; Uhlhorn et al., 1992]. The pore sizes of this type of membranes usually range from
25-100 Å . The major driving force of permeation of gaseous molecules through such
membranes is Knudsen diffusion. Although the permeability of such membranes is much
higher than that of dense polymeric membranes, their selectivity, in contrast, is extremely
low, making them unsuitable for application in efficient gas separation systems.
This problem has been addressed by the recent development of microporous
inorganic membranes. Such membranes usually act as sieves for large molecules and
particles. Common materials used for porous inorganic membrane are glass, metal,
alumina, zirconia, zeolite and carbon. They vary greatly in pore size, support material and
configuration. Various crystalline zeolitic membranes are prepared by hydrothermal
routes [Musuda et al., 1994; Yamazaki & Tsutsumi, 1995; Merizudean et al., 1995],
while amorphous silica membranes are made via sol-gel [de Lange et al., 1994, 1995;
Nair et al., 1996; Smaihi et al., 1996; Naito et al., 1997] and CVD techniques [Tsapatsis
& Gavalas, 1992; Lin et al., 1994; Wu et al., 1994; Kim & Gavalas, 1995; Morooka et
al., 1995; Levy et al., 1996]. These membranes have small pore diameters (typically < 10
Å ) relative to the size of molecules to be separated and, therefore, exhibit molecular
sieving behavior. The separation of various gaseous molecules is accomplished as a result
of differing characteristics in terms of entering the pore structure, as well as differences
in the transport and sorption behavior, once they have entered the porous structure. The

7
overall permeation process is normally activated, which means that it is greatly affected
by temperature. Higher temperatures usually favor the permeation of the more mobile
and less adsorbing species. At lower temperatures microporous membranes adsorb and
condense the more readily adsorbing species in a gaseous mixture within the pore
structure and, as a result, this phenomenon prevents the passage of lighter and more
mobile molecules through the membrane. The fact that the larger molecules, which are
often impurities or byproducts, pass through the membrane, while the smaller molecules
remain in the high-pressure retentate side, enables further processing without the extra
compression needed to pressurize the product stream [Funke et al., 1997]. This property
is useful in, for example, the separation of methane from carbon dioxide in reforming
units, and for upgrading landfill gas to salable gas. It is also used in the purification of
hydrogen streams containing hydrocarbon impurities. Unlike polymeric membranes,
inorganic membranes are quite stable at high temperature and, therefore, are suitable for
moderate to high-temperature applications. However, the challenge that remains is the
preparation of defect-free, large area microporous membranes, and also their sensitivity
to various gases commonly encountered in industrial streams must be overcome. For
example, amorphous silica-based membranes lose pore volume when water vapor is
present.
1.1.2. General Gas Separation Mechanisms
As mentioned earlier, the membrane permeation mechanisms can be divided into
two main categories: through the bulk of the material for dense membranes and through
the pores for porous membranes. In the following discussion, detailed description of

8
several membrane transport mechanisms is given. For dense membranes, the
solution/diffusion mechanism is the most common means of describing gas transport. A
gas molecule is first adsorbed on the feed side of the membrane, then dissolved in the
membrane material, followed by diffusion through the structure, and finally desorbed on
the other side of the membrane. Sometimes diffusion through the membrane occurs in the
form of ions, electrons, or constituent atoms, which means that the gas molecules are
required to split-up after adsorption, and to recombine after diffusing through the
membrane [Scott & Hughes, 1996; Shu et al., 1991].
For porous membranes, four types of mechanisms can be used to describe the
transport phenomena, as shown in Figure 1.3. In some cases, the gas molecules’ transport
is not limited to only one mechanism. The four mechanisms vary significantly in
selectivity. Knudsen diffusion dominates in the largest pores and yields relatively low
separation factors, whereas molecular sieving yields high selectivity for the smallest
pores. The separation factor of these mechanisms strongly depends on the pore size
distribution, temperature, pressure, and the interactions between gases being separated
and the membrane surfaces [Dalmon, 1997; Dixon, 1999; Drioli & Giorno, 1996; Drioli,
2004; Mulder, 1991; Noble, 2002, 2003].

9

Figure 1.3 Transport mechanisms of porous membranes: (a) Knudsen diffusion, (b)
surface diffusion, (c) capillary condensation, (d) molecular sieving

Knudsen diffusion, or the so-called free-molecular diffusion, takes place at large
Knudsen numbers. The Knudsen number is defined as the ratio of the mean free path of
the gas molecules and a representative physical length scale of the pore openings. If the
average pore radius is used as a representative physical length scale, Knudsen diffusion
prevails for Knudsen numbers larger than 10. Under such conditions, the lighter
molecules permeate through the pores faster, and the selectivity with respect to the
heavier molecules is calculated as the square root of the ratio of the molar masses of the
gas molecules involved. However, the selectivity is still relatively low compared to that
associated with the other transport mechanisms. For smaller Knudsen numbers,
corresponding to larger pore sizes, the selectivity is even lower. For Knudsen numbers
less than one, for example, the dominant transport mechanism is viscous flow, which is
non-selective.

10
Surface diffusion may happen simultaneously with Knudsen diffusion. Gas
molecules are adsorbed on the walls of membranes’ pores and migrate along the surface.
The strong adsorbing molecules are easier to permeate through the pores, while the
transport of non-adsorbing components is relatively more difficult. Consequently, the
separation factor is generally high. The only drawback of such a mechanism is the
restricted temperature range of operation, and the small pore diameters required for the
effect to become dominant.
Capillary condensation arises when a condensed phase partially fills the
membrane pores. If the pores are completely filled-up with the condensed phase, only the
components that are soluble in the condensed phase can permeate through the membrane.
Normally, capillary condensation exhibits high fluxes and selectivity. However, this
mechanism strongly depends on gas composition, pore size and uniformity of pore sizes.
Molecular sieving can be used to separate molecules with different kinetic diameters. For
molecular sieving to prevail, the pore size must be small enough, that only the smallest
gas molecules can permeate through the membranes, while the other molecules are
excluded [Dalmon, 1997; Dixon, 1999; Drioli & Giorno, 1996, Drioli, 2004; Mulder,
1991; Noble, 2002, 2003].
1.2. Scope of the Present Work
During the past few years, carbon molecular-sieve (CMS) membranes have
attracted considerable attention, due to their effective and selective gaseous separation.
Their performance in separating selected gaseous mixtures, such as H
2
/CH
4
[Sedigh et
al., 1998, 1999, 2000] has been outstanding. To fully exploit the versatility of the CMS

11
membranes, several techniques have been proposed and studied. Tuning of the CMS
membranes has previously been accomplished by the selection of polymeric precursors,
and of the appropriate pyrolysis conditions (temperature, pyrolysis duration, pyrolysis
atmosphere, etc.) This technique’s impact on the pore size and pore size distribution is,
however, not very strong and, therefore, it does not provide a generic means of
membrane development for a broad range of applications.
In this thesis, the focus is on the preparation of a novel class of CMS membranes
with tunable properties. The approach is based on the ability to independently tune the
pore size and the surface affinity of the CMS membranes, so that their performance is
potentially no longer limited by conventional permeance vs. selectivity trade-off
behavior. The pore size distribution is essential in determining the separation
characteristics of the CMS membranes. Sometimes the molecules to be separated vary in
size by only a few tenths of an Angstrom and, therefore, tight control on the pore sizes
cannot be achieved by simply adjusting the pyrolysis condition. Novel techniques must
be applied in order to achieve the kind of structural control that is needed. Besides
employing a tighter control of the pore structure, another focus of this thesis is on
modifying the surface affinity (or surface interaction of molecules with the pore walls)
during the transport process, in order to improve the selectivity characteristics of the
CMS membranes. In the previous mixed-gas permeation experiments by the USC group,
a strong surface affinity of carbon surface toward CO
2
was identified [Sedigh et al., 1998,
1999, 2000]. CO
2
is the only gas that has considerable interaction with the pure carbon
surface in the intermediate range of 120-300 °C. In this thesis, incorporation of metal or

12
metal oxide nanoparticles is employed with the goal of adjusting the surface affinity
toward specific gas components via surface interaction.
Thus, it is the objective of this thesis to fully exploit the versatile features of CMS
membranes. The thesis is divided into three parts:
 Transport and separation characterization of polymeric-based CMS membranes
 Pore size modification of the supported CMS membranes
 Study of the metal-incorporated CMS membranes
Chapter II presents the results of our preliminary experimental study of the
preparation and transport properties of CMS membranes prepared by the pyrolysis of
polyetherimide (PEI) polymeric precursors, supported by a meso-/macroporous support
substrate. Various steps were used to prepare the CMS membranes, including glazing,
polymer coating, and carbonization. Each step can effectively influence the final pore
structure and membrane transport properties.
Chapter III studies fine-tuning of the membrane pore size, and involves the
utilization of both steam and methane activation of the CMS structure as a post-treatment
step. In the first part of Chapter III the carbon surface is made to react with steam at high
temperature, in order to adjust the pore size characteristics of the CMS membrane. The
structural changes in the membrane due to steam activation are discussed and their
relation with the corresponding changes in the permeance and ideal separation factor of
gases through the membranes is studied. In the second part of the chapter, the CMS
membrane is treated with methane at elevated temperatures, with the purpose of

13
decomposing it within the membrane structure to carbon and, thus, favorably modifying
the membrane properties.
Chapter IV focuses on the surface affinity tuning of the CMS membranes toward
selected components, via modification of the carbon surface at the atomic scale with
metals or metal oxide nanoparticles. Several preparation techniques of the metal-
incorporated CMS membranes are discussed. They include (1) impregnation; (2) direct
incorporation, and (3) the so-called ―sandwiching‖ method. Nickel and palladium are
chosen as the test metal components, since they are well-known to exhibit significantly
strong affinity towards H
2
. The impact of metal particles on the permeance and ideal
separation factor of the metal-containing membranes is investigated and compared to the
base-CMS membranes.
Chapter V presents an economic analysis of the membranes for post-combustion
CO
2
capture applications. We describe a mathematical model for estimating the CO
2

recovery and correspondingly required membrane area as a function of the membrane
permeance and selectivity.

14
Chapter II
Carbon Molecular Sieve Membranes

2.1. Introduction
As pointed out in Chapter I, carbon molecular-sieve membranes have been
studied in the past few years as an alternative to both inorganic and polymeric
membranes. The separation performance of polymeric membranes is limited by an ―upper
boundary‖ of the permeability vs. selectivity trade-off relationship [Singh & Koros,
1996]. The CMS membrane, on the other hand, has the potential to exceed the trade-off
boundary.

Figure 2.1 Permeability-permselectivity trade-off curve for O
2
/N
2
separation properties of
polymeric materials

15

The CMS membranes are highly porous materials that contain a distribution of
pore sizes with constricted, microporous apertures having dimensions that are of the same
order of magnitude as the molecular sizes of the diffusing gas molecules. These
membranes have been identified as very promising candidates for gaseous separation,
both in terms of their selectivity and stability. The CMS membranes exhibit better
separation properties than polymeric membranes, while having permeabilities that are
similar to those of microporous inorganic membranes. Such properties make them
potentially beneficial to industrial applications, such as the separation of CO
2
from
gaseous mixtures. The latter separation is of importance in the processing of reformate
mixtures, the upgrading of biogas and landfill gas, and the treatment of flue gas [Sedigh
et al., 1999].
The CMS membranes have been generally prepared by one of following two
ways:
i. Pyrolysis of pre-existing polymeric substrates, such as hollow fibers or self-
supporting thin polymeric films
ii. Pyrolysis of films coated on underlying macro- and mesoporous supports (Figure
2.2)

Figure 2.2 Two-layer composite membrane form by coating a thin layer of a selective
polymer on a support

16

The porous support provides mechanical strength, whereas the separation is mainly
performed by the thin top layer. The supported membrane is generally prepared by
solution coating method, which involves deposition of a dilute polymeric solution onto
the surface of a porous membrane and the subsequent drying of the thin liquid film.
However, in the early stage of membrane development, it is generally very difficult to
produce defect-free thin-film composite membranes with thickness of less than 1 μm by
the solution casting process. Most defects are caused by incomplete coverage of surface
pores in the support membrane after complete evaporation of the solvent. The coating
solution penetrates into the porous support membrane structure and results in the partial
covering of surface pores. Several methods have been proposed to overcome problems
with the formation of the thin, selective layer by solution-coating process. One is to use
ultrahigh molecular weight polymers [Rezac & Koros, 1992]. Another approach is to
fabricate multi-layer composite membranes using a sealing layer to plug the pores in the
support membrane, and to provide a smooth surface onto which the thin coating layer can
be applied [Lundy & Cabasso, 1989].
The carbonization process usually takes place in an inert atmosphere, such as
argon and nitrogen, at temperatures from 500-1000 °C, depending on the polymer and the
desired final membrane structure. The pyrolysis process is conventionally used for the
generation of porous carbon fibers, and causes the product to have a certain
microporosity of molecular dimensions that is responsible for the molecular sieving
behavior. During the pyrolysis process, most of the heteroatoms, which are originally
presented in the polymeric macromolecules, are progressively removed, and only a cross-

17
linked and stiff carbon skeleton remains afterward. The carbon skeleton is responsible for
the membrane’s mechanical properties and separation performance, if a selective layer is
not subsequently added. By judicious control of the concentration of the polymeric
precursor solution, carbonization temperature, and the heating rate, it is possible to
generate defect-free films. Thus, the prepared CMS membranes have an amorphous
porous structure resulting from the evolution of gases, which generate a short-range order
of specific pore sizes [Vu & Koros, 2002]. Generally, the pore system of carbon
membranes is non-homogeneous, as it consists of relatively wide openings with a few
constrictions [Koresh & Soffer, 1980]. An idealized structure of a pore in a carbon
material after pyrolysis is shown in Figure 2.3 [Steel, 2000; Lagorsse et al.,2004].

Figure 2.3: Idealized structure of a pore in a carbon material

The larger micropores, ―D‖ (6-20 Å ) in Figure 2.3, are responsible for diffusion of gas
molecules through the carbon material and connected by the ultramicropores ―d‖ (<10 Å )
that perform the molecular sieving of the penetrating molecules [Centeno & Fuertes,
1999; Singh-Ghosal & Koros, 2002]. Therefore, such combination offers the ability to
perform molecular sieving and high flux characteristics of carbon materials.
Despite their amorphous structure, the structure of polymeric precursors can still
be investigated with the aid of electron microscopy. The sub-domain structures observed
correlate to the types of polymeric precursor utilized and the pyrolysis conditions.

18
Various polymeric precursors perform quite differently in gas separation. The higher the
pyrolysis temperature and the longer the pyrolysis period, the less similarity is found
between the final carbon matrix and the initial structure of the polymeric precursor. In
some instances, the final carbon structure consists of graphite layers at high carbonization
temperature [Sedigh et al., 1999]. Higher final pyrolysis temperatures also tend to yield
more selective membranes, whereas lower final pyrolysis temperatures yield membranes
with higher fluxes [Geiszler & Koros, 1996]. The increase in carbonization temperature
also leads to membranes with higher compactness, and higher crystallinity and density.
The rate of heating to the final pyrolysis temperature, during membrane preparation, also
determines the evolution rate of the volatile components from the polymer precursor
during the pyrolysis process, and consequently affects pore formation. Lower heating
rates are preferable for producing small pores; higher heating rates can lead to the
formation of pinholes and microscopic cracks. Though most of the CMS membranes are
prepared by pyrolysis in inert atmosphere, vacuum pyrolysis was found to produce
membranes with higher selectivity, but with lower permeability. The CMS membranes
with specific pore sizes designed for a particular separation system may not be
appropriate for other separations. In such instances the membranes can always be re-
tailored by varying the numbers of heating cycles, in order to obtain the desired pore
sizes for the given separation [Vu & Koros, 2002]. Research has revealed a range of the
CMS membranes morphologies and permeation properties, depending on the starting
polymeric precursor material (polyimide, polypyrrolone, poly(furfurl alcohol),
polyetherimide, etc.), the precursor geometry, and the pyrolysis conditions. The CMS

19
membranes derived by pyrolyzing polymeric precursors offer a very desirable feature, in
that the final CMS membrane morphology and structure can be adjusted to result in the
desired permeation properties by manipulating the preparation conditions [Vu & Koros,
2002].
As mentioned earlier, the molecular sieving mechanism is able to separate gas
molecules with similar sizes effectively. The carbon matrix in the CMS membranes is
assumed to be impervious, and the permeation is attributed to the pore system. The pore
system is thought to consist of relatively wide openings with narrow constrictions. The
openings contribute the major part of pore volume and are responsible for the adsorption
capacity. On the other hand, the constrictions are responsible for the selectivity and
kinetics of pore penetration [Ismail & David, 2001]. Due to the pore size constrictions,
the interaction energy between the gaseous molecules and the carbon surface is
comprised of both dispersive and repulsive interactions. When the pore dimension
becomes sufficiently small relative to the size of the diffusing molecules, the repulsive
forces dominate and, therefore, the molecule requires activation energy to pass through
the constrictions. In such cases, molecules with only slight differences in size can be
effectively separated by means of molecular sieving. The CMS membranes have
relatively high surface areas and adsorptive affinities towards many molecules, as do
amorphous carbon adsorbents. The selectivity of the CMS membranes depends,
therefore, significantly on the differences in the adsorptive affinity, mostly through Van
der Waals forces, between various species in a mixture. In addition, pore filling and
condensation, and hindered and surface diffusion also have impact on the permeation and

20
molecular sieving properties at lower temperatures. At higher temperatures, the
permeation mechanism is mainly activated diffusion and, as a result, the membrane
allows the smaller molecules to permeate through more efficiently.
In conjunction with theoretical and modeling work, numerous permeation studies
with the CMS membranes have been carried out. Koresh and Soffer [1983, 1987] were
the first to successfully prepare the CMS membranes by pyrolysis of organic polymeric
hollow fiber precursors (cellulosic and phenolic resins). They studied how high
temperature treatment influences the development of the pore structure and found that
high temperature (up to 700 °C) evacuation and mild temperature (up to 450 °C) air
oxidation treatments lead to the opening of the pore structure. A further increase in
temperature, under vacuum or an inert atmosphere, results in a gradual pore closure due
to sintering. This way, by choosing the appropriate sequence of thermal treatments, the
membrane’s pore aperture can be nearly continuously tuned. Based on single gas
permeation tests, they reported a separation factor of 2.0 for the He/N
2
gas pair and a
separation factor of 9.0 for the CO
2
/N
2
pair. Since then, numerous studies on the CMS
membranes have been done. Jones and Koros [1994] prepared the CMS membranes by
pyrolyzing commercially available, asymmetric hollow fiber polyimide membranes. The
carbonization process was performed at two different temperatures, 500 °C and 550 °C.
They reported a separation factor for CO
2
/CH
4
of 190 and 450 for H
2
/CH
4
. Exposure of
the membranes to volatile organic compounds (VOC) at ambient temperatures has the
consequence of losing both their permeability and selectivity [Jones & Koros, 1994].
Carbons generally have non-polar surfaces and, therefore, they are organophilic.

21
Ultramicroporous carbon membranes would be expected to be vulnerable to adverse
effects from exposure to organic contaminants, due to their adsorption characteristics
towards organics. In industrial applications, the gaseous feeds usually contain the VOC,
which may harm the membrane performance even at concentrations as low as 0.1 ppm of
the organics. It was shown, for example, that as the membranes adsorb the organics their
capacity to adsorb other compounds diminishes and membrane performance breaks down
rapidly. However, the same membranes can be regenerated using propylene [Jones &
Koros, 1994]. The propylene most likely acted as a solvent, removing other adsorbed
compounds from the carbon surface.
The same group also studied the effect of humidified feeds on O
2
/N
2
selectivity
and permeability of the CMS membranes, using feeds with relative humidity levels
between 23% and 85% [Jones & Koros, 1995]. The presence of water vapor degraded the
performances of the CMS membranes at all humidity levels. The adsorbed water
molecules attract additional water molecules through hydrogen bonding, leading to the
formation of clusters. The clusters grow and coalesce and, therefore, diminish the
permeation of the other species. Increasing the humidity level increases the losses in the
permeability and selectivity. The sensitivity to humidity at ambient conditions was
prevented by modifying the membrane surface to become more hydrophobic. The
modification is done by coating the membrane with a thin layer of Teflon [Jones &
Koros, 1995]. They also reported that the losses in performance with the presence of
water vapor are relatively smaller at higher temperatures.

22
Geiszler and Koros [1996] studied the effect of the pyrolysis conditions of
polyimide precursor on the CMS membrane properties. They compared the performance
of membranes prepared by vacuum pyrolysis and those prepared by inert gas purge
pyrolysis. In addition, the effects of using various processing temperatures, purge gas
flow rates, and the presence of residual oxygen concentration in the purge gas were
studied. They reported that both H
2
/N
2
and O
2
/N
2
selectivity increased when using
vacuum carbonization. In addition, increasing the purge gas flow rate resulted in
increasing the permeability of O
2
.
Petersen et al. [1997] also prepared a CMS membrane by pyrolysis of a Kapton
hollow fiber membrane at 950 °C. The membranes were subsequently treated by coating
them with a thin layer of poly(dimethyl siloxane) (PDMS) to minimize gas flow through
defects in the structure. Based on single gas permeation tests, the reported CO
2
/N
2

separation factor was approximately 5 at 200 °C. Shusen et al. [1996] reported a simple
method to fabricate asymmetric CMS membranes. A thin, self-supporting film of phenol
formaldehyde resin was pyrolyzed, followed by unequal oxidation. They proposed that
the key to creating an asymmetric CMS membrane is to keep different oxidation
atmospheres on the two sides of the membranes, strong oxidation conditions on one side
and relatively weak oxidation conditions on the opposite side. Based on single gas
permeation tests, a separation factor of 23.6 for the H
2
/N
2
mixture was reported at
ambient temperatures. Kita et al. [1998] synthesized an unsupported CMS membrane by
means of a casting method. They reported that a polypyrrolone thin, flat film, carbonized

23
at 700 °C for 1 h, gave the best performance. A CO
2
/CH
4
separation factor of around 230
was reported for this membrane.
Besides direct carbonization of polymeric films, numerous groups have also
reported the preparation of carbon membranes by carbonization of polymeric films
previously deposited on porous inorganic and metal substrates. Rao et al.

[1994]
fabricated a CMS membrane by carbonization at 1000 °C in nitrogen atmosphere of a
thin, uniform layer of poly(vinylidene chloride)-acrylate terpolymer latex film, deposited
on a macroporous graphite support from an aqueous suspension. The permeation test was
done using single gases (He and H
2
), as well as mixtures of H
2
/hydrocarbon and
H
2
/CO
2
/CH
4
. A separation factor of 2.4 was reported for the CO
2
/CH
4
gas pair in the
ternary mixture.
Hayashi et al. [1995] prepared the CMS membranes by dip coating the BPDA-
4,4’-oxydianiline (ODA) solution on the outer surface of α-alumina tubular porous
supports, followed by pyrolysis at 500 °C to 900 °C in an inert atmosphere. In their study,
the sorption capacity and diffusivity of penetrants in the CMS membranes were greatly
improved, due to the increased micropore volume and segmental stiffness. The separation
factor of CO
2
/CH
4
was about 100 at ambient temperatures. They studied the effect of
pyrolysis temperature on permselectivity and permeance of the membrane. At higher
carbonization temperatures, the selectivity of CO
2
/CH
4
and CO
2
/N
2
binary mixtures
increased, but the permeance for both decreased. They also studied [Hayashi et al., 1996]
the selectivity and permeability of the C
2
H
6
/C
2
H
4
and C
3
H
6
/C
3
H
8
binary mixtures with
membrane pyrolyzed at 700 °C. The reported separation factors were 7, 56 and 40 for the

24
C
2
H
4
/C
2
H
6
, C
3
H
6
/C
3
H
8
and CO
2
/N
2
binary mixtures, respectively. They also modified
[Hayashi et al., 1997] the resulting CMS membranes by the CVD method using
propylene as a carbon source at 650 °C. The CVD modification was effective for
increasing the CO
2
/N
2
selectivity to 73, because the pore structure was further controlled
and the micropores were narrowed.
Centeno and Fuertes [1998] prepared the CMS membranes by carbonization of a
thin phenolic resin films coated on macroporous carbon disks. After preparing a crack-
free carbon disk, they deposited a polymeric film on it by a means of a spin-coating
technique with polyamic acid solution. A rotating speed of 160 rpm was used for coating
the substrate. The polymeric film was then gelled by immersion into a coagulant bath,
either acetone or isopropyl alcohol. The resulting polymeric film was dried in air at room
temperature, followed by drying in air at 150 °C for 1 hour. The film was then imidized
at 380 °C for 1 h and carbonized in vacuum for an additional 1 h. They reported
separation factors of 15 and 25 for the CO
2
/N
2
and CO
2
/CH
4
mixtures, respectively.
Phase-inversion was another technique they used to prepare an asymmetric
polyetherimide-based CMS membrane on macroporous carbon disks [Centeno & Fuertes,
1998]. Acharya et al. [1997] prepared a CMS membrane by pyrolysis of the PFFA
[poly(furfuryl alcohol)] films, deposited on a macroporous sintered stainless steel flat
plate. They measured the permeation of single gases, such as H
2
, He, Ar, O
2
, N
2
, and SF
6
,
to determine the perm-selectivity of the membrane. Separation factors of 2-3 for the
O
2
/N
2
and 30 for H
2
/N
2
were reported at 20 °C.

25
In conclusion, various CMS membrane preparation techniques have been studied
by numerous research groups, and the resulting membranes have exhibited considerable
promise over polymeric membranes, both in terms of their high separation factors and
permeances. There is, however, room for further significant improvement both in terms
of the separation factors, as well as the permeances attained. In this chapter, we present
the results of our fundamental study on the preparation and transport properties of the
CMS membranes by the pyrolysis of polyetherimide polymeric precursors, supported on
meso-/macroporous support substrates. The experimental techniques for membrane
preparation and characterization are described. In addition, extensive permeability tests at
various temperatures and the stability of the resulting CMS membranes are investigated
to offer further insight into the fundamental phenomena that determine the transport
characteristics of the membranes.
2.2. Experimental Techniques
The description of the experimental techniques begins with the most important
step, namely, preparation of the membrane.
2.2.1. Membrane Preparation
As previously noted, the performance of the CMS membranes in terms of their
separation factor, permeance, and chemical and thermal stability is affected by the
molecular structure of the starting polymeric precursor, as well as the experimental
conditions of the coating and carbonization processes. For successful membrane
preparation all the steps must be optimized. Polyimides have attracted recent attention for
the preparation of the CMS membrane, as well as polymeric membranes, due to their

26
superior separation factors, permeance, and stability. Tanaka et al. [1992] prepared 18
different polyimides with various dianhydride and dianiline moieties and found that
changing the moieties would vary a free space in the range of 0.12-0.19. For example,
polyimides containing C(CF
3
)
2
groups in the dianhydride or dianiline structure exhibited
a free space of 0.16-0.19. Larger free volume typically improves the permeance through
the membrane. In most cases, however, the increase in the permeance is typically
accompanied by a simultaneous decrease in the separation factor. This may be avoided by
introducing stiff molecular units into the polymer, in order to restrict the rotation mobility
of the structure. Tanaka et al. [1992] also showed that the introduction of a rigid and
bulky dianiline structure inhibited the rotation around the -CH
2
- and -O- linkages, and
increased the permeance with minimal loss in the separation performance. Several
studies indicated that it is essential to have a molecular structure with a high free volume,
together with restricted intra- and inter-segmental mobility in the starting polymeric
precursor, in order to improve the performance of the membrane, both in terms of
permeance and separation factor [Kim et al., 1988; Sedigh et al., 1999; Tanaka et al.,
1992]. In the preliminary experimental study polyetherimide (PEI) was chosen as the
polymeric precursor for the preparation of the membranes. The PEI is an amorphous
polymer with a glass transition temperature T
g
near 200 °C. It offers high strength and
excellent flame and heat resistance. The PEI can be produced by either a nucleophilic
substitution reaction process, or by the more conventional technique involving the
condensation of diamines and dianhydrides. The general structure of the PEI is given in
Figure 2.4.

27

Figure 2.4 General structure of PEI

Many different resins can be prepared through the use of various groups in place of R and
R’. The most popular one is Ultem
®
1000, supplied by General Electric (GE), with the
structure given in Figure 2.5.

Figure 2.5 Ultem
®
1000 resin

The presence of a bulky C(CH
3
)
2
group, along with highly rigid supporting structure of
the polymer, suggests that this resin is a very promising candidate for the preparation of
polymeric membrane. After the carbonization, the carbon backbone of the starting
polymer remains, most likely, fairly intact, and as the USC group [Sedigh et al., 1999]
showed previously, the CMS membranes, fabricated by using the pyrolysis of the PEI,
maintain the superior performance of the polymeric precursor.
To prepare a PEI solution, the Ultrem
®
1000 resin was dissolved in 1, 2-
dichloroethane (DCE) using a stirrer for 24 h with a slight heat applied in the early stages
of dissolution. Two different solution concentrations were prepared, namely 6% and 2%
by weight. As previously observed [Sedigh et al., 1998, 1999], the concentration of the

28
polymeric precursor solution dominates the thickness and the location of the selective
layer that is formed. Thus, it is important to choose an appropriate concentration to
optimize the performance. The 6% solution was, typically, used for the initial coating,
followed by additional coatings with the 2% solution.
The ceramic tubular supports that we have used in our preliminary experimental
investigations were manufactured by Media and Process Technology, Inc., M & P
(3.5mm ID, and 6mm OD) with a thin inside layer of γ-alumina (about 1-2 µm in
thickness and an average pore diameter of 40 Å ). The substrates were cut into pieces,
each 8 cm long. Both their ends were glazed (approximately 1.5 cm from each end) with
Duncan GL Ultra clear glaze, and dried inside the oven to produce impermeable ends.
During the glazing procedure, the substrates were heated up to 550 °C with a heating rate
of 2 °C/min, and then to 850 °C with a rate of 1 °C/min for 1 h. Each substrate was
glazed twice in order to ensure that the ends were fully impermeable, before further
processing.
Before coating the resulting substrates with the polymeric precursor layers, the
permeances of such gases as He and Ar (as well as N
2
for some membranes) through the
support were measured. The ideal Knudsen selectivity for He/Ar is 3.16, but in selecting
substrates for further coating a He/Ar separation factor of >2.5 was deemed adequate
(when measuring the N
2
permeance a He/N
2
>2.2-2.4 is considered adequate - the ideal
Knudsen value for He/N
2
is 2.64). The lower limits in the separation factors are thought
to be indicative of relatively crack-free supports with adequately glazed membrane tips.
Membranes that failed such tests were re-glazed, and then tested again. If the same result

29
was obtained, the membranes were identified as having structural defects, and were
discarded.
To prepare the CMS membranes using the glazed substrates, their outer surface
was wrapped with Teflon tape. The substrate was then dip-coated with a 6% PEI/DCE
solution for 3 min and was pulled out from the solution at a constant rate of 2 cm/min. As
mentioned earlier, for the successive layers all the substrates were dip-coated with a 2%
PEI/DCE solution for 3 min. After coating with the PEI film, the membrane was stored in
the incubator and dried in air for 24 h. The membrane was then carbonized in a
cylindrical furnace (2 in diameter, and 2 ft long, controlled using two external heating
elements and a programmable Omega CN3000 controller, as seen in Figure 2.6) in the
presence of flowing ultra high pure (UHP) argon, at a rate of 60-75 cm
3
/min, to remove
all the gases evolved during the pyrolysis process and to maintain an inert atmosphere.
Ceramic Heater Ceramic Heater
600
Thermocouple
Temperature
Controller
Membrane
Gas in Gases out
Furnace

Figure 2.6 Schematic diagram of the furnace set-up

The carbonization process was carried out in two stages. The membrane was
initially heated up to 350 °C with a heating rate of 1 °C/min and held there at constant
temperature for half an hour. The temperature was then increased to 600 °C at a rate of 1

30
°C/min, where the membrane was held for 4 h. Subsequently, it was cooled down to 180
°C at a rate of 2 °C/min, and further down to room temperature at 5 °C/min. The
coating/carbonization procedures can be repeated as many times as needed to modify the
selective layers and achieve the desired membrane performance in terms of separation
factor and permeance. After carbonization, the membranes are ready to be tested. In these
preliminary investigations we only tested the membranes by single gas permeation using
CH
4
and CO
2
in order to evaluate their performance.
2.2.2. Transport Investigations
The experimental cell used to study the membrane transport properties is shown
schematically in Figure 2.7.

Figure 2.7 Model for membrane transport

The gas is fed on one side of the membrane, called the feed or upstream side, transports
through the membrane and emerges on the opposite side, the so-called permeate or
downstream side. The transport through the membrane is the result of a concentration or
a pressure gradient. Membrane performance and efficiency in this preliminary study are
tested in terms of the flux through the membrane and its ideal perm-selectivity towards
two test gases (CH
4
and CO
2
). The experimental apparatus is presented in Figure 2.8. It

31
consists of a gas delivery system, the membrane testing module, a heating and
temperature control system, and pressure measurement devices.

Figure 2.8 Schematic diagram of the experimental apparatus

The module is a stainless-steel tee casting in which the membrane is placed. The
membrane ends are sealed by silicon O-rings wrapped around the tube. The
measurements were taken both at room temperature (~25 °C), as well as at 120 °C. For
experiments at higher temperatures, a heating tape was wrapped around the outside
surface of the module, and a wire thermocouple (Omega) is inserted inside the membrane
tube. A programmable temperature controller (Omega CN8200) is utilized for ramping
up to measurement temperature at a constant rate of 1 °C/min.
Methane, carbon dioxide, argon, and helium were supplied from high-pressure
gas cylinders with pressure regulators. Although all the gas cylinders were UHP, they
were still passed through a purifier column, in order to remove any contaminants, such as
water vapor or hydrocarbons, that may be present. The gases were fed at a pressure of
around 40 psig; however, the tube side pressure was controlled with a needle valve on the
tube side outlet, and was normally adjusted at a pressure of around 30 psig. The pressure

32
on the permeate side was maintained at atmospheric value. The flow rate of the outlet gas
from the permeate side was measured with a soap bubble flow-meter.
Membrane permeances were calculated by the following equation:

where P
j
is the permeance of species j in units of m
3
/m
2
.bar.h, V the volumetric flow rate
of the gas across the membrane in m
3
/h, P
m
the pressure in psia at which the volumetric
flow of permeate side was measured, T
m
the temperature where the measurements were
taken in K, L the membrane length in m, R the membrane inner radius in m, P
o
= 14.696
psia, and T
o
= 273.15K. ΔP is the pressure difference inside the tube in bar. The
separation factor S
ij
of species i with respect to species j is defined as follows:

where P
i
and P
j
are the permeances of species i and j, respectively.
After each membrane was prepared, the permeance of single gases, CH
4
and CO
2
,
was measured at room temperature and at 120 °C.
2.3. Results and Discussion
All the membranes in this study were prepared based on the preparation method
that was described in earlier section, using substrates manufactured by the M & P. All the
membranes were characterized by single gas permeation. As previously noted, the
membranes were tested at room temperature, but also at 120 °C. The initial substrates
were also characterized by measuring their Ar and He permeances and the corresponding
ideal He/Ar separation factor. In the following, we discuss the results of our preliminary

33
investigations on the effect of the temperature and the number of coating/carbonzation
steps on the permeance and the separation factor for the CMS membranes.
Two batches of alumina substrates were obtained at different times. The first
batch contained 4 and the second 6 substrates. Table 2.1 summarizes the Ar and He
permeances and corresponding He/Ar separation factors for the initial substrates in the
first batch of membranes. All substrates had an ideal separation factor that was >2.5.
Membrane ΔPmax Ar (psig) Δpmax He (psig) P
Ar
P
He
S(He/Ar)
1 36.5 38 39.49 110.62 2.8
2 39 39 38.25 110.79 2.9
3 37.4 38 36.26 92.31 2.5
4 38 38 38.10 96.02 2.5

Table 2.1 Initial characterization of substrates of the first batch of membranes
(permeances in m
3
/m
2
.bar.h)

Though these substrates came from a single 25 cm membrane, note that their transport
characteristics are different. This may be due to leaks at the membrane seals (which are
convective and, therefore, non-selective), but also potentially due to the nonuniform
thickness of the γ-alumina layer along the length of the original tubular substrate. Such
thickness differences are of concern as they can potentially influence the quality of the
final prepared membranes.
After the initial characterization, the substrates were dip-coated with 6%
PEI/DCE solution and carbonized to create the first CMS membrane film, following the
procedure described in the previous section. After the first layer was created, they were
subsequently dip-coated with 2% solution and carbonized to prepare additional layers.
Single gas (CO
2
and CH
4
) permeation tests were carried out after each pyrolysis step at

34
room temperature (~25 °C) to evaluate membrane performance in terms of the permeance
and the corresponding separation factors.
Table 2.2 shows the CO
2
and CH
4
permeances and ideal CO
2
/CH
4
separation
factors after each CMS membrane layer was added for the four substrates in Table 2.1.
Membrane 1 Membrane 2
Layer 1
st
2
nd
3
rd
1
st
2
nd
3
rd

P
CO2
12.90 2.60 0.77 12.08 2.62 0.80
P
CH4
10.08 0.42 0.18 8.63 0.10 0.05
S(CO
2
/CH
4
) 1.3 6.2 4.2 1.4 26.4 17.4

Membrane 3 Membrane 4
Layer 1
st
2
nd
3
rd
1
st
2
nd
3
rd

P
CO2
8.74 3.37 1.21 8.38 2.50 0.84
P
CH4
3.45 0.21 0.08 4.11 0.18 0.03
S(CO
2
/CH
4
) 2.5 16.2 14.8 2.0 14.1 25.5

Table 2.2 Permeances and separation factors of the first batch of membranes (permeances
in m
3
/m
2
.bar.h)

Two of the prepared membranes achieved a CO
2
/CH
4
selectivity higher than 25.
However, adding additional CMS layers did not, in some cases, improve the CO
2
/CH
4

separation factor. For example, from the data shown in the table, three out of four
membranes attained their best performance after 2 cycles of carbonization. Only
membrane 4 reached its optimum performance after coating and carbonization of the
third carbon layer. This behavior was consistent with the previous results reported by the
USC group [Sedigh et al., 1998, 1999, 2000].
As expected, and also shown in Table 2.2, the membrane permeances and
separation factors are a strong function of the number of coating/carbonization steps.
Adding extra carbon layers to the CMS membrane does result in a decrease in the
permeance. However, the tradeoff between the separation factor and the permeance is not

35
always observed. Normally, as the permeances of CO
2
and CH
4
decrease, one would
expect the separation factor to increase, but this was not always true in our preliminary
investigation, as can be seen in Table 2.2.

Figure 2.9 CO
2
and CH
4
permeances of the first batch of membranes as a function of the
number of carbonized layers

36

Figure 2.10 CO
2
/CH
4
separation factor of the first batch of membranes as a function of
the number of carbonized layers

In the experiments with the first batch of the membranes we noted that the permenaces of
CO
2
and CH
4
for all the CMS membranes drop dramatically after the second carbon layer
is added. Between the second and third layer the permenaces also decrease, but not as
strongly as in between the first and the second layer. For all the membranes the CO
2
/CH
4

separation factor increases after two carbonization cycles. The increase in the separation
factor may be attributed to pore narrowing, but more likely because the additional
coatings tend to repair the existing pinholes and cracks in the previous layer.
In this initial study we also evaluated the ability to re-use some of the original
substrates. The reason for that is twofold. First, in practice it is important to be able to
repair the membranes after they are damaged, and being able to utilize the original
substrate will significantly decrease the cost of repairing the membranes. In addition, it is
0
5
10
15
20
25
30
1 2 3 4
CO
2
/CH
4
Separation Factor
Membrane Number
First layer
Second layer
Third layer

37
important for our own study to be able to re-use the substrates. In our study of the
feasibility of being able to reuse the substrates after regeneration, the carbon layers were
removed in three steps. The CMS membranes were placed inside the oven in air and
heated to 300 °C at a rate of 10 °C/min, and were held at the constant temperature for 1 h.
Subsequently, the temperature was raised up to 600 °C at a rate of 10 °C/min, at which
the membranes were held for an additional 3 h. After that, the substrates were gradually
cooled down to room temperature. The heating treatment resulted in the carbon films in
the CMS membranes to burn-off, as one can easily observe by a visible change in the
membrane color. Subsequently, the substrates were washed with acetone via sonication,
and were dried in air at 150 °C for an additional 5 h.
The regenerated substrates were again characterized by measuring their Ar and
He permeances and the corresponding ideal He/Ar separation factor. Table 2.3
summarizes the results after several regenerations of the substrates from the first batch.

38

Regeneration Initial 1
st
Time 2
nd
Time
1
P
Ar
39.49 34.06 36.85
P
He
110.62 85.81 85.96
S(He/Ar) 2.8 2.5 2.3
2
P
Ar
38.25 37.25 37.42
P
He
110.79 96.28 88.05
S(He/Ar) 2.9 2.6 2.4
3
P
Ar
36.26 32.74
P
He
92.31 81.56
S(He/Ar) 2.5 2.5
4
P
Ar
38.10 39.85
P
He
96.02 98.98
S(He/Ar) 2.5 2.5

Table 2.3 Initial characterization of the substrates from the first batch after each
regeneration step (permeances in m
3
/m
2
.bar.h)

Results from the table indicate that there is no particular trend that is followed by the
permeances of Ar and He. They may increase or decrease after each regeneration step.
For some of the substrates there is a diminution in the He/Ar separation factor after each
regeneration step. This phenomenon can possibly be the result of the modification of
structure of the γ-alumina layer after each heat treatment. Since this is likely to affect the
properties of the resulting CMS membranes, we further studied the issue using the
regenerated substrates in the preparation of the CMS membranes, following the
procedure described previously. To characterize the membranes we performed single-gas
permeation tests at 120 °C. The permeances of CO
2
and CH
4
and the corresponding ideal
separation factors are summarized in Table 2.4.

39

Membrane 1 Membrane 3
Layer 1
st
2
nd
1
st
2
nd
3
rd

P
CO2
2.74 0.35 1.20 0.23 0.06
P
CH4
1.84 0.24 0.53 0.07 0.03
S(CO
2
/CH
4
) 1.5 1.5 2.3 3.4 1.7

Table 2.4 Permeances and separation factors for the first batch of re-used membranes
(permeances in m
3
/m
2
.bar.h)

It is apparent that the permeance and selectivity are lower when compared to the ones in
Table 2.2. One reason for that is the effect of higher temperature of the measurement that
is discussed below. The lower selectivity is also probably due to the regeneration of the
substrate, due to defect formation on the γ-alumina layer within the substrate.
The second batch of the membranes from the M & P underwent the same set of
initial characterization experiments as the first batch. The Ar and He permeances, as well
as the He/Ar separation factors, for all the substrates in the second batch are presented in
Table 2.5.
Membrane P
Ar
P
He
S(He/Ar) P
N2
S(He/N
2
)
1 46.18 119.48 2.6
2 46.32 123.64 2.7
3 42.71 118.58 2.8
4 39.83 105.11 2.6
5 42.14 125.15 3.0 51.75 2.4
6 38.43 114.55 3.0 47.82 2.4

Table 2.5 Initial characterization of the second bath substrates (permeances in m
3
/m
2
.bar)

Once again, all the substrates had separation factors higher than 2.5. The permeances
from the second batch of the substrates are slightly higher than those in the first batch,

40
with the Ar permeances varying from 39 to 46 m
3
/m
2
.bar.h, compared to 35 to 39
m
3
/m
2
.bar.h from before.
After the initial characterization, all the substrates were again dip-coated with a
6% PEI/DCE solution and carbonized to prepare the first carbon layer. Subsequently,
they were dip-coated with a 2% solution and carbonized to prepare any additional layers.
The single-gas permeation tests with CO
2
and CH
4
for the second batch of the
membranes were also taken at 120 °C are shown in Tables 2.6, for the second batch of
the membranes after each different carbonization cycle. The same data are also shown
graphically in Figures 2.11 and 2.12.
Membrane 1 Membrane 2
Layer 1
st
2
nd
3
rd
1
st
2
nd
3
rd

P
CO2
6.32 1.05 0.97 5.66 0.79 0.33
P
CH4
5.68 0.96 0.62 5.41 0.58 0.11
S(CO
2
/CH
4
) 1.1 1.1 1.6 1.0 1.4 3.0

Membrane 3 Membrane 4
Layer 1
st
2
nd
3
rd
1
st
2
nd
3
rd

P
CO2
4.53 0.78 0.16 5.45 0.17 0.49
P
CH4
5.02 0.71 0.08 1.56 0.09 0.085
S(CO
2
/CH
4
) 0.9 1.1 2.2 3.5 1.9 5.7

Membrane 5 Membrane 6
Layer 1
st
2
nd
3
rd
1
st
2
nd
3
rd

P
CO2
10.11 6.78 5.59 1.22 0.77
P
CH4
10.10 5.85 2.98 0.26 0.15
S(CO
2
/CH
4
) 1.0 1.2 1.9 4.8 5.3

Table 2.6 Permeances and separation factors for the second batch of membranes at
120 °C (permeances in m
3
/m
2
.bar.h)

41

Figure 2.11 CO
2
and CH
4
permeances of the second batch of membranes as a function of
the number of carbonized layers at 120 °C

Figure 2.12 CO
2
/CH
4
separation factors of the second batch of membranes as a function
of the number of carbonized layers at 120 °C
0
2
4
6
8
10
12
1 2 3
Permeance (m
3
/m
2
.bar.h)
Number of Carbon Layers
M1_CO₂
M1_CH₄
M2_CO₂
M2_CH₄
M3_CO₂
M3_CH₄
M4_CO₂
M4_CH₄
M5_CO₂
M5_CH₄
M6_CO₂
M6_CH₄
0
1
2
3
4
5
6
7
1 2 3 4 5 6
CO
2
/CH
4
Separation Factor
Membrane Number
First layer
Second layer
Third layer

42

Unfortunately, as the data indicate, both the permeances and separation factors obtained
with the second batch of the substrates are lower than those attained in the first batch. It is
not clear to us why this is so, based simply on the characterization of the transport
properties of the initial substrates that appear to be very similar to each other. This
behavior may probably be due to the higher temperature of the measurements or any
defect formation during the carbonization process, which results in potential degradation
of the quality of the final CMS membranes.
The stability of the membrane over an extended time of operation is critical for
any type of industrial application. To study such effects we monitored CO
2
and CH
4

permeances over an extended period of time. For the membrane with transport properties
that are shown in Figure 2.13 and 2.14 (membrane 5 in Table 2.6 after the second
carbonization), the permeances are initially measured at room temperature at the
beginning of the experiment (time 0 on the figures). The membrane was then gradually
heated up to 120 °C in the presence of CO
2
or CH
4
, and the permeance was measured
again until it stabilized.

43

Figure 2.13 CO
2
permeance of membrane 5 as a function of time

Figure 2.14 CH
4
permeance of membrane 5 as a function of time

The CO
2
permeance decreased from 13.71 m
3
/m
2
.bar.h at room temperature to 6.92
m
3
/m
2
.bar.h at 120 °C during the first 2 h of heating. Subsequently, the permeance
0
2
4
6
8
10
12
14
16
0 2 4 6 8
CO
2
Permeance (m
3
/m
2
.bar.h)
Time (h)
0
2
4
6
8
10
12
14
0 5 10 15 20 25 30
CH
4
permeance (m
3
/m
2
.bar.h)
Time (h)

44
decreased from 6.92 m
3
/m
2
.bar.h to 6.78 m
3
/m
2
.bar.h over the next 5 h. It remained
approximately at this level afterwards. For CH
4
the permeance decreased from 12.57
m
3
/m
2
.bar.h at room temperature to 5.85 m
3
/m
2
.bar.h and stabilized afterwards.
Similar results were also obtained with other membranes, as shown in Figures
2.15 and 2.16 for a different membrane (membrane 6 in Table 2.6 after the second
carbonization). However, for this membrane it took much longer for the CO
2
and CH
4

permeances to stabilize.

Figure 2.15 CO
2
permeance of membrane 6 with two carbon layers as a function of time
0
1
2
3
4
0 2 4 6 8
CO
2
Permeance (m
3
/m
2
.bar.h)
Time (h)

45

Figure 2.16 CH
4
permeance of membrane 6 with two carbon layers as a function of time

Note that for both membranes the time for the CH
4
permeance to stabilize is much longer
than that of CO
2
, for which it takes typically a few hours. This behavior is probably due
to the fact that CO
2
is much more permeable than CH
4
and, as a result, it may require
more time for the CH
4
that is adsorbed within the membranes to equilibrate at its new
condition corresponding to the higher temperatures.
The CH
4
and CO
2
permeances were also measured as a function of temperature
for membrane 6, following the experiments shown in Figures 2.15 and 2.16, with the
results shown in Figures 2.17 and 2.18. In these experiments the temperature was lowered
in steps, and after it stabilized to a new value, measurement was carried out (an hour after
the temperature had stopped changing).
0
0.1
0.2
0.3
0.4
0 10 20 30 40 50
Permeance (m
3
/m
2
.bar.h)
Time (h)

46

Figure 2.17 CO
2
permeance of membrane 6 with two carbon layers as a function of
temperature

Figure 2.18 CH
4
permeance of membrane 6 with two carbon layers as a function of
temperature

0
0.5
1
1.5
2
0 20 40 60 80 100 120 140
CO
2
permeance (m
3
/m
2
.bar.h)
Temperature ( °C)
0
0.1
0.2
0.3
0 20 40 60 80 100 120 140
CH
4
permeance (m
3
/m
2
.bar.h)
Temperature ( °C)

47
The permeance of CH
4
increases with increasing temperature, indicative of an activated
diffusion process. The CO
2
permeance, on the other hand, decreases, with the net result
that the separation factor also decreases as one goes from room temperature to 120 °C.
This behavior was also previously reported for the CMS membranes [Sedigh et al., 1999]
and is indicative of the fact that CO
2
adsorbs more strongly within the membrane, but as
the temperature increases, the effect diminishes.
2.4. Conclusions
In the preliminary studies supported CMS membranes were prepared by dip-
coating tubular alumina supports by a polyetherimide precursor, followed by
carbonization. The transport characteristics of the resulting membranes were investigated
with single gas permeation tests. Membranes from the same batch of the support did not
necessarily have the same performance, due to the nonuniform thickness of γ-alumina
layer. This explains why we were not able to achieve as consistent results with those
membranes within the same batch. Membrane from the first batch yielded promising
results in terms of the CO
2
and CH
4
permeance, and CO
2
/CH
4
separation factors. The
permeation tests, after adding each additional layer, indicated that the separation factor of
CO
2
/CH
4
may or may not increased, but the permeances of CO
2
and CH
4
simultaneously
decreased. The stability of the membrane was also conducted. CO
2
permeances seemed to
be stabilized after a few hours, but CH
4
permeances took much longer time to achieve
stability, up to 24 h or longer. The increase in the measurement temperature caused the
CO
2
permeance to decrease because of less effective adsorption at higher temperature.

48
On the other hand, the permeances of CH
4
increased with increasing temperature,
indicative of an activated diffusion process.

49
Chapter III
Pore Size Modification of Carbon Molecular Sieve Membranes

3.1. Introduction
Porous carbons have been in use for thousands of years. Their application in water
purification can be traced back to 2000 BC, when the ancient Egyptians used charcoal to
purify water for medicinal purposes [Faust & Aly, 1983]. All carbon materials, except
highly-oriented graphite, contain pores because they are polycrystalline, and are created
from the thermal decomposition of organic precursors. During pyrolysis, the
decomposition reaction results in various gases, the nature of which strongly depends on
the precursors. The pores in carbon materials are distributed over a wide range of sizes
and shapes and may be classified as shown in Table 3.1 [Inagaki & Tascon, 2006]. The
International Union of Pure and Applied Chemistry (IUPC) proposed a classification of
the pores based on pore width, which represents the distance between the walls of a slit-
shaped pore or a radius of a cylindrical pore.

50

Classification of pores in solid materials
1) Based on their origin
Intraparticle pores Intrinsic intraparticle pores
Extrinsic intraparticle pores
Interparticle pores Rigid interparticle pores
Flexible interparticle pores
2) Based on their size
Micropores < 2 nm Ultramicropores <0.7 nm
Supermicropores 0.7-2 nm
Mesopores 2-50 nm
Macropores >50 nm
3) Based on their state
Open pores
Closed pores (Latent pores)

Table 3.1 Classification of pore system

Almost all porous carbons contain pores in a wide range of sizes and do not
exhibit any selectivity in the adsorption of molecules of varying dimensions. However,
carbon molecular-sieves have a unique microporous structure with a narrower pore size
distribution, in which the slit-like apertures or constrictions of its micropores are of a size
similar to the molecular dimensions of the adsorbing species. In the separation of gases,
molecules that are smaller than the size of the micropore constriction rapidly diffuse
through them into the associated micropore volume. On the other hand, a large molecule
is denied access to the volume behind the constriction. The CMS is defined by two
properties: the selectivity for the adsorption of one of the species in the gas mixture to be
separated, and the adsorption capacity for that species. Selectivity is provided by a
narrow pore size distribution around an appropriate pore size, determined by the
particular gas mixture, whereas capacity is related to a relatively high micropore volume.

51
In Chapter II, we investigated the single gas permeations and ideal
permselectivity of the CMS membranes. They are known as good candidates in gas
separation applications, separating gas molecules with similar size effectively with
considerable resistivity to high temperatures and versatility in the pore structure. Pore
size plays a critical role in determining the selectivity of CMS membranes. The CMS can
be produced from distinct raw materials by a variety of methods, such as [Freitas &
Figueiredo, 2001]:
 Carbonization, controlled activation and thermal treatment of carbonaceous
materials
 Pyrolysis of polymers
 Modification of coals by mixing with tars and resins and subsequent
carbonization
However, the implementation of the three methods must be carefully controlled,
in order to obtain materials with small micropores. A small change in the effective size of
the constriction in the CMS membranes can affect the rate of diffusion of an adsorbing
gas molecule to a considerable extent. The development of a given kind of pore size
distribution of the CMS membrane can be controlled and manipulated to specific
applications of separation or purification of various gas mixtures by choosing appropriate
polymeric precursors, varying the carbonization conditions, such as the atmosphere,
temperature and duration of treatment. However, tuning based on such factors does not
provide a general method for membrane development for a wide range of industrial
applications. In this research project novel techniques are applied and investigated in

52
order to achieve this kind of structural control. They are described in this chapter and are
applied to the CMS membranes.
3.1.1. Steam Activation
Steam activation is a technique commonly used for preparing activated carbon
with various pore structures. The steam reacts with the carbon atoms and removes some
of the mass of the internal surface of the solid in the incipient micropores, creating a well
developed microporous material [Bansal et al., 1988]. At some point during the
activation process, the pore walls begin to collapse though, resulting in the number of
pores and the micropore surface area to go through their maxima [Walker, 1996]. In the
actual physical process, the activation of CMS membranes in steam flow at elevated
temperature is due to the following endothermic chemical reaction [Menéndez-Dí az &
Martí n-Gullón, 2006]
C + H
2
O  CO + H
2
ΔH = 132 kJ/mol (1)
which has been extensively studied, not only for activating carbons, but also because it is
one of the main reactions during coal and char gasification processes. In addition to the
heterogeneous reaction, gaseous products may also react among themselves in gas phase,
as well as with the carbon material itself according to the following reactions:
CO + H
2
O  H
2
+ CO
2
ΔH = -41.5 kJ/mol (2)
C + 2H
2
 CH
4
ΔH = -87.5 kJ/mol (3)
Generally, such reactions are slower than the main gasification reaction, and therefore do
not contribute greatly towards pore formation. Typically, the largest increase in the
material’s porosity is produced in the early stages of the activation process, and is caused

53
by the opening of constrictions in the char’s porous structure and the development of new
interconnecting pores. As burn-off continues, the widening of the existing micropores
and external burning of the particle’s surface outweighs the creation of new porosity,
eventually resulting in a net destruction of the porosity [Molina-Sabio et al., 1996].
Numerous studies with steam-activated carbons have been carried out. Rodrí guez-
Reinoso et al. [1995] prepared activated carbon from carbonized olive stones and studied
the development of porosity in the char, subjected to gasification with steam under
various experimental conditions, namely, by varying the partial pressure and the
activation temperature. They reported that the activation with steam - both pure and
diluted in N
2
- resulted in the widening of the existing microporosity, but the widening
was more significant when the steam was diluted. As the reaction proceeds, it produces
CO and H
2
, which are inhibitors for the reaction. The increase in the concentration of the
inhibitors in the interior of the particles favors the increased reaction of steam at the
exterior surface of the material. Increasing the activation temperature from 750 to 800 °C
makes the inhibiting effect of hydrogen less effective and, consequently, results in a
larger development of porosity, with the porosity maximum displaced to larger burn-off
rates (around 50%).
Molina-Sabio et al. [1996] reported on the steam activation of a char with respect
to the development of microporosity and of the micropore size distribution using
immersion microcalorimetry with compounds with different molecular sizes (bezene, 2,2
dimethylbutane, iso-octane and α-pinene). They report that micropores with widths of
less than 0.5 nm dominated at <10% burn-off rates. At larger burn-offs, as the activation

54
process proceeded further, molecular sieving behavior disappeared. In addition, they
report that only in the early stages of the activation process there was a net development
of micropores, with the widening of micropores and external burning of the particles
becoming more important at burn-offs larger than 40% for the activation at 800 °C, and
20% for the activation at 700 °C.
Pastor-Villegas and Durán-Valle [2002] studied the influences of the activation
temperature (700-950 °C) on the pore structure of steam-activated carbons, prepared from
rockrose. The microporous structure of the activated carbon was mainly formed by
widening of the micropores. The pore volume first increased up 0.25 cm
3
/g during
treatment at 850 °C; upon increasing the temperature to 950 °C the pore volume first
continued to increase. However, at 40% burn-off, the meso- and macropore volumes
began to decrease. The activated carbons contained a significant open pore volume
accessible to helium V
P
compared to the starting char (0.409 cm
3
/g). The maxima value
of V
P
was for a char activated at 700 °C at 40% burn-off (0.957 cm
3
/g).
Macia-Garcia et al. [2004] studied the influence on the steam activation of
samples of a commercial Holmoak wood charcoal at 800-950 °C at 20-60% burn-off. The
specific surface area (S
BET
) of the active carbons increased with increasing burn-off rate,
and the maximum S
BET
of 987 m
2
/g was obtained for the sample activated at 850 °C to
60% burn-off (by comparison the S
BET
of the carbon precursor was 120 m
2
/g). As the
burn-off rate increased, so did the micro- and mesopore volumes. The highest micropore
volume of the samples prepared with 20 and 40% burn-off fractions was obtained with an
activation temperature of 850 °C. The micropore volumes of the samples prepared with

55
20 and 40% burn-off were 0.241 cm
3
/g and 0.301 cm
3
/g, respectively, compared to the
starting precursor of 0.06 cm
3
/g. On the other hand, for the samples prepared with 60%
burn-off, the best result was 0.403 cm
3
/g, which was obtained at a process temperature of
800 °C.
Jaseinko-Halat and Kedzior [2005] studied bituminous coal, and performed the
steam activation at 750 °C for low levels of burn-off of 5-25%. Benzene and carbon
dioxide adsorption was used to evaluate the porous structure. Progressing activation
resulted in a gradual decrease and complete disappearance of the submicropores (width
<0.4 nm), accompanied by a notable increase of the volume of micropores (widths 0.4-
2.0 nm) from 0.120 cm
3
/g to 0.237 cm
3
/g, and a slight increase in the volume of
mesopores (widths 2-50 nm). The adsorption isotherms of several organic vapors with
varying molecular diameters (dichloromethane, benzene, cyclohexane and
tetrachloromethane) were obtained to determine the size distribution of micropores.
During the early stage of activation (5% and 10% burn-off rates), the new micropores
were being created by elimination of constrictions present in the char with a slight
widening of the micropores. At more advanced levels of burn-off (10-25%), they found
that the opening of new micropores by elimination of the constrictions had come to an
end, and further activation resulted only in the widening of the already existing
micropores. The creation of narrow micropores (~0.4-0.5 nm) was pronounced for chars
activated to burn-off levels less than 10%. These samples were reported to exhibit
noticeable molecular sieving properties. Continued activation increased the micropore
volumes, due to the widening of narrow micropores, creating micropores bigger than 0.63

56
nm. As a result, the molecular sieving characteristics in the samples disappeared. The
fractional contribution in the total micropore volume (pores in the range of 0.33-2.0 nm)
in chars, of micropores with widths (0.33-0.41 nm), which are accessible to
dichloromethane and inaccessible to benzene, was 48% at 5% burn-off rate, and that
significantly decreased to 27% at 25% burn-off. On the other hand, the volume fraction
of micropores with widths exceeding 0.63 nm, which are accessible to
tetrachloromethane, increased from 5% to 23% as the activation process progressed.
The influence of steam as activating agent on the porosity development of
activated carbons produced from walnut shells was investigated by Gonzalez et al.
[2009]. Steam activation was carried out either at 700 °C for 60-120 min, or at 850-900
°C for 30-60 min. It was found that the adsorption capacity of the carbons increased with
increasing activation time. For activation time of 30 min (about 38% burn-off), higher
activation temperatures did not markedly change the textural characteristics of the
activated carbons. For an activation time of 60 min, the increase in temperature from 700
to 850 °C led to greater N
2
adsorption capacities, giving rise to a wider pore size
distribution. As the temperature increased, the BET surface area increased from 542 to
1361 m
2
/g, and the of the micropore volume increased from 0.30 to 0.74 cm
3
/g, while the
mesopore volume increased from 0.04 to 0.20 cm
3
/g. A further increase in the
temperature to 900 °C resulted in a decrease in the N
2
adsorption capacity due to the
external burning. In addition, increasing the activation time from 30 to 45 and finally to
60 min at 850 °C gave rise to a major increase in the N
2
adsorption capacity (699, 966,

57
and 1361 m
2
/g, respectively), whereas increasing the activation time at 700 °C had no
effect on the micropore volume, and only little effect on the mesopore volume.
Singh and Lal [2010] studied experimentally the effects of the processing
parameters, such as gasification time and temperature, on the adsorption properties of
steam-activated carbons. The activated carbons were prepared by carbonization of
polystyrene sulphonate beads at 800 °C, followed by steam activation of the resultant
chars. Steam activation was carried out at 800 °C for the gasification times 1, 2, and 3 h,
and at different gasification temperatures (750, 800, 850, and 900 °C) for 2 h. When the
gasification time increased from 1 to 3 h, the BET surface area, total pore volume, and
average pore size increased from 310 to 949 m
2
/g, 0.27 to 0.98 cm
3
/g, and 3.5 to 4.1 nm,
respectively. Both the surface area and the total pore volume of activated carbons
increased with increasing gasification temperature. The carbon samples exhibited a
surface area ranging from 239 to 620 m
2
/g, and a total pore volume from 0.26 to 0.62
cm
3
/g, when the temperature increased from 750 to 900 °C.
In conclusion, the pore characteristics of activated carbons produced by steam
activation are varied by choosing different types of carbon source, activation temperature,
and burn-off levels. In this chapter steam activation is utilized to modify the pore
characteristics of CMS membrane by tuning their pore size and volume. The operating
conditions, such as the temperature and period of activation, were studied in order to
achieve an improvement in the quality of the resulting membranes.

58
3.1.2. Carbon Deposition
The CMS can also be prepared by a post-treatment that narrows the pore-size
distribution to produce a material with a predominance of micropores in a limited range
of sizes. The dimensions of the pore entrances are carefully controlled by reducing the
pore diameter by pore blocking. Carbon deposition from cracking of carbon sources is
known to be an effective and suitable modification of the pore structure. The selectivity
of a carbon membrane may be increased through the introduction of organic species into
the pore system and their pyrolytic decomposition within the membrane [Hayashi et al.,
1995; Soffer et al., 1997; Cabrera et al., 1993]. In order to manufacture carbon molecular
sieves, the inherent pore structure of the carbonaceous precursor is initially tailored into a
suitable pore size range by controlling the thermal pretreatment, followed by a final
adjustment of the pore apertures by chemical vapor deposition (CVD) [Verma & Walker
Jr., 1992]. Chemical vapor deposition (CVD) of carbon onto a carbon membrane may
results in three distinct phenomena, namely homogeneous deposition, ad-layer deposition
and in-layer deposition as depicted in Figure 3.1 [Soffer et al., 1997].

Figure 3.1 Mechanisms of a carbon deposition in the pore system of carbon membranes:
(a) Homogeneous carbon deposition on membrane pore walls; (b) in-layer carbon
deposition on membrane pore wall entrances; and (c) adlayer carbon deposition outside
the membrane pores

59
The carbon sources that can be adopted for carbon deposition include various
organic compounds such as methane, benzene, ethylene, hexane, cyclohexane, and
methanol [Vyas et al., 1993; Lizzio & Rostam-Abadi, 1993; Cabrera et al., 1993]. Upon
pyrolysis, carbon atoms are deposited, resulting in modification of the pore structure of
the base carbonaceous material, and consequently improving the separation performance.
Benzene has been widely employed because it does not produce intermediate
species in the cracking process, which makes deposition easier to be controlled
[Kawabuchi et al., 1996, 1997; Becker & Hüttinger, 1998]. Moore and Trimm [1977] and
Cheredkova et al. [1989] modified the gaseous adsorption properties of carbons by
depositing carbon produced by the pyrolysis of benzene vapors. The resulting material
was a CMS with pore diameter between 0.3 and 0.6 nm, which allow nitrogen and
oxygen to be adsorbed at different extents. Barton and Koresh [1989] prepared molecular
sieve carbons from macadamia nut shell carbons by thermal decomposition of benzene
vapors on the carbon surface at 800 °C for various time periods. They reported that the
pore structure of the resulting carbon was modified with an improvement in the
separation of nitrogen and oxygen by the thermal decomposition of benzene vapors on
the carbon surface. The resulting molecular sieve carbons were characterized by water-
vapor adsorption and by immersion calorimetry. It was found that both the Gurvitch
values for the water vapor and the heat of immersion were constant and independent of
the benzene treatment time, indicating that the pore closure occurred in such a way that
neither the pore volume nor the energy of interaction with the adsorbate were diminished.
They concluded that the pyrolysis of benzene, which conferred molecular sieving

60
properties on the active carbons, appeared to proceed mainly by the formation of
constrictions at preferred sites in the pore structure. Manso et al. [1999] studied the
formation of CMS via carbon vapor deposition over activated carbons produced from
four different rank coals. The deposition of carbon was carried out by pyrolyzing benzene
vapors at 725 °C. This produced gradual closing of the micropores, due to the formation
of constrictions at their entrances. As a result, CMS materials with a narrow micropore-
size distribution between 0.35 to 0.5 nm were obtained. Samples with diameters smaller
than 0.33 nm obtained by a high degree of deposition were capable to separate O
2
/N
2
and
CO
2
/CH
4
mixtures.
Although benzene has been, by far, the most commonly used gas, its high toxicity
makes the process not very attractive. By comparison, methane is a non-toxic component,
that can be used instead of benzene. The pyrolysis of methane is a very complex process
involving a large number of reactions [Choudhary et al., 1991; Hidaka et al., 1999;
Legrand et al., 1999]. Temperature and dilution were suggested as two significant factors
during the pyrolysis of methane [Gueret et al., 1997; Lucas & Marchand, 1990].
Lucas and Marchand [1990] investigated the effects of three important
parameters, namely, temperature, flow-rate, and pressure, in the deposition process of
pyrolytic carbon from methane. They reported that acetylene was produced at a
temperature of 1200 °C and ethylene at 1300 °C. In contrast, in the range of 800-1000 °C,
it was found that carbon and hydrogen are the major products produced by the pyrolysis
of pure methane under a pressure of 20-500 mmHg. In addition, more intense pyrolysis
conditions (higher temperature and pressure or lower flow-rate) accelerated the

61
condensation of methane into polyaromatic hydrocarbons. When the methane was diluted
with hydrogen, the pyrolysis mechanism remained essentially the same. Earlier, Vnukov
et al. [1986] had prepared molecular sieve carbons by depositing carbon on the active
carbon surface by the pyrolysis of methane at temperatures between 600 °C and 900 °C.
However, the deposition of the carbon resulted in a decrease in the pore volume. Vyas et
al. [1993] prepared CMS materials from carbonizing bituminous coal, followed by
cracking of methane in the range of 750-780 °C for 5-14 min. An increase of the methane
flow rate was shown to retard the O
2
/N
2
uptake ratio due to the deposition of coke on the
pore mouths. When the methane cracking time was increased, the uptake ratio increased
only as far as 10 min, and decreased after that. Li et al. [2002] prepared carbon coated
ceramic membranes by pyrolysis of methane at a temperature of 1000 °C. They utilized a
modified CVD apparatus, which allowed them to periodically reverse the direction of the
gas flow in the reactor, in order to ensure a uniform deposition of carbon on the
membrane. It was found that the flow rate of the gas mixture and the alternation of the
direction of the gas flow were essential for determining the uniformity of the carbon
deposits in the membranes. An increase in the input flow rate was shown to increase the
average deposition rate and to generate a more uniform carbon layer on the surface. By
varying the methane concentration, the carbon film was found to be thicker at higher inlet
methane concentration. Results also showed that the pore size of the membranes
decreased as the methane concentration increased from 10% to 20%. In addition, longer
periods of deposition increased the amount deposited on the membrane and the thickness
of the carbon films, resulting in decreasing nitrogen permeability. de la Casa-Lillo et al.

62
[2002] modified the pore structures of activated carbon fibers via cracking of methane at
various temperatures and time periods. These materials had been evaluated for their
selectivity during CO
2
and CH
4
separation and their uptake ratio were also compared
with non-treated activated carbon fibers. Results indicated that the CVD treatment at
1173 K deposited carbon inside the pores of the activated carbon fibers, which reduced
the overall pore volume, and led to the improved kinetics of CO
2
/CH
4
separation. When
the temperature was increased to 1223 K, the deposition of carbon was found to occur
rapidly, and the results indicated that this deposition was primarily limited at the pore
mouths, thus preventing any significant deposition within the internal porous structure. In
addition, they reported that, with increasing time of CVD, both CO
2
and CH
4
uptakes
decreased, but the adsorption selectivity was improved. Villar-Rodil et al. [2005]
prepared microporous CMS materials through the CVD of methane at 1098 K on
activated carbon cloth obtained from a Nomex aramid fiber. The closure of the porosity
was achieved more quickly when using a pure methane stream than when employing a
diluted one. The samples, which were CVD-treated for short periods of time using both
pure and diluted methane, were found to have too large of an adsorption rate for CH
4
to
be selective towards the CO
2
/CH
4
separation. Prolonging the CVD treatment time
produced a decrease in the CH
4
adsorption rate, while high adsorption rate and capacity
were preserved for CO
2
, resulting in improved CO
2
/CH
4
separation.
The type of carbon substrate used in the methane cracking process was found to
be important. Lizzio and Rostam-Abadi [1993] produced CMS carbons from Illinois
coals by carbon deposition from methane at 1000 °C, diluted with nitrogen to a

63
concentration of 10%. Their study indicated that the molecular sieving properties of the
resultant CMS carbons varied greatly depending on the type of coal utilized: products
from one type of coal exhibited no sieving capacity, while products from other types of
coal were effective for gas separation.
In conclusion, the pore characteristics of CMS materials prepared by methane
pyrolysis depend on the choice of the initial carbon source, the activation temperature and
the duration of treatment. In this chapter, carbon deposition by methane pyrolysis is also
employed to tune the pore size of CMS membranes; the hope is that the methane will
enter the mesopores within the CMS membrane and deposit carbon within the pore
network during pyrolysis at high temperatures. As a result one expects the pores to
slowly close-up and the permeance of the membrane to decrease, while simultaneously
its selectivity improves.
3.2. Experimental Techniques
We first describe the preparation of the membrane, and then explain how we
modify its pore structure.
3.2.1. Membrane Preparation and Modification
The base-CMS membranes were fabricated using the technique which was
previously described in Chapter II. Each base-CMS membrane was prepared with three
layers of coatings, in order to ensure proper membrane performance, and subsequently
modified by a variety of post-treatments, in order to fine-tune their pore size
characteristics.

64
3.2.1.1. Steam Activation
During steam activation, as noted above, the carbon surface is made to react with
steam at high temperature, in order to adjust and optimize the pore size distribution of the
CMS membrane for the separation of specific gaseous mixtures. During our experiments,
the base-CMS membrane was placed in a stainless steel tubular furnace (2 in diameter
and 2 ft long, as depicted in Figure 3.2), and heated at a rate of 1 °C/min in the presence
of UHP argon. Once the desired activation temperature was reached, steam was mixed
with the argon flowing through the membrane. The flow of water was controlled by an
HPLC pump (LDC Analytical). The molar ratio of the steam/Ar mixture fed into the
reactor was 2 to 1. The mixture of argon and water was pre-heated to 250 °C before
introduction into the reactor. The membrane reacted with steam for several hr at various
pre-selected temperatures, before it was slowly cooled down to room temperature at a
rate of 2 °C/min.

Figure 3.2 Schematic diagram for steam activation module

65
3.2.1.2. Methane Activation
The base-CMS membranes were also treated with methane, with the purpose of
decomposing it within the membrane structure to carbon, and thus favorably modifying
the membrane properties. To accomplish this, each membrane was placed inside a
stainless steel module, with each of its ends sealed using graphite tape (Figure 3.3 is a
schematic of the apparatus). The membranes were then heated at a rate of 1 °C/min to
either 650 or 700 °C in the presence of UHP argon. Once the reaction temperature was
reached, the flow of Ar was ceased and UHP methane was then allowed to flow into the
module, with the retentate side closed in order to force all the methane gas to permeate
through the membrane. The membrane was held at the desired reaction temperature for
several hours before it was slowly cooled down to room temperature.

Figure 3.3 Schematic diagram of methane treatment module

3.2.2. Membrane Testing
The transport characteristics of the membranes prepared by the aforementioned
two post-treatment steps were measured using single gases (UHP hydrogen, carbon

66
dioxide, and methane) in order to determine the ideal permselectivity towards various gas
pairs. Details about the experimental apparatus and the experiments performed and the
equations utilized for data analysis were presented in Chapter II.
3.3. Results and Discussion
We first describe the result of pore modification by steam activation.
3.3.1 Pore Modification via Steam Activation
The base-CMS membrane is microporous [Sedigh et al., 1998, 1999]. As
previously noted, during the post-treatment step the carbon surface in the membranes
reacts with steam at high temperatures which, as expected, results in the opening of some
of the pores. The steam activation technique is gentle, in that though it typically enlarges
the pore size, and has a minimal impact on surface and mechanical properties. In this
section, the effect of steam activation on the properties of the treated CMS membranes is
discussed.
3.3.1.1. Permeation Tests
Several CMS membranes were prepared by steam activation at different
temperatures. The He and Ar permeances of the base-CMS membranes, prior to the
steam treatment, were initially measured (they are generally fairly similar) and utilized as
a guide to indicate whether pore widening has indeed occurred, as they are inert gases
with permeation rates that are known not to be affected by any potential changes in the
membrane surface characteristics. One of these membranes was subsequently activated
with steam in a tubular furnace at 600 °C and the other was activated at 700 °C, both for
1 h. After the steam treatment, the membranes were slowly cooled down to room

67
temperature, and their He and Ar permeances were measured at 120 °C. After the
measurements, the same membranes were placed in the furnace and further activated for
an additional 2h, then cooled down to room temperature again, and their He and Ar
permeances at 120 °C.. This activation/permeation measurement process was repeated
until the membranes were activated for a total of 15 h at 600 °C and 3 h at 700 °C. The
typical behavior observed, when activating the CMS membranes in steam, is shown in
Figures 3.4 and 3.5. (Note from the figures that these membranes are not permselective,
based on the He/Ar ideal separation factor; however, these measurements are still
important in trying to understand the steam activation process.) The activation
temperature and the duration of treatment are the two most important parameters for
adjusting the pore size of the CMS membranes. The membranes’ permeances increase
with increasing activation temperature and the duration of treatment. This signifies an
increase of the pore size during the activation process. Activation at 700 °C appeared to
affect the pore size more than the treatment at 600 °C, since the increase in the
permeation was significantly larger after 3 hr of activation at 700 °C. For the remaining
studies, an activation temperature of 600 °C was selected, as it allows one to gently adjust
the average pore size of the base-CMS membranes.

68

Figure 3.4 He permeance of membranes as a function of time and temperature in steam
activation

Figure 3.5 Ar permeance of membranes as a function of time and temperature in steam
activation
0
1
2
3
4
5
0 5 10 15
He Permeance (m
3
/m
2
.bar.h)
Time (h)
at 600 °C
at 700 °C
0
0.2
0.4
0.6
0.8
1
1.2
1.4
0 5 10 15
Ar Permeance (m
3
/m
2
.bar.h)
Time (h)
at 600 °C
at 700 °C

69
Another group of the base-CMS membranes was prepared and treated with steam
at 600 °C. The permeances of H
2
, CO
2
, and CH
4
and the corresponding ideal
permselectivity of H
2
/CO
2
and H
2
/CH
4
were measured as a function of the duration of the
treatment, as shown in Figure 3.6, 3.7, and 3.8. Once again, the permeances of the base-
CMS membranes were alike before further treatment by steam, as shown in Table 3.2.
Membrane 1 Membrane 2 Membrane 3 Membrane 4
P
H2
0.514 0.457 0.492 0.479
P
CO2
0.138 0.183 0.163 0.158
P
CH4
0.023 0.028 0.024 0.027
S(H
2
/CO
2
) 3.7 2.5 3.0 3.0
S(H
2
/CH
4
) 22.7 16.3 20.5 17.7

Table 3.2 H
2
, CO
2
, and CH
4
permeances and ideal separation factors of 3-layer base-CMS
membranes prior to steam treatment

As Figure 3.6 indicates, the permeances of all three gases increased, as expected, after
steam treatment. The steam activation process is known to remove carbon atoms from the
interior of the membrane, which results in the enlargement of the micropores and the
opening of closed (inaccessible) micropores (from the literature on activated carbons, see
prior discussion).

70

Figure 3.6 H
2
, CO
2
, and CH
4
permeances of membranes as a function of the duration of
treatment

0
0.5
1
1.5
2
2.5
0 6 12 18 24
Permeance (m
3
/m
2
.bar.h)
Time (h)
M1_H₂
M2_H₂
M3_H₂
M4_H₂
M1_CO₂
M2_CO₂
M3_CO₂
M4_CO₂
M1_CH₄
M2_CH₄
M3_CH₄
M4_CH₄

71

Figure 3.7 Separation factors of H
2
/CO
2
of membranes as a function of steam activation
period

Figure 3.8 Separation factors of H
2
/CH
4
of membranes as a function of steam activation
period

0
1
2
3
4
5
0 6 12 18 24
H
2
/CO
2
Separation Factor
Time (h)
M1
M2
M3
M4
0
10
20
30
40
0 6 12 18 24
H
2
/CH
4
Separation Factor
Time (h)
M1
M2
M3
M4

72
However, one interesting observation, as indicated by Figure 3.7, is that the
separation factor of H
2
/CO
2
remains almost unaffected by steam treatment. In addition, as
shown in Figure 3.8, the selectivity of H
2
/CH
4
initially increased after the treatment. This
is potentially due to initial development of microporosity, which significantly increased
the H
2
flux through the membrane. On the other hand, the initial creation of micropores
had a weaker effect on the CH
4
permeance, which results in a slight increase in H
2
/CH
4

separation factor during the early stage (up to the first 12 h) of activation. These
membranes exhibit a noticeable improvement of gaseous permeances without destroying
their molecular sieving characteristics. However, as the steam treatment continues, the
pore sizes significantly increase and the external burning of the carbon surface begins to
dominate. In this case, the separation performance begins to degrade. Thus, whereas in
the early stage of the activation process, there is a net development of micropores with
the modified membrane exhibiting much higher gaseous permeances without destruction
of separation characteristics, further activation causes a widening of narrow micropores
into mesopores, and the external burning of the CMS surface. As a result, the molecular
sieving characteristic of the membranes diminishes.
3.3.1.2. Gas Adsorption Analysis
The transport characteristics of CMS membranes provide valuable information
regarding the separation mechanisms and overall behavior. However, to gain a better
understanding of the separation and permeation mechanism, it is also important to study
the structure of the CMS membrane. Adsorption-based characterization techniques have
been widely used for evaluating the membranes’ pore structure in terms of the exposed

73
surface area, pore volume, mean pore size, and pore size distribution. Adsorption-based
techniques are capable of probing overall characterization data for various microporous
materials, unlike other microscopy techniques, such as STM and AFM. The major
drawback of the adsorption-based techniques is in matching the interpretation of the
adsorption data with an appropriate network pore model. In some instances, the
resolution of the pore size is restricted by the size of the probe molecules that can access
the pore structure, as the technique investigates the adsorption/condensation of different
types of probe gases in the pores at specified temperature and pressure. In the present
study we have utilized nitrogen adsorption.
The experiments were carried out at 77 K using an ASAP 2010 micropore
analyzer from Micromeritics, Inc., in a static mode. A turbo-molecular pump was used
along with the system to vacuum down to 10
-5
torr. The degassing of the samples took
about 24 h at 250 °C directly in the analysis port. An isothermal jacket around the sample
tube maintains the level of liquid N
2
constant for up to one day. The CMS membrane
normally takes a longer time to analyze, as a result, both the sample and the cold trap
dewers need to be refilled daily.
In order to fully characterize the pore structure, it was essential to obtain the pore
size of the initial support substrate. Figure 3.9 presents the data of the original substrate
obtained by the BJH method. It is important to note that gas adsorption mainly took place
in the γ-alumina layer, which is a small portion of the support substrate [Sedigh et al.,
1999]. As a result, the value reported for cumulative and differential pore volumes were

74
unusually low, due to the fact that the total weight of the support tube was used for pore
volume calculation.

Figure 3.9 Pore size distribution of the support substrate

A 3-layer supported CMS membrane was then prepared, and the transport and
separation characteristics of the membrane were evaluated by measuring the permeance
of H
2
, CO
2
, and CH
4
prior to the steam activation, shown in Table 3.3.
Permeance (m
3
/m
2
.bar.h) Separation Factor
P
H2
P
CO2
P
CH4
H
2
/CO
2
H
2
/CH
4

3.138 0.763 0.108 4.1 29.0

Table 3.3 H
2
, CO
2
, and CH
4
permeances and ideal separation factors of a 3-layer base-
CMS membrane prior to steam treatment

Figure 3.10 compares the adsorption isotherms of the CMS membrane after three
coating/carbonization cycles with the original support substrate. It can be seen that the
coating led to greater N
2
adsorption capacity. After each coating, the CMS layer formed

75
inside the pore network of the support, which led to the increase in the pore volume and,
most importantly, introduced the molecular sieving behavior of the supported membrane
[Sedigh et al., 1999].

Figure 3.10 Nitrogen adsorption isotherms of the support substrate and CMS membrane

This CMS membrane was subsequently exposed to steam at 600 °C for 6 h and
cooled down to the room temperature inside the furnace. The sample was then inserted
into the BET apparatus, and the porous structure of the membrane was evaluated after the
steam treatment. The steam activation process was repeated until the same membrane
was activated for a total of 24 h. Note that since the glazing tips of the CMS membrane
were cut in order to carry out the pore size characterization in the BET module, the
transport properties of the membrane were not measured after the steam treatment. Figure
3.11 shows the N
2
adsorption isotherms of the resulting CMS membrane after several
stages of steam activation (namely, 0 h, 6 h, 12 h, 18 h, and 24 h). According to the
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
0 0.2 0.4 0.6 0.8 1
Volume Adsorbed cm
3
/g STP
Relative Pressure (P/Po)
Support
CMS

76
classification of physisorption isotherms recommended by the International Union of
Pure and Applied Chemistry, the shapes of the adsorption isotherms in Figure 3.11 may
be considered to be a composite of types I and II. In this case, type I would be
representing the microporous carbon layers, and type II with the broader pore size would
have mainly resulted from the mesoporous support.

Figure 3.11 Nitrogen adsorption isotherms of steam-activated CMS membrane at each
activation time (0 h, 6 h, 12 h, 18 h, and 24 h)

As Figure 3.11 indicates, the shape of the isotherms does not change significantly
even after a 24 h activation period. This is due to the fact that the reaction mainly
happened in the carbon layers, which is only a small portion of the entire supported CMS
membrane. However, the amount of adsorbed N
2
increases significantly in the low
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
0 0.2 0.4 0.6 0.8 1
Volume Adsorbed cm
3
/g STP
Relative Pressure (P/Po)
CMS
6 h
12 h
18 h
24 h

77
relative P/P
o
range (up to 0.03) during the initial stage of activation. The adsorption of
nitrogen at the low relative pressure is commonly used to determine the volume of
micropores [Bansal et al., 1988]. As can been seen in Figure 3.12, the increase in
activation time from 0 to 12 h significantly increases the N
2
adsorption capacity in the
micropore region. However, as the activation process is continued for an additional 12 h,
only an insignificant increase in the development of the micropore volume is observed,
indicating no further increase in the number of micropores that are formed.

Figure 3.12 Nitrogen adsorption isotherms at low relative pressure of steam-activated
CMS membrane at each activation time (0 h, 6 h, 12 h, 18 h, and 24 h)

The data were further analyzed by the Horvath-Kawazoe (HK) method, assuming
slit-shape pore geometry (see Figure 3.13). Figure 3.14 shows the effect of steam
0
0.1
0.2
0.3
0.4
0.5
0.6
0 0.01 0.02 0.03
Volume Adsorbed cm
3
/g STP
Relative Pressure (P/Po)
CMS
6 h
12 h
18 h
24 h

78
activation on the average pore diameter in the microporous region (using the HK method)
as a function of the time of treatment, The effect of the steam activation on the mesopore
volume and average pore diameter (using the BJH method) is shown in Figures 3.15 and
3.16, respectively,. It can be seen that although the micropore volume increases with
increasing activation time, the mesopore volume decreases. In the first 12 h of the
activation, a significant increase in the micropore pore volume is attained, and the change
in the mesopore volume is relatively smaller. Activation in the early stages causes an
initial increase in the micropore volume, due to the widening of existing narrow
micropores, but most likely also due to the development of new micropores; the latter
phenomenon is manifested in Fig. 3.14, where one observes a decrease in the average
pore diameter in the microporous region. However, when the activation was continued
for another 12 h for a total of 24 h, it had a negligible effect on the micropore volume, but
increased the mesopore diameter. For longer steam activation period, the development of
micropores comes to an end.

79

Figure 3.13 The Horvath-Kawazoe cumulative pore volume plot of steam-activated CMS
membrane at each activation time (from bottom of graph: 0 h, 6 h, 12 h, 18 h, and 24 h)

Figure 3.14 The HK median pore diameter of steam-activated CMS membrane as
function of activation time

8.8
8.9
9
9.1
9.2
9.3
9.4
9.5
0 6 12 18 24
HK Median Pore Diameter (Å )
Time (h)

80

Figure 3.15 The BJH adsorption cumulative pore volume of steam-activated CMS
membrane at each activation time (0 h, 6 h, 12 h, 18 h, and 24 h)

Figure 3.16 The BJH adsorption average pore diameter of steam-activated CMS
membrane as function of activation time
0
0.0002
0.0004
0.0006
0.0008
0.001
0.0012
0.0014
0.0016
0.0018
0.002
10 100
Cumulative Pore Volume (cm
3
/g)
Pore Diameter (A)
CMS
6 h
12 h
18 h
24 h
52
54
56
58
60
62
64
0 6 12 18 24
BJH Adsorption Average Pore
Diameter (Å )
Time (h)

81

The initial increase in the N
2
adsorption capacity over the micropore region can
be explained by considering that steam primarily reacts with the active sites located at the
center of the pores, and thus, creates microporosity. As a result, the gaseous permeances
of the resulting CMS membrane increase initially without loss in the separation
performance. However, the reaction of the steam over a longer period of time caused the
external burning effect to dominate, resulting in a destruction of the molecular sieving
properties of the CMS membrane.
3.3.2. Pore Modification via Methane Activation
The CMS membranes are treated with methane in the hope that methane will
decompose into carbon at high temperature and patch any defects within the membranes.
In addition, methane activation is expected to decrease the average pore size and to
narrow the pore size distribution. In the experiments, two CMS membranes prepared by
pyrolysis of one layer using a 6% PEI solution were activated with methane at 700 °C.
During the experiments we continuously measured the permeance of CH
4
, in order to
attain a better understanding of the pore closure phenomena during the CH
4
decomposition process. The methane permeances were measured at 700 °C by keeping
the pressure gradient between the retentate and permeate sides constant, by means of
needle valve adjustment on the permeate side. Figure 3.17 shows the CH
4
permeances as
a function of activation time. (Table 3.4 indicates the estimated total amount of CH
4

decomposing within the membrane as a function of activation time, by assuming that the
decrease in the permeance is proportional to the loss in pore volume. Note from this
Table that the amount of methane reacted is negligible compared to the methane

82
permeation through the pores). It can be seen that the permeances of both membranes
decrease significantly during the initial stage of the methane activation process and
slowly stabilize after 48 h. The decrease in the gaseous permeances indicates that pore
closure results from methane decomposing into carbon and depositing inside the pores.

Figure 3.17 CH
4
permeances of membranes as a function of methane activation period

Estimated total amount of Methane Deposit (mol)
Time (h) Membrane 1 Membrane 2
2
4 1.78E-06 4.86E-06
6 3.93E-06 1.13E-05
8 6.25E-06 1.76E-05
20 1.46E-05 6.80E-05
22 1.86E-05 7.07E-05
24 2.21E-05 7.14E-05
48 5.37E-05 7.23E-05
72 5.45E-05 7.24E-05
144 5.44E-05 7.28E-05

Table 3.4 Estimated total amount of CH
4
decomposed within the membrane as a function
of activation period

0
1
2
3
4
5
6
0 24 48 72 96 120 144 168
CH
4
permeance (m
3
/m
2
.bar.h)
Time (h)
M1
M2

83
The permeances of H
2
, CO
2
, and CH
4
, after the methane treatment, were measured at 120
°C and are reported in Table 3.5. Note that the permeances of CH
4
for both membranes
appear to be higher than that of CO
2
, and the molecular sieving behavior vanishes after
methane treatment. However, the methane activation results in a significant improvement
in H
2
/CO
2
separation, as depicted in Table 3.5 where a separation factor of 6.2 was
achieved for Membrane 2. The CH
4
permeance of Membrane 1 in Figure 3.17 was higher
than the corresponding value in Table 3.5. This is potentially due to the higher
measurement temperature in Figure 3.17, but may also be indicative of poor sealing of
the membrane within the activation module at this high temperature.
Membrane 1 Membrane 2
P
H2
0.982 0.311
P
CO2
0.245 0.050
P
CH4
0.439 0.150
S(H
2
/CO
2
) 4.0 6.2
S(H
2
/CH
4
) 2.2 2.1

Table 3.5 Permeances and separation factors for membranes subjected to methane
activation at 700 °C (permeances in m
3
/m
2
.bar.h)

Table 3.6 shows the impact of the activation temperature on the transport
characteristics of two 3-layer base-CMS membranes which were activated with methane
continuously for 24 h. Membrane 1 in Table 3.6 was treated with methane gases at 650
°C, and Membrane 2 was treated at 700 °C. The permeances of H
2
, CO
2
, and CH
4
, before
and after methane treatment, were measured at 150 °C and are compared in Table 3.6.
Note that, as expected, the permeances of all species decreases as a result of activation in
methane. Activation at higher temperature has a more significant effect, as indicated by
the corresponding fractional changes. One interesting thing to see is that the fractional

84
permeance changes are not the same for all species, indicative of the different
mechanisms and/or parts of the pore structure that are effective during transport. Note
that methane treatment results in a more severe impact on CO
2
permeance, as compared
with hydrogen, resulting for membrane 2 in Table 3.6 into a separation factor of 10.
However, the effect of methane activation on methane permeance is significantly less.
This can be explained by the fact that methane cannot easily pass through the micropores
that are available for H
2
and CO
2
transport, but can instead decompose and deposit
carbon at the pore entrances. As a result, these micropores are no longer available for
transport for H
2
and CO
2
, thus their corresponding permeances are significantly reduced.
Membrane 1 (at 650 °C) Membrane 2 (at 700 °C)
Before After Change Before After Change
P
H2
1.059 0.258 -76 % 1.572 0.161 -90 %
P
CO2
0.610 0.046 -92 % 0.648 0.016 -98 %
P
CH4
0.073 0.054 -26 % 0.088 0.071 -19 %
S(H
2
/CO
2
) 1.7 5.6 229 % 2.4 10.0 317 %
S(H
2
/CH
4
) 14.5 4.7 -68 % 17.9 2.3 -87 %

Table 3.6 Permeances and separation factors for base-CMS membranes subjected to
methane activation (permeances in m
3
/m
2
.bar.h)

Another base-CMS membrane was treated with methane for 8 and 24 hr at 700
°C, and, again, the permeances of H
2
, CO
2
, and CH
4
, before and after each methane
treatment, were measured at 150 °C and compared in Table 3.7. Specifically, in these
experiments, after measuring the permeances of the base-CMS membrane, the membrane
was activated at 700 °C for 8 h and slowly cooled down to 150 °C for the permeance
measurement. Subsequently the temperature was raised to 700 °C for another 16 h of
activation (total of 24 h treatment) before the system was cooled down again to 150 °C.
Again, the changes in the permeance of H
2
and CO
2
were more significant than those for

85
CH
4
. One notes, that most of the changes in the H
2
and CO
2
permeances seem to occur
within the first 8 h, and relatively little change occurs after that. Again, this is consistent
with the idea that the carbon resulting from methane decomposition mostly blocks the
pore entrances of the pores associated with H
2
and CO
2
transport.
Before After 8 h After 24 h
P
H2
0.850 0.040 0.032
P
CO2
0.264 0.004 0.003
P
CH4
0.026 0.021 0.018
S(H
2
/CO
2
) 3.2 10.6 10.2
S(H
2
/CH
4
) 33.2 1.9 1.8

Table 3.7 Permeances and separation factors for base-CMS membranes subjected to
methane activation (permeances in m
3
/m
2
.bar.h)

3.4. Conclusions
The aim of the research presented in this chapter was to develop a post-treatment
technique for tuning the pore size properties of the CMS membranes. The techniques
utilized included steam activation and carbon deposition. An in-depth analysis of the
effects, in terms of pore structure and gaseous transport characteristics, was presented.
Steam activation is a gentle technique that allows one to increase the flux through the
carbon membrane by enlarging the pore size. The temperature and the duration of
treatment were seen to be the two important parameters one has to adjust in order to
attain desirable pore size characteristics. In order to obtain membranes exhibiting
molecular sieving properties, i.e. with a microporous structure characterized by a narrow
pore size distribution, the process of activation was limited to low levels of burn-off. The
steam activation process was shown to be effective in tailoring the pore size, as
manifested by the measured permeances and ideal selectivities of the treated membranes.

86
The nitrogen adsorption technique was employed to study the potential changes in the
pore structure characteristics following the steam treatment. The results show an increase
in microporosity at the beginning of steam treatment, which is accompanied by an initial
increase in the gas permeances while simultaneously maintaining reasonable selectivity.
It was observed that the mesopore volume decreased with progressively longer steam
treatment. The resulting membranes showed much higher permeances but this was
accompanied with a loss in selectivity. By appropriately manipulating the duration of
treatment, one can sensitively adjust the average pore size of the base-CMS membranes
for further post-treatment without simultaneously diminishing their molecular sieve
characteristics.
Methane activation, on the other hand, is shown to block the pores and, thus,
reduce the membrane permeance. However, the outcomes on the permeance for different
gases were not the same: the effect on CO
2
permeances is shown to be more significant
than that of H
2
, resulting in an enhancement of the separation factor for H
2
/CO
2
. The
experimental results also show that the impact is more severe at higher processing
temperature, as seen from the increase in the loss in permeability of all species. The data
are consistent with a model whereby carbon deposition happens mainly at the entrances
of the small pores which are accessible mostly to H
2
and CO
2
.

Once the mouths of such
pores close further declines in the H
2
and CO
2
permeances diminish, rendering methane
treatment ineffective.

87
Chapter IV
Surface Affinity Modification of Carbon Molecular Sieve Membranes

4.1. Introduction
As mentioned earlier in Chapters II and III, a high surface area and an appropriate
pore size distribution are essential for a carbon molecular-sieve membrane to perform
well in separation applications. As mentioned earlier, the molecular sieving mechanism is
able to effectively separate those gas molecules with similar sizes, and tuning of the pore
structure was investigated in Chapter III. However, in addition to size exclusion,
membrane separation can also be accomplished through chemical affinity, which means
that the nature and number of surface groups which may be present on the carbon surface
must also be taken into account. In CMS membranes with pores similar to those of the
molecular diameters of the transporting gases, both molecular sieving and surface affinity
are expected to play an important role. As long as the molecule is not excluded by the
membrane pore structure (molecular sieving), its interaction with the membrane surface
will significantly affect the permeation characteristics during separation.
Carbon atoms located at the edges of the basal planes are unsaturated atoms,
which are usually bonded to heteroatoms giving rise to important surface groups. The
importance of these surface groups lies in the fact that their presence can have a
significant impact on the interaction of the carbon surface with various adsorbates. The
previous studies by our group have identified CO
2
as being a molecule that exhibits a
significant and reversible surface affinity effect as a result of its π-bonding with the slit-

88
shaped, pure carbon pore surface [Sedigh et al., 1999, 2000]. Nevertheless, one may be
able to introduce surface affinity for the CMS membranes towards other selected gas
components (e.g. H
2
or O
2
), via modification of the carbon surface at the atomic level
with metals or metal oxides. For example, in order to effectively separate H
2
from water-
gas-shift and reformate mixtures, a process that modifies the surface affinity of the
carbon surface toward H
2
must be employed, which enhances the H
2
interaction with the
pure carbon surface and improves the efficacy of H
2
separation from gas mixtures. As a
result, the gas separation properties of the CMS membranes can be manipulated by
doping metal or metal oxide components into the membrane structure.
In the past, several techniques have been pursued in an attempt to enhance the
separation characteristics of carbon membranes. Barsema et al. [2003] prepared Ag-
containing CMS membranes using a co-polyimide (BTDA ‐TDA/MDI or P84) containing
Ag salts (AgNO
3
and AgAc). The solution was cast and pyrolyzed at temperatures
ranging from 350 to 800 °C. They reported that adding Ag salts into the co-polyimide
increased the selectivity of O
2
over N
2
for the resulting membranes by a factor of 1.6,
when compared to a non-functionalized CMS membranes prepared by the same pyrolysis
procedure. By measuring the permeabilities of pure He, CO
2
, O
2
, and N
2,
they determined
the effect of Ag nanoclusters in the carbon matrix, concluding that in the case of pure
gases, the Ag nanoclusters act primarily as a spacer at pyrolysis temperatures of up to 600
°C, increasing the O
2
permeability by a factor of 2.4. At higher pyrolysis temperatures
(700 and 800 °C), the separation of O
2
over N
2
was enhanced due to the build-up of an
Ag layer on the surface of the membrane that, however, also significantly reduced the

89
permeability of both gases. Barsema et al. [2005] also prepared Ag-functionalized CMS
membranes from blends of P84 and a sulfonated poy(ether ether ketone) with an Ag
+

counterion (Ag-SPEEK). They proposed that such blends offer the possibility of the
production of new functionalized precursor structures, for producing integrally-skinned
asymmetric hollow fibers. The CMS membranes prepared at a pyrolysis temperature of
800 °C were reported to have the maximum permeability and selectivity. The maximum
permeability was obtained for the CMS membranes that contained approximately 2.5
wt% of Ag (P
O2
= 91.8 Barrer, selectivity of O
2
over N
2
= 8.9), whereas the maximum
selectivity for O
2
over N
2
was 13.5 with a 4.5 wt% of Ag content.
Yoda et al. [2004] prepared Pt and Pd-doped CMS membranes via supercritical
impregnation, using polyimide as the polymeric precursor. Impregnation of Pt(II)
acetylacetonate and Pd(II) acetylacetonate, dissolved in supercritical CO
2
, into the
polymeric precursor was found to yield Pt and Pd nanoparticles that were highly
dispersed inside the polyimide films under optimized conditions. The CMS membranes
were fabricated by pyrolysis of the metal-impregnated polyimide films at 873-1273 K for
2 h, under a vacuum at very low pressure. Single gas permeation tests were performed to
characterize the membranes using H
2
and N
2
. They reported that the permeabilities of
both gases were decreased in the Pd-doped CMS samples, but blocking of N
2
was more
effective than that of H
2
. As a result, a H
2
/N
2
selectivity of 5640 was reported with the
membranes, which was 17 times higher than that of the CMS membranes without metal
nanoparticles.

90
Lie and Hägg [2005] prepared CMS membranes using cellulose as a precursor.
The membranes were doped with several metal oxides and nitrates, including oxides of
Ca, Mg, Fe(III) and Si, and nitrates of Ag, Cu, Fe(III). They were prepared by casting a
film from the mixture of polymer and metal precursors and subsequent carbonization. Lie
and Hagg performed single-gas permeation tests using various gases to study the effect of
the metal additives, and that the carbon membranes containing Fe nitrate had promising
separation performance for the pairs of O
2
/N
2
and CO
2
/CH
4
. The CMS membranes
containing nitrates of Cu or Ag showed high selectivity, but reduced O
2
and CO
2

permeability when compared to membranes prepared using Fe nitrate. The study reported
that the metal oxides, when added to the carbon membranes, retarded the transport of
easily condensable gases (e.g. CO
2
), and therefore failed to improve the CO
2
/CH
4

separation performance. A high H
2
permeability was reported for the Ag ‐ and Cu ‐
containing membranes, 1500 and 1100 Barrer, respectively. Apart from the spacer effect
of the additives, Lie and Hägg also proposed that the gases released upon nitrate
decomposition were porogens, creating pores by tunneling through the material.
Zhang et al. [2006] prepared nano ‐sized nickel particle ‐filled carbon membranes
on the inner surface of porous, sol-gel modified α-Al
2
O
3
support tubes using phenolic
resin as the precursor. Nano-sized nickel particles were prepared by a chemical reduction
method using nickel sulfate hexahydrate as the metal precursor. The nano-sized nickel
powders were then mixed with the polymeric precursor solution. The support was dipped
into that solution, and was subsequently carbonized in order to prepare the nickel-CMS
membranes. To characterize the membranes, single-gas permeation measurements were

91
carried out using H
2
, O
2
, N
2
, and CO
2
. Zhang et al. reported that for the membranes
prepared on the sol-gel modified supports, the Ni-filled carbon membrane generally
exhibited lower N
2
and H
2
permeances, but higher CO
2
/N
2
, O
2
/N
2
and CO
2
/H
2
ideal
permselectivities than the pure carbon membrane. The Ni-filled carbon membranes
showed lower H
2
permeances and higher CO
2
/H
2
ideal selectivities than the pure carbon
membrane, when the nickel content in the precursor solutions was 1 wt%. With the
increase of the nickel content from 1 to 5 wt%, they found that the corresponding carbon
membranes showed decreasing hydrogen permeances and increasing CO
2
/H
2
ideal
selectivity. Further increase of the nickel content to 7.5 and 10 wt%, however, resulted in
the increase of the H
2
permeance, and the decrease of the CO
2
/H
2
ideal selectivity, when
compared to the corresponding carbon membranes.
Grainger and Hägg [2007] prepared unsupported CMS membranes from
cellulose-hemicellulose, and then Cu(II) nitrate was added to the precursor in the range of
0-6 wt%. The carbonization temperature was varied from 400 to 700 °C to investigate the
effect on the permeation properties of the membrane. Mixed gas tests were performed
using H
2
, CO
2
, C
1
-C
4
, and N
2
, in order to characterize the prepared membrane. A similar
trend was found between pure carbon and metal additive membrane, which the peak of
H
2
and CO
2
permeability occurred between 550 and 650 °C. They also reported that the
selectivity of the 4 wt% copper nitrate-doped membrane was higher than for the pure
carbon membrane, decreasing as the final temperature increased. The hydrogen/methane
permselectivity of >1000 was reported and the hydrogen/carbon dioxide permselectivity

92
was found to be approximately 23 at 25 °C. A loading of 6 wt% caused a surface layer to
appear and subsequent retardation of hydrogen caused the selectivity to decrease again.
Liu et al. [2008] prepared nanoporous Ni ‐ containing carbon membranes on
alumina supports using nickel acetate as the metal precursor. A Ni/polyamideimide sol
was prepared by dissolving Ni salts in a polymeric precursor. The support was coated
with the precursor and carbonized at high temperature, followed by catalytic
decomposition of methane within the membrane structure in a nitrogen atmosphere. Pure
gas permeation measurements were performed using N
2
, CH
4
, CO
2
, He, and H
2
in order
to characterize the resulting membranes. Liu et al. found that the separation performance
of these membranes was significantly better than that of pure CMS membranes fabricated
by a single coating and conventional pyrolysis, even as good as or better than those of
membranes prepared using three carbon layer coatings. They suggested that this
procedure significantly reduced the cost of membrane production, and that it provided a
promising technical way for the preparation of carbon membranes.
In summary, a few prior studies attempted to tune the properties of the CMS
membranes by incorporating metal or oxide nanoparticles into the membrane structure.
The main technique utilized was to mix the nanoparticles into the polymeric precursor,
prior to carbonization in order to fabricate the hybrid CMS membranes. The results were
encouraging with ideal selectivity of selected gas pairs, such as O
2
/N
2
, H
2
/N
2
and
CO
2
/CH
4
, showing improvement with the metal incorporation. In this chapter, the
preparation of metal-carbon membranes is studied using several different approaches.

93
The results are compared among the various techniques in order establish which
technique is the most optimal
4.2. Experimental Approach
4.2.1. Membrane Preparation and Modification
Incorporation of metal or metal oxide nano-particles is the primary technique
utilized here, which enables us to adjust the surface affinity towards specific gas
components via surface interaction. Nickel (Ni) and Palladium (Pd) were chosen in this
study as the test metal components, because they are well-known to have significant
strong affinity toward H
2
. The deposition of Ni/Pd nano-particles within the CMS
membrane will potentially tune the carbon surface properties, in a way that the CO
2
-π-
bonding interaction is inhibited as compared to that of H
2
-Ni/Pd interaction. As part of
our study we have developed and tested several techniques in order to incorporate the
Ni/Pd nano-particles into the CMS membrane structure. These include (1) impregnation;
(2) direct incorporation; and (3) the so-called sandwiching method.
4.2.1.1. Impregnation
To prepare metal-impregnated CMS membranes using this technique, the base-
CMS membranes were fabricated using the technique discussed in Chapter II. Each base-
CMS membrane was prepared with three layers, in order to ensure proper membrane
performance. The membrane performance and transport properties of the base-CMS
membranes were tested in terms of the flux through the membranes, and its
permselectivity towards three test-gases, namely H
2
, CO
2
, and CH
4
, before further
processing. Nickel formate was chosen as the metal precursor for the preparation of the

94
metal-impregnated CMS membranes due to the fact that it is known to self-decompose to
nickel nano-particles at elevated temperature [Edwards & Garner, 1997]. In order to
prepare the metal precursor solution, the Nickel(II) formate dehydrate salt (Alfa Aesar)
was dissolved in water together with additional formic acid (Alfa Aesar, 96+%). The
addition of formic acid assists the decomposition of the Ni-formate compound to nickel
nano-particles, and increases its solubility in water. To prepare the Ni-impregnated CMS
membranes, the outer surface of base-CMS membrane was wrapped with Teflon tape. It
was then impregnated with the 0.004 M Ni-formate solution for different periods of time,
depending on the desired amount of Ni nano-particles to be deposited on the membranes.
After impregnation with the Ni-formate solution, the membrane was stored in an
incubator and was dried in air for 24 h. The Ni-formate adsorbed on the membrane was
then thermally decomposed in a stainless steel tubular furnace in the presence of flowing
UHP hydrogen and helium (flowing at a rate of 75-100 cm
3
/min with a He/H
2
ratio of
1:1), in order to remove all the gases evolved during the decomposition process, and to
maintain an inert atmosphere. The thermal decomposition was carried out in two stages.
The membrane was initially heated-up to 150 °C, with a heating rate of 1 °C/min, and
was held there at that temperature for 2 h, in order to evaporate all the water contained
within the salt. The temperature was then raised to 300 °C with a heating rate of 1
°C/min, and the membrane was held at this temperature for an additional 15 min.
Subsequently, it was cooled down to 120 °C at a rate of 1 °C/min. The temperature of
300 °C was chosen for the decomposition, since the Ni salt is known to self-decompose
rapidly to Ni nano-particles at this temperature [Edwards & Garner, 1997; Xia et al.,

95
2001; Geng et al., 2004]. The membranes were exposed to flowing hydrogen, since its
presence is known to assist in the decomposition process [Xia et al., 2001; Geng et al.,
2004]. Measuring the sample weight before and after the metal impregnation determines
the estimated amount of Ni nano-particles that are loaded on the membranes.
4.2.1.2. Direct Incorporation
Metal-loaded CMS membranes were also prepared by direct incorporation of the
precursor metal compounds into the polymeric precursor prior to its pyrolysis for CMS
membrane formation. Nickel-acetate, nickel-acetylacetonate, and palladium-
acetylacetonate were chosen as the metal precursors to self-decompose to Ni/Pd
nanoparticles at elevated temperatures in H
2
atmosphere. In order to prepare the
metal/polymeric solution, the nickel(II)-acetate tetrahydrate salt (Acros Organics, 99+%)
was initially dissolved in N-methypyrolidone (NMP) (Burdick & Jackson, 99.5% min).
The nickel and palladium acetylacetonate salts (Acros Organics, 96%) were dissolved in
N,N-dimethylacetamide (DMAc) (Acros Organics). The PEI polymer was subsequently
added into the metal precursor solution. Different solution concentrations were prepared
(categorized in terms of their PEI and Ni/Pd wt% content).
Similar types of ceramic tubular supports (provided by our industrial partner M &
P) were used as those utilized for the preparation of the base-CMS membranes in Chapter
II. The ends of each substrate were glazed, and their outer surface was again wrapped
with Teflon tape. To fabricate a Ni/Pd-carbon layer, the substrate was then dip-coated
with the Ni-(or Pd) PEI solution for 3 min, and was pulled from the solution at a constant
rate of 2 cm/min in order to create a layer with uniform thickness. After coating with the

96
Ni (or Pd)-PEI film, the membrane was stored in the incubator, and dried in air for 24 h.
The membrane was then carbonized in a cylindrical furnace in the presence of UHP
argon, flowing at a rate of 60-75 cm
3
/min, in order to remove all the gases evolved during
the pyrolysis process, and to maintain an inert atmosphere during the pyrolysis step. The
carbonization process was carried out in two stages in an Ar atmosphere. The membrane
was initially heated up to 350 °C with a heating rate of 1 °C/min, and was held there at a
constant temperature for 0.5 h. The temperature was then raised to 600 °C, at a rate of 1
°C/min, and the membrane was held there for 4 h. Subsequently, the membrane was held
at this temperature for 2 h with H
2
flowing at a rate of 60-75 cm
3
/min for the thermal
decomposition of the Ni (or Pd) precursor to occur. Treatment with H
2
is thought to result
in particles with a smaller particle size and better dispersion. It was then cooled down to
180 °C, at a rate of 2 °C/min, and further down to room temperature at a rate of 5 °C/min.
The coating, carbonization, and decomposition procedure can be repeated as many times
as needed in order to modify the selective layers to achieve the desired membrane
performance.
4.2.1.3. The Sandwiching Method
In addition to the direct-incorporation method, metal-CMS membranes were also
prepared by sandwiching Ni layers in between the carbon layers. Nickel acetylacetonate
was chosen as the metal precursor. In order to prepare the metal solution, the nickel
acetylacetonate salt was initially dissolved in DMAc. The PEI polymeric solution was
prepared as described in Chapter II. Several individual metal and polymeric precursor
solutions with different concentrations were prepared (categorized, again, in terms of the

97
PEI and Ni wt% content). Similar types of ceramic tubular supports were used as those
for the base-CMS membranes in Chapter II. The ends of each substrate were glazed with
their outer surface being wrapped in Teflon tape. The substrate was then dip-coated with
the PEI or Ni precursor solution for 3 min, and was pulled from the solution at a constant
rate of 2 cm/min in order to create a layer with uniform thickness. For successive layers,
the same dip-coating method was applied. After coating with the first PEI film, the
membrane was dried and carbonized using the same carbonization method as those for
base-CMS membranes. After the first layer of carbon coating, the membrane was then
coated with a layer of the Ni precursor solution, which was then either dried in the oven
with air at 120 °C for 2 h, or decomposed in the furnace at 600 °C for 2 h with H
2
,
flowing at a rate of 60-75 cm
3
/min, prior to another PEI polymeric coating. After that, an
additional layer of PEI was placed on the membrane which was pyrolyzed following the
procedure described in Chapter II. This coating and carbonization procedure can be
repeated as many times as needed in order to modify the selective layers to achieve the
desired membrane performance.
4.2.2. Membrane Testing
The transport characteristics of the membranes prepared were evaluated in terms
of the flux through the membrane using single gases (UHP hydrogen, carbon dioxide, and
methane) and its ideal permselectivity towards various test gas pairs. Detailed description
of the experimental apparatus as well as details about the performance of the experiments
together with the equations used for data analysis were presented in Chapter II.

98
4.3. Results and Discussion
4.3.1. Preparing Ni-CMS Membranes via Impregnation
Several Ni-CMS membranes have been prepared using the metal impregnation
technique. Table 4.1-4.4 summarized the results in terms of permeance and selectivity of
the impregnated Ni-CMS membranes. For these Tables, as well as for the remainder of
this chapter, the names of the membranes are abbreviated in the following manner. The
initial letter signifies the method of metal addition, namely I for impregnation, D for
direct incorporation, and S for the sandwiching method. This is followed by the symbol
of the batch number (corresponding to the base-CMS membranes prepared and
carbonized at the same time), which is then followed by the number corresponding to the
membrane (for example, I-B4-M1 corresponds to the first membrane prepared in batch 4
by impregnation). The membranes in Table 4.1-4.3 were only impregnated once, while
membranes in Table 4.4 were impregnated twice.

99

Before After Change
M1 (~3.02 wt% Ni)
P
H2
1.147 0.883 - 23%
P
CO2
0.500 0.164 - 67%
P
CH4
0.046 0.143 + 200%
S(H
2
/CO
2
) 2.3 5.4 + 135%
S(H
2
/CH
4
) 24.7 6.2 - 75%

M2 (~15 wt% Ni)
P
H2
1.098 0.660 - 40%
P
CO2
0.359 0.109 - 70%
P
CH4
0.023 0.023 + 4%
S(H
2
/CO
2
) 3.1 6.1 +100%
S(H
2
/CH
4
) 47.8 28.6 - 40%

M3 (~26.6 wt% Ni)
P
H2
1.351 1.433 + 6%
P
CO2
0.490 0.327 - 33%
P
CH4
0.062 0.013 - 78%
S(H
2
/CO
2
) 2.8 4.4 + 60%
S(H
2
/CH
4
) 21.6 6.6 - 70%

Table 4.1 Permeances and separation factors measured at 120 °C for Ni-impregnated
membrane –first batch (I-B1) (permeances in m
3
/m
2
.bar.h)

Before After Change
M1 (~0.8 wt% Ni)
P
H2
1.648 0.154 -6%
P
CO2
0.367 0.274 -25%
P
CH4
0.133 0.214 +60%
S(H
2
/CO
2
) 4.5 5.6 +24%
S(H
2
/CH
4
) 12.4 7.2 -40%

M2 (~27.8 wt% Ni)
P
H2
1.247 0.892 - 28%
P
CO2
0.334 0.168 - 50 %
P
CH4
0.101 0.111 + 9 %
S(H
2
/CO
2
) 3.7 5.2 + 40%
S(H
2
/CH
4
) 12.3 8.1 - 34%

Table 4.2 Permeances and separation factors measured at 120 °C for Ni-impregnated
membrane –second batch (I-B2) (permeances in m
3
/m
2
.bar.h)

100

Before After Change
M1 (~1.5 wt% Ni)
P
H2
2.246 1.733 - 23%
P
CO2
0.659 0.486 - 26%
P
CH4
0.068 0.091 + 35%
S(H
2
/CO
2
) 3.4 3.6 + 4.7%
S(H
2
/CH
4
) 33.2 19.0 - 43%

M2 (~7.7 wt% Ni)
P
H2
0.964 1.110 +12.8%
P
CO2
0.148 0.134 - 9 %
P
CH4
0.039 0.048 + 23.6 %
S(H
2
/CO
2
) 6.5 8.3 +27%
S(H
2
/CH
4
) 24.8 23.2 - 6.6%

M3 (~20 wt% Ni)
P
H2
0.920 0.639 -30%
P
CO2
0.252 0.228 -10%
P
CH4
0.036 0.034 -5.3%
S(H
2
/CO
2
) 3.6 2.8 -23%
S(H
2
/CH
4
) 25.3 18.6 -27%

M4 (~23 wt% Ni)
P
H2
1.190 0.976 - 18%
P
CO2
0.194 0.163 - 16%
P
CH4
0.017 0.034 + 100%
S(H
2
/CO
2
) 6.1 6.0 - 1.8%
S(H
2
/CH
4
) 70.5 29.0 - 57%

Table 4.3 Permeances and separation factors measured at 120 °C for Ni-impregnated
membrane –third batch (I-B3) (permeances in m
3
/m
2
.bar.h)

101
M1
CMS 1
st
Impregnation
(~0.5 wt%)
Change 2
nd
Impregnation
(~1.5 wt%)
Change
P
H2
0.535 0.466 -13% 0.435 -19%
P
CO2
0.170 0.161 -5% 0.146 -14%
P
CH4
0.015 0.035 +138% 0.026 +75%
S(H
2
/CO
2
) 3.2 2.9 -7.9% 3.0 -5.1%
S(H
2
/CH
4
) 36.1 13.2 -64% 16.7 -54%
M2
CMS 1
st
Impregnation
(~0.4 wt%)
Change 2
nd
Impregnation
(~8.34 wt%)
Change
P
H2
1.022 0.629 -38% 0.371 -64%
P
CO2
0.198 0.152 -23% 0.064 -68%
P
CH4
0.009 0.019 +107% 0.024 +172%
S(H
2
/CO
2
) 5.1 4.1 -19.8% 5.8 +13%
S(H
2
/CH
4
) 114.0 33.9 -70% 15.2 -86%
M3
CMS 1
st
Impregnation
(~1.5 wt%)
Change 2
nd
Impregnation
(~15 wt%)
Change
P
H2
0.859 0.076 -11% 0.489 -43%
P
CO2
0.225 0.218 -3% 0.098 -56%
P
CH4
0.015 0.217 44% 0.012 -24%
S(H
2
/CO
2
) 3.8 3.5 -8.6% 5.0 +30%
S(H
2
/CH
4
) 56.8 35.0 -38% 42.5 -25%
M4
CMS 1
st
Impregnation
(~4.4 wt%)
Change 2
nd
Impregnation
(~18.3 wt%)
Change
P
H2
0.477 0.285 -40% 0.215 -55%
P
CO2
0.158 0.089 - 44 % 0.050 -69%
P
CH4
0.007 0.033 +360% 0.011 +59%
S(H
2
/CO
2
) 3.0 3.2 +6.3% 4.3 +43%
S(H
2
/CH
4
) 67.6 8.7 - 87% 19.1 -72%
M5
CMS 1
st
Impregnation
(~17 wt%)
Change 2
nd
Impregnation
(~25 wt%)
Change
P
H2
2.533 1.546 -39% 1.790 -29%
P
CO2
0.762 0.299 -61% 0.428 -44%
P
CH4
0.034 0.029 -15% 0.055 +63%
S(H
2
/CO
2
) 3.3 5.2 +55% 4.2 +26%
S(H
2
/CH
4
) 74.9 53.8 -28% 32.5 -57%

Table 4. 4 Permeances and separation factors measured at 120 °C for Ni-impregnated
membrane –fourth batch (I-B4) (permeances in m
3
/m
2
.bar.h)

102
Comparing the performance of the membranes before and after the impregnation
process, the permeances of both H
2
and CO
2
decrease upon deposition of the metal
(though exceptions also exist, notably membranes I-B1-M3 and I-B3-M2, for which the
hydrogen permeance increases 6-13%, while the CO
2
permeance still decreases).
Generally the permeance of CO
2
decreases faster than the permeance of H
2
resulting in
ideal separation factors of H
2
/CO
2
which are generally higher for the Ni-impregnated
membranes; this is particularly true for the membrane with the higher Ni loadings, for
which the separation factor of H
2
/CO
2
is almost twice as high as that of the original base-
CMS membrane (e.g. membrane I-B1-M2). These results are consistent with the concept
that Ni impregnation enhances the membrane surface affinity towards hydrogen, though
some of the beneficial effects are masked by the fact that Ni deposition itself seems to
decrease the average micropore diameter, as manifested by decreases in both the H
2
and
CO
2
transport. The permeance of H
2
and CO
2
and the corresponding ideal selectivity are
shown in Figure 4.1 and 4.2 as a function of wt% Ni loading for all the membranes in
Batch 4. That the Ni enhances the surface affinity towards hydrogen is confirmed. The
higher selectivity is achieved at higher Ni loading; however, the permeance still remains
lower than the permeance of the base-CMS membrane due to the pore constriction from
deposition of Ni within the pores.

103

Figure 4.1 Percent change in the permeances of H
2
and CO
2
as a function of Ni loading
for Ni-impregnated membranes (I-B4)

Figure 4.2 Percent change in the separation factor of H
2
/CO
2
as a function of Ni loading
for Ni-impregnated membranes (I-B4)

-80
-60
-40
-20
0
0 5 10 15 20 25 30
% change in permeance
wt% Ni
H₂
-20
-10
0
10
20
30
40
50
0 5 10 15 20 25 30
% change in SF(H
2
/CO
2
)
wt% Ni

104
The methane permeance generally increases with addition of Ni. It signifies that
the addition of Ni somehow opens the larger channels through which CH
4
is transported
during the impregnation/decomposition process. The H
2
/CO
2
separation factors of those
membranes in Table 4.1-4.4 are shown to be higher than the base-CMS membranes;
however, the permeances were also significantly decreased due to the narrowing of the
pore structure from Ni impregnation. Preparing CMS membrane with high initial H
2
/CO
2

separation values is clearly possible through the judicious use of methane activation,
followed by Ni impregnation, if the gases permeances are not as critical as separation
properties. On the other hand, Ni impregnation does not appear to be a technique
appropriate for preparing selective membranes for CO
2
/CH
4
gases pair, since addition of
Ni appears to enhance the CH
4
permeation rate.
4.3.2. Preparing Metal-CMS Membranes via Direct Incorporation
4.3.2.1. Preparing Ni-CMS Membranes via Direct Incorporation
A number of Ni-CMS membranes were prepared using the direct incorporation
method. Using the Ni-acetate tetrahydrate as the metal precursor generally proved
ineffective in preparing high quality molecular sieve type membranes. For example,
Table 4.5 indicates the transport characteristics of three membranes (columns 2, 4, and 6)
prepared by first coating the substrates once with a solution containing 6 wt% of PEI and
0.5 wt% of Ni (referred to as the 6%PEI/0.5%Ni solution), and then twice with a
2%PEI/0.5%Ni solution, the carbonization/decomposition step carried out before
applying each additional coating. For these membranes, the permeances of CH
4
were
slightly higher than those of CO
2
, which is indicative that these membranes do not show

105
molecular sieving characteristics, potentially due to the existing cracks within the
membranes or poor adhesion between Ni nano-particles and CMS membranes.
Membrane D-B1-M1 and D-B1-M2 were additionally coated with a pure PEI layer (a 6%
PEI solution membrane D-B1-M1, and a 2% PEI solution for membrane D-B1-M2) and
pyrolyzed again in order to repair any such cracks in the membranes. As one can see
from Table 4.5, the addition of the polymeric coatings and the subsequent carbonization
does seem to repair the cracks, thus resulting in a significant decrease in the permeances
of CH
4
and in improvement of the separation of H
2
/CH
4
value. Increasing the PEI content
of the initial coating did not seem to have a beneficial improvement, as the data in Table
4.6 indicate. The two membranes in this Table were prepared by a single coating of a
higher initial concentration of polymer (10%PEI/0.5%Ni). As shown in Table 4.6, the
permeances of CH
4
are still slightly higher than those of CO
2
,

indicating that the
membrane does not exhibit molecular sieving characteristics. In conclusion, from this
part of the study, nickel acetate is not an ideal metal precursor compound to be used for
preparing metal-containing CMS membranes. Using additional polymeric precursor
coatings and carbonizations appears to be an effective way to repair some of the defects
that these membranes develop (e.g., Table 4.5); however, this introduces additional
processing steps, and it does not appear to improve the hydrogen separation performance
of Ni-CMS membranes compared to base-CMS membranes.

106
Membrane
1
Membrane
2
Membrane
3
Ni-CMS Ni-CMS with
6% PEI
Ni-CMS Ni-CMS with
2% PEI
Ni-CMS
P
H2
18.97 5.92 18.27 9.83 18.91
P
CO2
7.13 2.79 6.47 3.73 7.70
P
CH4
8.95 1.00 7.28 2.37 8.82
S(H
2
/CO
2
) 2.7 2.1 2.8 2.6 2.5
S(H
2
/CH
4
) 2.1 5.9 2.5 4.2 2.1

Table 4.5 Permeances and separation factors measured at 120 °C of Ni-CMS membranes
prepared by the direct incorporation method – first batch (D-B1) (permeances in
m
3
/m
2
.bar.h)

Membrane 1 Membrane 2
P
H2
13.24 13.92
P
CO2
6.28 6.29
P
CH4
7.22 7.27
S(H
2
/CO
2
) 2.1 2.2
S(H
2
/CH
4
) 1.8 1.9

Table 4.6 Permeances and separation factors measured at 120 °C of Ni-CMS membranes
prepared by the direct incorporation method – second batch (D-B2) (permeances in
m
3
/m
2
.bar.h)

Another set of metal-CMS membranes was prepared using a different Ni
precursor, namely nickel-acetylacetonate. Table 4.7 shows the results of two membranes
that were prepared by coating the substrates once with a 10%PEI/0.5%Ni solution, and
then twice with a 2%PEI/0.5%Ni solution, again with a carbonization and decomposition
step being carried out before applying each additional coating. These membranes, in
contrast with the membranes prepared by using the nickel-acetate tetrahydrate precursor
with same concentration of Ni in PEI solution, do show molecular sieving characteristics,
with the membrane’s methane permeance being significantly smaller than the CO
2

permeance. Therefore, nickel acetylacetonate appears to be a more appropriate precursor

107
for the preparation of hydrogen permselective membranes and will be use in further
study.
Membrane 1 Membrane 2
1
st
layer 2
nd
layer 3
rd
layer 1
st
layer 2
nd
layer 3
rd
layer
P
H2
3.10 3.08 3.30 3.75 4.49 4.39
P
CO2
2.46 2.09 2.57 2.48 2.63 2.67
P
CH4
0.93 0.77 1.31 1.36 1.53 1.76
S(H
2
/CO
2
) 1.3 1.5 1.3 1.5 1.7 1.6
S(H
2
/CH
4
) 3.4 4 2.5 2.8 2.9 2.5

Table 4.7 Permeances and separation factors measured at 120 °C of Ni-CMS membranes
prepared by the direct incorporation method – third batch (D-B3) (permeances in
m
3
/m
2
.bar.h)

The effect of Ni content on membrane properties was also investigated. Higher Ni
content appears to be detrimental to membrane performance, as shown in Table 4.8,
which shows the transport characteristics of membranes prepared by deposition and
carbonization of a 6%PEI/1%Ni layer, followed by deposition and carbonization of
additional 2%PEI/1%Ni layers. These membranes shown in Table 4.8 appear to have
much higher permeances of H
2
, CO
2
, and CH
4
, compared to those membranes prepared in
Table 4.7; however, these membranes generally do not appear to have molecular sieving
properties. This is potentially due to the fact that a higher concentration of Ni gives rise
to poor adhesion between the CMS membrane and the Ni nano-particles, and to the
creation of bypass, thus resulting in the increase of the gas permeances with simultaneous
elimination of the molecular sieving behavior. Therefore, higher Ni content may have a
detrimental effect on membrane characteristics.

108

Membrane 1 Membrane 2
1
st
layer 2
nd
layer 3
rd
layer 1
st
layer 2
nd
layer 3
rd
layer
P
H2
14.60 7.74 8.46 15.26 4.90 7.71
P
CO2
6.29 3.54 4.01 6.31 2.32 3.29
P
CH4
7.37 3.51 3.54 7.41 1.26 1.78
S(H
2
/CO
2
) 2.3 2.2 2.1 2.4 2.1 2.3
S(H
2
/CH
4
) 2.0 2.2 2.4 2.1 3.9 4.3

Membrane 3 Membrane 4
1
st
layer 2
nd
layer 3
rd
layer 1
st
layer 2
nd
layer 3
rd
layer
P
H2
6.61 7.49 16.57 19.76
P
CO2
2.78 3.44 6.20 6.85
P
CH4
2.19 2.63 7.17 7.58
S(H
2
/CO
2
) 2.4 2.2 2.7 2.9
S(H
2
/CH
4
) 3.0 2.9 2.3 2.6

Table 4.8 Permeances and separation factors measured at 120 °C of Ni-CMS membranes
prepared by the direct incorporation method – fourth batch (D-B4) (permeances in
m
3
/m
2
.bar.h)

In addition, a hybrid approach was also developed and utilized in order to prepare
Ni-CMS membranes. The membrane support was first coated with a pure polymeric
precursor (6% PEI) and carbonized, followed by coating with two 2%PEI/0.1%Ni layers
and subsequent carbonization/decomposition (membrane D-B5-M1 and D-B5-M2 in
Table 4.9), and in some instances (membrane D-B5-M3 and D-B5-M4 in Table 4.9) with
an additional 2%PEI/0.1%Ni layer (2
nd
measurement in Table 4.9 is for the membranes
with the additional layer).

109

Membrane 1 Membrane 2
P
H2
7.08 9.06
P
CO2
2.26 2.84
P
CH4
0.30 0.75
S(H
2
/CO
2
) 3.1 3.2
S(H
2
/CH
4
) 24.2 12.1
Membrane 3 Membrane 4
Measurement 1
st
2
nd
1
st
2
nd

P
H2
4.38 2.85 10.51 5.89
P
CO2
1.14 0.68 3.13 1.64
P
CH4
0.23 0.10 0.73 0.45
S(H
2
/CO
2
) 3.8 4.2 3.4 3.6
S(H
2
/CH
4
) 19.5 28.0 14.4 13.0

Table 4.9 Permeances and separation factors measured at 120 °C of Ni-CMS membranes
prepared by the direct incorporation method – fifth batch (D-B5) (permeances in
m
3
/m
2
.bar.h)

These membranes do show molecular sieve characteristics, similar to Ni-loaded
CMS membranes, prepared by impregnation. However, they show much higher
permeances by utilizing a very small metal amount. In addition, these membranes in
Table 4.9 appear to have significantly better separation performance comparing to those
Ni-CMS membranes prepared utilizing PEI/Ni solution in the first layer. This is most
likely due to the fact that the first carbon layer potentially repairs any microscopic cracks
initially present in the support substrates and provides a smooth surface for additional
coating.
Another batch of Ni-loaded CMS membranes was prepared by carbonization of a
6% PEI layer, followed by deposition and carbonization of additional 2%PEI/0.1%Ni
layers. The permeances of H
2
, CO
2
, and CH
4
and ideal H
2
/CO
2
and H
2
/CH
4
separation
factors after each coating are shown in Figure 4.3 and 4.4. As expected, the membrane

110
performances and separation factors are a strong function of the number of
coating/carbonization steps. Adding extra Ni-carbon layers to the membranes does result
in a decrease in the permeances, but a simultaneous increase in the separation factors.
This behavior is consistent with those base-CMS membranes described Chapter II.

Figure 4.3 Permeances measured at 120 °C as a function of number of layers for
membranes prepared by direct incorporation method – sixth batch (D-B6) (permeances in
m
3
/m
2
.bar.h)

0
2
4
6
8
10
12
1 2 3 4
Permeance (m
3
/m
2
.bar.h)
Number of Layers
H₂
CO₂
CH₄

111

Figure 4.4 Separation factor measured at 120 °C as a function of number of layers of
membranes prepared by direct incorporation method – sixth batch (D-B6)

The effect of Ni content on membrane properties was also investigated utilizing
this hybrid approach. The membranes were initially coated with 6% PEI and carbonized,
followed by additional coatings with various concentration of nickel in 2% PEI solutions,
again with a carbonization/decomposition step being carried out before applying each
additional coating. After pyrolyzing the initial 6%PEI coating, the gaseous permeances of
the membranes were measured, in order to ensure that all membranes have similar
transport properties, namely the permeances and ideal separation factors, before any
further processing, as shown in Table 4.10.

0
5
10
15
20
25
1 2 3 4
Separation Factor
Number of Layers
SF(H₂/ CO₂)
SF(H₂/ CH₄)

112

Membrane 1 Membrane 2 Membrane 3 Membrane 4
P
H2
11.60 11.95 10.38 10.47
P
CO2
5.36 4.77 4.45 4.88
P
CH4
2.50 2.04 2.11 2.49
S(H
2
/CO
2
) 2.2 2.5 2.3 2.1
S(H
2
/CH
4
) 4.6 5.8 4.9 4.2

Table 4.10 Permeances and separation factors measured at 120 °C of one-layer carbon
membranes before additional Ni/PEI coatings (permeances in m
3
/m
2
.bar.h)

Figure 4.5 and 4.6 show the permeances of H
2
, CO
2
, and CH
4
, and separation
factors as a function of wt% Ni in 2% PEI solution.

Figure 4.5 Permeances measured at 120 °C as a function of Ni content in 2% PEI solution
of membranes prepared by direct incorporation method (permeances in m
3
/m
2
.bar.h)

0
1
2
3
4
5
6
0 0.2 0.4 0.6 0.8 1 1.2
Permeance (m
3
/m
2
.bar.h)
% Ni in 2% PEI Solution
M1_H₂
M2_H₂
M3_H₂
M4_H₂
M1_CO₂
M2_CO₂
M3_CO₂
M4_CO₂
M1_CH₄
M2_CH₄
M3_CH₄
M4_CH₄

113

Figure 4.6 Separation factors measured at 120 °C as a function of Ni content in 2% PEI
solution of membranes prepared by direct incorporation method

Note that the permeances of all three gases (H
2
, CO
2
, and CH
4
) appear to increase as the
amount of nickel in the 2% PEI solution increases. This is probably due an increase in the
number of bypasses between the Ni nano-particles and the CMS membrane structure as
the concentration of Ni increases. However, as seen from Figure 4.6, the separation
factors stay fairly constant as one increases the nickel concentration. A further increase
in the Ni loading in 2% PEI solution does seem to negatively impact the membrane’s
separation characteristics. This is potentially due to the fact that higher concentration of
Ni loading causes poor adhesion between the CMS membrane and the Ni nano-particles,
thus destroying the molecular sieving behavior.
4.3.2.2. Preparing Pd-CMS Membranes via Direct Incorporation
Another metal, namely Pd, was also utilized in membrane preparation. For the
preparation of such membranes we used Pd-acetylacetonate as the metal precursor. To
0
5
10
15
20
25
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Separation Factor
%Ni in 2% PEI Solution
M1_SF(H₂/CO₂)
M2_SF(H₂/CO₂)
M3_SF(H₂/CO₂)
M4_SF(H₂/CO₂)
M1_SF(H₂/CH₄)
M2_SF(H₂/CH₄)
M3_SF(H₂/CH₄)
M4_SF(H₂/CH₄)

114
prepare the membrane, the support was first coated with the pure polymeric precursor
(6% PEI) and carbonized, followed by coating with a 2%PEI/0.1%Pd layer and
subsequent carbonization/decomposition, the latter step repeated twice. Table 4.11
shows the results with two membranes prepared using this metal precursor. These
membranes do exhibit molecular sieve characteristics and extremely high gas
permeances. However, the hydrogen separation properties do not appear to be enhanced
by Pd metal additions. This is potentially due to the poor adhesion between the Pd nano-
particles and the CMS membranes. Both membranes were coated with an additional
2%PEI/0.1%Pd layer, as shown in Table 4.11. The permeances of all three gases
generally decreased after the additional layer with slight improvement in the H
2
/CH
4

selectivity.
Membrane 1 Membrane 2
Pd-CMS Additional coating Pd-CMS Additional coating
P
H2
7.00 7.07 15.24 6.20
P
CO2
3.87 3.65 5.96 2.65
P
CH4
1.89 1.53 5.54 1.45
S(H
2
/CO
2
) 1.8 1.9 2.6 2.3
S(H
2
/CH
4
) 3.7 4.6 2.8 4.3

Table 4.11 Permeances and separation factors measured at 120 °C of Pd-CMS
membranes prepared by the direct incorporation method – seventh batch (D-B7)
(permeances in m
3
/m
2
.bar.h)

As mentioned earlier, the metal content within the CMS membranes can affect the
transport characteristics. An excessive metal content appears to be detrimental to
membrane performance, as shown in Figure 4.6. Two additional batches of membranes
were therefore prepared using a smaller amount of palladium, namely 2%PEI/0.05%Pd
and 2%PEI/0.01%Pd solutions. Again, the support is initially coated with a 6% PEI

115
solution and carbonized, followed by coating with 2%PEI/0.05%Pd (D-B8-M1) or
2%PEI/0.01%Pd (D-B8-M2) solutions and subsequent carbonization/decomposition (the
latter step being repeated twice). The results of these membranes are shown in Table
4.12.
Membrane 1
(0.05%Pd)
Membrane 2
(0.01%Pd)
P
H2
6.06 5.97
P
CO2
2.21 2.24
P
CH4
0.51 0.50
S(H
2
/CO
2
) 2.8 2.7
S(H
2
/CH
4
) 12.9 12.3

Table 4.12 Permeances and separation factors measured at 120 °C of Pd-CMS
membranes prepared by the direct incorporation method – eight batch (D-B8)
(permeances in m
3
/m
2
.bar.h)

The decrease in the amount of Pd loaded in the CMS membranes seems to
decrease the formation of cracks during the carbonization/decomposition process, as
indicated by the increase in the H
2
/CH
4
separation factors in Table 4.12. However, as one
decreases the Pd concentration from 0.05 wt% to 0.01 wt%, there is no further
improvement in the separation properties. Such small amount of Pd metal addition
appears to be ineffective in the enhancement of surface affinity toward hydrogen. As a
result, there is no further increase in the hydrogen permeances and separation properties
of these Pd-CMS membranes.
4.3.3. Preparing Ni-CMS Membranes via the Sandwiching Method
Several Ni-CMS membranes were prepared by the sandwiching method in the
following manner: The support was first coated with a 6% PEI layer and carbonized,
followed by dip-coating into the nickel-acetylacetonate solution (0.5 wt% Ni) and drying

116
at 120 °C, followed by coating with a 2 wt% PEI solution and carbonization, and then
followed, once more, by dip-coating into the nickel-acetylacetonate solution (0.5 wt% Ni)
and drying at 120 °C, followed by coating with a 2 wt% PEI solution and a final
carbonization. The permeances of H
2
, CO
2
, and CH
4
(noted as 1
st
measurement) are
shown in Table 4.13.
Membrane 1 Membrane 2 Membrane 3 Membrane 4
Measurement 1
st
2
nd
1
st
2
nd
1
st
2
nd
1
st
2
nd

P
H2
11.32 0.46 7.80 2.39 5.63 1.31 15.00 1.79
P
CO2
4.71 0.13 3.07 0.54 2.47 0.41 5.29 0.45
P
CH4
3.21 0.047 1.51 0.064 1.35 0.040 4.59 0.10
S(H
2
/CO
2
) 2.4 3.6 2.5 4.4 2.3 3.2 2.8 4.0
S(H
2
/CH
4
) 3.5 9.8 5.2 37.1 4.2 32.7 3.3 17.3

Table 4.13 Permeances and separation factors measured at 120 °C of Ni-CMS
membranes prepared by the conventional sandwiching method (S-B1) (permeances in
m
3
/m
2
.bar.h)

These membranes show molecular sieve characteristics; however, the presence of
defects appears to be the main issue. Coating these membranes with 2 layers of pure 2
wt% PEI solution and carbonization after coating each layer in order to repair such
defects, if present, significantly enhanced (2
nd
measurement in Table 4.13) both the
S(H
2
/CO
2
) as well as the S(H
2
/CH
4
) values. We also investigated whether it would be
beneficial, if instead of drying, we decompose the nickel precursor solution prior to the
dip-coating within the PEI solution. This did not appear to have any major beneficial
effects, as Table 4.14 indicates, which shows the transport properties of two membranes
prepared by this approach, as compared with the more traditional method. Since
decomposition is a more time- and resource-consuming step, as compared to drying, this
approach does not appear promising or warranting further investigation.

117

Membrane 1 Membrane 2
P
H2
4.17 3.43
P
CO2
1.35 1.26
P
CH4
1.34 1.16
S(H
2
/CO
2
) 3.1 2.7
S(H
2
/CH
4
) 3.1 3.0

Table 4.14 Permeances and separation factors measured at 120 °C of Ni-CMS
membranes prepared by the conventional sandwiching method using pyrolysis of metal
solution instead of drying (S-B2) (permeances in m
3
/m
2
.bar.h)

4.4. Conclusions
In this chapter we have studied the incorporation of metal nanoparticles within the
membrane nanostructure to manipulate the surface affinity. Nickel and palladium
nanoparticles were chosen as metal precursors and introduced in the CMS structure. This
is done in order to modify the surface affinity of the carbon surface towards H
2
, and to
enhance its interaction with H
2
, and to therefore improve its ability to separate H
2
from
gaseous mixtures. During our investigation, metal-containing CMS membranes were
prepared using three different techniques. They include (1) impregnation; (2) direct
incorporation; and (3) the so-called ―sandwiching‖ method. Most of the metal-containing
CMS membranes that were prepared using the above techniques show molecular sieving
characteristics.
Using the impregnation method, we have found that the ideal permselectivity for
the H
2
and CO
2
molecular pair is generally higher for the metal-impregnated membranes,
particularly for the membranes with the higher metal loadings. On the other hand, both
the H
2
and CO
2
permeances of those membranes generally decrease upon the deposition
of the metal. The permeance of CO
2
decreases faster than the permeance of H
2
, especially

118
for the membrane with the higher metal loading. In addition, the permeance of CH
4

generally increases after the metal is loaded on the membranes via the impregnation
technique, which indicates that such technique is not appropriate for preparing selective
membranes for CO
2
/CH
4
gas pair.
Both the direct incorporation and sandwiching methods have the potential to
enhance the permeability of H
2
towards other gases; however, the formation of cracks
during the carbonization and decomposition process is the critical issue that needs to be
resolved, in order to improve the separation performance of H
2
from gaseous mixtures.
Using the direct-incorporation method, we have found that the ideal separation factors for
the H
2
/CO
2
and H
2
/CH
4
molecular pairs are higher for the Ni-CMS membranes with
0.1% Ni concentration in 2% PEI solution, compared to those of the base-CMS
membranes. However, higher Ni content may have a detrimental effect on membrane
characteristics, as seen from the decrease in separation performance. This is potentially
due to the poor adhesion between the Ni nano-particles and CMS membranes. Using Pd
as metal precursors Pd-CMS membranes were prepared which exhibit molecular sieving
characteristics and high gas permeances; the downside of the use of this additive is its
high cost, and the fact that the H
2
separation properties do not seem to improve as
significantly as with the Ni-CMS membranes.
Overall, the direct incorporation technique seems to be the most promising
method, since the hydrogen permeances of the metal-loaded CMS membranes prepared
by this technique significantly exceed those prepared by other techniques, namely the
impregnation and the sandwiching methods.

119
Chapter V
Economic Analysis of Membrane Applications in Post-Combustion CO
2

Capture

5.1. Introduction
This chapter focuses on a preliminary evaluation on the use of these membranes
for post-combustion CO
2
capture applications. These membranes can potentially apply to
the treatment of a number of refinery streams, and for IGCC power generation uses.
5.2. Model Development
A simple mathematical model was developed to evaluate the economic and
commercial viability of the membranes. Since membrane transport and separation is
driven by the pressure gradient between the feedside and permeate sides of the
membrane, the following equations can be written to describe membrane operation:

where

120

U
i
= permeance of component i (mol/m
2
.bar.hr)
F
i

F
= molar flow rate of component i in feedside (mol/h)
F
i

P
= molar flow rate of component i in permeate side (mol/h)
P
i
F
= partial pressure of component i in feedside (bar)
P
i
P
= partial pressure of component i in permeate side (bar)
A = membrane area (m
2
)
P
F
= total pressure in feedside (bar)
P
P
= total pressure in permeate side (bar)
Using these design equations, we have completed a preliminary evaluation on the
use of these membranes for post-combustion CO
2
capture applications. We have made an
assumption concerning the CO
2
purity of the sequestered stream that is necessary for the
proposed application. We then calculate the corresponding membrane area that is needed
for a 500 MW power plant as a function of CO
2
permeance and selectivity of CO
2
to N
2
.
5.3. Results and Discussion
In this study we have completed a preliminary evaluation on the use of these
membranes for a post-combustion CO
2
capture application. We focus as a case study on a
500 MW pulverized coal power plant, for which according to White et al. [2003], the

121
total CO
2
flow rate is 12,000 tons/day. We assume for simplicity that the flue gas that is
used as the feed into our membrane unit consists of 15 vol% CO
2
in N
2
. Pure N
2
is used
as the sweep flow rate in the permeate side at a molar flow rate which is 1% of the N
2

flow rate in the feed side; the membrane area required to obtain a CO
2
-rich stream in the
permeate side which is ready for sequestration with at least 70% purity of CO
2
(at a
condition of maximum CO
2
recovery), and the corresponding amount of CO
2
recovered
is given in the following Table.
P
CO2

(m
3
/m
2
.bar.hr)
SF
CO2/N2
Area
(A)
(1x10
4

m
2
)
CO
2

Recovered
(1x10
6

mol/h)
CO
2

Recovery
(%)
CO
2

Recovered/membrane
area (kg/m
2
.day)
1 50 17.2 6.193 60.13 38.03
2 50 8.92 6.193 60.13 73.33
5 50 3.434 6.188 60.08 190.33
1 100 41.27 7.317 71.04 18.73
1 250 104.48 7.496 72.78 7.58

Table 5.1 CO
2
recovery per membrane area (kg/m
2
.day) as a function of CO
2
permeance
and separation factor of CO
2
/N
2
(permeances in m
3
/m
2
.bar.h)

Based on the results in Table 5.1, increasing the permeance of CO
2
significantly
reduced the required membrane area (in fact, permeance and membrane area are
inversely proportional to each other, as expected). However, there is an upper limit in the
fraction of CO
2
in the flue gas for a fixed product purity (in this case 70% CO
2
) that can
be recovered. The only way to increase the recovery is by using a membrane which is
more permselective (one that has a higher SF
CO2/N2
) but the additional gains in recovery
come at a very steep cost in terms of the required membrane surface areas. In terms of the
capital costs for the operation, those generally relate to the surface area of the membrane,
the larger the membrane area the higher the cost. It is clear, therefore, that in the further

122
development of these systems the primary focus should be on improving the permeance
of these membranes, which can potentially improve the efficiency and reduce the cost of
separating CO
2
from flue gas.
There are currently 996 power plants operating in California, utilizing different
types of fuels, such as coal, geothermal, hydroelectric, oil/gas, solar, wind, and waste.
According to the Energy Information Administration, these plants generated 62,780,197
metric tons of CO
2
in 2007. We have calculated the total membrane area that is required
in order to recover CO
2
generated from these power plants with 70% purity at maximum
possible % recovery, and the values are shown in Table 5.2. These numbers indicate the
enormity of the challenge that lies ahead in terms of addressing the CO
2
issue, but also
the great opportunity for the membrane industry.
P
CO2
(m
3
/m
2
.bar.hr) SF
CO2/N2
CO
2
Recovered (kg/day) Area (m
2
)
1 50 1.03E+08 2.72E+06
2 50 1.03E+08 1.41E+06
5 50 1.03E+08 5.43E+05
1 100 1.22E+08 6.52E+06
1 250 1.25E+08 1.65E+07

Table 5.2 Amount of CO
2
recovered (kg/day) and its corresponding membrane area (m
2
)
as a function of CO
2
permeance and separation factor of CO
2
/N
2
(permeances in
m
3
/m
2
.bar.h)

5.4. Conclusions
In order to evaluate the economics of using the membranes prepared, we focused
our attention on a post-combustion CO
2
capture application. A mathematical model was
developed to estimate the CO
2
recovery and corresponding required membrane area as a
function of membrane permeance and selectivity. We have found that enhancing

123
membrane permeability (throughput) is more important for the CO
2
capture process than
improving selectivity of CO
2
towards other components in the flue-gas mixture.

124
Bibliography
Acharya, M.; Raich, B.A.; Harold, M.P.; Lerou, J.J.; Foley, H.C., Metal-Supported
Carbogenic Molecular Sieve Membranes: Synthesis and Applications, Ind. Eng. Chem.
Res. 1997, 36, 2924-2930.
Bansal R.C.; Donnet, J.B.; Stoeckli, H.F., Active Carbon, Marcel Dekker, New York,
1988, 14.
Barsema, J.N.; Balster, J.; Jordan, V.; van der Vegt, N.F.A.; Wessling, M.,
Functionalized Carbon Molecular Sieve Membranes Containing Ag-nanoclusters, J.
Membr. Sci. 2003, 219(1-2), 47-57.
Barsema, J.N.; Balster, J.; van der Vegt, N.F.A.; Koops, G.H.; Jordan, V.; Wessling, M.,
Ag Functionalized Carbon Molecular Sieves Membranes for Separating O
2
and N
2
,
Mater. Res. Soc. Symp. Proc. 2003, 752.
Barsema, J.N.; van der Vegt, N.F.A.; Koops, G.H.; Wessling, M., Ag-Functionalized
Carbon Molecular-Sieve Membranes Based on Polyelectrolyte/Polyimide Blend
Precursors, Adv. Funct. Mater. 2005, 15, 1, 69-75.
Barton, S.S.; Koresh, J.E., Extended Abstracts and Program, 19
th
Bienn. Conf. on Carbon
1989, 6-7
Becker, A.; Hüttinger, K.J., Chemistry and Kinetics of Chemical Vapor Deposition of
Pyrocarbon — III Pyrocarbon Deposition from Propylene and Benzene in the Low
Temperature Regime, Carbon 1998, 36(3), 201-211.
Bhave, R.R., Inorganic Membranes: Synthesis, Characteristics and Applications, Van
Nostrand Reinhold, New York, 1991.
Cabrera, A.L.; Zehner, J.E.; Coe, C.G.; Gaffney, T.R.; Farris, T.S., Preparation of Carbon
Molecular Sieves I. Two-step Hydrocarbon Deposition with a Single Hydrocarbon,
Carbon 1993, 31(6), 969-976.
Centeno, T.A.; Fuertes, A.B., Supported Carbon Molecular Sieve Membranes Based on a
Phenolic Resin, J. Membr. Sci. 1999, 160, 201-211.
Centeno, T.A.; Fuertes, A.B., Preparation of Carbon Molecular Sieve Composite
Membranes, Proc. Int. Conf. Intell. Mater. 6
th
1998, 90.
Cheredkova, K.I.; Golovina, G.S.; Tolstykh, T., Yu. Khim. Tverd. Topl., (Moscow) 1989,
1, 90-93.
Choudhary, V.R.; Chaudhari, S.T.; Rajput, A.M., Oxidative Pyrolysis of Methane to
Higher Hydrocarbons—Effects of Water in Feed, AICHE J. 1991, 37, 915-922.

125
Dalmon, J.A., Catalytic Membrane Reactors, in: G. Ertl, H. Knozinger, J. Weitkamp
(Eds.), Handbook of Heterogeneous Catalysis, Wiley-VCH, Weinheim, 1997, 1387-1398.
de la Casa-Lillo, M.A.; Moore, B.C.; Cazorla-Amorós, D.; Linares-Solano, A., Molecular
Sieve Properties Obtained by Cracking of Methane on Activated Carbon Fibers, Carbon
2002, 40, 2489-2494.
de Lange, R.S.A.; Keizer, K.; Burggraaf, A.J., Ageing and Stability of Microporous Sol-
Gel-Modified Ceramic Membranes, Ind. Eng. Chem. Res. 1995, 34(11), 3838-3847.
de Lange, R.S.A.; Kumar, K.N.P.; Hekkink, J.H.A.; Van de Velde, G.M.H.; Keizer, K.;
Burggraaf, A.J., Microporous SiO
2
and SiO
2
/MO
x
(M=Ti, Zr, Al) for Ceramic Membrane
Applications: A Microstructural Study of the Sol-Stage and the Consolidated State, J.
Sol-Gel Sci. Technol. 1994, 2, 489-495.
Dixon, A.G., Innovations in Catalytic Inorganic Membrane Reactors. In Specialist
Periodical Reports: Catalysis, Spivey, J. J. (ed.), Specialist Periodical. Reports: Catalysis,
Vol. 14, Royal Society of Chemistry: London 1999, 14, 40-92.
Drioli, E., Membrane Reactors, Chemical Engineering and Processing: Process
Intensification 2004, 43, 9, 1101-1102.
Drioli, E.; Giorno, L., Catalytic Membrane Reactors, Chemistry & Industry 1996, 1, 19-
22.
Edwards, A.B.; Garner, C.D., In Situ QXAFS Study of the Pyrolytic Decomposition of
Nickel Formate Dihydrate, J. Phys. Chem. B 1997, 101(1), 20-26.
Faust, S.D.; Aly, O.M., Chemistry of Water Treatment, London: Butterworth Publisheres
1983.
Freitas, M.M.A.; Figueiredo, J.L., Preparation of Carbon Molecular Sieves for Gas
Separations by Modification of the Pore Size of Activated Carbons, Fuel 2001, 80(1), 1-6.
Fuertes, A.B.; Centeno, T.A., Preparation of Supported Asymmetric Carbon Molecular
Sieve Membranes, J. Membr. Sci. 1998, 144(1-2), 105-111.
Funke, H.H.; Argo, A.M.; Falconer, J.L.; Noble, R.D., Separation of Cyclic, Branched,
and Linear Hydrocarbon Mixtures through Silicalite Membranes, Ind. Eng. Chem. Res.
1997, 36(1), 137-143.
Geiszler, V.C.; Koros, W.J., Effect of Polyimide Pyrolysis Conditions on Carbon
Molecular Sieve Membrane Properties, Ind. Eng. Chem. Res. 1996, 35(9), 2999-3003.

126
Geng, J.; Li, H.; Golovko, V.B.; Shephard, D.S.; Jefferson, D.A.; Johnson, B.F.G.;
Hofmann, S.; Kleinsorge, B.; Robertson, J.; Ducati, C., Nickel Formate Route to the
Growth of Carbon Nanotubes, J. Phy. Chem. B 2004, 108(48), 18446-18450.
Gonzalez, J.F.; Roman, S.; Gonzalez-Garcia, C.M.; Valente Nabais, J.M.; Ortiz, A.L.,
Porosity Development in Activated Carbons Prepared from Walnut Shells by Carbon
Dioxide or Steam Activation, Ind. Eng. Chem. Res. 2009, 48(16), 7471-7481.
Grainger, D.; Hägg, M.B., Evaluation of Cellulose-Derived Carbon Molecular Sieve
Membranes for Hydrogen Separation from Light Hydrocarbons, J. Membr. Sci. 2007,
306(1-2), 307-317.
Gueret, C.; Daroux, M.; Billaud, F., Methane Pyrolysis: Thermodynamics, Chem. Eng.
Sci. 1997, 52(5), 815-827.
Hayashi, J.; Mizuta, H.; Yamamoto, M.; Kusakabe, K.; Morooka, S., Separation of
Ethane/Ethylene and Propane/Propylene Systems with a Carbonized BPDA-PP’ODA
Polyimide Membrane, Ind. Eng. Chem. Res. 1996, 35(11), 4176-4181.
Hayashi, J.; Mizuta, H.; Yamamoto, M.; Kusakabe, K.; Morooka, S., Pore Size Control of
Carbonized BPDA-PP’ODA Polyimide Membrane by Chemical Vapor Deposition of
Carbon, J. Membr. Sci. 1997, 124(2), 243-251.
Hayashi, J.; Yamamoto, M.; Kusakabe, K.; Morooka, S., Simultaneous Improvement of
Permeance and Permselectivity of 3,3’,4,4’-Biphenyl Tetracarboxylic Dianhydride-4,4’-
oxydianiline Polyimide Membrane by Carbonization, Ind. Eng. Chem. Res. 1995, 34,
4364-4370.
Hidaka, Y.; Sato, K.; Henmi, Y.; Tanaka, H.; Inami, K., Shock-tube and Modeling Study
of Methane Pyrolysis and Oxidation, Combustion and Flame 1999, 118(3), 340-358.
Hsieh, H.P., Inorganic Membranes, Membra. Mater. Proc. 1990, 84, 1.
Inagaki, M.; Tascon, M.D., Pore Formation and Control in Carbon Materials, Activated
Carbon Surfaces in Environmental Remediation, Elsevier Ltd. 2006.
Ismail, A.F.; David, L.I.B., A Review on the Latest Development of Carbon Membranes
for Gas Separation, J. Membr. Sci. 2001, 193(1), 1-18.
Ismail, A.F.; Ridzuan, N.; Rahman, S.A., Latest development on the membrane formation
for gas separation, Songklanakarin J. Sci. Technol. 2002, 24, 1025-1043.
Jaseińko-Hałat, M.; Kędzior, K., Comparison of Molecular Sieve Properties in
Microporous Chars from Low-rank Bituminous Coal Activated by Steam and Carbon
Dioxide, Carbon 2005, 43(5), 944-953.

127
Jones, C.W.; Koros, W.J., Carbon Composite Membranes. A Solution to Adverse
Humidity Effects, Ind. Eng. Chem. Res. 1995, 34, 164-167.
Jones, C.W.; Koros, W.J., Carbon Molecular Sieve Gas Separation Membranes-1:
Preparation and Characterization Based on Polyimide Precursors, Carbon 1994, 32(8),
1419-1425.
Jones, C.W.; Koros, W.J., Carbon Molecular Sieve Gas Separation Membranes-II:
Regeneration Following Organic Exposure, Carbon 1994, 32(8), 1427-1432.
Jones, C.W.; Koros, W.J., Characterization of Ultramicroporous Carbon Membranes with
Humidified Feeds, Ind. Eng. Chem. Res. 1995, 34, 158-163.
Karger, J.; Ruthven, D.M., Diffusion in Zeolites and other Microporous Solids, John
Wiley, New York 1992.
Kawabuchi, Y.; Chiaki, S.; Kishino, M.; Kawano, S.; Whitehurst, D.D.; Mochida, I.,
Chemical Vapor Deposition of Heterocyclic Comounds over Active Carbon Fiber To
Control Its Porosity and Surface Function, Langmuir 1997, 13(8), 2314-2317.
Kawabuchi, Y.; Kishino, M.; Kawano, S.; Whitehurst, D.D.; Mochida, I., Langmuir
1996, 12, 4281-4285.
Keizer, K.; Uhlhorn, R.J.R.; Van Vuren, R.J.; Burggraaf, A.J., Gas Separation
Mechanisms in Microporous Modified γ-Al
2
O
3
Membranes, J. Membr. Sci. 1988, 39,
285-300.
Kim, S.; Gavalas, G..R., Preparation of H
2
Permselective Silica Membranes by
Alternating Reactant Vapor Deposition, Ind. Eng. Chem. Res. 1995, 34, 168-176.
Kim, T.H.; Koros, W.J.; Husk, G.R.; O’Brien, K.C., Relationship between Gas
Separation Properties and Chemical Structure in a Series of Aromatic Polyimides, J.
Membr. Sci. 1988, 37(1), 45-62.
Kita, H.; Yamada, T.; Tanaka, K.; Okamata, K., Preparation of Carbonized Polypyrrolone
Membranes and Their Gas Separation Properteis, Proc. Int. Conf. Intell. Mater. 6
th
1998,
215.
Koresh, J.; Soffer, A., Study of Molecular Sieve Carbons. Part 1: Pore Structure Gradual
Pore Opening and Mechanism of Molecular Sieving, J. Chem. Soc., Faraday Trans I
1980, 76, 2457-2471.
Koresh, J.; Soffer, A., The Carbon Molecular Sieve Membranes. General Properties and
the Permeability of CH
4
/H
2
Mixture, Sept. Sci. Technol. 1987, 22, 973-982.

128
Koresh, J.E.; Soffer, A., Molecular Sieve Carbon Permselective Membrane Part I.
Presentation of a New Device for Gas Mixture Separation, Sep. Sci. Technol. 1983, 18,
723-734.
Lagorsse, S.; Magalhaes, F.D.; Mendes, A., Carbon Molecular Sieve Membranes:
Sorption, Kinetic and Structural Characterization, J. Membr. Sci. 2004, 241(2), 275-287.
Leenaars, A.F.M.; Keizer, K.; Burggraaf, A.J., The Preparation and Characterization of
Alumina Membranes with Ultrafine Pores. Part 1. Microstructural Investigations on Non-
Supported Membranes, J. Mater. Sci. 1984, 19, 1077-1088.
Legrand, J.C.; Diamy, A.M.; Hrach, R.; Hrachova, V., Mechanisms of Methane
Decomposition in Nitrogen Afterglow Plasma, Vacuum 1999, 52(1-2), 27-32.
Levy, R.A.; Ramos, E.S.; Krasnoperov, L.N.; Datta, A.; Grow, J.M., Microporous
SiO
2
/Vycor Membranes for Gas Separation, J. Mater. Res. 1996, 11, 3164.
Li, Y.Y.; Nomura, T.; Sakoda, A.; Suzuki, M., Fabrication of Carbon Coated Ceramic
Membranes by Pyrolysis of Methane Using a Modified Chemical Vapor Deposition
Apparatus, J. Membr. Sci. 2002, 197(1-2), 23-35.
Lie, J.A.; Hägg, M.B., Carbon Membranes from Cellulose and Metal Loaded Cellulose,
Carbon 2005, 43, 2600-2607.
Lin, C.L.; Flowers, D.L.; Liu, P.K.T., Characterization of Ceramic Membranes II.
Modified Commercial Membranes with Pore Size Under 40Ǻ, J. Membr. Sci. 1994,
92(1), 45-58.
Liu, B.S.; Wang, N.; He, F.; Chu, J.X., Separation Performance of Nanoporous Carbon
Membranes Fabricated by Catalytic Decomposition of CH
4
Using Ni/Polyamideimide
Templates, Ind. Eng. Chem. Res. 2008, 47(6), 1896-1902.
Lizzio, A.A.; Rostam-Abadi, M., Production of Carbon Molecular Sieves from Illinois
Coal, Fuel Process. Technol. 1993, 34(2), 97-122.
Lucas, P.; Marchand, A., Pyrolytic Carbon Deposition from Methane: An Analytical
Approach to the Chemical Process, Carbon 1990, 28(1), 207-219.
Lundy, K.A; Cabasso, I., Analysis and Construction of Multilayer Composite Membranes
for the Separation of Gas Mixtures, Ind. Eng. Chem. Res. 1989, 28, 742-756.
Manso, R.; Pajares, J.A.; Bronick. E.; Jankowska, A.; Kaczmarezyk, J., Extended
Abstracts and Program, 24
th
Bienn. Conf. on Carbon 1999, 686.

129
Macia-Garcia, A.; Bernalte Garcia, M.J.; Diaz-Diez, M.A.; Hernandez Jimenez, A.,
Preparation of Active Carbons from a Commercial Holm-Oak Charcoal: Study of Micro-
and Meso-porosity, Wood Sci. Technol. 2004, 37, 385-394.
Menéndez-Dí az, J.A.; Martí n-Gullón, I., Types of Carbon Adsorbents and Their
Production, Interface Sci. Technol., Activated Carbon Surface in Environmental
Remediation, Elsevier Ltd. 2006, 7, 1-47.
Meriaudean, P.; Thangaraj, A.; Naccache, C., Preparation and Characterization of
Silicalite Molecular Sieve Membranes Over Supported Porous Sintered Glass,
Microporous Mater. 1995, 4(2-3), 213-219.
Molina-Sabio, M.; González, M.T.; Rodriguez-Reinoso, F.; Sepúlveda-Escribano, A.,
Effect of Steam and Carbon Dioxide Activation in the Micropore Size Distribution of
Activated Carbon, Carbon 1996, 34(4), 505-509.
Moore, S.V.; Trimm, D.L., The Preparation of Carbon Molecular sieves by Pore
Blocking, Carbon 1977, 15(3), 177-180.
Morooka, S.; Yan, S.; Kusakabe, K.; Akiyama, Y., Formation of Hydrogen-
Permselective SiO
2
Membrane in Macropores of α-Alumina Support Tube by Thermal
Decomposition of TEOS, J. Membr. Sci. 1995, 101(1-2), 89-98.
Mulder, M., Basic Principles of Membrane Technology, Kluwer Academic Publishers:
Dordrecht, The Netherlands, 1991.
Musuda, T.; Sato, A.; Hara, H.; Kouno, M.; Hushimoto, K., Appl. Catal. A 1994, 111,
143.
Nair, B.N.; Keizer, K.; Elferink, W.J.; Gilde, M.J.; Verweij, H.; Burggraaf, A.J.;
Synthesis, Characterization and Gas Permeation Studies on Microporous Silica and
Alumina-Silica Membranes for Separation of Propane and Propylene, J. Membr. Sci.
1996, 116(2), 161-169.
Naito, M.; Nakahira, K.; Fukuda, Y.; Mori, H.; Tsubaki, J., Process Conditions on the
Preparation of Supported Microporous SiO
2
Membranes by Sol-Gel Modification
Techniques, J. Membr. Sci. 1997, 129(2), 263-269.
Noble, R.D., A Perspective on Reactive Membranes, Abstracts of Papers of the American
Chemical Society 2002, 223, U647.
Noble, R.D., Perspectives on Membrane Technology for Advanced Gas Separations,
Abstracts of Papers of the American Chemical Society 2003, 225, U875.

130
Pastor-Villegas, J.; Durán-Valle, C.J., Pore Structure of Activated Carbons Prepared by
Carbon Dioxide and Steam Activation at Different Temperatures from Extracted
Rockrose, Carbon 2002, 40(3), 397-402.
Petersen, J.; Matsuda, M.; Haraya, K., Capillary Carbon Molecular Sieve Membranes
Derived from Kapton for High Temperature Gas Separation, J. Membr. Sci. 1997, 131(1-
2), 85-94.
Rao, M.B.; Sircar, S.; Anand, M., Novel Nanoporous Carbon Membrane for Gas
Separation, Paper presented at the 3
rd
International Conference on Intelligent Materials,
Worcester, MA, 1994.
Rezac, M.E.; Koros, W.J., Preparation of Polymer-Ceramic Composite Membranes with
Thin Defect-Free Polymer Layers, J. Appl. Polym. Sci. 1992, 46, 1927-1938.
Rios, G.M.; Sanchez, J.; Belleville, M.P.; Sarrade, S., An Overview on Microporous
Membrane Behavior in Various Fluid State Environments: Applications to Separation
and Reaction, Chemical Engineering Reviews 2002, 18, 1, 49.
Rodrí guez-Reinoso, F; Molina-Sabio, M.; González, M.T., The Use of Steam and CO
2
as
Activating Agents in the Preparation of Activated Carbons, Carbon 1995, 33(1), 15-23.
Scott, K.; Hughes, R.; Industrial Membrane Separation Technology, Blackie Academic
and Professional, Glasgow 1996.
Sedigh, M.; Jahangiri, M.; Lin, P.K.T.; Sahimi, M.; Tsotsis, T.T., Structural
Characterization of Polyetherimide-based Carbon Molecular Sieve Membranes, AICHE J.
2000, 46.
Sedigh, M.G.; Onstot, W.J.; Xu, L.; Peng, W.L.; Tsotsis, T.T.; Sahimi, M., Experiments
and Simulation of Transport and Separation of Gas Mixtures in Carbon Molecular Sieve
Membranes, J. Phys. Chem. A 1998, 102(44), 8580-8589.
Sedigh, M.G.; Xu, L.; Tsotsis, T.T.; Sahimi, M., Transport and Morphological
Characteristics of Polyetherimide-Based Carbon Molecular Sieve Membranes, Ind. Eng.
Chem. Res. 1999, 38(9), 3367-3380.
Shu, J.; Grandjean, B.P.A.; Van Neste, A.; Kaliaguine, S., Catalytic Palladium-Based
Membrane Reactors: A Review, Canadian J. Chem. Eng. 1991, 69(5), 1036-1060.
Shusen, W.; Meiyun, Z.; Zhizhong, W., Asymmetric Molecular Sieve Carbon
Membranes, J. Membr. Sci. 1996, 109(2), 267-270.
Singh, A.; Koros, W.J., Significance of Entropic Selectivity for Advanced Gas Separation
Membranes, Ind. Eng. Chem. Res. 1996, 35(4), 1231-1234.

131
Singh, A.; Lal, D., Preparation and Characterization of Activated Carbon Spheres from
Polystyrene Sulphonate Beads by Steam and Carbon Dioxide Activation, J. App. Polym.
Sci. 2010, 115, 2409-2415.
Singh-Ghosal, A.; Koros, W.J., Air Separation Properties of Flat Sheet Homogeneous
Pyrolytic Carbon Membranes, J. Membr. Sci. 2002, 174(2), 177-188.
Smaihi, M.; Jermoumi, T.; Marignan, J.; Noble, R.D., Organic-Inorganic Gas Separation
Membranes: Preparation and Characterization, J. Membr. Sci. 1996, 116(2), 211-220.
Soffer, A.; Azariah, M.; Amar, A.; Cohen, H.; Golub, D.; Saguee, S; Tobias, H., Method
of Improving the Selectivity of Carbon Membranes by Chemical Vapor Deposition, US
patent 5695818, 1997.
Steel, K., Carbon Membranes for Challenging Gas Separations, University of Texas at
Austin, Ph.D. Dissertation, 2000.
Tanaka, K.; Kita, H.; Okano, M.; Okamoto, K., Permeability and Permselectivity of
Gases in Fluorinated and Non-Fluorinated Polyimides, Polymer 1992, 33(3), 585-592.
Tanaka, K.; Okano, M.; Toshino, H.; Kita, H.; Okamoto, K., Effect of Methyl
Substituents on Permeability and Permselectivity of Gases in Polyimides Prepared from
Methyl-Substituted Phenylenediamines, J. Polym. Sci. Part B: Polym. Phys. 1992, 30(8),
907-914.
Tsapatsis, M.; Gavalas, G.R., A Kinetic Model of Membranes Formation by CVD of
SiO
2
and Al
2
O
3
, AICHE J. 1992, 38, 847-856.
Uhlhorn, R.J.R.; Keizer, K.; Burggraaf, A.J., J. Membr. Sci. 1992, 66, 259.
Verma, S.K.; Walker Jr., P.L.; Preparation of Carbon Molecular Sieves by Propylene
Pyrolysis over Microporous Carbons, Carbon 1992, 30(6), 829-836.
Villar-Rodil, S.; Navarrete, R.; Denoyel, R.; Albiniak, A.; Paredes, J.I.; Martinez-Alonso,
A.; Tascon, J.M.D., Carbon Molecular Sieve Cloths Prepared by Chemical Vapour
Deposition of Methane for Separation of Gas Mixtures, Microporous and Mesoporous
Materials 2005, 77(2-3), 109-118.
Vnukov, S.P.; Polyakov, N.S.; Dubinin M.M.; Fedosaev, D.V., Izv. Akad. Nauk SSSR,
Ser. Khim. 1986, 2, 267-273.
Vu, D.Q.; Koros, W.J., High Pressure CO
2
/CH
4
Separation Using Carbon Molecular
Sieve Hollow Fiber Membranes, Ind. Eng. Chem. Res. 2002, 41(3), 367-380.
Vyas, S.N.; Patwardhan, S.R.; Gangadhar, B., Synthesis of Carbon Molecular Sieves by
Activation and Coke Deposition, Fuel 1993, 72(4), 551-555.

132
Walker Jr., P.L., Production of Activated Carbons: Use of CO
2
versus H
2
O as Activating
Agent, Carbon 1996, 34(10), 1297-1299.
Webb, P.A.; Orr, C., Analytical Methods in Fine Particle Technology, Micromeritics
Instrument Corporation, Norcross GA, USA 1997.
White, C.M.; Strazisar, B.R.; Granite, E.J.; Hoffman, J.S.; Pennline, H.W., Separation
and Capture of CO
2
from Large Stationary Sources and Sequestration in Geological
Formations – Coalbeds and Deep Saline Aquifers, J. Air and Waste Manage. Assoc. 2003,
53, 645-715.
Wu, J.C.S.; Sabol, H.; Smith, G.W.; Flowers, D.L.; Liu, P.K.T., Characterization of
Hydrogen-Permselective Microporous Ceramic Membranes, J. Membr. Sci. 1994, 96(3),
275-287.
Xia, B.; Lenggoro, I.W.; Okuyama, K., Preparation of Nickel Powders by Spray
Pyrolysis of Nickel Formate, J. American Ceramic Soc. 2001, 84(7), 1425-1432.
Yamauchi, M.; Kitagawa, H., Hydrogen Absorption of the Polymer-coated Pd
Nanoparticles, Synthetic Metals 2005, 153(1-3), 353-356.
Yamazaki, S.; Tsutsumi, K., Synthesis of an A-Type Zeolite Membrane on Silicon Oxide
Film-Silicon, Quartz Plate and Quartz Fiber Filter, Microporous Mater. 1995, 4(2-3),
205-212.
Yoda, S.; Hasegawa, A.; Suda, H.; Uchimaru, Y.; Haraya, K.; Tsuji, T.; Otake, K.,
Preparation of a Platinum and Palladium/Polyimide Nanocomposite Film as a Precursor
of Metal-Doped Carbon Molecular Sieve Membrane via Supercritical Impregnation,
Chem. Mater. 2004, 16(12), 2363-2368.
Zhang, L.; Chen, X.; Zeng, C.; Xu, N., Preparation and Gas Separation of Nano-Sized
Nickel Particle-Filled Carbon Membranes, J. Membr. Sci. 2006, 281(1-2), 429-434.
Zhang, T.; Walawender, W.P.; Fan, L.T., Preparation of Carbon Molecular Sieves by
Carbon Deposition from Methane, Bioresource Technol. 2005, 96(17), 1929-1935.

Abstract (if available)

Abstract Carbon molecular-sieve (CMS) membranes have been studied in the past few years as an alternative to both inorganic and polymeric membranes. They are known to have considerable resistance to high temperatures and pressures for gas separation applications, such as those involving mixtures that contain H2, CO2, and CH4.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Fabrication of nanoporous silicon carbide membranes for gas separation applications

PDF

Preparation of polyetherimide nanoparticles by electrospray drying, and their use in the preparation of mixed-matrix carbon molecular-sieve (CMS) membranes

PDF

The use of carbon molecule sieve and Pd membranes for conventional and reactive applications

PDF

Experimental studies and computer simulation of the preparation of nanoporous silicon-carbide membranes by chemical-vapor infiltration/chemical-vapor deposition techniques

PDF

A study of the application of membrane-based reactive separation to the carbon dioxide methanation

PDF

Continuum and pore netwok modeling of preparation of silicon-carbide membranes by chemical-vapor deposition and chemical-vapor infiltration

PDF

On the use of membrane reactors in biomass utilization

PDF

Molecular modeling of silicon carbide nanoporous membranes and transport and adsorption of gaseous mixtures therein

PDF

Adsorption of trace levels of arsenic and selenium from aqueous solutions by conditioned layered double hydroxides

PDF

The roles of surface and pore properties in wetting resistance for membrane distillation membranes

PDF

Transport and separation of gas mixtures through carbon

PDF

A process-based molecular model of nano-porous silicon carbide membranes

PDF

Studies of transport phenomena in hydrotalcite membranes, and their use in direct methanol fuel cells

PDF

Exploring properties of silicon-carbide nanotubes and their composites with polymers

PDF

Fabrication of nanoporous silicon oxycarbide materials via a sacrificial template technique

PDF

Hydrogen storage in carbon and silicon carbide nanotubes

PDF

Fabrication of silicon-based membranes via vapor-phase deposition and pyrolysis of organosilicon polymers

Asset Metadata

Creator Lee, Hui-Chun Jocelyn (author)

Core Title Development of carbon molecular-sieve membranes with tunable properties: modification of the pore size and surface affinity

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Chemical Engineering

Publication Date 09/21/2010

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag carbon deposition,carbon molecular sieve,gas separation,membrane,metal incorporation,methane,OAI-PMH Harvest,steam activation

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Sahimi, Muhammad (committee chair), Tsotsis, Theodore T. (committee chair), Pirbazari, Massoud M. (committee member)

Creator Email huichun@usc.edu,jocelyn621@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m3461

Unique identifier UC1425937

Identifier etd-Lee-4053 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-387873 (legacy record id),usctheses-m3461 (legacy record id)

Legacy Identifier etd-Lee-4053.pdf

Dmrecord 387873

Document Type Dissertation

Rights Lee, Hui-Chun Jocelyn

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email cisadmin@lib.usc.edu