Membranes For Hydrogen Separation 2007 PDF

4078 Chem. Rev.
2007, 107, 4078−4110
Membranes for Hydrogen Separation

Nathan W. Ockwig† and Tina M. Nenoff*,‡
Geochemistry, and Surface and Interface Sciences, Sandia National Laboratories, P.O. Box 5800, M.S. 1415, Albuquerque, New Mexico 87185
Received November 27, 2006
Contents 6.3. H2 versus CO2 Selective Polymeric 4103

Membranes
1. Introduction 4078 6.4. Ionic and Ion Exchange Polymer Membranes 4103
2. Metallic Membranes 4081 7. Conclusion 4104
2.1. Pure Metals 4083 8. Comparisons and Perspectives 4105
2.2. Alloys 4083 9. Acknowledgments 4105
2.3. Amorphous Metals 4084 10. References 4105
2.4. Membrane Fabrication and Processing 4086
2.5. Modeling/Simulation and Characterization 4086
2.6. Membrane Fabrication 4087 1. Introduction
2.7. Catalytic Surface Coatings 4087 The increasing demand for “clean” and efficient energy
3. Silica Membranes 4088 has resulted in an increased global willingness to embrace
3.1. Membrane Layer Synthesis 4089 the proposed “hydrogen economy” as a potential long term
3.1.1. Sol−Gel Processing of a Membrane 4089 solution to the growing energy crisis. With global energy
Layer consumption predicted to nearly double by 2050 and our
3.1.2. Chemical Vapor Deposition (CVD) of a 4089 present fossil fuel reserves under increasingly urgent envi-
Membrane Layer ronmental, political, and economic pressures, we must
3.2. Preparation 4090 unambiguously overcome the many scientific and techno-
3.3. Intermediate Layers 4090 logical hurdles that exist between the present state of
3.4. Support 4090 hydrogen production, utilization, and storage capabilities and
3.5. Modification 4091 those required for a competitive sustainable hydrogen
3.5.1. Silica Membrane Modification 4091 economy.1 Although many multifaceted technological bar-
3.5.2. Membrane Structure Modification 4092 riers exist, before we can completely realize the full potential
3.6. Operational Stability 4092 of a hydrogen economy two economic barriers, namely the
cost of fuel cells and the cost of hydrogen production, must
4. Zeolite Membranes 4092
be reduced by factors of 10 and 4, respectively.2 In an
4.1. Membrane Growth Methods 4093 extensive effort to address these goals, the $1.2B Hydrogen
4.2. Permeation and Gas Transport 4093 Fuel Initiative was announced in January of 2003 as a
4.3. Defect Site Diffusion/Nonzeolitic Pores 4094 presidential directive. Since then tremendous cooperative
4.4. Thin Films 4094 efforts have been brought to bear on the safe economic
4.5. Zeolite Membrane Modification 4094 production and storage H2.
4.6. CO2 Sequestration in H2 Separations 4095 Nearly 2% or ∼6 Exajoules (1 EJ ) 1018 joules) of the
4.7. Manufacturing 4095 world’s primary energy is stored in the 41 MM tons of H2
5. Carbon-Based Membranes 4096 which is produced industrially on a yearly basis. Over 90%
5.1. Carbon Membrane Preparations 4097 of this 0.85 trillion m3/year is generated from fossil fuel
5.2. Carbon Membrane Post-treatment 4097 sources (mainly steam reforming of natural gas) while the
remaining fraction (∼8%) is produced through electrolysis
5.3. Carbon Membrane Module Construction 4097
of water. Much of this H2 is used for large-scale processes
5.4. Selective Surface Flow Membranes 4097 in the metallurgical, chemical, petrochemical, pharmaceutical,
5.5. Disadvantages of Carbon Membranes 4098 and textile industries to manufacture a diverse range of
5.6. Molecular Sieving Carbon Membranes 4098 products from semiconductors and steel alloys to vitamins
5.7. Carbon Nanotubes 4099 and raw chemical materials such as ammonia, methanol, and
6. Polymer Membranes for H2 Separations 4100 hydrogen peroxide.3 However, large-scale production of H2
6.1. Dense Polymeric Membranes 4100 for these industries often requires an almost prohibitively
6.2. Hydrogen Selective Polymeric Membranes 4101 large capital investment for the separation and purification
processes which significantly drives up the cost of H2.
Regardless of which method is used to produce H2, the need
* To whom correspondence should be addressed. Telephone: 505-844-0340. will always exist for a cost effective and efficient means to
Fax: 505-844-5470. E-mail: tmnenof@sandia.gov; Internet: www. separate it from other less desirable species. Currently, H2
sandia.gov/nenoff.
† Geochemistry. can be purified through one (or a combination) of three major
‡ Surface and Interface Sciences. processes: (1) pressure swing adsorption (PSA),4,5 (2)
10.1021/cr0501792 CCC: $65.00 © 2007 American Chemical Society
Published on Web 10/10/2007
Membranes for Hydrogen Separation Chemical Reviews, 2007, Vol. 107, No. 10 4079
Many H2 membrane separation technologies are based on

the most widely used method of hydrogen production, that
is, the steam reforming of light hydrocarbons, mainly
methane.9 This process, called steam-methane reforming
(SMR), consists of two basic steps. In the initial reforming
step, methane (CH4) and excess steam (H2O) react to form
carbon monoxide (CO) and hydrogen (H2) at ∼820 °C
(reaction 1). Additional H2 is obtained by the subsequent
Initial reforming reaction:

CH4 + H2O f CO + 3H2 (reaction 1)
Water-gas shift (WGS) reaction:

CO + H2O f CO2 + H2 (reaction 2)
Tina M. Nenoff obtained her B.A. degree in Chemistry in 1987 from the
University of Pennsylvania. She obtained her Ph.D. in 1993 at the
University of California, Santa Barbara, in the Chemistry Department under Steam-methane reforming (SMR) reaction:
the guidance of Dr. Galen D. Stucky. She then joined the staff at Sandia
National Laboratories. Her research has been directed toward the synthesis
CH4 + 2H2O f CO2 + 4H2 (reaction 3)
and application of novel condensed and microporous oxide phases for
catalysis and separations. She is focused on defect-free inorganic thin- reaction of CO with H2O in the water-gas shift (WGS)
film membranes for H2 purification and on hydrocarbon feedstock reaction (reaction 2). For each mole of CH4 consumed, the
separations. Another area of focus is on the synthesis and characterization overall SMR process (reaction 3) theoretically yields 4 mol
of oxide ion exchangers for the removal and fixation of radioisotopes from of H2 and 1 mol of CO2, although in practice this is seldom
caustic solutions containing many competing ions. Dr. Nenoff has published
over 100 papers in various material science and chemistry journals and achieved. The H2 product composition prior to purification
has presented at over 50 national and international conferences. She is depends on the exact nature of the shift process used.
a member of both the MRS and ACS societies. Typically, in a high-temperature shift reactor operating at
350 °C, a product stream composition is 73.9% H2, 17.7%
CO2, 6.9% CH4, and 1.0% CO.10 However, a second shift
process involving a lower temperature (190-210 °C) shift
reaction is often used with a resulting product composition
of 74.1% H2, 18.5% CO2, 6.9% CH4, and 0.1% CO.10
Regardless of the method, H2 purification ultimately equates
to a CO2 removal process.
Within the arena of gaseous H2 separations, membrane
compositions span the entire periodic table and range from
metallic alloys and organic polymers to inorganic oxides and
composites (i.e., cermets, metal-organic frameworks, and
composites).11 The diversity of structures synthesized for gas
separation applications cannot possibly be encompassed in
a single review. However, some generalized principles can
be extracted if we categorize them in a deliberate way to
highlight their compositions and distinct performance char-
Nathan W. Ockwig was born in Granite Falls, Minnesota, and obtained
his B.S. in Chemistry (1996) from St. Cloud State University in St. Cloud, acteristics. This article is intended to provide a critical and
Minnesota. He earned both an M.S. (2002) and Ph.D. (2005) in the comprehensive review of the diverse membrane materials
Department of Chemistry at the University of Michigan while working on which are under investigation for H2 separation and purifica-
the synthesis, classification, and characterization of new Metal-Organic tion technologies. We have chosen to put limits on the scope
Framework (MOF) materials under the guidance of Professor Omar M. of materials presented herein. In particular, we have chosen
Yaghi. Currently, Dr. Ockwig is a postdoctoral appointee in the Geochem- to report on recent research into membrane categories that
istry Department of Sandia National Laboratories under the technical
guidance of Randall T. Cygan and Tina M. Nenoff. His research at Sandia encompass broad and thematic trends of structure/property
is focused on the dynamic behavior of small molecules and solvated ions relationships between membrane class and H2 separation
in microporous zeolite and MOF materials using a variety of molecular ability. Older technology, individual phases, and less devel-
modeling, structural, and spectroscopic characterization techniques. oped categories of membranes, such as mixed ionic-
electronic proton conductors for H2 separation, remain vital
fractional/cryogenic distillation, or (3) membrane separa- to the research field and are covered in detail elsewhere.12,13
tion.6,7 While PSA and fractional/cryogenic distillation Classification by composition is perhaps the simplest way
systems are in commercial operation, they are generally not of categorizing membrane materials, and they are delineated
cost effective and are quite energetically demanding for the as follows: metallic (pure metals or alloys), inorganics
separation and purification of H2. In addition, neither of these (including oxides, zeolites, glasses, and ceramics), porous
methods provides sufficient purity for the targeted applica- carbons, purely organic polymers, and hybrids or composites.
tions in the hydrogen economy. The third method, membrane Beyond composition, the properties (mechanical, thermal,
separation, is currently considered to be the most promising and chemical stabilities) and performance characteristics
because of low energy consumption, possibility for continu- (processability, maximum H2 flux, permeability, selectivity,
ous operation, dramatically lower investment cost, its ease transport mechanism, lifetime) of a given membrane material
of operation, and ultimately cost effectiveness.8 are the most critical issues for any given application. The
4080 Chemical Reviews, 2007, Vol. 107, No. 10 Ockwig and Nenoff
Table 1. Current Status and Future H2 Membrane Property

Targets1,2,14
property 2003 2007 2010 2015
cost (USD/ft2) 178 150 100 <100
operating T (°C) 300-600 400-700 300-600 250-500
operating ∆P (MPa) 0.69 1.38 e2.75 2.75-6.90
H2 recovery 60 70 80 90
(% gas processed)
H2 purity (% of dry gas) >99.9 >99.9 >99.95 99.99
durability (years) <1 1 3 >5
combined result of these composition and performance issues

ultimately determines the cost and viability of a given
material for application in commercial H2 separation tech-
nologies. The five performance targets for H2 separation set
forth by the U.S. Department of Energy reflect the present
capabilities and highlight the distinct research and develop- Figure 1. Illustration of five H2 separation mechanisms: (i)
ment opportunities which are necessary components to fully Knudson diffusion; (ii) surface diffusion; (iii) capillary condensa-
realize the hydrogen economy.14 The specific targets are as tion; (iv) molecular sieving; (v) solution diffusion.
follows: (1) higher H2 flux rates; (2) lower material costs;
(3) improved durability; (4) lower parasitic power require- (or more) of five separation mechanisms (Figure 1):18,19 (i)
ments; and (5) lower membrane production/fabrication costs Knudson diffusion,20 (ii) surface diffusion, (iii) capillary
(see Table 1 for target values).1,2,14 condensation, (iv) molecular sieving, and (v) solution dif-
Although efficient and cost effective fuel cells utilizing fusion.21,22 Ultimately, the contribution of these mechanisms
H2 have taken the center stage of global energy interests, in a specific material culminates in its overall performance
research surrounding both the production and storage of H2 and efficiency characteristics.
is gaining international attention. This is because the
purification and separation stages of nearly all large-scale The most commonly reported and compared performance
manufacturing processes are often the most technological characteristics of gas separation membranes are permeance
challenging and economically limiting factors. Hydrogen has (or flux) and selectivity: the flux, J, is the amount (mass or
been reported as “A Clean and Secure Energy Future”15 due moles) of gas which permeates through the membrane (i.e.,
to its natural abundance and the nonpolluting nature of its flow or flux) per unit time and unit surface area; the
combustion products (H2O). However, significantly less permeability coefficient, F, is the quantitative expression of
attention has been drawn to the fact that other forms of a specific measure of gas moving through a membrane; and
energy (nuclear, fossil, solar, etc.)16 must be consumed to the selectivity, R, is the separating ability of a given
manufacture and purify H2 for various fuel and energy membrane.23
applications. Virtually all naturally occurring hydrogen is a
substituent of a more complex molecule (i.e., H2O or CH4), Diffusion through dense membranes is driven by an
and as such, a specific amount of energy is required to underlying chemical potential or concentration gradient
liberate the hydrogen from these compounds, plus the energy across the membrane and is well described by Ficks’ first
necessary for its purification, compression, and/or liquefac- law (eq 1):24
tion. In addition, many of these processes are known to
produce undesirable greenhouse gases as a byproduct and
therefore must be combined with (carbon) sequestration ∂C ∂C ∂C
JH2 ) -DH2∇C(x,y,z) ∇C(x,y,z) ) ı̂ + ĵ + k̂ (1)
technologies to significantly reduce the level of emissions. ∂x ∂y ∂z
This complex set of criteria culminates into a very energeti-
cally demanding and technically challenging obstacle which
must be fully addressed before the hydrogen economy can where DH2 is the diffusion coefficient and the differential
be embraced at local, national, and global levels. Despite vector operator, ∇C(x,y,z), is the three-dimensional equilibrium
these challenges and regardless of the advances in H2 concentration in Cartesian coordinates. However, since we
production methods, the need will always exist for cheaper are primarily interested in the steady-state flux across the
and more efficient ways to purify and separate it from other membrane itself, this equation is simplified to a single
gases.17 Currently, the most promising of these separation dimension. Permeability becomes important when the surface
technologies are based on membranes which are capable of concentrations of the gas are not known. In these cases,
operating under a wide variety of conditions while maintain- Henry’s law (SH ) Cgas/Pgas) is used, where SH is a constant
ing their efficiency. However, each class of membranes offers relating the vapor pressure of a nondissociative gas to its
its own unique advantages and disadvantages to H2 separation dilute concentration in a liquid or solid (i.e., the solution
and purification which are primarily governed by their phase). Cgas and Pgas are the concentration and pressure of
inherent chemical, thermal, and mechanical stabilities. the gas, respectively. Since inlet and outlet pressures are
In the broadest sense, a membrane is simply a barrier easily measured, pressure is substituted into Fick’s first law.
which selectively allows certain molecules to permeate across In the case of diatomic molecules such as H2, which
it. In terms of gaseous H2 purification and separation, this dissociate prior to dissolution (i.e., in metals), a modification
means that either H2 molecules or impurities selectively of Henry’s law is needed; this is called Sieverts’ law (SH )
interact with or permeate the membrane. Either of these very Cgas/Pgas1/2). This is then used to convert Fick’s law into a
simplistic H2 separation processes can be attributed to one usable form (eq 2):
-DH2∂CH2 -DH2SH2∂PH1/22 -DH2SH2∆PH1/22

JH2 ) ) = )
∂l ∂l ∆l
-FH2(PH1/22,l - PH1/22,0)
(2)
l
where JH2 is the hydrogen flux, DH2 is the concentration

independent diffusion coefficient (not universally true), SH2
is the Sieverts’ law constant or solubility, l is the membrane
thickness, and FH2 is DH2SH2, the hydrogen permeability, while
PH2,0 and PH2,l are the measured pressures of H2 on the
feedstock and product sides of the membrane, respectively.
If the individual permeabilities (Fi) of a given gas pair are
known, the ratio of these values is defined as the ideal
/
selectivity of the membrane, symbolized as Ri,j . The sepa-
ration factor (Ri,j) is given by the mole fractions of both
components on the feed stock (χfi or χfj ) and product sides
(χpi or χpj ) and is related to the ideal selectivity through the
follow expression (eq 3 and Table 2):
( ) ( ) ∆Pi ∆Pi
Ri,j ) /
(
χfi Pfi
Ri,j f f
χj Pj
-
- )( ) ( )( ) ( )( )( )
χpi Ppi
χpj Ppj
χfj
χfi
)
Fi Pfi
Fj ∆Pj
)
Di Si Pfi
Dj Sj ∆Pj
Pfj Pfj
(3)
The permeation and selectivity values associated with any

and all membranes can be related to each other for direct
comparison on performance. However, much of the similarity
ends there. The fundamental science of each membrane type
is unique and has its own set of questions to address in order
to make membranes specifically selective for H2 or related
gases found in the production processes. Furthermore, the
ability to take the laboratory bench-scale research to produc-
tion scale with defect-free, highly selective membranes for Figure 2. Seven-step diffusion mechanism in dense metal
large-scale applications is an involved and detailed endeavor. phases: (1) movement of the raw gas (mixture of H2 and undesired)
The ability to take concept to commercialization is the route to the feed stream surface of the membrane; (2) dissociation of
necessary for success in any new membrane technology. chemisorbed H2 into hydrogen ions (H+) and electrons (e-1); (3)
adsorption of H+ ions into the membrane bulk; (4) diffusion of the
H+ ions and electrons through the membrane; (5) desorption of
2. Metallic Membranes H+ ions from the membrane bulk to the product stream surface of
the membrane; (6) reassociation of the H+ ions and the electrons
This section of our review focuses specifically on metallic into discrete molecules of H2; and, finally, (7) diffusion of the H2
membranes for the separation of H2. Metallic membranes from the product surface of the membrane.
are typically dense sheets or films which H2 permeates
through as its component protons and electrons. The funda- Table 2. Engineering Strategies for Hydrogen Separation
mental mechanism of action in these dense metallic mem- Membranes
branes requires the conduction of free electrons and the selectivity general H2 selective H2 rejective
presence of specific catalytic surfaces to dissociate H2 on DH2 >>>1 >>>>1 >>1
the raw feed stream side and reassociate the protons and Dgas
electrons on the product side (Figure 2). Hydrogen selectivity SH2 <<1 <1 <<<1
is typically very high in these systems, since the dense Sgas
structure prevents the passage of large atoms and molecules
PH2 >1 >>>1 <1
such as CO, CO2, O2, N2, etc.). This high selectivity translates
Pgas
to very high purity H2 and the increased thermal stabilities
allow higher operating temperatures. These are the primary
advantages that metallic membranes offer over other materi-
als. The metals which are most suitable for H2 separation cheap. Historically, H2 separations were performed with Pd-
membranes typically have high H2 permeabilities,25 high based membranes, since they naturally catalyze the surface
diffusivities or solubilities,26 and good thermal stability at dissociation/reassociation processes and are highly permeable
elevated temperatures.27 These include but are not limited to H2. There is extensive information in the literature
exclusively to tantalum, niobium, and vanadium, and unlike regarding many years of research into Pd membranes. For a
platinum and palladium, they are abundant and comparatively review of recent advances in these membranes, see refs 28-
30. In particular, Ma’s research into Pd-based membranes The fundamental concepts, mechanisms, and equations
has been at the forefront for years.31-34 From an economic governing the performance of dense metallic membranes
perspective, Pd-based membranes are generally considered have been the subject of numerous reviews.25,61-65 In a dense
prohibitively expensive for finding global integration for metallic membrane, H2 permeates through the solid material
hydrogen production through large-scale industrial pro- via the solution diffusion mechanism outlined in the first
cesses,30,35 though a recent economic study has shown section (type v).18 The solution diffusion mechanism involves
economic competitiveness for steam reforming using Pd- a total of seven steps, which are illustrated in Figure 2: (1)
based membranes versus a conventional plant.36 Pd- and Pt- movement of the raw gas (mixture of H2 and undesired) to
based membranes are plagued by a very high sensitivity to the feed stream surface of the membrane; (2) dissociation
surface contamination from a wide variety of impurities (H2S, of chemisorbed H2 into hydrogen ions (H+) and electrons
CO, thiophene, chlorine, and iodine), which severely reduce (e-1); (3) adsorption of H+ ions into the membrane bulk;
their performance.37 This dramatic reduction in performance (4) diffusion of the H+ ions and electrons through the
is primarily due to the more favorable interaction energies membrane; (5) desorption of H+ ions from the membrane
between the membrane and the contaminant than H2 or bulk to the product stream surface of the membrane; (6)
irreversible chemisorptive reactions. For example, palladium reassociation of the H+ ions and the electrons into discrete
membranes and catalysts have a well documented history molecules of H2; and, finally, (7) diffusion of the H2 from
of poisoning in the presence of sulfur containing species.38-40 the product surface of the membrane. The most commonly
Additionally, while Pd membranes still out-perform many compared performance characteristic of H2 selective mem-
other prospective materials, the Pd-H phase transition at branes is the steady-state flux (J)23,66 of hydrogen atoms
∼300 °C often leads to membrane degradation in the through a membrane. This steady-state flux is simply the
presence of H2 due to a significant difference in their lattice quantity (typically given in moles) of H2 permeating through
constants.41-43 This problem of hydrogen embrittlement can a certain area (cm2) over a given period of time (s) at a
be minimized in Pd membranes by alloying them with Ag, specified temperature and applied pressure differential (∆P),
Cu, or Au or controlling the operating conditions to avoid a and it is typically expressed in units of mol H2/(cm2 s) for
two-phase region.44 A more detailed description of alloys is H2 permeation through metal membranes. Reported J values
given below, and recent advances in Pd-based membranes (H2 flux) through metallic membranes commonly range from
are readily available in a variety of reviews.28,29,30,45,46 10-4 to 10-1 mol H2/(cm2 s) and are closely dependent on
Recently, several efforts have made significant advances in the elemental composition, the underlying chemical structure,
non-Pd metallic membranes, and this review will largely and the fabrication method(s) used to produce the mem-
focus on these systems.47,48 Membranes made from metals brane.37,54,55,64,67,68
with high diffusivity or solubility are more prone to degrada-
Ficks’ first law (eq 1) describes the atomic permeation
tion by hydrogen embrittlement27,49 and are consequently less
flux of hydrogen through a homogeneous metal phase as a
durable. Of course, each H2 separation/production process
function of the concentration gradients and a diffusion
has its own unique performance characteristics and require-
coefficient, DH2 (cm2/s), which is the concentration gradient
ments which must be met by the membrane employed. For
resulting from ∆P across the membrane. Sieverts’ law (eq
example, the mildest thermal requirements (300-500 °C)
2) may be used under certain conditions to describe the
are present in processes based on the water-gas shift (WGS)
relationship between the concentration (CH2) and the square
reaction.50,51 Significantly higher thermal conditions are
root of pressure (PH21/2). This model assumes that the bulk
required for the reformation of natural gas (800-950 °C),52
while the gasification of coal often requires temperatures diffusion of H2 (step 4) occurs very quickly and it does not
exceeding 1000 °C.53 However, these operating conditions adversely affect the overall rate of membrane permeation.
are continually changing with improvements of these pro- It should be noted however that Sieverts’ law is limited to
cesses. For a detailed example, see the reports by Amadeo54 systems where the H2 concentrations are low and the M-H
and Andreeva55 of the low-temperature and -pressure (180- interaction is significantly less than 1.42 Although beyond
230 °C and 101.325 kPa) WGS reactions. One further the scope of this review, there are several modifications to
consideration which should not be overlooked is performance this model which can account for grain boundaries complica-
effects and possible interactions between the WGS catalysts tions,69 contamination of the feed stream, and various surface
and undesired products with the specific metal membrane. phenomena.25,70-73 Corrections to account for other H2
diffusion modes can also be implemented.74,75 In addition,
Following the taxonomy used by Wipf et al.,26 a metallic Ward64 proposed an elaborate model which attempts to
membrane material can be classified as (1) pure (single account for all individual steps in the “solution diffusion”
element), (2) crystalline, or (3) amorphous. This allows for mechanism.
the direct comparison of H2 performance characteristics in
terms of both underlying chemical structure and basic For the purposes of this review, any crystalline, single-
elemental composition.26 The characteristic H2 flux of a element metal is defined as a “pure” metal. The permeability
metallic membrane is measured directly using a standard gas of H2 through these types of membranes is a function of the
permeation cell37,56,57 which places a gas pressure differential underlying lattice structure and various types of lattice defects
(∆P) on the membrane with a variety of gas mixtures and (i.e., vacancies, contaminant atoms, or dislocations) and
under various operating conditions (typically temperature, reactivity toward H2 or other feed stream gases. Body
pressure, and cycling to determine lifetime). The analysis is centered cubic (bcc) forms of Fe, V, Nb, and Ta commonly
typically coupled to a gas chromatograph (GC), a mass exhibit exceptionally high H2 permeabilities.25,26 Face cen-
spectrometer (MS), or both (GC-MS) to determine exact gas tered cubic (fcc) metals such as Ni and Pd also exhibit
compositions, to detect membrane leaks, and, most impor- favorable H2 permeabilities, with Pd possessing significantly
tantly, to quantify the permeability under a given set of higher H2 permeability than Ni.76 Because Ni is far cheaper,
conditions.25,58,59,60 its alloys are being actively investigated in a range of
Table 3. Interaction Properties of H2 for Pure Metals46,107,469,470 Table 4. Select Crystalline Single-Phase Body-Centered Cubic
Binary and Ternary Alloy Compositions of V, Zr, Nb, and Ta
H solubility hydride ∆H H2 permeability
hydride (H/M @ formation @ 500 °C [M1-r-βM′rM′′β]81
packing metal composition 27 °C) (kJ/mol) (mol/ms Pa1/2)a M′ R-max M′′ β-max
fcc Ni Ni2H ∼7.6 × 10-5 -6 7.8 × 10-11 Al 0.35 Fe 0.40
Cu ∼8 × 10-7 4.9 × 10-12 0.40 Ge 0.03
Pd PdH 0.03 +20 1.9 × 10-8 0.40 Cu 0.05
Pt PtH ∼1 × 10-5 +26 2.0 × 10-12 0.40 Zr 0.05
bcc V VH2 0.05 -54 1.9 × 10-7 0.40 Ni 0.08
Fe FeH 3 × 10-8 +14 1.8 × 10-10
0.40 Ga 0.12
Nb NbH2 0.05 -60 1.6 × 10-6
0.40 Mn 0.53
Ta Ta2H 0.20 -78 1.3 × 10-7
0.40 Mo 1
hcp Ti γ-TiH2 R ∼ 0.0014 -126
β ∼ 1.0 0.40 Nb 1
Zr ZrH2 <0.01 -165 0.40 Ta 1
Hf HfH2 R ∼ 0.01 -133 0.50 Ti 0.90
β ∼ 1.0 Co 0.10 Fe 0.30
0.12 Si 0.07
a
For unit conversions, please see p 19 of ref 46. 0.12 Ni 0.10
0.12 Ga 0.12
Cr 1 Hf 0.02
compositions for more favorable H2 separation proper- 1 Zr 0.04
ties.30,72,77,78 1 Ni 0.09
1 Ta 0.10
1 Fe 0.25
2.1. Pure Metals 1 Ti 0.80
1 Mo 1
The fundamental properties of pure metals critical to H2 1 Nb 1
separation membranes are summarized in Table 3. Higher 1 W 1
permeation rates result from higher H2 solubilities and Ga 0.10 Ge 0.04
lowered activation energies; decreased permeation rates 0.10 Si 0.04
originate with increased hydride formation enthalpies, result- 0.10 Ni 0.08
ing in the formation of stable hydrides and consequently 0.10 Mn 0.50
0.10 Nb 1
increasing the risk of hydrogen embrittlement.79 This em- 0.15 Ln 0.15
brittlement is primarily a result of changes in chemical Mo 1 Si 0.04
structure and unit cell dimensions which introduce stress 1 Ni 0.22
through abrupt changes in lattice constants. The slow 1 Ti 0.25
dissociation and reassociation of H2 for group IV or V metal 1 Nb 1
1 Ta 1
surfaces such as V, Nb, and Ta precludes reasonably high
flux rates.56 Decreased permeation in metals also results from
the formation of very stable passive oxides on the surface pies.85,89 Alloying with Cu, Ni, Ag, or Fe is one of the
which consequently hinder H2 molecule dissociation, dis- methods employed to reduce surface susceptibility to gaseous
solution, and H absorption by the bulk.56,58 Without sub- impurities (e.g., H2S, CO, H2O) and subsequent surface
stantially removing or modifying these metallic surfaces, H2 contamination.28 There are many examples of such alloys
separation membranes based on these particular metals are in the journal and patent literature.78,79,90,91
severely limited. Another area of intense research is directed toward
understanding the effects of microcrystalline or polycrystal-
2.2. Alloys line grain size (typically 0.5-20 µm) on the H2 permeation
rates. Since alloy grain size directly correlates to the volume
Group IV (Zr, Ti, Hf) and V (V, Nb, Ta) metallic crystal- and morphology of its grain boundaries, it is predicted to
line alloys are known to exhibit high H2 permeabilities.80-82 directly influence the specific H2 permeation rates and
Alloying is primarily employed to improve a pure metal’s embrittlement resistance. The production and processing
physical characteristics (e.g., strength, durability, degradation methods employed to synthesize a specific metallic alloy
resistance) while maintaining a single-phase bcc structure directly affect both the nucleation and size of the individual
that is required for high H2 permeation. Alloying is a very grains. These processes include, but are not limited to,
well established process and commonly includes a vast chemical vapor deposition (CVD), plating, sputtering, and
variety of elements: Fe, Mn, Mo, Cu, Ni, Ga, Ge, Sn, Si, melt cooling, all of which typically increase the grain size
W, La, and Be.79 However, Co, Cr, and Al are the most of an alloy. Cold working methods such as rolling, drawing,
commonly used for binary and ternary systems.78,79 The pressing, spinning, extruding, and heading can actually
atomic percentages of second or third elements in binary and reduce an alloy’s specific grain size.
ternary systems to form bcc single-phase alloys with V, Nb, Since alloys with very small grains have a higher volume
Ta, or Zr are easily established from the binary and ternary percentage of boundaries and more significant defects, they
alloy phase diagrams (Table 4).79,83,84 Certain alloys of the are expected to exhibit atypical diffusion mechanisms.92
highly permeable group IV and V metals have been con- These alloys have the potential of producing diffusion rates
sidered, because of their ability to reduce the susceptibility which exceed those of traditional lattice diffusion.93-95 One
to hydride formation and increase their resistance to H2 em- such example, nanostructured Pd-Fe supported mem-
brittlement caused through hydride formation pathways.79,85-88 branes,37,96-98 exhibit higher H2 fluxes (attributed to grain
In particular, small percentages of metals such as Zr, Mo, boundary diffusion) than their coarse-grained counterparts.37
Ru, and Rh have been shown to suppress the embrittlement Conversely, the Pd-Ag membranes reported by Ying and
mechanism caused by increased hydride formation enthal- co-workers37,96-98 and Lin et al.99,100 showed an increase in
the H2 permeation flux with increasing grain size. However, Table 5. Permeability Data for Some Recent Alloys Reported in
in this particular case, the elevated H2 permeation was also the Literature46
accompanied by an increase in He permeation and gaps in H2 permeability
the grain boundaries.97 Unfortunately, direct comparison of alloy (mol/m‚s‚Pa1/2)a temp (°C)
these two studies is not meaningful because of the differences VCr4Ti4 472 1 × 10-5 to 1.3 × 10-8 500-650
in manner of preparation and associated sample thicknesses. Ni3Al-6Fe 473 4 × 10-12 375
However, the difference in diffusion behaviors is tentatively Ni3Al-Zr 473 1 × 10-12 375
V99.98Al0.02 116 0.7-1.8 × 10-9 250-400
attributed to the differing nanostructures and grain boundary V99.1Al0.9 116 0.7-1.8 × 10-9 250-400
regions. In yet another set of studies of the Pd-Ag alloys, V97.1Al2.9 116 0.7-1.8 × 10-9 250-400
McCool and Lin99 describe the preparation of dense thin- V90.2Al9.8 116 2-3 × 10-9 250-400
film Pd-Ag membranes via dc magnetron sputtering. Heinze V81.3Al18.7 116 3.7-6 × 10-8 250-400
V71.8Al28.2 116 0.7-1.8 × 10-9 250-400
and co-workers69 investigated the effect of grain size on H2 V90Al10 115 1.3-2 × 10-7 250-400
diffusion in commercially available Pd-Ag foils (Ag23Pd77, V70Al30 115 0.7-1.8 × 10-9 250-400
23% Ag) and found that grain size had no significant effect V85Ni14.91Al0.09 114 3-4.5 × 10-7 250-400
on the overall H2 diffusion rates despite observing different V85Ni14.1Al0.9 114 3-4.5 × 10-7 250-400
V85Ni12.4Al2.6 114 4-6 × 10-7 250-400
operating mechanisms with different sized grains. V85Ni10.5Al4.5 114 5-7 × 10-7 250-400
There are at least two other factors within crystalline alloys Nb10Zr45Ni45 473 ∼2.5 × 10-8 350
which directly affect the diffusivity and permeation of H2. Nb95Zr5 474 ∼1.3 × 10-7 300
These are (1) specific H2 interactions with chemical or Nb95Mo5 475 ∼1.3 × 10-7 300
Nb95Ru5 475 ∼1.3 × 10-7 300
structural defects and (2) quasi-crystallinity. The H2 interac- Nb95Pd5 475 ∼1.3 × 10-7 300
tions ultimately lead to H2 trapping within or around the Fe3Al 475 0.6-1.01 × 10-10 25
specific chemical and/or structural defects in the alloy.101 Nb29Ti31Ni40 476,477 1.5-7 × 10-9 250-400
This factor becomes increasingly significant when H2 Nb17Ti42Ni41 477,478 1.1-6 × 10-9 250-400
Nb10Ti50Ni40 477,478 0.55-4.5 × 10-9 250-400
concentrations decrease due to reduced H2 fluxes through Nb39Ti31Ni30 477,478 0.3-2 × 10-8 250-400
immobilization of hydrogen or by H2 degradation of the alloy Nb28Ti42Ni30 477,478 0.3-1 × 10-8 250-400
itself. The effect of quasi-crystallinity102-104 on the behavior Nb21Ti50Ni29 477,478 0.09-2 × 10-8 250-400
of H2 in alloys remains completely unaddressed in the open V90Ti10 81 2.7 × 10-7 400
V85Ti15 81 3.6 × 10-7 435
literature. Quasi-crystals possess forbidden symmetry of five- V85Ni15 81 3 × 10-8 400
fold or greater than six-fold rotational symmetry in a periodic V90Co10 81 1.2 × 10-7 400
system.105 Further work is needed to understand the effects V85Al15 81 6 × 10-8 435
of processing and preparatory methods on the viability of R-Zr36Ni6447,478 1.2 × 10-9 350
(Zr36Ni64)1-R(Ti39Ni61)R 68 0.1-3.5 × 10-9 200-400
both noble metal- and non-noble metal-based alloys for H2 (Zr36Ni64)1-R(Ti36Ni64)R 68 0.15-3.5 × 10-9 200-400
separation. Zr36-xHfxNi64 66 0.6-3 × 10-9 200-400
Ni65Nb25Zr10 110 ∼5 × 10-9 400
2.3. Amorphous Metals Ni45Nb45Zr10 110 ∼3 × 10-9 400
Ni50Nb50 110 ∼2 × 10-9 400
H2 separation membranes based on amorphous metals are
a
generally more attractive than their crystalline equivalents For unit conversions, please see p 19 of ref 46.
because they typically exhibit improved mechanical and
structural properties without concern for defect-free film
growth. This is primarily a result of the fact that these have been made and this area is still an entirely open field
structured materials are readily stabilized in alloy form. (Table 5). The variety of amorphous alloys available to
Amorphous metals are commonly reported to exhibit in- explore for H2 separation membranes is limited only by the
creased strength, ductility, corrosion resistance, and, more imagination, and there remains many unexplored composi-
importantly, H2 solubility106 than their crystalline analogues. tions to consider. To date, most alloys examined for H2
Furthermore, they usually contain a more open lattice66 which separations have been V, Nb, Ta, or Zr based because of
decreases the embrittlement dangers associated with H2 their relatively high pure metal H2 permeabilities.107 A range
purification.47 Amorphous metallic H2 membranes are ca- of other Zr-Ni alloys have been investigated and shown to
pable of withstanding repeated cycling, high temperatures,
have relatively good mechanical and thermal stability.108,109
and high pressures, all of which are common operating
conditions for industrial scale H2 separations. This class of The ternary Ni-Nb-Zr alloys have also been studied,
membrane material offers the additional advantage of including the effects of additional elements (e.g., quaternary
outstanding compositional flexibility and homogeneity and phases including Al, Co, Cu, P, Pd, Si, Sn, Ta, and Ti), and
high catalytic surface activities for enhanced H2-surface they have been shown to be reasonably successful for H2
interactions.73 This can be highly composition dependent, separations.47,110 Amorphous Fe-based alloys have also been
as is the case for the amorphous nickel-based alloys: explored, although the surface behaviors of these particular
(Zr36Ni64)1-a(Ti39Ni61)a and (Zr36Ni64)1-a(Hf36Ni64)a, where alloys severely inhibited the adsorption/desorption of H2 and
0 < a < 1 and which required catalytic surface coatings to dramatically reduced permeation.111,112 Another heavily
lower the surface activation energies. It is important to note researched group of alloys is based on V because of its high
that the durability of the surface coating needs further H2 permeability and solubility. However, the severe H2
research, as intermetallic diffusion of coating metals into bulk embrittlement characteristics of V require the addition of
metals is commonly observed at high temperatures. In other elements, which can dramatically change the perme-
contrast, (Zr36Ni64) did not require surface coatings and Ti39- ability, mechanical, and thermal properties.113-117
Ni61 was far too brittle for use in H2 separation applica-
tions.45,64,66 To efficiently optimize amorphous alloys for H2 separation
While measured H2 permeabilities for amorphous alloys membranes, multiple physical and chemical characteristics
have yet to equal or exceed that of Pd, considerable advances must be known and ideally understood. These characteristics
Scheme 1. Namboodhiri’s469 Taxonomy of H2 Damage in Solids
span a wide variety of properties, from H2 diffusivity, example, amorphous alloys exhibit higher H2 permeation
solubility, and adsorption/desorption, to alloying and H2 rates, but this may be significantly offset by their slower
exposure effects on thermal and mechanical stability, and diffusivities. Conversely, crystalline alloys have lower H2
mechanisms of H2 damage (Scheme 1). The transport and permeation rates, but this is offset by their faster diffusion
separation of H2 is governed purely by the absorption, rates.
diffusion, and desorption energies91 and by the enthalpy of Hydrogen diffusivity in amorphous metals and alloys
hydride formation for a given alloy. Theoretical models increases with the absorption of increasing amounts of
which account for variations in the amorphous structure are hydrogen. According to Wu et al.,123 this is due to weakening
useful for describing H2 occupancies and distributions in an of metal-metal bonds. However, this also corresponded with
amorphous alloy.23,91,118,119 Direct measurements of hydrogen an increasing population of low-energy sites and increased
vibrations within an alloy lattice have demonstrated that hydrogen mobility, which ultimately translates to higher
topological changes are insignificant in amorphous alloys diffusivities. Generally, greater mobilities and smaller aver-
and that specific polyhedral sites are preferentially occupied age activation energies have been observed for hydrogen in
by hydrogen atoms.120 This direct observation is further amorphous metals and alloys.116,117,124,127,128 In contrast to this
supported by models using Fermi-Dirac statistics which work, Dos Santos et al.122,126 have argued that hydrogen
successfully describe the hydrogen distribution throughout diffusivities decrease in amorphous metals and alloys as
a predefined energy landscape or density of site ener- compared to the crystalline structure. This is due to a higher
gies.117,121 According to Dos Santos and co-workers,122 at density of defects, which is heavily dependent on the method
lower concentrations, the hydrogen atoms occupying high- of preparation. Although highly dependent on composition
energy sites have restricted mobility, but as the concentration and structure, the hydrogen diffusivities of an amorphous
increases, the low-energy sites become increasingly popu- or crystalline metal can easily be distinguished from the
lated, which increases mobility and diffusivity, thus raising standard Arrhenius behavior in amorphous metals, which is
permeability and flux values. This provides a realistic model linked to the temperature dependence of hydrogen diffusion
which can be used to evaluate potential membrane candidates on the short-range order.129
from a variety of alloys. Hydrogen trapping, short range Optimal structures for maximum diffusivity in amorphous
ordering, and structural and chemical defects are additional metals remain to be established, and modeling the hydrogen
factors which affect hydrogen diffusivity in amorphous distribution and diffusion in amorphous alloys has proved
alloys.123 to be more complicated than for crystalline alloys.124 Recent
H2 permeability through a metallic membrane (crystalline Monte Carlo simulations have provided a useful model which
or amorphous) is fundamentally governed by the solubility can predict hydrogen diffusion in amorphous metals based
of hydrogen within that particular metal or alloy. This on the dispersion of interstitial sites and a distribution of
solubility depends on both the solution activation energy and their sizes.130 Since they are thermodynamically metastable,75
the operating temperature.26,27 Hydrogen absorption capacities amorphous alloys have the distinct disadvantage that they
are the major method by which H2 solubilities are quantified. have a tendency to crystallize when heated to temperatures
Typically, larger values (for example amorphous Ni64Zr36 > 500 °C (dependent on time, temperature, and composition).
has a H/M solubility of 0.4 H/M) are reported for amorphous This limits the operating conditions (low temperature) and
alloys than for their crystalline counterparts,26,116,120,124,125,126 subsequent applications where this class of membrane can
and this is commonly attributed to the “matrix of defects” be used. Unfortunately, since hydrogen permeation is slower
within a specific amorphous alloy. This provides a consider- at lower temperatures, the fluxes are not sufficient to be
able density of defects (a distribution of high energy sorptive industrially attractive.
sites) which can be occupied by the hydrogen over a wide Other factors influence permeation in amorphous metals.
range of potential energy levels. However, depending on the First, amorphous metals exhibit exothermic enthalpies of H2
mechanism by H2 dissolution, the hydrogen solubilities can absorption, which has the potential to generate sufficient
vary considerably. Although there is a definitive dependence energy to crystallize, decompose, or change the local
of H2 permeability on solubility, it is also heavily dependent structure near the absorption site.131 The physical differences
upon the hydrogen diffusivity, which is directly correlated between amorphous and crystalline metals of similar chemi-
to the membrane’s crystalline or amorphous nature. For cal composition for application as H2 separation membranes
Table 6. Distinctions between Crystalline and Amorphous by atomic diffusion.139 The applied cooling rate should
Metallic Membranes with Comparable Compositions remain not only high enough to prevent bulk grain growth
crystalline amorphous but also lower than that required to form a glass in order to
plateau in PH2 vs [H2] isotherm no plateau in PH2 vs [H2] isotherm retain a nanocrystalline state.
Sieverts’ law obeyed at positive deviation from Sieverts’ law An important aspect of the formation of nanocrystalline
elevated [H2] metals is the role of minor alloying additions. It has been
Arrhenius H2 diffusion behavior no-Arrhenius H2 diffusion behavior
H2 embrittlement from dislocation H2 embrittlement from free volume demonstrated that adding a small amount of elements with
filling nearly zero or positive heat of mixing to alloy components
constant diffusivity with no constant diffusivity with dissolved based on Zr and Hf (e.g. Ag, Pd, Au, Pt, Ir, Re, Zn, Mo, V,
dissolved [H2] [H2]
stable at high temperatures potential crystallization at high
Nb, Ta, and Cr) causes the precipitation of a primary
temperatures nanoscale icosahedral phase,145,146 resulting in the formation
mechanically weak mechanically strong of homogeneously distributed nano-quasi-crystalline (nq)
particles within the bulk glassy alloy. However, the addition
are summarized in Table 6. Second is the presence (crystal- of an element with an extremely negative heat of mixing to
line) or absence (amorphous) of a plateau in the pressure- a Zr-based amorphous alloy may result in the production of
concentration isotherm. This lack of a plateau in the a nanocrystalline alloy where the nanocrystalline particles
amorphous isotherm suggests that no plastic deformation are dispersed throughout the glassy phase.140
accompanies hydride formation.132 However, hydrogenation
of amorphous and crystalline metals and alloys usually leads 2.5. Modeling/Simulation and Characterization
to a significant volume expansion of similar magnitude.122 Modeling is becoming an acceptable method for identify-
Somewhat paradoxically, there are no differences in the ing candidates for metal membranes. In particular, many
enthalpies of solution as hydrogen concentrations are in- recent efforts have focused on predictive modeling of
creased in either the crystalline or amorphous materials. candidate metallic alloys from first principles. Ideally,
While H2 embrittlement is observed for both types of metallic specific performance features such as hydrogen permeability,
membrane, it is less substantial in amorphous than crystalline hydrogen embrittlement, and possibly thermal and chemical
metals because the mechanisms which cause the embrittle- stabilities are modeled as a function of metal composition
ment are slightly different.133 That is, dislocation transport or structure. A recent strategy to predict the hydrogen flux
is thought to be the major H2 embrittlement pathway in through various metal alloys was based on density functional
crystalline membranes while filling of free volumes is the theory (DFT) ab initio calculations and coarse grain
mechanism believed to cause H2 embrittlement in amorphous modeling.147-150 An alternative method for predicting metal
membranes.117 candidates may be based on the enthalpy of solution of
hydrogen within a material such as disordered transition-
2.4. Membrane Fabrication and Processing metal alloys. For these metals, the enthalpy of solution has
New and novel alloys are being produced by sputtering, been predicted using a semiempirical embedded-cluster
thermal evaporation, arc-melting,71 die-casting techniques,134 model and is based on the local band structure model, which
and electrodeposition. However, the most commonly em- incorporates a coupling between local-site volumes and the
ployed methods for preparing novel alloys of variable average site volume in the alloy.151
structure and diverse composition are from melt-spinning47 In combination with predictive structural models, it is
and arc-melting.71 Ni-based alloys,106,124,135,136 Ti-based al- useful to have experimental techniques to characterize or
loys,137 Zr-based alloys,138 and, to a lesser extent, Cu-based confirm their structure. For instance, energy selective electron
alloys139,140 have all been developed as bulk metallic glasses diffraction can provide diffraction data of amorphous metals
(BMGs) and BMG matrix composites.134 To enhance the subsequently used to refine models based on molecular
utility of grain boundaries, biphasic or multiphase alloys dynamics simulations.152 One recently developed experi-
could be developed (where the presence of a bcc structure mental technique, fluctuation electron microscopy (FEM),
may or may not be required) to promote specific types of can verify the medium-range order of amorphous alloys
grain boundary conditions. Such alloys may include nanoc- predicted by mathematical models such as that proposed by
rystalline alloys or those designed with mixed crystallites Miracle et al.153 The “medium-range order” of amorphous
of variable sizes and structure. Nanocrystalline alloys are alloys is not easily determined through traditional scattering
particularly attractive because of their high resilience to methods because its pair correlations have a relatively small
degradation37 and their preparation through a variety of contribution.154 The advantage of FEM is that it is quite
techniques, such as melt-quenching,141 devitrification,142,143 sensitive to spatial variations in the scattered intensity which
or, more traditionally, high-energy ball milling, electrodepo- are caused by the “medium-range order”.155 As a conse-
sition,37 and laser ablation. The range of nanocrystalline quence, it is receptive to higher-order correlations such as
alloys formed following the devitrification pathway is less the more common triple (three atom) and pair-pair (four
studied than bulk metallic glasses (BMGs). atom) correlations which have been observed in amorphous
To date, no established methodology has been formulated metals.
for the formation of a nanocrystalline structure. However, Numerous other general material design strategies exist
the methods generally include (1) a multistage crystallization which have not yet been extended to the design of novel
process, (2) high nucleation frequency, (3) slow growth rate, metallic membranes for H2 separation. These include the
and (4) thermal stabilization of the remaining amorphous utilization of trained neural network models which have led
phase by the solute element redistributing along the nanoc- to optimization of Ni-based polycrystalline superalloys by
rystal/amorphous interface.144 Other authors have cited the using composition iterations to predict tensile properties as
need for high rates of homogeneous nucleation, normally a function of temperature.156,157 However, the use of neural
associated with stoichiometric compositions and governed networks to predict H2 permeability or durability as a
function of composition has not yet been reported in the open to increase H2 flux while maintaining mechanical strength,
literature. Alternatively, recent application of combinatorial thermal stability, and reliability. The current deposition
materials screening and synthetic methods for accelerating methods used are electroless plating,172 electrodeposition,
the process of technology discovery and application could spray pyrolysis, sol-gel dip coating, physical and chemical
play a significant role in the rapid development of H2 vapor deposition (PVD/CVD), or sputtering.173,37 While CVD
separation membranes.158-161 Notably, combinatorial materi- offers a thin layer with more efficient hydrogen permeation
als preparation has already been used to produce continuous compared to that prepared by electroless plating, this
phase diagrams useful for describing structure-property technique is less industrially attractive.174 Reports in the
relationships of well established Ni1-xFex alloys.162,163 The literature show CVD layers having higher H2 permeability
challenge of using the combinatorial method lies in deter- because hydrogen transported through the membrane via a
mining the H2 permeabilities and operating durabilities of surface diffusion mechanism rather than the traditional
the metals on such small scales. This is an essential solution diffusion (i.e., bulk diffusion) mechanism of dense
component of screening the candidates as a function of metals, as observed for Pt electroless layers.65 However,
composition, structure, and H2 interaction characteristics. researchers have reported that electroless coated amorphous
Significant advances have been made in utilizing thin-film Ni-B alloys exhibit a Knudsen diffusion mechanism (i.e.,
deposition and masking techniques, such as molecular beam the mean free path of the diffusing H2 molecule is much
epitaxy (MBE), to incorporate spatially variable or selective larger than the pore size) for hydrogen and speculate that a
deposition, needed in making combinatorial databases and surface diffusion mechanism may also be operating here
wide composition candidates.156,164 rather than a solution diffusion mechanism.175
Compositional libraries can also be created from “diffusion Thin layer metallic coatings have been classified into four
multiples”, an assembly of three or more metal blocks, which types:29 (1) a thin metal layer (dense or porous) is formed
undergo high temperature interdiffusion to generate complete on the surface and extraneous to the support; (2) a thin metal
phase diagrams, as has been done for Ni-based alloys.165 layer is formed on the walls within a porous support; (3) a
Compositional databases are generated by solution-based microporous ceramic layer is formed on the supporting layer
combinatorial synthetic methods using simple and cost by finely distributing metal particles within the pores of the
effective multicomponent inkjet delivery systems, which have support; and (4) a microporous ceramic layer is formed on
many obvious advantages over the thermal interdiffusion the supporting layer by sintering metal-coated particles onto
methods.166 Property characterization of samples generated the surface. For thin metal membranes, the permeability
by the combinatorial approach requires highly sensitive, behavior may differ significantly from that of bulk metals
sophisticated, and typically nondestructive equipment. For of similar composition. This is due to the increase in the
example, scanning microwave microscopy (SMM) can dominance of surface phenomena, defects, grain boundaries,
characterize and map conducting and electromagnetic materi- thermal dilatation, and lattice defects on the permeability
als,156,167 while electron probe microanalysis (EPM) can be behavior (for instance, have been observed in Pd).74 This
used for compositional mapping.168 Electron backscatter divergence emphasizes how critical it is to characterize the
diffraction (EBD) can provide crystal structure analyses, permeability behavior for thin metal structures as opposed
while nanoindentation determines some of the mechanical to mere inference from bulk metal behavior.
properties. As the amount of information describing novel
bulk metallic glasses, crystalline metals, and nanocrystalline 2.7. Catalytic Surface Coatings
metals increases, new databases are developed to allow more
systematic comparisons to inform future alloy design direc- Aside from Pd-based supported thin films, electroless and
tions and eventually develop methods for rational alloy electrolytically deposited films of alternative compositions,
design.169 Although currently in its infancy, rational alloy such as Ni-based alloy films, have been reported.172 It is
design is now starting to be achieved through prediction of expected that Ni-based films will be useful for catalyzing
bulk composition ranges using a range of thermodynamic the dissociation and reassociation reactions of H2, since they
models such as Miedema’s semiempirical model, qua- have already been shown to catalyze the hydrogen evolution
sichemical models, or kinetic models (e.g., phase field reaction.176 Typically, amorphous alloys have the premier
model).151,170,171 catalytic activity. In particular, high catalytic activity has been
Having established a list of useful candidates with high found for Ni with S, P, and B existing as amorphous alloys
performance characteristics, methods for large-scale process- with some pure Ni nanocrystals.177 Initial results indicate that
ing, production, and fabrication of industrial membranes can Ni-P alloys are also very efficient for separating H2 due to
then be addressed. Alloy compositions are selected and permeation via both surface diffusion and Knudsen diffusion
designed such that they can be manufactured from more mechanisms. Both mechanisms are available because of the
conventional casting or high-pressure die casting techniques NiP “cluster” structure and the amount of interstitial space
to ease large-scale membrane production. However, if these available for diffusion.72 While Ni-P films, particularly
fabrication and production methods are underdeveloped or electroless deposited films, are relatively easy to synthesize
unavailable for utilization, this will further delay the process and do not require special equipment for preparation or
and ultimately make complete assessments unavailable for pretreatment for surface catalytic activity, they do have a
commercially promising membranes. Furthermore, in situ few drawbacks. Namely, H2 separation must be performed
membrane repair and regeneration costs must also be at low temperatures (<10 °C); otherwise, the structures may
considered during membrane development before imple- crystallize. In addition, the surface area of the alloys can be
menting large-scale fabrication methods. quite small, and the storage of the amorphous alloys can be
difficult, since certain compositions are sensitive to oxida-
2.6. Membrane Fabrication tion.178
Metals are being deposited as thin layers on various Catalytic thin-film layers (e.g., Pd) can be applied by vapor
supports (e.g., glasses, ceramics, or other metals), in an effort deposition,59,74 by electro- and electroless-plating,56 or by roll
Table 7. Table 8.
M-H bond energy M-CO bond energy metal (indices) specific site binding energy (eV)
metal (indices) (kJ/mol) (kJ/mol)
V(110) 3-fold -3.29
Ag(111) 218 25 Ta(110) 3-fold -3.24
Pt(100) 247 134 W(110) 3-fold -3.15
Pt(111) 247 126 Mo(110) 3-fold -3.05
Co(0001) 251 105 Fe(110) 3-fold -2.99
Co(1010) 251 Ru(0001) fcc -2.97
Cu(111) 251 70 Ni(111) fcc -2.89
Ni(110) 259 Co(0001) fcc -2.89
Ni(100) 263 109 Pd(111) fcc -2.88
Ni(111) 263 109 Rh(111) fcc -2.81
Pd(111) 259 142 Ir(111) fcc -2.74
Pd(100) 268 151 Pt(111) fcc -2.72
Pd(110) 268
Fe(100) 265 105
Fe(110) 273 Table 7), while Ag and Pt have the lowest adsorption
Nb(100) 273 energies. In addition, a substantial amount of work has
Mo(110) 273 investigated and modeled surface segregation in multicom-
Mo(100) 277 ponent systems for the purpose of designing catalytic layers
with seemingly contradictory properties but that are optimal
cladding. This has been effective for bcc materials such as for a specific application.179-181
V, Nb, and Ta57,59 and for Ni-Zr-based amorphous alloys.64 Near surface alloys (i.e., alloys that have a different surface
The addition of surface layers or treatment of the metal solute metal concentration from that of the bulk) have
membrane surface is becoming increasingly common to presented themselves useful for providing a surface that
improve the catalytic activity (for dissociation and reasso- simultaneously allows both weak hydrogen binding and low
ciation of hydrogen) and chemical stability of the metal hydrogen dissociation barriers.182 According to Table 8,
membrane in use. Dissociation and reassociation of H2 on metals with high binding energies such as V (-3.29 eV)
the metal membrane surface can be accomplished by several and Ta (-3.24 eV) can be alloyed with Pt, for example, to
means, including the inherent catalytic nature of the metal yield surfaces with low H2 dissociative transition state
itself, the addition of a catalytic layer, or electrochemical energies, calculated to be around about 0.5 and 0.6 eV,
reactions. This process is directly controlled by the atomic- respectively.183 However, under certain conditions, such as
scale tomography of the surface, the quantity and distribution strong oxidizing conditions, inverse segregation occurs
of impurity species, and any one of these steps can be depending on the relative affinities of the alloy components
kinetically controlled. A single material can be effective at for gaseous oxygen.184 Oxide layer formation on the alloy
catalyzing the dissociation/reassociation processes in addition surface may inhibit the catalytic advantages of the surface.
to having high bulk diffusivity, as is the case for Pd and Pd
alloys such as Pd-Cu and Pd-Ag and Ni-P.72 Other
candidate membrane alloys, for example amorphous Ni-
3. Silica Membranes
Zr, could be activated by exposure to H2 at elevated Due to some of the inherent limitations of metal mem-
temperatures.64 Alloys containing one of the following branes, research is underway for alternative membrane
elements, Fe, Ru, Rh, W, Mo, Pt, Co, or Ni, can also be materials for H2-based applications. Silica membranes are
sufficient to act as a catalyst to dissociate the hydrogen.78 one of the candidates for hydrogen separation due to their
Alternatively, catalytic layers could be applied to the surfaces ease of fabrication, low cost of production, and scalability.
of metals which readily diffuses hydrogen ions but have Because of their porosity and composition, silica membranes
inadequate catalytic properties. are also less expensive than metals (due to the lack of
There is considerable room for developing non-Pd alterna- precious elements) and not susceptible to H2 embrittlement.
tives for catalytic layers, with numerous transition-metal- They are inorganic membranes that have a network of
based catalysts having already been tried, e.g., Pt, Ir, Co, connected micropores of approximately 0.5 nm diameter and
Co-Mo, Fe, magnetite (Fe3O4), La-Sr-Co-O, WS2, or can accommodate the separations of small molecules such
MoS2.78 The main requirements for such catalytic surface as H2, He, CO2, CO, N2, and O2. In fact, these membranes
coatings are that reactive surface sites should be sufficiently have yielded exceptional H2 selectivities, with reported H2/
close and concentrated to assist the dissociative adsorption N2 values exceeding 10,000.185-187,199 Summaries of the
of hydrogen and are not readily blocked by the adsorption preparation of inorganic membranes have been presented by
of contaminants such as S, CO, or other adsorbates. When many research groups, including those of Morooka and
hydrogen reassociates to leave the surface of the metal having Kusakabe,188 Tsapatsis and Gavalas,189 Omaya,190 and Ver-
passed through the membrane, the surface of a metal with weij.191 Contrary to dense metal, alloy, and ceramic mem-
the lowest possible desorption energy will favor the process. branes, microporous silica membranes are not 100% selective
Table 7 provides a list of the adsorption/desorption energies for one component. Their separation of molecular mixtures
for H2 and CO onto the crystal face of a pure metal. Despite is based on a competitive process in which individual
the fact that the energies depend on the crystal surface molecules move by site-hopping diffusion in the connected
indices, if the metal is alloyed with Ag or Pt, then H2 will micropore network. A full description is found in ref 191.
more readily desorb than the pure metal, particularly if Ag Silica membranes are generally comprised of three lay-
segregates to the surface. On the other hand, if a contaminant ers: (3.1) a membrane layer, (3.2) an intermediate layer, and
such as CO has a high binding energy, it will most likely (3.3) a support. Much research has focused on each
persist and interfere with H2 surface dissociation. Thus, Pd component to determine the structure/property relationship
and Pt are the most susceptible to CO contamination (see between material and light gas permeation ability. Each of
the three layers that comprise a silica membrane will be more

fully explained below.
3.1. Membrane Layer Synthesis
Silica membranes are synthesized primarily through two
different methods: sol-gel modification192-198 and chemical
vapor deposition (CVD).188,199-210 Sol-gel modification
provides good selectivity and permeability, as opposed to
CVD methods, where there is an attendant loss of perme-
ability, though the selectivity is enhanced. The sol-gel
method, however, suffers from a lack of reproducibility. CVD
methods usually require substantial capital investment and
controlled conditions of deposition. More detailed descrip-
tions of each method are given below.
3.1.1. Sol−Gel Processing of a Membrane Layer
Sol-gel processing can be done via three different synthe-
tic methods: the silica polymers, particulate-sol, and tem-
plate methods.221 The silica polymers route involves the hy-
drolysis and condensation of alkoxysilane precursors, such
as tetraethyloxosilane (TEOS), under controlled condi-
tions.211-214 Furthermore, de Lange showed ultrathin 60 nm
microporous membranes with pores of 0.5-0.7 nm. Gas
transport was activated for H2 (Eact ) 21.7 kJ/mol) and
molecular sieve-like separation factors of 200 for mixtures
of H2/C3H6 at 260 °C.195 The particulate-sol route is based
on the packing of nanoparticles to make a highly porous
structure.215-217 Silica particles of different sizes are packed
into the support substrate to process membranes with
different pores sizes. Added binder material or hierarchical
size packing aids in packing the particles to avoid defects.
The template route uses organic molecules as templates in Figure 3. Schematic of the three important sol-gel routes used
for preparation of microporous membranes.246,299 (Reprinted with
the sol matrix that are burned out upon calcination. The permission from ref 246. Copyright 2002 Taylor & Francis Group,
organic molecule’s size and shape can be imprinted in the LLC, http://www.taylorandfrancis.com.)
sol for a tuned porosity. Surfactants, organic ligands, and
polymers have been reported as templates (Figure 3).218-221 configuration of the precursors.197,198 In the first type, the
Silica membranes by sol-gel deposition are made by dip- precursors are provided from one side of the substrate188,201
coating an aqueous silica polymer sol on a mesoporous while the other side of the substrate is usually vacuumed to
support surface, followed by drying and calcination at 400 obtain a pinhole-free membrane.205 Prabhu and Oyama187,190
< T < 800 °C.222 The silica polymers are formed by acid- reported that a stable silica membrane was prepared by CVD
catalyzed hydrolysis and polymerization at pH < 7 of TEOS treatment at 600 °C. However, hydrogen permeance was less
and MTES.275 Reaction parameters such as time, temperature, than 1.8 × 10-8 mol/(m2 s Pa).
pH, and mixing must be closely controlled in all stages of The second method is counterdiffusion CVD, where two
the process.223 However, the effect of these parameters on kinds of reactants are supplied from the opposite sides of
the final microporous structure is limited. The micropores the substrates.197,200-202,206 Pore sizes and effective membrane
in sol-gel silica are likely formed around original solvent thickness can be controlled by changing reactants and
molecules such as H2O and C2H5OH. Examples of prede- reaction conditions. One of the first gas-phase methods to
signed templating have been reported with other alcohols,224 be developed was generation of a silica-modified membrane
HTEAB,225 and methacryl oxypropyl trimethoxy silane.226 by a high temperature atmospheric CVD process on Vycor
Amorphous silica contains many “nonbridging” Si-O bonds glass.188,247 The new membrane (Nanosil) showed unprec-
that surround the micropores. These are normally terminated edented selectivity to hydrogen (100%), without loss of
with protons to form Si-OH, but introduction of MTES permeability compared to the porous Vycor precursor. The
before hydrolysis results in the formation of Si-CH3- membrane also showed high stability under hydrothermal
terminated groups.227 The latter makes the structure more conditions over prolonged times. Contrary to various other
stable and more open. The terminal groups gradually silica membranes, this Vycor membrane showed high stabil-
disappear upon heating by condensation and carbonization ity under hydrothermal conditions over prolonged times. In
below 600 °C, and a “dense” silica structure is formed at particular, the membrane was used in a catalytic reactor with
800 °C. In ref 228 an alternative to the sol-gel method is 1% Rh/Al2O3 for the dry reforming of methane. Conversions
reported, where polysilazane was spin coated, cross-linked were higher than those in the bulk packed-bed reactor. The
in N2 at 270 °C, and pyrolyzed in air at 600 °C. incorporation of the inorganic membrane into the catalytic
process resulted in the circumventing of thermodynamic
3.1.2. Chemical Vapor Deposition (CVD) of a Membrane limitations normally found in the bulk process.
Layer Other CVD methods use simple thermal decomposition186
CVD methods to prepare a membrane on a porous or oxidation of the precursor with oxygen or ozone within
substrate are classified into two types, based on the supplying similar temperature ranges.229
Figure 4. Steam stability for a silica membrane prepared at 600 °C.199 (Reprinted from ref 199, copyright 2005, with permission from
Elsevier.)
Nakao et al. reported silica membranes having an excellent layer thickness and delamination. The silica membranes are
H2/N2 permeance ratio (about 1000).231 They were obtained repaired by impregnation with TEOS ethanol, followed by
on porous γ-alumina substrates by the counterdiffusion CVD thermolysis.230 CVD methods have also been proposed to
method using a TMOS/O2 system at 600 °C. This membrane repair residual connected defects in wet-chemical layers.231
was stable under 76 kPa of steam vapor at 500 °C for 21 h
without any reduction in the H2/N2 permeance ratio. This 3.3. Intermediate Layers
membrane can be applied for H2 production from the steam
reforming of methane, because 76 kPa of steam vapor at Intermediate layers are prepared by dip coating of nano-
500 °C is one of the target conditions for the H permselective particle dispersions, followed by drying and calcination.
membrane reactors of the reaction (Figure 4). The reaction Typical compositions include transition aluminas, silica, and
species at 500 °C of the counterdiffusion CVD method was zirconia. Precursor particles are made by precipitation from
confirmed as TMOS/O2 by comparing the permeation results simple salt solutions or hydrolysis of organo-metal reagents.
of thermal decomposition of TMOS, the TMOS/O2 system, To obtain homogeneously packed layers with little shrinkage,
and the TMOS/O3 system. The activation energy of H2 was it is important to control particle agglomeration during
∼20 kJ/mol through the membrane. The H2 permeance of synthesis and to remove any agglomeration after synthesis.
the 600 °C permeation test was 1.5 × 10-10 mol/(m2 s kPa). A well-known synthesis of γ-alumina layers starts with
The H2/N2 permeance ratio was kept for 21 h under the hydrolysis of ATSB at 90 °C, followed by HNO3 addition,
typical steam-reforming conditions of methane for a mem- resulting in the partial dissolution of the Boehmite precipitate
brane reactor (76 kPa of steam at 500 °C). The silica and redispersion of agglomerates.230 The HNO3 addition also
membrane was damaged by the heat treatment at higher than ensures colloidal charge stabilization by preferential proton
the deposition temperature. The membrane should be pre- sorption. The hydrolysis/peptization method has a favorable
pared at higher temperature than the application temperature. yield but suffers from the presence of residual agglomerates.
The agglomerates can be removed by high-g centrifuga-
3.2. Preparation tion,232 which leads to stable and homogeneous layers.
Sonochemical and modified emulsion precipitation methods
There are a variety of coating methods commonly used help avoid formation of the agglomerates. More recently,
and continuously optimized for making thin films of silica methods were developed in which agglomeration is com-
membranes. For an overview, see ref 6. In dip-coating, a pletely avoided up-front, such as sonochemical and modified
support is contacted briefly with a sol or dispersion. Film emulsion precipitation.233,234 Intermediate layer formation
formation occurs by two mechanisms: slip-casting and film- from a nanoparticle dispersion is often assisted by additions
casting. In slip-casting, the dispersion liquid penetrates into of linear chain polymers such as PVA to the dispersion or
the support under the action of capillary forces. The dispersed by pretreatment of the support with polymers to minimize
particles (polymers) form a dense-packed film on the surface penetration. An example includes the repair of a commercial
while dissolved additives disappear into the support. In film- supported γ-alumina membrane by dip-coating with a
coating, a dispersion layer is formed on the slip cast layer Boehmite precursor nanoparticle dispersion.235
and maintained by surface tension. To avoid the frequently
present defects in dip-coating, researchers need to avoid air- 3.4. Support
borne contamination, agglomeration, and particulate con-
tamination during synthesis, and microbubbles by controlling A variety of methods for fabricating support layers have
shear, ultrasonic treatment, and additives. Finally, particulate been employed. One common method is the dip-coating of
contamination needs to be removed after synthesis by agglomerate-free submicron particle dispersions, made from
screening or centrifugation. commercially available R-Al2O3 powers. Colloidal stabiliza-
The effect of connected coating defects is often diminished tion can be adjusted such that coherent dense-packed layers
by application of two or more coatings. This approach, are formed with 25 nm surface roughness after slight
however, does not work for surface defects in the support, sintering at temperatures around 1000 °C.222 Other methods
and it may affect the operational lifetime due to excessive result in thicker support layers and include colloidal filtra-
tion222 and centrifugal casting,236 resulting in high strength,

excellent surface properties, and roundness of the tubes.
These structures can also be considered for applications
where f HKn2,l < 10-8 mol/(m2 s kPa) is acceptable.
Carrier structures are generally made with conventional
ceramic forming methods using commercially available,
coarse R-Al2O3 powders. Dry-pressing is used to make small
disks for research purposes. Extrusion is used for tubes and
multichannel honeycomb structures.237,238 These forming
methods are very suitable for fairly cheap mass-scale
production but have limited near-net-shape capabilities.
Nonroundness and other dimensional limitations may result
in sealing and construction problems in high-temperature
membrane reactors. It is for this reason that gel-casting
methods239 are also considered. Such methods allow for a
higher initial solid load, better control of overall homogeneity
during forming, and, hence, better dimensional specifications.
The large pore diameter requires very high sintering tem-
peratures and may result in poor mechanical strength and
reliability. This problem might be addressed by application
of wet-chemical techniques such as phosphate bonding,240
which provides thermochemical stability up to 900 °C.
3.5. Modification
An extension of the silica membranes work is the metal-
coated silica systems.208 Al-coated SiO2 permselective mem-
branes have been studied and reported.188 They are prepared Figure 5. (a) Cross section of a Ni-doped silica membrane (Si/Ni
by chemical vapor deposition of a thin SiO2 layer on a porous ) 2/1). (b) Observed permeance ratio of H2/H2O vs the permeance
alumina substrate, resulting in a noncontinuous network of ratio of He/H2 for nine membranes (Si/Ni ) 0-1/1) at 500 °C.
solubility sites. The submicron thick silica-on-alumina (Reprinted from ref 244, copyright 2006, Elsevier.)
composite membranes utilize size gradation in their layering,
Table 9. Reported Values of High-temperature Hydrogen
allowing for enhanced permeability for hydrogen over CO2, Permeation and Separation through Silica Membranes237
N2, CO, and CH4. However, silica modified membranes
prep measurement H2 permeation
developed by several researchers suffer from loss of perme- method temp (°C) (mol/m2‚s‚kPa) separation ref
ability (as much as 50% or greater in the first 12 h) on
exposure to moisture. This has been attributed to the removal CVD 400 4× 10-10 H2/HBr ) 1000 480
CVD 427 6 × 10-10 H2/N2 ) 160 480
of Si-OH groups leading to the formation of Si-O-Si CVD 600 1.8 × 10-10 H2/CH4 ) 4200 187
bonds which close pore channels.241 This phenomenon is sol-gel 300 1.3 × 10-9 H2/CH4 ) 150 217
termed densification. Moisture catalyzes this reaction, par- sol-gel 350 2.2 × 10-9 H2/CH4 ) 35 481
ticularly at higher temperatures.242 Densification not only sol-gel 600 2.5 × 10-9 H2/C3H8 ) 75 482
leads to lower permeability but also causes embrittlement sol-gel 500 2.6 × 10-10 H2/N2 ) 87 483
CVD 500 1.3 × 10-10 H2/N2 ) 2300 199
of the silica film that compromises selectivity.
3.5.1. Silica Membrane Modification permeance for He and for H2 with a high selectivity of 1450
(He/N2) and 400 (H2/N2) even after being kept in steam
In order to improve the stability of silica membranes in (steam: 90 kPa) at 500 °C for about 6 days. The permeance
steam, inorganic oxides, such as TiO2, ZrO2, Fe2O3, Al2O3, ratio of H2/H2O was found to be dependent not only on the
NiO, etc., were added to silica.239,243,244 Aseada et al. have permeance ratio of He/H2 (Figure 5) but also on the Ni
shown the Ni-doped silica membranes exhibited relatively content, while the maximum permeance ratio observed at
high H2-permeance and high stability against water vapor at 37 for a Ni-doped silica membrane (Si/Ni ) 1/1).
35-300 °C,242 suggesting the effectiveness of the addition The silica membranes shown in Table 9237 have excellent
of nickel oxides to silica for the membrane stability against separation performance for hydrogen and helium in dry
steam at higher temperatures. The Ni-doped silica mem- conditions in a wide temperature range, 50-600 °C. Their
branes242 were fabricated in this work by the sol-gel stability against water or water vapor, however, is rather
techniques under various conditions of Ni contents and firing poor at high temperatures or even at room tempera-
temperatures. The H2-selective permeation characteristics and ture.188,221,226,241,245,246 Separate measurements have shown that
hydrothermal stability of the membranes were tested in steam H2 gas permeation available with the present silica membrane
at 500. The hydrothermal treatments of the membranes before system has a maximum value of 5 × 10-9 mol/(m2 s kPa).
exposure to H2 were quite effective to prevent the further This value was around 50% of the H2 permeation of the
densification of Ni-doped amorphous silica networks. This γ-alumina membrane layer under the same measurement
is due to reduction in H2 and sintering in steam (500 °C, 70 conditions. The permeation through the support substrate was
kPa). Ni-doped silica membranes (Si/Ni ) 2/1) fired in the therefore considered as the major resistance for gas perme-
steamed atmosphere (partial pressure: 90 kPa) at 650 °C, ation. An increase in gas permeation of the substrate by
for example, were found to show an asymptotic steady increasing the porosity or by reducing the thickness of the
substrate should not be difficult. Yet, retaining the strength adhesion. In refs 249 and 252, the stability is reported on
of the support substrate is important for membrane reactor sol-gel silica membranes in synthetic reformate: H2O:H2:
applications, where a thermal gradient exists between the N2:CO2:CO ) 34:28:24:8:6 at 200 °C and a total pressure
inlet and other parts of the reformer. of 200 kPa. These fairly mild conditions likely did not lead
to any significant intermediate layer degradation. It was
3.5.2. Membrane Structure Modification found that membranes made with HTEAB (hexyl triethy-
The most popular configuration of these membrane lammonium bromide) templating were more stable than
systems is a three-layer asymmetrical structure247 with an conventional sol-gel membranes. They exhibited a gradual
R-alumina-based substrate and a γ-alumina-based intermedi- decline of total flux for H2, but that effect could be undone
ate layer as supports for the top coated silica microporous by regeneration at 500 °C in air. The conventional sol-gel
layer. Though the structure is simple, the smaller particle membranes showed a similar decline but could not be
size of γ-alumina mostly limits the pore size of the alumina regenerated.
substrate. The resultant permeation of the system is limited Amorphous silica is emerging as a valuable material for
due to the large resistance of the support substrate. H2 production membranes as long as the operational stability
Nair in ref 237 recently reported a four-layer membrane; can be optimized. The thermochemical stability of the
the thickness of the γ-alumina membrane could also be membrane and the intermediate layers is probably best if
reduced because of the better surface smoothness of the dip their thickness remains well below 1 µm. This limits the build
coated R-alumina intermediate layer compared to the ex- up of stresses due to (dynamic) thermal expansion differ-
truded substrate, which normally supports the γ-alumina layer ences. The current cost price of lab prototypes indicates that
in the three-layer configuration. In this configuration, the dramatic cost reductions are needed to realize viable
intermediate support layer was made from R-alumina par- membrane designs. These reductions might be realized by
ticles of average size 300 nm. These alumina particles were replacing conventional synthesis and ceramic firing by rapid
made into a water-based slurry. Alumina tubular substrates processing methods, instead of sintering for support and
were dip-coated with this slurry and sintered at 1077 °C to carrier structures. Wet-chemical deposition of films provides
make intermediate layers. The slurry was also dried (and in favorable cycle times but requires supports with excellent
some cases powdered and pressed into pellet form) and heat- surface quality. The occurrence of structural defects remains
treated to make samples for characterization of pore size, a major source of irreproducibility and poor performance.
porosity, and thermal expansion properties. They conclude Further defect minimization requires characterization of the
that the pore size of the membrane is probably the easiest overall and local defect population by any of the application
and most trouble-free way of improving the gas permeation methods.
of the substrate while retaining its strength, stability, and Amorphous silica shows primarily a structural densifica-
durability. tion that is determined by the presence of terminating groups
and temperature, as discussed before. There is little evidence
3.6. Operational Stability that this process is influenced by modest steam pressures.
The microporous silicas are very promising due to their Optimization of the industrial processes to lower pressures,
low cost, high stability, and high permeance.226 Systematic plus taking into account the chemical stability of the
studies of the operational stability of supported membranes amorphous silica, will enable widespread industrial use of
with good zero hour properties are scarce. The micro- and these materials as membranes.
mesoporous structures used for the membranes and inter-
mediate layers have a very high surface area, which means 4. Zeolite Membranes
that further densification, phase transformations, and struc-
tural disintegration may occur at elevated temperatures and Much effort has recently been devoted to the synthesis
steam pressures. Rapid pressure fluctuations may lead to and potential application of inorganic membranes in the
disruptive tensile stresses in the interfaces between the domains of gas separation, pervaporation, and reverse
layers.247 osmosis or in the development of chemical sensors and
In a recent review by Verweij et al.,247 different studies catalytic membranes. Inorganic membranes, which have good
were described regarding optimization of silica membranes thermal stability and chemical inertness, have advantages
for enhanced operational stability.248 Supported silica mem- over polymer membranes for many industrial applications.
branes are affected by the delamination and structural Improved membrane integrity and manufacturing costs are
instability of the intermediate layer. As described in ref 252, constant factors which are the focus of many research efforts.
the formation of macrodefects in the γ-alumina layer was Zeolite membranes, in particular, combine pore size and
found after 23 h at 475 °C and a partial pressure of H2O of shape selectivity with the inherent mechanical, thermal, and
40 kPa. Further studies showed partial delamination of chemical stability necessary for continuous long-term separa-
γ-alumina after a 100 h steam reforming treatment.227 The tion processes. The effective pore size distribution of the
instability of the γ-alumina layers is ascribed to poor zeolite membrane, and hence its separation performance, is
adhesion in combination with structural densification. The intrinsically governed by the choice of the zeolitic phase(s).
latter, in turn, is caused by γ-alumina packing defects and This applies when molecular size exclusion sieving is the
sintering. These phenomena are confirmed in a recent study dominant mechanism and no other diffusion pathways bypass
of the stability of a tubular four-layer structure.237 It was the network of well-defined zeolitic pores/channels; other-
also shown that these problems can be addressed by wise, viscous flow through grain boundaries prevails. The
improving adhesion by phosphate bonding and γ-alumina optimum thickness of the zeolite film is always a compromise
surface modification with La2O3. Improvements in the between separation performance and overall trans-membrane
γ-alumina packing homogeneity as demonstrated in ref 236 flux and is often tailored to the specific needs of the
are also expected to result in less shrinkage and better envisioned application.
industrial applications. Currently, research is ongoing to

resolve the intercrystalline diffusion path issue by using
mixed matrix membranes. Readers are directed to refs 284
and 285 for further information on this area.
4.1. Membrane Growth Methods

Syntheses of zeolite membranes, described recently, can
be broadly classified into two categories: in situ and
secondary (or seeded) growth.286,287 In the in situ technique,
the support surface is directly contacted with an alkaline
solution containing the zeolite precursors and subjected to
hydrothermal conditions. Under appropriate conditions, zeo-
lite crystals nucleate on the support and grow to form a
continuous zeolite layer. At the same time, reactions occur-
Figure 6. Not-to-scale representation of a zeolite membrane on a ring in the solution lead to deposition of nuclei and crystals
nonselective oxide support (ie., Al2O3), showing possible permeation on the surface followed by their incorporation into the
pathways, either interzeolite crystals or intrazeolite crystals. membrane, thus minimizing intercrystal pore contributions.
MFI films grown in situ may exhibit a preferred orientation
Zeolites are crystalline inorganic framework structures that that depends on the synthetic protocol and associated
have uniform, molecular-sized pores. They have been used interplay of nucleation and growth phenomena.286 However,
extensively as bulk catalysts and adsorbents. The zeolite because of the insufficient understanding of nucleation and
structure is made up of TO2 units, with T ) a tetrahedral growth processes in hydrothermal systems, the success of
framework atom (Si, Al, B, Ge, etc.). In all cases other than in situ methods in yielding uniformly oriented MFI films is
neutral silica zeolite frameworks, the net overall charge of limited.
the framework is negative and is charge balanced by cations In the secondary (or seeded) growth technique, zeolite
(either inorganic or organic). The cations reside in the pores nucleation is largely decoupled from zeolite growth by
of the framework; the size of the pore is catagorized by the depositing a layer of zeolite seed crystals on the support
number of T atoms in that ring. Small-pore zeolites include surface prior to membrane growth. The layer of seed crystals
those structures made up of eight-member oxygen rings, can be deposited with precise control over which crystal-
medium-pore zeolites have 10-member rings, and large-pore lographic axis is oriented perpendicular to the support (see
zeolites have 12-member rings.250 More recently, membranes Figure 7).225 The seeded surface is then exposed to the
of continuous polycrystalline zeolite layers have been membrane growth solution and hydrothermal conditions,
deposited on porous supports. The first zeolite membranes whereupon the seed crystals grow into a continuous film.
were reported in 1987,251 and since then, significant progress Although this method offers greater flexibility in controlling
has been made to expand the types of zeolites utilized in the orientation of the zeolite crystals and the microstructure
membranes, improve membrane quality, and widen their of the zeolite membrane (since it decouples nucleation from
range of applications. Today, more than 14 zeolite structures, growth), it is done so at the expense of additional processing
including MFI,252-260 LTA,261-263 MOR,264-266 and FAU,267-270 steps. In principle, the orientation and morphology of the
have been employed as H2 selective separation membranes. membrane can be manipulated by changing the morphology
The MFI structure is typically used in zeolite membranes and orientation of the deposited seed layer and then perform-
because of its pore size and ease of preparation, and this ing secondary growth under appropriate conditions.
structure includes silicalite-1 and ZSM-5. Silicalite-1 is made
up of pure silica, and ZSM-5 has Al substituted for some of
the Si atoms.
4.2. Permeation and Gas Transport
Significant progress has been made in developing new The following five-step model can be used to describe the
membranes, optimizing their synthetic preparation, and gas-molecule transport through a zeolite membrane:288,289 (1)
understanding transport and separation fundamentals over adsorption from the bulk phase to the zeolite external surface;
thepastdecade.Severalreviewsofzeolitemembranes243,246,250,271-282 (2) diffusion from the surface to the inside of the zeolite
have focused mainly on membrane synthesis and gas channels; (3) diffusion inside the zeolite channels; (4)
separation applications. This progress suggests that many diffusion from the zeolite channel to the external surface;
applications of zeolite membranes in commercially valuable and (5) desorption from the external surface to the gas phase.
enterprises, such as separations, are promising. Gas and liquid The actual mechanism of gas permeation through an MFI-
separation on zeolite membranes is primarily governed by type zeolite membrane depends on the gas adsorption
competitive adsorption and diffusion mechanisms. When the properties on the zeolite. For nonadsorbing gases, molecules
zeolite pore size distribution falls between the molecular sizes may directly enter the zeolite pores from the gas phase. The
of the feed components, a size exclusion mechanism can separation factor of a nonadsorbing gas mixture is determined
dominate the separation process.225,283 However, one of the by the mobility of the molecules inside the zeolite pores and
main challenges in zeolite membrane development is the the probability of the molecules entering the zeolitic pores.256
minimization of intercrystal pores formed inherently in Gas molecules with small size and high mobility tend to
polycrystalline zeolite films (see Figure 6). The existence permeate through the zeolite membrane, while those with
of intercrystal pores with sizes larger than the zeolitic pores larger size and lower mobility tend not to permeate. For
is the major cause for decline in molecular separation strongly adsorbing gases, permeation through an MFI
efficiency.231 The elimination of intercrystalline pores is membrane is controlled by either adsorption or activated
essential for having high separation selectivity viable for diffusion, or both, depending on the operation conditions
(y1/y2)p
R) (5)
(y1/y2)f
The separation selectivity, R, is defined as the enrichment

factor of one component in the permeate as compared to the
feed composition ratio, where y1 and y2 are the mole fractions
of components 1 and 2, respectively, and the subscripts p
and f refer to the permeate side and feed side, respectively.289
When one or two strongly adsorbing components are
involved, there is no correlation between the permselectivity
and the separation factor. For gas mixtures containing
strongly adsorbing components, the separation factor strongly
depends on the operation conditions, that is, temperature and
pressure.289 Molecular simulations and Maxwell-Stefan
(M-S) modeling of multicomponent diffusion through
zeolite pores indicate that, in some mixtures, slower larger
molecules inhibit the diffusion of faster smaller mol-
ecules.290,291 In addition, detailed studies by several groups
have been reported over the last 5 years292-296 and reviewed
by Sholl last year.297
4.3. Defect Site Diffusion/Nonzeolitic Pores

Polycrystalline zeolite membranes contain transport path-
ways within the intercrystalline regions, or nonzeolite pores.
The synthesis procedure, the type of zeolite, and the
calcination conditions affect the number and size of the
nonzeolite pores. Molecules which interact with nonzeolite
pores have different adsorption and diffusion properties from
those in zeolite pores. The differences, however, are difficult
to quantify because of the irregularities in both shape and
size of the nonzeolite pores. Usually, only nonzeolite pores
that are larger than the zeolite pores are considered, but
nonzeolite pores have a size distribution and pores smaller
than the zeolite pores may also affect flux and selectivity.254
Transport through nonzeolite pores that are larger than zeolite
pores has contributions from both surface diffusion and
Knudsen diffusion, and it might also have viscous flow
contributions. Knudsen diffusion requires that the pores are
smaller than the mean free path of the diffusing molecules.298
Viscous flow requires a pressure gradient across the mem-
brane and sufficient interactions between diffusing molecules
that their motions are driven by the pressure gradient.299
4.4. Thin Films

Figure 7. Cross-sectional SEM photographs of (a) a nonoriented Recent advances in preparing thin zeolite membranes have
B-ZSM-5 zeolite membrane on an R-Al2O3 coated SiC porous dramatically increased gas permeation fluxes while maintain-
support;192 (b) an oriented silicalite-1 membrane on a silica coated ing good selectivities. Recently, ultrathin silicalite-1 mem-
R-Al2O3 porous support223 (Part b is from ref 225 (http://www- branes with a thickness of 0.5 µm were made, and they had
.sciencemag.org). Reprinted with permission from AAAS.); and (c)
a nonoriented silicalite-1 MFI membrane on tubular R-Al2O3 light gas fluxes that are 1 to 2 orders of magnitude higher
substrates (Pall Corp., New York).241 than those of other silicalite-1 membranes reported in the
literature.253 In another report, Lai et al.225 prepared 1 µm
thick oriented silicalite-1 membranes that performed signifi-
cantly better for xylene isomer gas-phase separations than
(temperature and pressure) on both sides of the membrane. previously reported membranes. They obtained p-/o-xylene
The maximum value of flux with respect to temperature can separation factors as high as 500 with a permeance of 2 ×
be observed for strongly adsorbing gases, because the 10-7 mol/(m2 s Pa) at 200 °C.
apparent activation energy is the sum of adsorption heat
(negative) and diffusion activation energy (positive).257,289 4.5. Zeolite Membrane Modification
The temperature of maximum flux increases with the
adsorption strength of the substance.252 For permeation of In an effort to further improve zeolite membranes, surface
binary gas mixtures, when both components are nonadsorb- modification techniques have been developed by a number
ing, the separation factor is R: of research groups.71 The majority of the techniques are post-
treatment methods that include inorganic silylation to was 24, and the separation selectivity for an equimolar H2/
decrease pore size300 and to increase hydrophobicity301,302 and CO2 mixture was 9.7 at 35 °C.
defect treatments to fill nonzeolite pores by chemical vapor More recently, we (Nenoff et al.) explored using zeolite
deposition (CVD),283 atomic layer deposition (ALD),303 or membranes for the separation of hydrogen from multicom-
coking.304,305 Recently, we (Nenoff et al.) reported on a new ponent reforming streams.310-312 Using methods developed
method of online membrane modification by carbonization by Dong et al.,185 we synthesized silicalite-1 membranes and
of 1,3,5-triisopropylbenzene in the feed stream, which was tested their H2 separation abilities with varying temperatures
found to be effective for reducing the MFI intercrystalline (70-300 °C) and feed compositions.313 The composition of
pores and improving the PX separation (see Figure 7c).306 the dry stream was H2, CO2, CO, CH4, and H2S in the ratio
For an eight-component mixture containing hydrogen, meth- 70.8:8.7:5.79:14.69:0.03; the wet stream was H2, CO2, CO,
ane, benzene, toluene, ethylbenzene (EB), PX, MX, and OX, CH4, H2S, and H2O in the ratio 50:10:6:4:0.02:30. At lower
a PX/(MX + OX) selectivity of 7.71 with a PX flux of 6.8 × temperatures in both experiments, H2 had low permeation
10-6 mol/(m2 s) was obtained at 250 °C and atmospheric due to pore blockage by adsorbing components such as H2O,
feed pressure (87 kPa). CO2, CH4, and CO. H2 permeance increased with temperature
throughout the range 70-300 °C with a separation factor
The silylation method of modifying the effective pore varying from 0.13 to 0.4. However, the H2 separation value
opening of a zeolite was first reported by Masuda et al.300 for the five-component stream increases to 2 when water is
In this method, methyldiethoxysilane (MDES) compounds not included, with permeances around 3 × 10-11 mol/
are preadsorbed on active sites within the MFI zeolite, and (m2 s kPa).
then they are catalytically cracked, leaving coke that contains
Si atoms on the active sites. After calcination, mono-SiO2 4.6. CO2 Sequestration in H2 Separations
units are formed on active sites, thereby reducing the size
of the pores. After the silation modification of the membrane, CO2 separation is one of the most studied applications for
a mixture of varying gas ratios of H2/N2 was tested (fraction FAU-type zeolite membranes314,315 due to its significance,
of H2 in retentate gas: H2/(H2 + N2) ) 0.2-0.8; 110 °C, such as CO2 capture for carbon sequestration, natural gas
101.9 kPa steady-state pressure). The separation factor of purification, and separation of product streams from water-
H2 was calculated at 90-140 for the treated membrane. This gas-shift (WGS) reactions for hydrogen production, to name
is about 50 times larger than that of the fresh membrane a few. We (Dong and Nenoff et al.) investigated FAU
(1.5-4.5). Similar results were obtained for mixture gases membranes for the purification of CO2 from 50/50 mixtures
of H2 and O2 (separation factor ) 110-120).300 of CO2/N2 under dry and moist conditions in the temperature
range 23-200 °C at atmospheric pressure.316 At room
This method was borrowed and applied307 to modification temperature, the CO2 selectivity was about 31.2 for the CO2/
of B-ZSM-5 and SAPO-34, whose pores are approximately N2 dry gas mixture with a CO2 permeance of 2.1 × 10-11
0.4 nm and, thus, too small for the silyation compound to mol/(m2 s kPa). The addition of water to the stream
penetrate. The MDES reacted in the B-ZSM-5 pores and significantly enhanced the CO2 selectivity at 110-200 °C
reduced their effective pore diameter, and their H2 selectivity but drastically lowered the CO2 selectivity below 80 °C. At
greatly increased. The H2/CO2 separation selectivity at 473 200 °C, with increasing water partial pressure, the CO2
K increased from 1.4 to 37, whereas the H2/CH4 separation selectivity increased and then decreased after reaching a
selectivity increased from 1.6 to 33. Though silylation maximum of 4.6 at a water partial pressure of 12.3 kPa.
decreased the H2 permeances in the B-ZSM-5 membranes, In another study, Noble and Falconer have shown that their
at 673 K, the H2 permeance increases and the H2/CO2 silica/aluminophosphate (SAPO-4) zeolite membranes can
separation selectivity was 47. In contrast, MDES does not be made and used to separate CO2 from CH4 under a variety
fit into SAPO-34 pores, but silylation apparently decreased of pressures and temperatures, with high selectivities at 3.04
the pore size of the nonzeolite pores in the SAPO-34 MPa and 50 °C.317 Permselectivity for H2 by zeolite
membranes. After silylation, the H2 permeances and H2/CO2 membranes from more complex eight-component simulated
and H2/N2 separation selectivities were almost unchanged refinery gas steam has also been reported.185 The steam
in the SAPO-34 membranes because H2, CO2, and N2 included hydrogen (≈84 mol %) and light hydrocarbons
permeate mainly through SAPO-34 pores. In contrast, the (C1-C4, 7.5-0.3 mol %). An R-alumina-supported poly-
H2/CH4 separation selectivity increased from 35 to 59, and crystalline MFI zeolite membrane was tested between 25 and
the CO2/CH4 separation selectivity increased from 73 to 110, 500 °C and at feed pressures of 0.1-0.4 MPa. The zeolite
apparently because CH4 permeates mainly through non- membrane showed excellent separation properties for rejec-
SAPO-34 pores. tion of hydrogen from the hydrogen/hydrocarbon mixture
The synthesis of small-pore zeolite membranes has also at <100 °C. At room temperature and atmospheric pressure
been pursued for the separation of small light gas molecules. on both feed and permeate sides, the hydrogen permeation
rate is almost zero, while the hydrocarbon permeation rate
Zeolite A membranes have shown H2 permeances ranging
is 2-4 × 10-7 mol/(m2 s kPa). At 500 °C, the zeolite
from 10-10 to <10-11 mol/(m2 s kPa), with a maximum of
membrane becomes permselective for hydrogen over hy-
H2/N2 separation selectivity of 4.8 between 35 and 125 °C.262
drogen (C1-C4), with a separation factor over 3.
Changes in the charge balancing cation result in changes in
the H2 permeance and followed the order of K < Na < Ca,
which is consistent with the order of the pore size of the A
4.7 Manufacturing
zeolite.308 The highest H2/N2 separation selectivity of 9.9 was Zeolite membrane manufacturing is still an industry in the
obtained for a KA membrane. For AlPO4-5 membranes (pore making. Most membranes are still fabricated in lab scale sizes
size of 0.73 nm), the H2 permeance was 2 × 10-10 mol/(m2 and quantities. Furthermore, the technology needs to be able
s kPa) at 35 °C.309 The H2/CO2 ideal selectivity (RH2, CO2) to commercialize large-scale continuous films without in-
tercrystalline pores for successful high separation selectivity. of CO2 could be sequestered and ultimately removed
However, recently, there has been a big leap forward in the (presumably through recompression) from coal gasification
commercialization of this technology. Currently, only Mitsui products to prevent its escape into the atmosphere.319-321 This
Engineering and Shipbuilding Company in Japan has com- process would require a rejective type H2 membrane which,
mercialized a process using zeolite membranes.282 It is a unlike the selective membranes, must possess decreased H2
pervaporation process using NaA zeolite membranes for diffusion but increased solubility and selectivities (Table 14).
organic dehydration. The membranes are 20-30 µm thick Nonpolymer carbon-based membranes are rejective H2
on porous, tubular ceramic supports. The plant processes membranes that can be categorized into three classes: carbon
alcohols up to 530 L/h with separation factors up to 10,000.
membranes, carbon molecular sieve membranes (CMSMs),
Manufacturers of zeolite membranes listed by Bowen et al.282
and carbon nanotubes (CNTs). The separation ability of each
include Smart Chemical Co., Ltd and Christison Scientific,
class of material is dependent both on the chemistry of the
both in the U.K., and Artisan Industries Inc., USA.
material and on the fabrication/implementation (ie., module
Current estimated costs per zeolite membrane gas separa-
design). As a result, there are serious advantages and
tion module have been approximated around $400/ft2.
Though this is an estimate, it compares favorably with metal disadvantages to each.
membranes and modules (≈ $1500/ft2).318 It is safe to assume When compared to polymeric membranes, the cost of
that, once in mass use and production, those costs will decline carbon-based membranes is 1 to 3 orders of magnitude
significantly to about $100/ft2, allowing zeolite membranes greater per unit area. Only when they achieve higher
to compete both in economics and on performance. performance than polymeric membranes is the high invest-
Significant progress has been made in the synthesis or ment cost justified. The most popular precursor for carbon
various types of crystalline zeolite membranes. Good quality membranes is currently polyimide, which contributes largely
zeolite membranes can be prepared by several methods, to the high manufacturing cost. Therefore, attempts have been
including in situ synthesis, secondary growth, and vapor- made to use less expensive starting materials, such as
phase transportation. To be considered useful for gas polyacrylonitrile. However, the performance of these mem-
separations applications, these membranes will have to be branes still remains inferior.322
synthesized without macropore-sized defects or pinholes. Carbon membranes have been prepared in both unsup-
Furthermore, the ability to surface modify zeolite membranes ported and supported forms of materials. Typical precursors
(both internal pore surfaces and external surfaces) through are organic polymers that are converted to pure carbon
silation or carbonization allows for the fine-tuning in materials by treatment at high temperature in an inert
selectivity. Gas separations246 through the membranes are
atmosphere (carbonization). Among the unsupported mem-
governed by mechanisms of preferential adsorption, selec-
branes, capillary tubes or hollow fibers and flat membranes
tively configurational diffusion, or molecular sieving. Gas
have been prepared. Supported membranes are flat or tubular
permeation through these microporous inorganic membranes
and are grafted onto macroporous materials.323 The major
is an activated process that can be predicted through gas
diffusion theories (i.e., Maxwell-Stephans equations that disadvantage with both types is that they suffer from
govern gas permeation and separation). mechanical performance problems, specifically brittleness.
Zeolite membranes have chemical, mechanical, and ther- Brittleness presents a problem for unsupported membranes,
mal stability not observed in many types of membranes. The whereas multiple polymer deposition and carbonization
trends in zeolite membrane research show clearly the cycles must be repeated to obtain crack-free supported
improvements in selectivity, fabrication methodology, and membranes. The complexity of the latter procedure presents
energy-production applications. In the near future, the ability an impediment to practical applications.
to inexpensively fabricate these membranes for tuned Different configurations exist for unsupported and sup-
selectivity will put them at the forefront of separations ported carbon membranes. Unsupported membranes have
technology. For the time being, their stability at high three different configurations: flat (film), hollow fiber, and
temperatures and their ability to be regenerated without loss capillary. Supported membranes can adopt two configura-
to performance make them interesting candidates for stream- tions: flat and tube. Detailed descriptions of these two
lined hydrogen production via natural gas reformation. categories can be found in Ismail and David’s review.324 In
most cases, supported polymeric membranes are produced
5. Carbon-Based Membranes because of the poor mechanical stability (i.e., brittleness) of
unsupported carbon membranes. In making the supported
Hydrogen rejection and contaminate permeation is yet carbon membranes, various options are available for coating
another method which is being intensely explored as a new the supports with thin polymeric films, such as ultrasonic
approach for H2 purification, primarily using carbon-based
deposition,325,326 dip coating,327 vapor deposition,328 spin
membranes. Because of hydrogen’s low critical temperature
coating,329 and spray coating.330
and small kinetic diameter, a rejective membrane process
allows for H2 purification via contaminant permeation with The six major steps in carbon-based membrane fabrication
respect to H2. Such rejective membranes have the very are briefly detailed in this section of the review but are more
distinct economic advantage that maintaining/collecting H2 fully described elsewhere.324 These six steps are precursor
in the retentate reduces the need for costly H2 recompression selection, polymeric membrane preparation, pretreatment of
steps, though it potentially adds a CO2 compression step. the precursor, pyrolysis process, post-treatment of pyrolyzed
Therefore, this may only find limited utility in certain membranes, and module construction. The manipulation of
applications, such as refinery use of H2, where only medium the pretreatment variables, pyrolysis process parameters, and
or low pressure hydrogen is required. Furthermore, in 2003 post-treatment conditions was shown to provide an op-
a project was started by the United States Department of portunity to enhance the separation performance of carbon
Energy, called FutureGen, which proposed that large amounts membranes in the future.
5.1. Carbon Membrane Preparations applications of membranes, it is preferable to fabricate a

module with an asymmetric structure and capillary or hollow
Carbon membranes can be produced through the carbon- fiber configurations in order to increase the rate of permeation
ization or pyrolysis process of suitable precursor carbon of the products.339 In general, the characteristics of modules
containing materials, such as thermosetting resin, graphite, must be considered in all system designs.
coal, pitch, and plants, under inert atmosphere or vacuum.331
Numerous synthetic precursors have been used to form 5.4. Selective Surface Flow Membranes
carbon membranes, such as polyimide and derivatives,
polyacrylonitrile (PAN), phenolic resin, polyfurfuryl alcohol Rao and Sircar introduced new membranes in 1993 for
(PFA), polyvinylidenechloride-acrylate terpolymer (PVDC- the separation of gas mixtures that they called selective
AC), phenolformaldehyde, cellulose, and others. A thermo- surface flow membranes (SSF).340-342 These membranes were
setting polymer can often withstand high temperatures332 and synthesized by coating a macroporous graphite disk with
neither liquefies nor softens during any stage of pyrolysis. layers of polyvinylidene chloride-acrylate terpolymer latex
Suitable precursor materials for carbon membrane production which contained 0.1 mm polymer beads in aqueous solution.
will not cause any pore-holes or cracks to appear after the After deposition of each layer, the disk is dried and heated
pyrolysis step.333 to 1000 °C, which produces a porous carbon membrane
Pretreatment methods can be divided into physical and through sequential cross-linking and carbonization of the
chemical methods. The polymer membrane preparation underlying polymer. The resulting thickness of each layer
involves the physical pretreatments of stretching or drawing was ∼0.5 mm, and between two and five layers were applied,
hollow fiber membranes prior to pyrolytic processing. In to yield a complete module between 1.0 and 2.5 mm. The
contrast, chemical pretreatments involve chemical reagents, permeability for H2 in a mixture with hydrocarbons was
which are applied to the polymeric precursor to alter its reduced by several orders of magnitude over that of pure
properties or behavior. Sometimes the precursor is subjected hydrogen. These membranes became promising for H2
to repeated and varied pretreatment methods to achieve the separation because the hydrocarbon selective adsorption
desired properties in a carbon membrane. Perhaps the most hindered pore diffusion by hydrogen.
important and popular pretreatment method employed has Further fine-tuning of the pore structure can be facilitated
been the oxidation treatment.324 through various synthetic methods. They include an increase
Pyrolysis, the process in which a suitable carbon precursor in the oxidation time and temperature, allowing for controlled
is heated in a controlled atmosphere to its pyrolysis tem- increase of the pore size and permeabilities of all compo-
perature, is conventionally used for the production of porous nents, and a variation in the kinetic selectivity.343 Excessive
carbon fibers, and it causes the product to have a microporous oxidation, however, may render the pores too large to be
structure. Control over the molecular dimensions of these selective.
micropores and the subsequent molecular sieving properties Optimization of these membranes has led to even further
is one of the primary methods being actively researched.334 gas separation advantages, such as hydrogen gas purification.
The pores vary in size, shape, and degree of connectivity, However, this process works opposite to other inorganic
depending on the morphology of the organic precursor and membranes. Since adsorption occurs on the high pressure
the chemistry of its pyrolysis. The pore structure is essentially side, the partial pressure of the component to be adsorbed
retained and can be controlled selectively by adjusting the can be low. The partial pressure gradient across the mem-
various process parameters (i.e., chemical pretreatment, brane does not need to be high to attain separation, since
precursor identity, etc.).334 While pyrolytic treatment methods the driving force for mass transfer across the membrane is
do have a definitive effect on the performance behaviors of the difference in the concentration of the adsorbed species
carbon membranes, the temperature is almost always between (i.e., concentration gradient). It should also be noted that the
500 and 1000 °C.335,336 activation energy for surface diffusion is typically lower than
that for transport across the membrane. Furthermore, adsorp-
5.2. Carbon Membrane Post-treatment tion decreases the effective pore volume, hindering the
Knudsen diffusion mechanism of nonadsorbed molecules,
As a result of pyrolytic processing, polymeric membranes which would ultimately diminish the degree of separation.
are transformed into carbon membranes with varying degrees Separation processes utilizing SSF membranes are based on
of porosity, structure, and separation properties that depend the adsorption properties of the component(s), and larger or
to an extent on the carbonization conditions employed. In more polar species can be separated from the mixture. For
most cases, it is found to be an advantage that the pore example, in the methane reforming process, hydrogen
dimensions and distribution in the carbon membrane can be remains on the high-pressure side of the membrane while
finely adjusted by simple thermochemical treatment(s) to unwanted species are passed through, eliminating subsequent
meet different separation needs and objectives.331 The various compression of the H2 gas for many applications. A further
post-treatment methods include postoxidation, chemical advantage of the technique lies in the fact that adsorption
vapor deposition (CVD), postpyrolysis, and coating. These capacity and selectivity increase with decreasing temperature,
post-treatments can also repair the defects and cracks that reducing operational cost. This is the reverse of molecular
exist in the carbon membrane. sieving.342
Furthermore, the pressure swing adsorption (PSA) process
5.3. Carbon Membrane Module Construction has been integrated with SSF membranes for enhanced
The geometry and installation of a membrane in a suitable performance in the extraction of hydrogen from steam-
device (i.e., a module) are also important337 to its separation methane reformer gas.344 In the commercial production of
abilities. The selection of a membrane module is mainly hydrogen, this reformer gas is subjected to water gas shift
determined by economic considerations, including all the cost (WGS) reactions followed by hydrogen purification by PSA.
factors plus the cost of the module.338 For commercial Typical PSA cycles consist of alternating pressurization and
depressurization of feed gas and hydrogen enriched gas to Subsequent mild oxidation, known as activation, may
augment hydrogen recovery. If a SSF membrane is used for enhance the development of new pores and the widening of
purification of the waste gas of the PSA process, hydrogen existing ones.347 It is possible to interrupt the pore develop-
recovery can be increased from 78% to 85% in the integrated ment process when the pore widths have reached molecular
process. Due to a reduction in the compression duty and the dimensions. Under these conditions, however, the capacities
membrane area, the process is particularly economic when are generally low, as the overall pore dimension remains low.
the PSA waste gas is first fractionated, and only the hydrogen Another problem may be the low degree of carbonization.
rich portion is used as feed for the SSF membrane.344 Alternatively, carbon molecular sieves can be prepared from
Viera-Linhares and Seaton used molecular dynamics activated carbons with high adsorption capacities. Tuning
calculations343 and critical path analysis345 to model the of pore opening sizes is possible through chemical vapor
separation process in selective surface flow membranes for deposition (CVD) processing of certain organic compounds,
methane/hydrogen mixtures. They showed that pore width thus increasing the selectivity. Nevertheless, the deposition
is critical for the separation process, since it controls control at pore mouths is difficult to achieve, and often the
adsorption capacity and transport properties through the process results in a shift of the pore size distribution to overall
material. They defined three distinct regions, characterized smaller values.347
by pore size. If the size is smaller than 6 Å, a sieving effect MSCMs produced through pyrolysis of polymeric materi-
results and separation occurs solely based on molecular size. als have proved very effective for gas separation in adsorp-
Between 6 and 10 Å, selective adsorption of methane tion applications.348-351 Molecular sieving carbon can be
occurred with very little dependence on pressure, due to the obtained by pyrolysis of many thermosetting polymers, such
pores being filled almost to capacity. Maximum selectivity as poly(vinylidene chloride) (PVDC), poly(furfuryl alcohol)
was achieved at 7-8 Å, while permeability was optimal at (PFA), cellulose, cellulose triacetate, saran copolymer, poly-
9 Å. Larger pore sizes gave rise to a regime in which methane acrylonitrile (PAN), and phenol formaldehyde, and various
was preferentially adsorbed on the pore walls and hydrogen coals, such as coconut shell.348 The pore dimensions of
occupied the center of the pore. It is less effective for H2 carbon depend on the morphology of the organic precursor
purification, as the hydrogen can diffuse through the low- and the chemistry of pyrolysis. In particular, nongraphitizing
density region of the porous network. Further research carbons were extremely specific and adjustable by mild
showed that, for the separation of methane from hydrogen,345 activation and sintering steps to the discrimination range
both species pass through distinct pore subnetworks of the 2.8-5.2 Å.352,353 Pyrolysis of thermosetting polymers typi-
membranes, with methane populating the larger pores. The cally yielded an exact mimic of the morphology of the parent
importance of this work lies in the fact that a selective material without proceeding through a melt or softening
blockage of smaller pores would reduce the permeability to during any stage of the pyrolysis process.352 Koresh and
hydrogen and enhance the effectiveness of the separation. Soffer346,352,353 successfully prepared crack-free molecular
sieving hollow fiber membranes by carbonizing cellulose
5.5. Disadvantages of Carbon Membranes hollow fibers. They have shown the dependence of perme-
abilities and selectivities on temperature, pressure, and extent
Though carbon-based membranes show much promise in of pore for both adsorbing and nonadsorbing permeates.352,353
the area of light gas separations, they still possess problems They recognized that the adsorption followed a sequence of
that influence their introduction to market. First, they are CO2 > H2 > N2 > Xe > SF6 for the molecules studied and
very brittle and fragile. Therefore, they require more careful that hydrogen and methane permeabilities exhibited a
handling.324,331,346,374, This may be avoided to a certain degree maxima between 600 and 700 °C, which they attribute to a
by optimizing precursors and preparation methods.331 Second, molecular-sieving permeation mechanism. However, those
their difficulty to process results in high expenses to membranes would lack sufficient mechanical strength and
fabricate.374 Carbon membranes require a prepurifier for durability for practical applications.
removing traces of strongly adsorbing vapors, which can clog
up the pores due to the transport being through a pore system Fuertes and Centeno323 reported in 1998 the preparation
rather than through the bulk system. This is typical of many of a flat, asymmetric carbon membrane supported on a
industrial adsorption separators. This problem may be macroporous carbon support, made by carbonization of
avoided by operating at sufficiently high temperatures.346 agglomerated graphite particles blended with a phenolic
Third, they only demonstrated high selectivities for certain resin.340 The support had a porosity of 30% and a mean pore
gas mixtures, with gases of molecular sizes smaller than 4.0- diameter of 1 mm. In an effort to stop translation of support
4.5 Å. Carbon membranes are not suitable to separate certain cracks to the membranes, the support was coated with an
industrially relevant gas mixtures, such as branched pentanes intermediate carbon layer made from fine graphite particles
versus linear hydrocarbon molecules or gas-vapor mixtures blended with a polyamide-imide resin. The disk was then
(i.e., H2/hydrocarbon).374 carbonized and polished before deposition of the polymeric
membrane with an asymmetric structure through a polyamic
acid precursor. Gelling of the polymer film, drying, and
5.6. Molecular Sieving Carbon Membranes subsequent thermal treatment led to carbonization. The
Molecular sieving carbon membranes (MSCMs) are the resulting membrane had better permselectivities for carbon
second class of carbon-based membranes on which we will dioxide and methane than the multilayer SSF.323,340-342
focus. These membranes are able to achieve both capacity However, the permeabilities of the pure gases decreased, and
and selectivity, primarily due to having pore openings of no data on hydrogen have yet been reported.
optimal molecular dimensions and high pore volume(s). The In a recent study by Hatori,354 H2 and CO2 data were
broad variety of carbonaceous precursors and different reported for their polyimide molecular sieving membranes.
processing procedures allow for a wide range of variability. Precise pore size and structure control were reportedly
The development of porosity begins in the pyrolysis process. achieved by adjusting the heat-treatment temperature for
carbonization. They used commercially available polyimide

films that were subsequently carbonized under a flow of dry
argon. Their data showed that, with increasing carbonization
temperatures, the micropores’ size of their membranes
decreased. After treatment at 1000 °C, the H2/CO selectivity
increased to above 1000. Above 1000 °C, an obvious
decrease in CO2 adsorption was observed due to the decrease
in pore size to below that of a CO2 molecule. The authors
concluded that a surface diffusion mechanism was respon-
sible for the achieved separation.
The commercial availability of activated polyaramid
carbon fibers led Villar-Rodil et al.347 to investigate their
gas separation properties. These fibers have narrow pore size
distributions. After chemical vapor deposition of benzene
with simultaneous pyrolysis at 800 °C, these membranes
were appropriate for N2/O2 separation based on capacity and
selectivity. It is anticipated that a careful control of the pore
mouth treatment might result in membranes suitable for the
separation of hydrogen purification.
5.7. Carbon Nanotubes

The discovery355,356 of carbon nanotubes (CNTs) in 1991 Figure 8. (A) An as-grown, dense, multiwalled CNT array. Scale
stimulated a great deal of interest due to their unusual bar, 50 µm. (B) Schematic of the target membrane structure. With
mechanic and electronic properties. With a hundred times a polymer embedded between the CNTs, a viable membrane
structure can be readily produced, with the pore being the rigid
the strength of steel and their high gas uptake capacity, inner-tube diameter of the CNT.370 (From ref 370 (http://www-
applications ranging from lightweight fuel tanks to cables .sciencemag.org). Reprinted with permission from AAAS.)
for elevators into space were envisioned.357
CNTs can be thought of as graphite sheets that have been tubes. O’Connell et al.360 introduced a method to create a
wrapped into a tube and capped at each end with half a stable suspension of individual SWNTs using ultrasound.
fullerene (C60) sphere.340 Their electronic properties depend Most methods for separation of metallic and semiconducting
on tube diameter and helicity.358,359 Depending on the degree SWNTs require initial debundling or solubilization of the
of twist along their length, nanotubes encompass many nanotubes, which is generally achieved through surfactant
structural types ranging from the chiral “armchair” (metallic) aided dispersion by ultrasonication or chemical functional-
over other chiral types to the achiral “zigzag” (semiconduct- ization.361
ing) tubes. Oxidation procedures allow the selective removal SWNTs possess several possible adsorption sites. While
of end caps, yielding open-ended CNTs. Tube diameters lie the binding energy is highest for interstitial channel sites,
typically in the range of several angstroms to a few the surface area of pore sites is largest. These sites, however,
nanometers. Besides the single-walled variety (SWNTs), are only accessible when the tube walls or caps are broken,
multiwalled CNTs (MWNTs) exist which are composed of as typically happens under harsh oxidation conditions.362
concentric layers of single-walled tubes separated roughly Since multiwalled tubes rarely bundle, groove and interstitial
by the same distance as the planar sheets in graphite (ca. channel sites do not exist. The interlayer spaces may be
3.4 Å). Here, each individual layer can have a different possible adsorption sites for small molecules.363 The pores
helicity. of MWNTs may be occupied, but the volume fraction of
Carbon nanotubes are typically synthesized by one of three these sites is very small compared with SWNTs, considering
major production methods: laser ablation, chemical vapor their thick walls.
deposition, and electric arc discharge. Metal catalysts such Recent breakthroughs in molecular dynamics (MD)
as Fe, Co, Mo, and Ni and combinations thereof are simulations364-368 and membrane fabrications369,370 have put
necessary to grow SWNTs and are generally also used for carbon nanotubes at the forefront of carbon-based mem-
making MWNTs. These metallic particles remain as impuri- branes. Because the walls of the nanotubes are considered
ties in the final product. In addition, other forms of carbon very smooth as compared to the other materials (i.e., zeolite
such as amorphous carbonaceous materials, graphitic par- membranes), they have been predicted by Sholl and Johnson
ticles, graphitic carbon fibers, nano-onions, and the like are to contain rapid transport rates for gases.364,365 Holt et al.369
formed. and Hinds et al.370 fabricated single- and double-walled
In order to remove these impurities, the CNTs are nanotube membranes (1.3-2 nm diameter), and multiwalled
subjected to post-treatment. Most purification techniques nanotube (6-10 nm diameter) membranes, respectively
exploit the fact that the nanotube carbon network is extremely (Figure 8).
stable and usually not affected by oxidation, while other Diffusion of single gas molecules (both light gases and
forms of carbon are readily oxidized to CO2. Simultaneously, hydrocarbons) was studied by both groups. Rapid transport
metal particles are converted into their oxides, which can of gases was recorded. Transport rates by Holt et al.369 were
then be dissolved by acid. It should be noted that, in this 1-2 orders of magnitude larger than Knudsen diffusion
process, tubes may be cut into shorter pieces, possibly leaving predictions (Figure 9).
tube ends opened and tube walls damaged or exfoliated. Multiple gas diffusion experiments have still not been
Due to van der Waals forces between the tubes, SWNTs reported. However, MD calculations have been reported for
typically aggregate in the form of bundles with 50 or more CH4/H2 mixtures.367 They predict a preferential adsorption
Table 10. Hydrogen Separation Ability of First-Generation,

Commercial Membranes for Gas Separations
membrane
material(s) developer H2/N2 H2/CO H2/CH4 ref
polysulfone
Monsanto 39 23 24 375
silicone rubber
polyimide Ube 35.4 30 484
cellulose acetate Separex 33 21 26 372, 376
Table 11. Hydrogen Purity Required in Industry380

hydrogen
purposes purity (%)
rocket engine fuel
99.999999
semiconductor manufacture
Figure 9. Pure gas selectivity (defined as permeability relative to polymer electrolyte fuel cell
He) data for sub-2-nm DWNT (double-walled nanotube; triangles) on-site hydrogen generating equipment
99.99
and MWNT (circles) membranes. Open symbols denote nonhy- hydrodesulfurization 90
drocarbon gases (H2, He, Ne, N2, O2, Ar, CO2, Xe); solid symbols
adjustment of molecular weight distribution 70-80
denote hydrocarbon gases (CH4, C2H6, C3H6, C4H6, C4H8). The solid
fuel gas 54-60
line is a power-law fit of the nonhydrocarbon gas selectivity data,
showing a scaling predicted by the Knudsen diffusion model
(exponent of -0.49 ( 0.01). The dashed line is a power-law fit of evolved (Table 10), they became more commonly utilized
the hydrocarbon gas data, showing a deviation from the Knudsen
model (exponent of -0.37 ( 0.02). The inset shows the full mass on commercial scales for various recovery processes that
range of the nonhydrocarbon gas data, illustrating agreement with eventually included H2 reclamation from recycled refinery
the Knudsen model scaling.369 (From ref 369 (http://www.sciencema- gas.377,378
g.org). Reprinted with permission from AAAS.) One of the primary polymer-based technologies for energy
production is proton exchange membrane (PEM)-based fuel
cells, which convert the chemical energy of H2 directly and
of CH4 over H2, with selectivities greater than 10, which is efficiently to electrical energy with dramatically reduced
much higher than Knudsen diffusion predictions (<3).368 emissions of greenhouse gases (hydrocarbons, CO, CO2,
NOx, and SOx).379 However, for these fuel cells to become
6. Polymer Membranes for H2 Separations widely applicable, a well distributed supply of hydrogen is
required.
Organic polymers have a long rich history which can be
While polymer-based H2 separation membranes are ca-
traced back to the 1840s with the discovery of nitrocellulose
pable of H2 production at very high purities, palladium
and the vulcanization of polyisoprene (natural rubber). Since
then, the amazing variety and ubiquitous presence of composites, inorganic membranes, or more advanced separa-
polymeric materials in modern society has had an enormous tion processes are required to produce the highest of purities
impact on virtually all facets of large-scale industry and the for many applications (Table 11).380 However, it is quite clear
global economy. It follows then that these materials be that further scientific developments are needed to fully realize
considered and exploited as a potential cornerstone of the the hydrogen economy, since the current technologies do not
proposed hydrogen economy. Membrane separations have yet meet the performance criteria set forth by the U.S.
been considered for a variety of gas separations since at least Department of Energy. Gas separation membranes have
the 1950s;371 however, it was not until the mid-1970s that significant potential for application in this growing process.
DuPont pioneered the use of small-diameter hollow fibers
as a viable gas separation membrane.372,373 Among the 6.1. Dense Polymeric Membranes
numerous industrial targets, H2 recovery, separation, and Dense type polymer membranes can be divided into glassy
purification remains one of the most highly prized yet most
and rubbery polymeric membranes. The former have higher
elusive applications of polymer separation membranes.373
selectivity and lower flux, whereas the latter have higher
Typical strategies for these separations include variation of
the H2/CO ratio in synthesis gas (commonly referred to as flux but lower selectivity.19 According to Kluiters,19 operating
syn-gas), removal of H2 from hydrocarbon streams, and, temperatures for polymer membranes are ∼100 °C. Several
finally, removal from purge gases in ammonia production key advantages are that they possess the ability to cope with
and other large-scale/commercial petrochemical pro- high-pressure drops and low cost. However, limited me-
cesses.372-376 This particular section of the review examines chanical strength, relatively high sensitivity to swelling and
various polymeric membranes for H2 separations, provides compaction, and susceptibility to certain chemicals such as
comparative data, and addresses active areas in need of hydrochloric acid (HCl), sulfur oxides (SOx), and CO2 make
expansion. polymeric membranes less attractive. Polymer membranes
While DuPont’s hollow fibers were groundbreaking, their used for separation processes operate according to the
low permeance was not efficient or productive enough to solution diffusion mechanism. An in depth study on polymer
provide economically sustainable gas separations. Several membranes can be found elsewhere.19 Table 14 gives
years later, this performance shortfall was addressed by hydrogen permeabilities (at 27 °C temperature and 206.8 kPa
Monsanto when they developed asymmetric polysulfone feed gas pressure) for the selected polymer membranes and
hollow fiber membranes for H2 recovery from ammonia the selectivities for nitrogen (N2), methane (CH4), and CO2.
purge gases.375 The next advance was introduced by Separex The polystyrene shows the best combination of hydrogen
in the form of spiral wound cellulose acetate membranes permeability and selectivities for N2, CH4, and CO2.6,19,372
(Separex) for H2 and natural gas separations.376 As the Polymeric membranes are separated into porous and
transport properties of polymeric H2 separation membranes nonporous, and the hydrogen transport mechanisms of these
Table 12. Common Hydrogen Sources and Their Impurities380 Table 13. Properties of Several Common Gas Molecules
hydrogen source impurities ka σb TCc
compound molecule (×1010 m) (×1010 m) (°C)
electrolysis of water CH4, O2, N2, CO2, and CO
steam reforming gas CO, CO2, and CH4 helium He 2.6 2.551 -277.0
petroleum refining C1-C6, and BTXa ammonia NH3 2.6 2.900 132.4
ammonia purge gas NH3, N2, and CH4 water H2O 2.65 2.641 374.2
coke oven gas CH4, N2, BTX,a CO, CO2, and O2 hydrogen H2 2.89 2.827 -240.0
carbon dioxide CO2 3.3 3.941 31.0
a
BTX ) benzene, toluene, and xylene. carbon monoxide CO 3.73 3.690 -140.3
oxygen O2 3.46 3.467 -118.6
membranes may occur through five different diffusion nitrogen N2 3.64 3.798 -147.0
processes (see Figure 1). If the polymeric membrane is methane CH4 3.8 3.758 -82.8
propane C3H8 4.3 5.118 96.7
porous, then diffusion occurs through mechanisms which BTXd d g5.85 g5.349 288-357
depend largely on pore size and the size(s) of the diffusing a
gas molecule(s): (i) Knudsen diffusion, (ii) surface diffusion, Kinetic diameter calculated from the minimum equilibrium cross-
sectional diameter.485 b Lennard-Jones collision diameter.382 c Critical
(iii) capillary condensation, and (iv) molecular sieving. In temperature.381 d BTX ) benzene (C6H6), toluene (C7H8), and xylene
Knudsen diffusion (i), the diffusing gaseous molecules collide (C8H10).
more frequently with the pore walls than with other diffusing
molecules, thus facilitating differential retention times.381
The idea of an “upper bound”, originally introduced by
With surface diffusion (ii), gaseous molecules adsorb onto
Robeson in 1991,393 is the carefully modulated balance
the pore surfaces (walls) and then move from along a specific
between permeability and selectivity. This proposed upper
decreasing concentration gradient from one site to the
bound provides insight into the maximum selectivity that is
next.382,383 Capillary condensation (iii) occurs under very
attainable for a given membrane permeability while using
specific circumstances when diffusing gas molecules con-
polymeric membranes for a given composition of gases in
dense within a given pore to generate capillary forces which
the H2 feed stream. For example, at the low end of H2
inhibit diffusion rates.384 Finally, the molecular sieving (iv)
permeability, the separation of H2 from a mixture of H2/N2
mechanism is again a specific case where the diffusing gas
has been established by various poly(methyl methacrylate)s
molecules and the pore size are sufficiently close in size to
(PMMs), and at the high end of H2 permeability, poly(1-
require an energy of activation (directly related to molecule
trimethylsilyl-1-propyne) (PTMSP) has been used. Figure 10
size).385
illustrates the substantial amount of research directed toward
In nonporous, or dense, polymeric membranes, transport
more effective gas separation membranes, but Robeson’s
is controlled by the (v) solution diffusion.18,386-388 In the
upper bound hypothesis, which was established over a decade
solution diffusion mechanism, the gaseous molecules absorb
ago, still remains.394-414 This has been further explained by
to the surface, dissolve into the bulk of the polymer
Benny Freeman, who presents415 a clear and physically
membrane, and are finally transported across the membrane
meaningful rationale behind the existence of this upper
by a gradient of chemical potential from the feed stream to
bound. Thus, membranes are now being designed from
the product stream. The equations which govern this
polymers along this upper bound. However, if separations
particular diffusion mechanism are discussed at length in the
exceeding this upper bound are required, then alternative
introduction of this review and in refs 388-390.
membranes (zeolitic, metallic, etc.) and technologies must
be employed.
6.2. Hydrogen Selective Polymeric Membranes The use of cross-linkable polymers is another more recent
Polymeric membranes which are selective for hydrogen methodology which has been used to improve the perfor-
are designed such that the concentration of hydrogen is mance of polymeric membranes.237,416-418 These cross-linking
increased in the product stream (permeate) and the remaining moieties have been shown to provide a selectivity improve-
components of the gas mixture remain in the feed stream or ment for H2 at least ten times the magnitude of their non-
a secondary waste stream (retentate). cross-linked counterparts. However, the complexity of
The variety of H2 sources (feed stock streams) provides a implementing this approach on large industrial scale mem-
modest number of impurities (Table 12) which could branes has yet to be overcome and must be addressed before
potentially interfere with the separation membrane. But as they find widespread commercial use. Although polymeric
Table 13 shows, H2 has the lowest critical temperature (Tc) membrane research in the open literature appears to be
compared to those of most of the other gases, and it has one shifting toward membrane processing419,420 membrane sys-
of the smallest kinetic diameters of any gas molecule. This tems and new supports,421-428 the patent literature still reveals
low Tc indicates lower potential hydrogen solubility while substantial efforts toward the development of new membrane
the small kinetic diameter suggests substantially higher materials.401,429-433
diffusivities.382,391,392 Thus, current research on selective A specific application for polymer membranes is as
polymeric H2 membranes is aimed at the exploitation of high hydrogen rejective membranes (as mentioned above). Rejec-
diffusivities while minimizing the consequential effects of tive membranes use the significantly higher sorption of other
the lower solubilities. Since dense membranes of this type gases to overcome the potential selective preference of the
operate entirely on the solution diffusion mechanism, poly- small size of the hydrogen molecule.434 Rubbery polymers
meric membranes are engineered in an attempt to use these such as polydimethylsiloxane (PDMS) have been shown to
two major components to their advantage. Improving the reduce the diffusion selectivity.435,436 The higher mobility of
performance of polymeric H2 selective membranes is largely the chain structures in these rubbery polymers increases the
based on targeted separation, which rejects more condensable diffusivity of all gaseous species. Since smaller molecules
compounds (impurities) and allows the less condensable H2 are already highly mobile, larger molecules benefit most from
to permeate. this chain mobility, which ultimately causes reductions in
Table 14. Permselective Properties of CO2-selective Polymeric Facilitated Transport Membranes

other CO2 CO2 CO2
membrane gas (kPa) permeancea selectivity ref
sulfonated polystryene-EDAH N2 0.29 4.13 × 10-6 600 449, 450
Nafion-EDAH CH4 0.16 7.99 × 10-7 550 378, 452
Nafion-EDAH H2 101 3.63 × 10-7 6.8 453, 486
Nafion-EDAH H2 1.88 × 10-6 55 379, 454
sulfonated polystryene-divinyl benzene-EDAH N2 0.407 4.97 × 10-8 524 456, 487
EDAH-alginate N2 1 1.05 × 10-8 b 50 457, 488
EDAH-polyacrylate N2 4.76 7.50 × 10-6 4700 458, 459, 489, 490
poly(acrylate-EDAH)/poly(vinyl alcohol) N2 6.18 5.25 × 10-6 1900 380, 460
poly(vinyl alcohol)-amino acid salts H2 76.0 6.38 × 10-7 30 381, 461
poly(ethylenimine)-lithium glycinate H2 76.0 2.80 × 10-6 75 382, 462
polyvinylalcohol-tetramethylammonium fluoride H2 76.0 5.33 × 10-7 19 383, 463
polyvinylalcohol-cesium fluoride H2 4.4 5.96 × 10-7 60 384, 464
cesium polyacrylate-cesium fluoride H2 4.4 6.09 × 10-7 61 384, 464
poly(diallydimethylammonium fluoride) H2 40.0 1.35 × 10-7 81 385, 465
poly(vinylbenzyltrimethylammonium fluoride) H2 4.21 4.52 × 10-7 87 386, 466
poly(vinylbenzyltrimethylammonium fluoride) H2 113.9 2.22 × 10-7 207 18, 467
poly(vinylbenzyltrimethylammonium fluoride)-cesium fluoride H2 4.08 1.93 × 10-6 127 387, 468
poly-2-(N,N-dimethyl)aminoethyl methacrylate N2 4.76 3.75 × 10-7 130 18, 388
poly-(2-(N,N-dimethyl)aminoethyl acrylate-co-acrylonitrile) N2 0.48 1.53 × 10-9 90 389, 491
hydrolyzed polyvinylpyrrolidone N2 1.62 1.27 × 10-5 48.1 390,492
poly(ethylenimine)/poly(vinyl alcohol)- N2 6.59 2.93 × 10-7 230 392,493
poly(vinylamine)-ammonium fluoride CH4 200.0 2.33 × 10-8 1143 393,494
poly(vinylamine)-cesium fluoride H2 4.29 5.03 × 10-7 120 384,464
Biomimetic carbonic anhydrase N2 0.10 1.28 × 10-5 >1000 398,495
a
Permeance in units of m3/(m2 s kPa); literature P0 values converted to permeance using reported membrane thicknesses. b Membrane thickness
not reported, assumed a value of 50 µm.
on poly(amido-amine) dendrimers. This membrane has a

unique CO2 gating property and is predicted to provide
unmatched CO2/H2 selectivity in polymeric materials. The
rejective properties of polymeric membranes have also been
achieved through careful production and processing control
of the fractional free volumes.442,443 For example, the
fractional free volumes of H2 selectiVe PTMSP membranes
can be modified to produce reverse selective membranes that
favor hydrocarbons over hydrogen.444-448
Because the distinct properties of each gas (size, shape,
chemistry, etc.) play a major role in the sorption selectivity
and solubilities of a polymer, it is often not possible to
generate a membrane which performs uniformly for all gas
compositions. Therefore, it is often necessary to design these
Figure 10. Permeability and selectivity data for hydrogen/nitrogen types of membranes for specific pairs of target gases.
separation.394-414 The upper bound was developed by Robeson393
[1 Barrer ) 7.50062 × 10-18 m3(STP) m/(m2 s Pa)]. (Reprinted The separation and purification of H2 in SMR (steam-
from ref 393, copyright 1991, with permission from Elsevier.) methane reforming) plants is presently accomplished through
PSA (pressure swing adsorption) and/or amine-based acid
the diffusion selectivity. Unfortunately, this is not the case gas scrubbers.9 As implied above, polymeric membranes are
in glassy polymers because the low mobility of the polymer well suited to remove bulk CO2 and to retain H2, and they
chains prevents any significant reduction in diffusion selec- offer a particularly attractive economic alternative to these
tivities. well-established technologies. While there are various classes
As alluded to above, increasing the solubility selectivity of membranes whose function may be considered for H2
is another technique to improve the properties of H2 rejective purification at SMR plants, the temperature of the product
polymeric membranes. There have been several methods gas stream (450-650 °C) prevents conventional polymeric
reported to increase the solubility selectivities. These include
membranes from being used. However, they can be used if
the incorporation of polar groups such as poly(propylene
the temperature of the gas stream is sufficiently cooled. If
oxide),321,435,437 poly(ether oxide),438-440 poly(ester-ether),432
or poly(urethane-ether).433 These polymers provide an im- product gas stream cooling is not a viable option, then
proved environment to solubilize polarizable feed-stream microporous inorganic or palladium (Pd)-based proton
molecules such as CO, CO2, and NOx while the nonpolar- conducting or carbon molecular sieving membranes are
izable feed-stream components such as H2 do not experience used.6 The H2 is removed from the minor component (CO2),
this improved solubility. A very significant development in and the product is recovered. However, since these types of
these types of polymeric membranes was reported by the membranes usually permeate both H2 and CO2, the recovered
Research Institute of Innovative Technology for the Earth H2 will also contain CO2, since it also permeates to some
(RITE).441 RITE developed a novel gating membrane based extent.
6.3. H2 versus CO2 Selective Polymeric density than that of Nafion was shown to permeate CO2
Membranes through a facilitated transport mechanism.455 A water-swollen
sulfonated styrene-divinylbenzene (2.2 mequiv/g) containing
A CO2-selective polymeric membrane permeates CO2 by EDAH+ exhibited a high CO2 permeance of 5.0 × 10-8 m3/
means of a reversible reaction, where an amine-based carrier (m2 s kPa) at 0.41 kPa CO2 feed pressure. Although H2 data
gas species reacts with CO2 on the high-pressure feed side were not reported, the CO2/N2 selectivity was still quite
of the membrane to form an adduct, R2N-CO2. The R2N- promising.456 A novel ion exchange membrane was obtained
CO2 adduct diffuses through the membrane to the low- by cross-linking a polysaccharide, sodium alginate, followed
pressure product stream side, where it reforms free carrier by exchange to form the EDAH+ containing membrane.457
species and CO2 is liberated. Both CO2 and H2 permeate by Exponential increases in both CO2 permeance and selectivity
Fickian diffusion, in which gas molecules dissolve at the resulted from increasing the EDAH+ concentration in the
high-pressure feed side, diffuse across the membrane, and membrane, but a modest CO2/N2 selectivity of about 50
desorb into the gas phase on the lower pressure permeate strongly suggests that this membrane may not be suitable
side (see eq 1 and Figure 1).9 for CO2/H2 separations.
The ideal polymeric H2 separation membrane is one that
maximizes both CO2 permeability and CO2/H2 selectivity Ion exchange membranes prepared by polymerizing of acid
without reducing the relative H2 fluxes. The increase in CO2 containing monomers onto microporous or highly permeable
permeability translates to low membrane area and, hence, surfaces have shown great promise.458,459 Plasma graft
lower costs. High permeability, however, cannot be achieved polymerization of acrylic acid gave membranes with higher
at the expense of selectivity, since H2 losses are commercially exchange capacities than those of typical ion exchange
and economically unacceptable. While H2 permeates only membranes. The result is a higher CO2-carrier density and a
by a solution diffusion mechanism (v), the key to resolving more highly ionic barrier to H2 permeation. Following
this issue is to avoid a membrane that has very low H2 neutralization with EDA to form EDAH+, facilitated trans-
solubility because it functions as a barrier to H2 permeation. port of CO2 with very high selectivity was observed: CO2
A wide range of polymeric materials have been investigated permeance, 7.5 × 10-6 m3/(m2 s kPa); CO2/N2 selectivity,
as candidates, including ion exchange resins, hydrophilic 4700 at 4.8 kPa CO2 feed pressure.458,459 With such a high
polymers, blended polymers with CO2-reactive salts, poly- selectivity, these membranes are expected to be effective for
electrolytic membranes, and polyanilines. Membranes con- removal of CO2 from H2.
sisting of ion exchange resins, polyelectrolytes, and polymer/ Another novel class of polymeric membranes consists of
salt blends contain mobile carrier species which preferentially polyvinyl alcohol/poly(acrylic acid) block copolymer blends
react with CO2 and diffuse across the membranes. The neutralized with EDA.460 Depending on the nature and ratio
following sections are brief summaries of these classes of of polymers, ion exchange capacities ranged from 1.3 to 4.5
membranes that have been reported for CO2/H2 separations. mequiv/g. A maximum CO2/N2 selectivity of 1500 at 6.2
kPa CO2 feed pressure was obtained for the highest exchange
6.4. Ionic and Ion Exchange Polymer Membranes capacity membrane. Such a high CO2/N2 selectivity would
suggest utility for CO2/H2 separations.
Pioneering work in this area was carried out in the early Hydrophilic polymers blended with basic or CO2-reactive
1980s using ion exchange membranes.449,450 Membranes salts have also proved effective for CO2/H2 separations. These
consisting of polyvinylpyridines and simple anions (carbonate membranes contain mobile carrier species that react to form
(CO3-) or glycinate (NH2CH2CO2-)) resulted in high CO2 bicarbonate or carbamates. Membranes consisting of poly-
permeabilities relative to standard chloride containing polyvinylalcohol blended with various amino acid salts exhibited
vinylpyridine membranes. Polystryenesulfonic acid (PSSA) CO2 permeabilities ranging from 0.72 × 10-11 to 1.6 × 10-11
membranes neutralized with ethylenediamine (EDA) im- m3/(m2 s kPa) with CO2/H2 selectivities of 13-30 at a CO2
proved the permselective properties. However, the resulting feed pressure of 76 kPa.461 Membranes consisting of poly-
monoprotonated ethylenediamine, NH2CH2CH2NH3+ ethylenimine alone, or blended with polyvinylalcohol, have
(EDAH+), can form carbamates (R2NCO2-) based on the been utilized. When blended with 50 wt % lithium glycinate,
reversible reaction of R2NH with CO2, which provides a the resulting CO2 permeabilities range from 1.4 to 1.6 ×
robust pathway for reversible carbon dioxide sequestration. 10-11 m3/(m2 s kPa) and CO2/H2 selectivities range from 28
Following diffusion to the low-pressure side of the to 37 at ambient temperature and 1230 Barrers [1 Barrer )
membrane, the carbamate decomposes, or rather dissociates, 7.50062 × 10-18 m3(STP) m/(m2 s Pa)] and a selectivity of
to form the original components, EDAH+ and CO2 in the 75 is obtained at 80 °C.462 Polyvinylalcohol membranes
gas phase. However, the CO2 permeabilities were not containing 50 wt % tetramethylammonium fluoride tetrahy-
constant because of carrier saturation (carbamate formation) drate had a reported CO2/H2 selectivity of 19 at 76 kPa CO2
at higher CO2 pressures.451 The ionic nature of this membrane feed pressure.463 Membranes consisting of blends of cesium
provides a substantial barrier to permeation of nonpolarizable fluoride with polyvinylalcohol or cesium polyacrylate ex-
gases and, by extension, H2. Ion exchange membranes, hibited CO2/H2 selectivities of about 60 at a CO2 feed
containing EDAH+, have fueled intense research on these pressure of about 4.4 kPa.464
membrane technologies. One particular group of membranes, Polyelectrolytes are polymers that have high ionic contents
Nafion, consists of poly(perfluorosulfonic acid), which up to one ionic unit per polymer repeat unit. Unlike ion
contains EDAH+, and was reported to mediate the transport exchange polymers, polyelectrolytes are water soluble. The
of CO2,452 with only a modest CO2/H2 selectivity.453 Since high ionic site density of polyelectrolytes provides a high
Nafion has a relatively low ion exchange capacity, it provides concentration of CO2-reactive sites and an ionic medium with
only a moderate H2 permeation barrier. However, highly a low H2 solubility. Polyelectrolytes that have utility for H2
hydrated Nafion-EDAH membranes exhibited increased purification have cationic groups on the polymer backbone
CO2/H2 selectivities.454 A sulfonated polybenzimidazole- coupled with anions, particularly fluoride (F-) and acetate
EDAH (PBI-EDAH) membrane with a higher ionic site (CH3CO2-). One common example is poly(vinylbenzyltri-
methylammonium fluoride) (PVBTAF).465,466 PVBTAF has selectivity will put them at the forefront of separations
quaternary nitrogen groups on the polymer backbone. These technology. Currently, they are interesting candidates for
functional groups cannot interact with CO2; rather, CO2 streamlined hydrogen production via natural gas reformation
reactivity arises from the counteranion. As described in because of their stability at high temperatures and their ability
reaction 4, such anions act as bases to promote formation of to be regenerated without loss to performance. Concerns
HCO3-. associated with inorganic membranes center on their fabrica-
tion reproducibility. Compared with organic membranes,
2F-‚nH2O + CO2 f HCO3- + HF2-‚(2n-1)H2O (4) inorganic membranes are currently expensive to manufacture.
However, introduction into large-scale production facilities
PVBTAF and related membranes exhibited facilitated should result in more competitive production costs.
transport451-455 of CO2 and CO2/H2 selectivities greater than Carbon-based membranes have the potential for a wide
80.465,466 Permselective properties were, not surprisingly, variety of applications associated with the separation and
strongly dependent on the feed and sweep gas relative purification of hydrogen gas. A recent review33 summarizes
humidities. Membranes consisting of two polyelectrolyte the literature for performances of carbon membranes for the
layers exhibited an unexpected improvement in selectivity separation of mixtures of permanent gases. However, to
to 207 without sacrificing CO2 permeance.467 Much like the become economically viable as commercial products, these
hydrophilic polymers discussed above, incorporation of membranes will have to have better selectivity, thermal
various fluoride and acetate containing salts into polyelec- stability, and chemical stability. Today, the production of
trolyte membranes yielded improved permselective proper- carbon membranes involves a very high cost; it is on the
ties.464,468 Incorporating 4 mol of cesium fluoride/mole of order of 1-3 orders of magnitude greater than that of the
repeat unit gave membranes with a 4- to 6-fold higher CO2 typical polymeric membranes. Therefore, it must achieve a
permeance and a CO2/H2 selectivity of 127 at a CO2 pressure superior performance in order to compensate for the higher
of 4.1 kPa. cost. Areas of fundamental research aimed at achieving that
superior performance include8 (1) optimization of fabrication
parameters during the pyrolysis process, (2) development of
7. Conclusion effective yet inexpensive carbon membrane precursors, and
We have outlined some of the basic requirements and (3) enhancing the stability of these membranes when exposed
concepts central to the application of metallic membranes to water vapor (as found in reforming processes).
for H2 separation into a clean, useable energy source, for Over the past 25 years, substantial progress had been
the Hydrogen economy. Metallic membranes containing achieved in developing polymeric membranes for hydrogen
group IV (Ti, Zr, Hf) and group V (V, Nb, Ta) metals are purification. However, much remains to be done before such
of great focus in the research community for H2 separations. membranes become a commercial reality. While polymeric
As shown in Table 5, amorphous alloys offer greater membranes are successfully employed for large-scale com-
mechanical durability and resistance to H2 embrittlement, mercial H2 recovery programs, there is still plenty of room
although it is not yet known whether they can produce higher for further improvement. Advanced polymeric membranes
H2 fluxes or display higher thermal stability than their with improved selectivity, diffusivity, H2 fluxes, and per-
crystalline counterparts. The modeling of H2 permeation meabilities are being developed. A wider range of operating
through a metal membrane and complications associated with conditions, specifically temperature and pressure, in addition
thin membranes, amorphous metals, and membranes with to higher chemical resistance to hydrocarbons and other
complex microstructures have been discussed, and a brief aggressive feed streams are all important properties where
summary of the key chemical and physical properties of non- research efforts are being conducted by the global research
Pd-based metals and alloys is highlighted. Despite this, the community. Cross-linkable polymers show promise in ad-
preparation of amorphous and nanocrystalline alloys through dressing some of these concerns; however, there is still a
melt techniques provides a readily accessible avenue for lot of research needed with those materials. Further develop-
novel membrane design. In addition, it should be noted that ment in the area of membrane processing and applications
the processing and production techniques must also be should improve the effectiveness of polymeric membranes
sufficiently able to manipulate the defect structure if in some cases, and focusing this research on very specific
amorphous metals are to be useful for H2 separations. applications should allow membranes to play an increasing
Undoubtedly, the advances in the fundamental science and role in the growing hydrogen separation market.
understanding will continue to play a significant role in While many of the challenges are associated with poly-
improving the performance characteristics of metallic H2 meric H2 separations, it may eventually prove necessary to
separation membranes. blend multiple membrane technologies together into mixed-
Inorganic silica and zeolite membranes hold the potential matrix systems such as hybrid membranes (metal-polymer,
for full and near-term industrial implementation due to their ceramic-polymer, metal-ceramic), which attempt to incor-
tunable nature and high-temperature and high-pressure stabil- porate the benefits of each component to improve the overall
ity. Silica membranes are one of the candidates for hydrogen membrane performance. While there are still some major
separation due to their ease of fabrication, low cost of technical barriers that must be overcome, the future of
production, and scalability. Because of their porosity and polymeric membranes for hydrogen separations is very
composition, silica membranes are also less expensive than promising and has excellent potential for growth, especially
metals (due to the lack of precious elements) and not in the area of functionalization and chemical resistivity.
susceptible to H2 embrittlement. Similarly, zeolite membranes Polymeric membranes are a dense type of membrane,
have inherent chemical, mechanical, and thermal stability. transporting species through the bulk of the material.
The trends in zeolite membrane research show clearly the Depending on their state, polymeric membranes can be
improvements in selectivity, fabrication methodology, and subdivided into glassy (prepared at temperatures below the
energy-production applications. In the near future, the ability glass transition temperature) and rubbery (prepared at tem-
to inexpensively fabricate these membranes for tuned peratures above the glass transition temperature) polymeric
Table 15. Comparison of Membrane Classes

ceramic: carbon: porous
metallic: silica, alumina, zirconia, carbons,” single-wall polymer: polyesters,
Pd, Ta, V, Nb, and alloys titania, and zeolites carbon nanotubes ethers, imides, urethanes, etc.
dense porous dense porous porous dense porous
T range (°C) 300-600 600-900 200-600 500-900 <100
selectivity >1000 >1000 5-139 4-20 low
flux 60-300 6-80 60-300 10-200 low
mechanical phase transitions brittle very brittle swelling and compaction
issues
chemical poisoned by H2S, potential degradation oxidizing and susceptible degraded
stability HCl, CO2, SOx with H2O, H2S or CO2 to organic vapors H2S, HCl, CO2, SOx
transport SolD SolD/MS SolD MS SolD/MS SolD MS
mechanisma
a
KD, Knudson diffusion; SurD, surface diffusion; CC, capillary condensation; MS, molecular sieving; SolD, solution diffusion.
membranes. Glassy membranes have relatively high selectiv- flux capacities are on par with those of zeolite and silica
ity and low flux, whereas rubbery membranes have increased membranes. However, the lack of robust mechanical proper-
flux but lower selectivity. In absolute terms, both types have ties makes processing and modular design very difficult and
moderate fluxes and selectivity. They are usually produced adversely effects the performance lifetime of the membrane.
using the phase inversion method. Operating temperatures Metal and zeolite separation membranes operating on the
are limited to 90-100 °C. Advantages in most applications solution diffusion mechanism easily provide the highest
are a good ability to cope with large pressure drops, low selectivities and flux capacities of all the membrane classes.
cost, and good scalability. Possible problems are limited The typically higher operating temperature ranges are ad-
chemical resistance to certain chemicals, such as HCl, SOx, ditional parasitic energy costs which must be considered over
but also CO2, limited mechanical strength, and relatively high the lifetime of the membrane. In addition, the potential phase
sensitivity to swelling and compaction. Polymeric membranes transitions and hydride embrittlement experienced by many
are in an advanced stage of development and are being pure metals and alloys offer a substantial obstacle for a
considered for industrial commercialization by gas producing permanent membrane solution.
companies such as Air Products, Linde, BOC, and Air Perhaps the biggest hurdle which is faced by all classes
Liquide.19 of H2 separation membranes is the lack of chemical stability.
Water, sulfur containing species, acidic vapors, and CO2 are
8. Comparisons and Perspectives the most commonly encountered problematic contaminants
which must be dealt with. The combined result of these
Given the tremendous body of literature and the ever chemical and thermal performance issues ultimately deter-
growing global research efforts on H2 separation membranes, mines the cost and viability of a given material for application
it is not surprising to find such diverse materials and in commercial H2 separation technologies and how it
engineering approaches. This review is intended to provide addresses the five performance targets for H2 separation set
a distinct cross section of these major efforts and current forth by the U.S. Department of Energy listed in Table 1.14
state-of-the art materials. Beyond this, there is a compara- With the predicted doubling of global energy consumption
tively small but growing body of literature on composite or by 2050, our research efforts must unambiguously overcome
hybrid materials which attempt to integrate the ideal per- the many scientific and technological hurdles that exist in
formance of several classes of H2 membrane materials. We H2 separation membranes.1 Regardless of which method is
suspect that research on these types of materials will become used, separation and purification of the nearly 6 Exajoules
an increasingly important effort toward overcoming the many (1 EJ ) 1018 joules) of H2 which is produced industrially
scientific and technological hurdles that exist between the per year is a paramount task which the membrane research
present state of hydrogen production, utilization, and storage community must endeavor to address before the “hydrogen
capabilities and those required for a competitive sustainable economy” can become a reality.
hydrogen economy. Table 15 provides a comparison of the
classes of the H2 separation membrane materials presented 9. Acknowledgments
throughout this review. When this information is compared
Sandia is a multiprogram laboratory operated by Sandia
against the 2015 targets, we quickly see that each individual
Corporation, a Lockheed Martin Company, for the U.S.
class has its own distinct advantages and disadvantages.
Department of Energy’s National Nuclear Security Admin-
While polymer-based membranes are arguably the cheap- istration, under Contract DE-AC04-94AL85000. T.M.N.
est and easiest processed of the materials, they are less thanks the U.S. DOE Hydrogen, Fuel Cells, & Infrastructure
thermally robust and lack sufficient selectivity and flux Technologies Program for continued support.
capacities. Unlike other classes of materials, polymers
possess the greatest flexibility in their synthetic compositions 10. References
and the available organic chemistries for component pre- and
postfabrication modification. This offers a distinctly unique (1) Report from the Basic Energy Sciences Advisory Committee (Feb
2003); available electronically at http://www.sc.doe.gov/bes/besac/
opportunity over zeolite, metal, and carbon membranes in Basic_Research_Needs_To_Assure_A_Secure_Energy_Future_FEB2003.pdf.
that precise synthetic control, which clearly has an impact (2) Report from the Basic Energy Sciences Workshop on Hydrogen
on their performance, can be systematically tailored as the Production, Storage, and Use (Feb 2004); available electronically at
knowledge and performance base grows. Although closely http://www.sc.doe.gov/bes/hydrogen.pdf.
(3) Ogden, J. Lecture (Oct 2004); available electronically at http://
related, carbon-based separation membranes do not suffer www.its.ucdavis.edu/education/classes/pathwaysclass/7-StationaryH2-
from the thermal limitations of the polymers. Generally, their (Ogden).pdf.
(4) Sircar, S.; Golden, T. C. Sep. Sci. Technol. 2000, 35, 667. (55) Andreeva, D.; Idakiev, V.; Tabakova, T.; Andreev, A.; Giovanoli,
(5) Stocker, J.; Whysall, M.; Miller, G. Q. 30 years of PSA technology R. Appl. Catal., A 1996, 134, 275.
for hydrogen purification 2005; UOP LLC: Des Plaines, IL, 1998. (56) Morreale, B. D.; Ciocco, M. V.; Enick, R. M.; Morsi, B. I.; Howard,
(6) Bredesen, R.; Jordal, K.; Bollard, O. Chem. Eng Process. 2004, 43, B. H.; Cugini, A. V.; Rothenberger, K. S. J. Membr. Sci. 2003, 212,
1129. 87.
(7) Adhikari, S.; Fernando, S. Ind. Eng. Chem. Res. 2006, 45, 875. (57) Kajiwara, M.; Uemiya, S.; Kojima, T. Int. J. Hydrogen Energy 1999,
(8) Spillman, R. W.; Grace, W. R. Chem. Eng. Prog. 1989, 85, 41. 24, 839.
(9) Gunardson, H. Industrial Gases in Petrochemical Processing; Marcel (58) Buxbaum, R. E.; Marker, T. L. J. Membr. Sci. 1993, 85, 29.
Dekker, Inc.: 1998. (59) Castello, G.; Biagini, E.; Munari, S. J. Chromatogr., A 1965, 20,
(10) Molburg, J. C.; Doctor, R. D. 20th Annual International Pittsburg 447.
Coal Conference (2003). (60) Rothenberger, K. S.; Howard, B. H.; Killmeyer, R. P.; Cugini, A.
(11) Lin, Y. S. Sep. Purif. Technol. 2001, 25, 39. V.; Enick, R. M.; Bustamante, F.; Ciocco, M. V.; Morreale, B. D.;
(12) Norby, T. Solid State Ionics 1999, 125, 1. Buxbaum, R. M. J. Membr. Sci. 2003, 218, 19.
(13) Guan, J.; Dorris, S. E.; Balachandran, U.; Liu, M. Solid State Ionics (61) Makrides, A. C.; Wright, M. A.; Jewett, D. A. U.S. Patent 3,350,-
1997, 100, 45. 846.
(14) U.S. Department of Energy, Small Business Innovation Research (62) Moss, T. S.; Peachey, N. M.; Snow, R. C.; Dye, R. C. Int. J. Hydrogen
Program and Small Business Technology Transfer Program FY 2005 Energy 1998, 23, 99.
Solicitations, Technical Topic Descriptions, 15. Materials Research, (63) Uemiya, S.; Kato, W.; Uyama, A.; Kajiwara, M.; Kojima, T.; Kikuchi,
Office of Fossil Energy, 2005.www.science.doe.gov/sbir/solicitations/ E. Sep. Purif. Technol. 2001, 22-3, 309.
fy%202005/15_FE3.htm. (64) Ward, T. L.; Dao, T. J. Membr. Sci. 1999, 153, 211.
(15) The President Hydrogen Fuel Initiative: A Clean and Secure Energy (65) Shu, J.; Grandjean, B. P. A.; Vanneste, A.; Kaliaguine, S. Can. J.
Future; available electronically at http://www.hydrogen.energy.gov/ Chem. Eng. 1991, 69, 1036.
presidents_initiative.html. (66) Hara, S.; Hatakeyama, N.; Itoh, N.; Kimura, H. M.; Inoue, A. J.
(16) Broehl, J. Energy Bulletin 2004, http://www.energybulletin.net/ Membr. Sci. 2003, 211, 149.
3342.html. (67) Kajiwara, M.; Uemiya, S.; Kojima, T.; Kikuchi, E. Catal. Today 2000,
(17) Rosen, M. A.; Scott, D. S. Int. J. Hydrogen Energy 1998, 23, 631. 56, 65.
(18) Koros, W. J.; Fleming, G. K. J. Membr. Sci. 1993, 83, 1. (68) Hara, S.; Hatakeyama, N.; Itoh, N.; Kimura, H. M.; Inoue, A.
(19) Kluiters, S. C. A. Status reView on membrane systems for hydrogen Desalination 2002, 144, 115.
separation; Energy Center of Netherlands: Petten, The Netherlands, (69) Heinze, S.; Vuillemin, B.; Colson, J. C.; Giroux, P.; Leterq, D. Solid
2004. State Ionics 1999, 122, 51.
(20) Malek, K.; Coppens, M. O. J. Chem. Phys. 2003, 119, 2801. (70) Pick, M. A. J. Nucl. Mater. 1987, 147, 297.
(21) Jaguste, D. N.; Bhatia, S. K. Chem. Eng. Sci. 1995, 50, 167. (71) Makhlouf, M. M.; Sisson, R. D. Metall. Mater. Trans. A 1991, 22,
(22) Noble, R. D.; Stern, S. A. Membrane Separations Techologys 1001.
Principles and Applications; Elsevier: Amsterdam, 1995. (72) Fukada, S.; Nakahara, T.; Mitsuishi, N. J. Nucl. Mater. 1990, 171,
(23) Mulder, M.; Mulder, J. Basic principles of membrane technology; 399.
Kluwer Academic Publishers: Dordretch, The Netherlands, 1996. (73) Zhang, Y.; Ozaki, T.; Komaki, A.; Nishimura, C. J. Membr. Sci.
(24) Fick, A. Ann. Phys., Leipzig 1855, 170, 59. 2003, 224, 81.
(25) Buxbaum, R. E.; Kinney, A. B. Ind. Eng. Chem. Res. 1996, 35, 530. (74) Liu, B. S.; Li, H. X.; Cao, Y.; Deng, J. F.; Sheng, C.; Zhou, S. Y. J.
(26) Wipf, H. Phys. Scr. 2001, T94, 43. Membr. Sci. 1997, 135, 33.
(27) Lewis, F. A. Pure Appl. Chem. 1990, 62, 2091 and references therein. (75) Liu, B. S.; Zhang, W. D.; Dai, W. L.; Deng, J. F. J. Membr. Sci.
(28) Gao, H. Y.; Lin, Y. S.; Li, Y. D.; Zhang, B. Q. Ind. Eng. Chem. 2004, 244, 243.
Res. 2004, 43, 6920. (76) Peachey, N. M.; Snow, R. C.; Dye, R. C. J. Membr. Sci. 1996, 111,
(29) Uemiya, S. Sep. Purif. Method 1999, 28, 51. 123.
(30) Paglieri, S. N.; Way, J. D. Sep. Purif. Method 2002, 31, 1. (77) Lu, K.; Wang, J. T.; Wei, W. D. J. Phys. D: Appl. Phys. 1992, 25,
(31) Ma, Y. H.; Mardilovich, I. P.; Engwall, E. E. Ann. N. Y. Acad. Sci. 808.
2003, 984, 346 and references therein. (78) Siriwardane, R. V.; Poston, J. A.; Fisher, E. P.; Lee, T. H.; Dorris,
(32) Sholl, D. S. MRS Bull. 2006, 31, 770 and references therein. S. E.; Balachandran, U. Appl. Surf. Sci. 2000, 167, 34.
(33) Ma, Y. H.; Mardilovich, P. P.; She, Y. U.S. Patent 6,152,987. (79) Birnbaum, H. K. J. Less-Common. Met. 1984, 104, 31.
(34) Ma, Y. H.; Mardilovich, I. P.; Engwall, E. E. U.S. Patent Applications (80) Mundschau, M. V. Hydrogen transport membranes. U.S. Patent 6,-
0,244,590; 0,244,583; and 0,237,779. 899,744.
(35) Criscuoli, A.; Basile, A.; Drioli, E.; Loiacono, O. J. Membr. Sci. (81) Roark, S. E.; Mackay, R.; Mundschau, M. V. Dense, layered
2001, 181, 21. membranes for hydrogen separation. U.S. Patent 7,001,446.
(36) Middelton, P.; Solgaard-Anderson, H.; Rostrup-Nielson, H. T. (82) Paglieri, S. N.; Birdsell, S. A.; Barbero, R. S.; Snow, R. C.; Smith,
Proceedings of the 14th World Energy Conference, Montreal, Canada, F. M. Tubular hydrogen permeable metal foil membrane and method
June 9-13, 2000. of fabrication. U.S. Patent 0,045,034 A1.
(37) Bryden, K. J.; Ying, J. Y. J. Membr. Sci. 2002, 203, 29. (83) Villars, P.; Prince, A.; Okamoto, H. Handbook of Ternary Alloy Phase
(38) Castro, F. J.; Meyer, G.; Zampieri, G. J. Alloys Compd. 2002, 330, Diagrams; ASM International: Materials Park, OH, 1995.
612. (84) Massalski, T. B.; Okamoto, H.; Subramanian, P. R.; Kacprzak, L.
(39) Hoyos, L. J.; Primet, P.; Praliaud, H. J. Chem. Soc., Faraday Trans. Binary Alloy Phase Diagrams; ASM International: Materials Park,
1992, 22, 3367. OH, 1990.
(40) Rutlowski, W.; Wetzig, D.; Zacharias, H. Phys. ReV. Lett. 2001, 87, (85) Dos Santos, D. S.; De Miranda, P. E. V. J. Non-Cryst. Solids 1998,
246101. 234, 133.
(41) Roa, F.; Block, M. J.; Way, J. D. Desalination 2002, 147, 411. (86) Yukawa, H.; Yamashita, D.; Ito, S.; Morinaga, M.; Yamaguchi, S.
(42) Frieske, H.; Wicke, E. Ber. Bunsen-Ges. Phys. Chem. 1973, 77, 48. Mater. Trans., JIM 2002, 43, 2757.
(43) Wicke, E.; Nernst, G. H. Ber. Bunsen-Ges. Phys. Chem. 1964, 68, (87) Yukawa, H.; Teshima, A.; Yamashita, D.; Ito, S.; Yamaguchi, S.;
224. Morinaga, M. J. Alloys Compd. 2002, 337, 264.
(44) Ma, Y. H.; Akis, B. C.; Ayturk, M. E.; Guazzone, F.; Engwall, E. (88) Yukawa, H.; Yamashita, D.; Ito, S.; Morinaga, M.; Yamaguchi, S.
E.; Mardilovich, I. P. Ind. Eng. Chem. Res. 2004, 43, 2936. J. Alloys Compd. 2003, 356, 45.
(45) Sholl, D. S.; Ma, Y. H. MRS Bull. 2006, 31, 770. (89) Komiya, K.; Ito, S.; Yukawa, H.; Morinaga, M.; Nagata, K.; Nambu,
(46) Phair, J. W.; Donelson, R. Ind. Eng. Chem. Res. 2006, 45, 5657. T.; Ezaki, H. Mater. Trans., JIM 2003, 44, 1686.
(47) Hara, S.; Sakaki, K.; Itoh, N.; Kimura, H. M.; Asami, K.; Inoue, A. (90) Dorris, S. E.; Lee, T. H.; Balachandran, U. Metal/ceramic composites
J. Membr. Sci. 2000, 164, 289. with high hydrogen permeability. U.S. Patent 6,569,226.
(48) Hara, S.; Sakaki, K.; Itoh, N. Amorphous Ni alloy membrane for (91) Edlund, D. J.; Newbold, D. D.; Frost, B. Composite hydrogen
separation/dissociation of hydrogen. U.S. Patent 6,478,853. separation element and module. U.S. Patent 5,645,626.
(49) Yamaura, S.-I.; Shimpo, Y.; Okouchi, H.; Nishida, M.; Kajita, O.; (92) Gleiter, H. Phys. Status Solidi B 1992, 172, 41.
Kimura, H.; Inoue, A. Mater. Trans., JIM 2003, 44, 1885. (93) Kirchheim, R. Phys. Scr. 2001, T94, 58.
(50) Newsome, D. S. Catal. ReV. 1980, 21, 275. (94) Gleiter, H. Prog. Mater. Sci. 1989, 33, 223.
(51) Pasel, J.; Samsun, R. C.; Schmitt, D.; Peters, R.; Stolten, D. J. Power (95) Gleiter, H. Nanostruct. Acta Mater. 2000, 48, 1.
Sources 2005, 152, 189. (96) Bryden, K. J.; Ying, J. Y. Mater. Sci. Eng., A 1995, 204, 140.
(52) Rostrup-Nielsen, J. R.; Rostrup-Nielsen, T. Caltech 2002, 6, 150. (97) Bryden, K. J.; Ying, J. Y. Nanostruct. Mater. 1997, 9, 485.
(53) Ruth, L. A. Mater. High Temp. 2003, 20, 7. (98) Papaefthymiou, G. C.; Bryden, K. J.; Ying, J. Y. Physica B 2002,
(54) Amadeo, N. E.; Laborde, M. A. Int. J. Hydrogen Energy 1995, 20, 311, 279.
949. (99) McCool, B. A.; Lin, Y. S. J. Mater. Sci. 2001, 36, 3221.
(100) Xomeritakis, G.; Lin, Y. S. AIChE J. 1998, 44, 174. (152) Cockayne, D. J. H.; McKenzie, D. R.; McBride, W.; Goringe, C.;
(101) Yang, K.; Cao, M. Z. Scr. Metall. Mater. 1991, 25, 2139. McCulloch, D. Microsc. Microanal. 2000, 6, 329.
(102) Galano, M.; Audebert, F.; Cantor, B.; Stone, I. Mater. Sci. Eng., A (153) Miracle, D. B. Nat. Mater. 2004, 3, 697.
2004, 375, 1206. (154) Treacy, M. M. J.; Gibson, J. M. Acta Crystallogr., A 1996, 52, 212.
(103) Eliaz, N.; Eliezer, D.; Abramov, E.; Zander, D.; Koster, U. J. Alloys (155) Hufnagel, T. C. Finding order in disorder. Nat. Mater. 2004, 3, 666.
Compd. 2000, 305, 272. (156) Warde, J.; Knowles, D. M. ISIJ Int. 1999, 39, 1006.
(104) Luck, R. Quasicrystals. An Introduction to Structure, Physical (157) Warde, J.; Knowles, D. M. ISIJ Int. 1999, 39, 1015.
Properties, and Applications; Springer: Berlin, 2002. (158) Amis, E. J. Nat. Mater. 2004, 3, 83.
(105) Senechal, M. Notes AMS 2006, 53, 886. (159) Amis, E. J.; Xiang, X. D.; Zhao, J. C. MRS Bull. 2002, 27, 295.
(106) Gapontsev, A. V.; Kondrat’ev, V. V. Phys.-Usp. 2003, 46, 1077. (160) Xiang, X. D.; Sun, X. D.; Briceno, G.; Lou, Y. L.; Wang, K. A.;
(107) Steward, S. A. ReView of Hydrogen Isotope Permeabilities Through Chang, H. Y.; WallaceFreedman, W. G.; Chen, S. W.; Schultz, P.
Materials; Lawrence Livermore National Laboratory: Livermore, G. Science 1995, 268, 1738.
CA, 1983. (161) Koinuma, H.; Takeuchi, I. Nat. Mater. 2004, 3, 429.
(108) Liu, L.; Zhang, J. MRS Bull. 2001, 36, 2073. (162) Yoo, Y. K.; Xiang, X. D. J. Phys.: Condens. Matter 2002, 14, R49.
(109) Shimpo, Y.; Yamaura, S.; Okouchi, H.; Nishida, M.; Kajita, O.; (163) Yoo, Y. K.; Ohnishi, T.; Wang, G.; Duewer, F.; Xiang, X. D.; Chu,
Kimura, H.; Inoue, A. J. Alloys Compd. 2004, 372, 197. Y. S.; Mancini, D. C.; Li, Y. Q.; O’Handley, R. C. Intermetallics
(110) Yamaura, S.-I.; Shimpo, Y.; Okouchi, H.; Nishida, M.; Kajita, O.; 2001, 9, 541.
Inoue, A. Mater. Trans., JIM 2004, 45, 330. (164) Xiang, X. D. Appl. Surf. Sci. 2004, 223, 54.
(111) Evard, E. A.; Kurdumov, A. A.; Berseneva, F. N.; Gabis, I. E. Int. (165) Zhao, J. C. Annu. ReV. Mater. Res. 2005, 35, 51.
J. Hydrogen Energy 2001, 26, 457. (166) Xiang, X. D.; Schultz, P. G. Physica C 1997, 282, 428.
(112) Evard, E. A.; Sidorov, N. I.; Gabis, I. E. Tech. Phys. 2000, 45, 377. (167) Wei, T.; Xiang, X. D.; WallaceFreedman, W. G.; Schultz, P. G. Appl.
(113) Nishimura, C.; Komaki, M.; Hwang, S.; Amano, M. J. Alloys Compd. Phys. Lett. 1996, 68, 3506.
2002, 330-332, 902. (168) Zhao, J. C. J. Mater. Res. 2001, 16, 1565.
(114) Ozaki, T.; Zhang, Y.; Komaki, M.; Nishimura, C. Int. J. Hydrogen (169) Rao, X.; Xia, Q. F.; Li, X. J.; Si, P. C. Intermetallics 2000, 8, 499.
Energy 2003, 28, 1229. (170) Takeuchi, A.; Inoue, A. J. Optoelectron. AdV. M 2004, 6, 533.
(115) Nishimura, C.; Ozaki, T.; Komaki, M.; Zhang, Y. J. Alloys Compd. (171) Miracle, D. B.; Senkov, O. N.; Sanders, W. S.; Kendig, K. L. Mater.
2003, 356-357, 295. Sci. Eng., A 2004, 375-377, 150.
(116) Zhang, Y.; Ozaki, T.; Komaki, M.; Nishimura, C. Scr. Metall. Mater. (172) Yeung, K. L.; Varma, A. AIChE J. 1995, 41, 2131.
2002, 47, 601. (173) Jayaraman, N.; Lin, Y. S.; Pakala, M.; Lin, R. Y. J. Membr. Sci.
(117) Amano, M.; Komaki, M.; Nishimura, C. J. Less-Common. Met. 1991, 1995, 99, 89.
172, 727. (174) Saracco, G.; Specchia, V. Catal. ReV. 1994, 36, 305.
(118) Eliaz, N.; Eliezer, D. AdV. Perform. Mater. 1999, 6, 5. (175) Xue, D.; Chen, H.; Wu, G. H.; Deng, J. F. Appl. Catal., A 2001,
(119) Kirchheim, R.; Mutschele, T.; Kieninger, W.; Gleiter, H.; Birringer, 214, 87.
R.; Koble, T. D. Mater. Sci. Eng. A 1988, 99, 457. (176) Burchardt, T. Int. J. Hydrogen Energy 2000, 25, 627.
(120) Rush, J. J.; Rowe, J. M.; Maeland, A. J. J. Phys. F: Met. Phys. (177) Burchardt, T.; Hansen, V.; Valand, T. Electrochim. Acta 2001, 46,
1980, 10, L283. 2761.
(121) Bankmann, J.; Pundt, A.; Kirchheim, R. J. Alloys Compd. 2003, 356- (178) Deng, J. F.; Li, H.; Wang, W. Catal. Today 1999, 51, 113.
357, 566. (179) Bozzolo, G.; Ferrante, J.; Noebe, R. D.; Good, B.; Honecy, F. S.;
(122) Dos Santos, D. S.; De Miranda, P. E. V. J. Mater. Sci. 1997, 32, Abel, P. Comput. Mater. Sci. 1999, 15, 169.
6311. (180) Good, B.; Bozzolo, G. Surf. Sci. 2002, 507-510, 730.
(123) Wu, Q. Y.; Xu, J.; Sun, X. K.; Hu, Z. Q. J. Mater. Sci. Technol. (181) Besenbacher, F.; Chorkendorff, I.; Clausen, B. S.; Hammer, B.;
1997, 13, 443. Molenbroek, A. M.; Nørskov, J. K.; Stensgaard, I. Science 1998,
(124) Maeland, A. J.; Tanner, L. E.; Libowitz, G. G. J. Less-Common. 279, 1913.
Met. 1980, 74, 279. (182) Greeley, J.; Mavrikakis, M. Nat. Mater. 2004, 3, 810.
(125) Suzuki, K. J. Less-Common. Met. 1983, 89, 183. (183) Symons, D. M.; Young, G. A.; Scully, J. R. Metall. Mater. Trans. A
(126) Dos Santos, D. S.; De Miranda, P. E. V. Int. J. Hydrogen Energy 2001, 32, 369.
1998, 23, 1011. (184) Song, Z.; Tan, D.; He, F.; Bao, X. Appl. Surf. Sci. 1999, 137, 142.
(127) Shimizu, E.; Aoki, K.; Masumoto, T. J. Alloys Compd. 1999, 295, (185) Dong, J.; Liu, W.; Lin, Y. S. AIChE J. 2000, 46, 1957.
526. (186) Prabhu, A. K.; Radhakrishnan, R.; Oyama, S. T. Appl. Catal., A 1999,
(128) Kim, J. J.; Stevenson, D. A. J. Non-Cryst. Solids 1988, 101, 187. 183, 241.
(129) Eliaz, N.; Fuks, D.; Eliezer, D. Mater. Lett. 1999, 39, 255. (187) Prabhu, A. K.; Oyama, S. T. Chem. Lett. 1999, 213.
(130) Kirchheim, R.; Stolz, U. Acta Metall. Mater. 1987, 35, 281. (188) Morooka, S.; Kusakabe, K. MRS Bull. 1999, 24, 25.
(131) Zander, D.; Leptien, H.; Koster, U.; Eliaz, N.; Eliezer, D. J. Non- (189) Tsapatis, M.; Gavalas, G. R. MRS Bull. 1999, 24, 30.
Cryst. Solids 1999, 252, 893. (190) Prabhu, A. K.; Oyama, S. T. J. Membr. Sci. 2000, 176, 233.
(132) Aoki, K.; Kamachi, M.; Masumoto, T. J. Non-Cryst. Solids 1984, (191) Verweij, H. J. Mater. Sci. 2003, 38, 4677.
61-62, 679. (192) Tsuru, T.; Tsuge, T.; Kubota, S.; Yoshida, K.; Yoshioka, T.; Asaeda,
(133) Eliaz, N.; Moshe, E.; Eliezer, S.; Eliezer, D. Metall. Mater. Trans. M. Sep. Sci. Technol. 2001, 38, 3721.
A 2000, 31, 1085. (193) Kurungot, S.; Yamaguchi, T.; Nakao, S.-I. Catal. Lett. 2003, 86, 273.
(134) Ishida, M.; Takeda, H.; Watanabe, D.; Amiya, K.; Nishiyama, N.; (194) Kurungot, S.; Yamaguchi, T. Catal. Lett. 2004, 92, 181.
Kita, K.; Saotome, Y.; Inoue, A. Mater. Trans., JIM 2004, 45, 1239. (195) Nair, B. N.; Okubo, T.; Nakao, S.-I. Membranes 2000, 25, 73.
(135) Kim, W. B.; Ye, B. J.; Yi, S. Met. Mater. Int. 2004, 10, 1. (196) Kitao, S.; Kameda, H.; Asaeda, M. Membranes 1990, 15, 222.
(136) Park, E. S.; Kim, D. H. Met. Mater. Int. 2005, 11, 19. (197) de Lange, R. S. A.; Keizer, K.; Burggraaf, A. J. J. Membr. Sci. 1995,
(137) Inoue, A.; Nishiyama, N.; Amiya, K.; Zhang, T.; Masumoto, T. Mater. 104, 81.
Lett. 1994, 19, 131. (198) Asaeda M.; Kashimoto, M. Proceedings of the Fifth International
(138) Peker, A.; Johnson, W. L. Appl. Phys. Lett. 1993, 63, 2342. Conference Inorganic Membranes, Nagoya, Japan, 1998; p A-405.
(139) Wesseling, P.; Nieh, T. G.; Wang, W. H.; Lewandowski, J. J. Scr. (199) Nomura, M.; Ono, K.; Gopalakrishnan, S.; Sugawara, T.; Nakao,
Metall. Mater. 2004, 51, 151. S.-I. J. Membr. Sci. 2005, 251, 151.
(140) Zhang, Q. S.; Zhang, H. F.; Deng, Y. F.; Ding, B. Z.; Hu, Z. Q. Scr. (200) Nomura, M.; Yamaguchi, T.; Kumakiri, I.; Nakao, S.-I. Membranes
Metall. Mater. 2003, 49, 273. 2001, 23, 124.
(141) Bakonyi, I.; Cziraki, A. Nanostruct. Mater. 1999, 11, 9. (201) Okubo, T.; Inoue, H. J. Membr. Sci. 1989, 42, 109.
(142) Fan, C.; Takeuchi, A.; Inoue, A. Mater. Trans., JIM 1999, 40, 42. (202) Gavalas, G. R.; Megiris, C. E.; Nam, S.-W. Chem. Eng. Sci. 1989,
(143) Li, J. M.; Du, Y. W.; Feng, D.; Quan, M. X.; Hu, Z. Q. AdV. Eng. 44, 1829.
Mater. 1999, 1, 137. (203) Megiris, C. E.; Glezer, J. H. E. Ind. Eng. Chem. Res. 1992, 31, 1293.
(144) Inoue, A. Mater. Sci. Eng., A 2001, 304, 1. (204) Tsapatsis, M.; Gavalas, G. R. AIChE J. 1992, 38, 847.
(145) Fan, C.; Li, C. F.; Inoue, A. J. Non-Cryst. Solids 2000, 270, 28. (205) Ha, H.-Y.; Nam, S.-W.; Hong, S.-A.; Lee, W.-K. J. Membr. Sci.
(146) Murty, B. S.; Hono, K. Appl. Phys. Lett. 2004, 84, 1674. 1993, 85, 279.
(147) Kamakoti, P.; Morreale, B. D.; Ciocco, M. V.; Howard, B. H.; (206) Xomeritakis, G.; Lin, Y.-S. Ind. Eng. Chem. Res. 1994, 33, 2607.
Killmeyer, R. P.; Cugini, A. V.; Sholl, D. S. Science 2005, 307, 569. (207) Morooka, S.; Yan, S.-C.; Kusakabe, K.; Akiyama, Y. J. Membr. Sci.
(148) Kamakoti, P.; Sholl, D. S. J. Membr. Sci. 2006, 279, 94. 1995, 101, 89.
(149) Kamakoti, P.; Sholl, D. S. Phys. ReV. B 2005, 71, 14301. (208) Nakao, S.-I.; Suzuki, T.; Sugawara, T.; Tsuru, T.; Kimura, S.
(150) Sonwane, C. G.; Wilcox, J.; Ma, Y. H. J. Chem. Phys. 2006, 125, Microporous Mesoporous Mater. 2000, 37, 145.
184714. (209) Yamaguchi, T.; Ying, X.; Tokimasa, Y.; Nair, B. N.; Sugawara, T.;
(151) Brouwer, R. C.; Griessen, R. Phys. ReV. B 1989, 40, 1481. Nakao, S.-I. Phys. Chem. Chem. Phys. 2000, 2, 4465.
(210) Nenoff, T. M.; Spontak, R. J.; Aberg, C. M. MRS Bull. 2006, 31, (250) den Exter, M. J.; Jansen, J. C.; van de Graaf, J. M.; Kapteijn, F.;
735. Moulijn, J. A.; van Bekkum, H. Recent AdV. New Horizons Zeolite
(211) Fotou, G. P.; Lin, Y. S.; Pratsinis, S. E. J. Membr. Sci. 1995, 30, Sci. Technol. 1996, 102, 413.
2803. (251) Suzuki, H. Composite membrane having a surface layer of an ultrathin
(212) Nair, B. N.; Elferink, J. W.; Keizer, K.; Verweij, H. J. Colloid film of cage-shaped zeolite and processes for production thereof. U.S.
Interface Sci. 1996, 178, 565. Patent 4,699,892.
(213) Nair, B. N.; Keizer, K.; Maene, N.; Okubo, T.; Nakao, S.-I. In Surface (252) Bakker, W. J. W.; van den Broeke, L. J. P.; Kapteijn, F.; Moulijn, J.
Chemistry and Electrochemistry of Membranes; Sorenson, T. S., Ed.; A. AIChE J. 1997, 43, 2203.
Marcel Dekker: New York, 1999; p 125. (253) Hedlund, J.; Sterte, J.; Anthonis, M.; Bons, A.-J.; Carstensen, B.;
(214) Maier, W. F.; Schramm, H. O. Mater. Res. Soc. Symp. Proc. 1992, Corcoran, N.; Cox, D.; Deckman, H.; Gijnst, W. D.; de Moor, P.-P.;
271, 493. Lai, F.; McHenry, J.; Mortier, W.; Reinoso, J.; Peeters, J. Mi-
(215) Asaeda, M.; Yamamichi, A.; Satoh, M.; Kamakura, M. In Proceedings cropororous Mesoporous Mater. 2002, 52, 179.
of ICIM3; Ma, Y. H., Ed.; 1994; p 315. (254) Bowen, T. C.; Kalipcilar, H.; Falconer, J. L.; Noble, R. D. J. Membr.
(216) Asaeda, M. Proceedings of 1999 International Conference on Sci. 2003, 215, 235.
Membranes, Toronto, 1999. (255) Sano, T.; Yanagishita, H.; Kiyozumi, Y.; Mizukami, F.; Haraya, K.
(217) Asaeda, M.; Yamasaki, S. Sep. Purif. Technol. 2001, 25, 151-159. J. Membr. Sci. 1994, 95, 221.
(218) Julbe, A.; Balzer, C.; Guizard, C.; Cot, L. J. Sol-Gel Sci. Technol. (256) Lovallo, M. C.; Gouzinis, A.; Tsapatsis, M. AIChE J. 1998, 44, 1903.
1995, 4, 89. (257) Burggraaf, A. J.; Vroon, Z. A. E. P.; Keizer, K.; Verweij, H. J.
(219) Cao, G. Z.; Lu, Y.; Delattre, L.; Brinker, C. J.; Lopez, G. P. AdV. Membr. Sci. 1998, 44, 77.
Mater. 1996, 8, 588. (258) Noack, M.; Kolsch, P.; Caro, J.; Schneider, M.; Toussaint, P.; Sieber,
(220) Raman, N. K.; Anderson, M. T.; Brinker, C. J. Chem. Mater. 1996, I. Microporous Mesoporous Mater. 2000, 35, 253.
8, 1682. (259) Wegner, K.; Dong, J. H.; Lin, Y. S. J. Membr. Sci. 1999, 158, 17.
(221) de Vos, R. M.; Maier, W. F.; Verweij, H. J. Membr. Sci. 1999, 158, (260) Lin, X.; Kita, H.; Okamoto, K. Chem. Commun. 2000, 19, 1889.
277. (261) Gardner, T. Q.; Flores, A. I.; Noble, R. D.; Falconer, J. L. AIChE J.
(222) Brinker, C. J.; Ward, T. L.; Sehgal, R.; Raman, N. K.; Hietala, S. 2002, 48, 1155.
L.; Smith, D. M.; Hua, D.-W.; Headley, T. J. J. Membr. Sci. 1993, (262) Aoki, K.; Kusakabe, K.; Morooka, S. J. Membr. Sci. 1998, 141, 197.
77, 165. (263) Kita, H.; Horii, K.; Ohtoshi, Y.; Tanaka, K.; Okamoto, K. I. J. Mater.
(223) Koros, W. J.; Ma, Y. H.; Shimidzu, T. Pure Appl. Chem. 1996, 68, Sci. Lett. 1995, 14, 206.
1479. (264) Hedlund, J.; Schoeman, B.; Sterte, J. Chem. Commun. 1997, 13, 1193.
(224) Lin, Y. S.; Baxbaum, R. E. In Encyclopedia of Separation Science; (265) Navajas, A.; Mallada, R.; Téllez, C.; Coronas, J.; Menéndez, M.;
Wilson, I. D., Adlard, T. R., Poole, C. F., Cooke, M., Eds.; Academic Santamarı́a, J. Desalination 2002, 148, 25.
Press: San Diego, CA, 2000; p 3365. (266) Nishiyama, N.; Korekazu, U.; Masahiko, M. Chem. Commun. 1995,
(225) Lai, Z. P.; Bonilla, G.; Diaz, I.; Nery, J. G.; Sujaoti, K.; Amat, A. 19, 1967.
M.; Kokkoli, E.; Terasaki, O.; Thompson, R. W.; Tsapatsis, M.; (267) Yamazaki, S.; Tsutsumi, K. Adsorption 1997, 3, 165.
Vlachos, D. G. Science 2003, 300, 456. (268) Kita, H.; Asamura, H.; Tanaka, K.; Okamoto, K.-I.; Kondo, M. Abstr.
(226) Tsai, C. Y.; Tam, S. Y.; Lu, Y. F.; Brinker, C. J. J. Membr. Sci. Papers Am. Chem. Soc. 1997, 214, 269.
2000, 169, 255. (269) Kusakabe, K.; Kuroda, T.; Uchino, K.; Hasegawa, Y.; Morooka, S.
(227) Hoenicke D.; Dietzsch, E. In Handbook of Porous Solids; Schueth, AIChE J. 1999, 45, 1220.
F., Sing, K. S. W., Weitkamp, J., Eds.; Wiley: Weinheim, 2002; p (270) Li, S.; Tuan, V. A.; Falconer, J. L.; Noble, R. D. Microporous
1395. Mesoporous Mater. 2002, 53, 59.
(228) van Rijn, C. J. M.; Nijdam, W.; Kuiper, S.; Veldhuis, G. J.; van (271) Geus, E. R.; Den Exter, M. J.; Van Bekkum, H. J. Chem. Soc.,
Wolferen, H. A. G. M.; Elwenspoek, M. C. J. Micromech. Microeng. Faraday Trans. 1992, 88, 3101.
1999, 9, 170. (272) van Bekkum, H.; Geus, E. R.; Kouwenhoven, H. W. Stud. Surf. Sci.
(229) Richardson, J. T.; Paripatyadar, S. A. Appl. Catal. 1990, 61, 293. Catal. 1994, 85, 509.
(230) Kusakabe, K.; Shibao, F.; Zhao, G.; Sotowa, K.-I.; Watanabe, K.; (273) Coronas, J.; Santamaria, J. Sep. Purif. Methods 1999, 28, 127.
Saito, T. J. Membr. Sci. 2003, 215, 321. (274) Bein, T. Chem. Mater. 1996, 8, 1636.
(231) Nomura, M.; Yamaguchi, T.; Nakao, S. Ind. Eng. Chem. Res. 1997,
(275) Caro, J.; Noack, M.; Kolsch, P.; Schafer, R. Microporous Mesoporous
36, 4217.
Mater. 2000, 38, 3.
(232) Mottern, M. L.; Quickel, G. T.; Shi, J. Y.; Yu, D.; Verweij, H. In
(276) Chiang, A. S. T.; Chao, K.-J. J. Phys. Chem. Solids 2001, 62, 1899.
Proceedings of the 8th International Conference on Inorganic
Membranes, Cincinnati, OH, July 18-22, 2004; Lin, Y. S., Ed.; p (277) Matsukata, M.; Kikuchi, E. Bull. Chem. Soc. Jpn. 1997, 70, 2341.
26. (278) Mizukami, F. Stud. Surf. Sci. Catal. 1999, 125, 1.
(233) Liang, J.; Jiang, X.; Liu, G.; Deng, Z.; Zhuang, J.; Li, F.; Li, Y. (279) Tavolaro, A.; Drioli, E. AdV. Mater. 1999, 11, 975.
Mater. Res. Bull. 2003, 38, 161. (280) Davis, M. E. Nature 2002, 417, 813.
(234) Shi, J. Y.; Verweij, H. Langmuir 2005, 21, 5570. (281) Nair, S.; Tsapatsis, M. In Handbook of Zeolite Science and Technol-
(235) Lee, D.; Zhang, L.; Oyama, S. T.; Niu, S.; Saraf, R. F. J. Membr. ogy; Auerbach, S. M., Carrado, K. A., Dutta, P. K., Eds.; Marcel
Sci. 2004, 231, 117. Dekker: New York, 2003.
(236) Nijmeijer, A.; Huiskes, C.; Sibelt, N. G. M.; Kruidhof, H.; Verweij, (282) Bowen, T. C.; Noble, R. D.; Falconer, J. L. J. Membr. Sci. 2004,
H. Am. Ceram. Soc. Bull. 1998, 77, 95. 245, 1 and references therein.
(237) Yoshino, Y.; Suzuki, T.; Nair, B. N.; Taguchi, H.; Itoh, N. J. Membr. (283) Cui, Y.; Kita, H.; Okamoto, K. J. Mater. Chem. 2004, 14, 924.
Sci. 2005, 267, 8. (284) Husain, S.; Koros, W. J. J. Membr. Sci. 2007, 288, 195.
(238) Brinkman, H. W.; van Eijk, J. P. G. M.; Meinema, H. A.; Terpstra, (285) Pechar, T. W.; Kim, S.; Vaughan, B.; Marand, E.; Tsapatsis, M.;
R. A. Am. Ceram. Soc. Bull. 1999, 78, 51. Jeong, H. K.; Cornelius, C. J. J. Membr. Sci. 2006, 277, 195.
(239) Millan, A. J.; Nieto, M. I.; Moreno, R.; Baudin, C. J. Eur. Ceram. (286) Choi, J.; Ghosh, S.; Lai, Z.; Tsapatsis, M. Angew. Chem., Int. Ed.
Soc. 2002, 22, 2223. 2006, 45, 1154.
(240) Toy, C.; Whittemore, O. J. Ceram. Int. 1989, 15, 167. (287) Lai, Z.; Tsapatsis, M.; Nicolich, J. P. AdV. Funct. Mater. 2004, 14,
(241) Giessler, S.; Jordan, L.; da Costa, J. C. D.; Lu, G. Q. Sep. Purif. 716.
Technol. 2003, 32, 255. (288) Barrer, R. M. J. Chem. Soc., Faraday Trans. 1990, 86, 1123.
(242) Yoshida, K.; Hirano, Y.; Fujii, H.; Tsuru, T.; Asaeda, M. J. Chem. (289) Bakker, W. J. W.; Kapteijn, F.; Poppe, J.; Moulijn, J. A. J. Membr.
Eng. Jpn. 2001, 34, 523. Sci. 1996, 117, 57.
(243) Iler, R. K. The Chemistry of Silicas; Wiley: New York, 1979. (290) Jost, S.; Bar, N.-K.; Fritzsche, S.; Haberlandt, R.; Karger, J. J. Phys.
(244) (a) Kanezashi, M.; Fujita, T.; Asaeda, M. Sep. Sci. Technol. 2005, Chem. 1992, B102, 6375.
40, 225. (b) Kanezashi, M.; Asaeda, M. J. Membr. Sci. 2006, 271, (291) Krishna, R.; Paschek, D. Phys. Chem. Chem. Phys. 2002, 4, 1891.
86. (292) Sanborn, M. J.; Snurr, R. Q. Sep. Purif. Technol. 2000, 20, 1.
(245) Hasegawa, Y.; Kusakabe, K.; Morooka, S. J. Membr. Sci. 2001, 190, (293) Sanborn, M. J.; Snurr, R. Q. AIChE J. 2001, 47, 2032.
1. (294) Skoulidas, A. I.; Sholl, D. S.; Krishna, R. Langmuir 2003, 19, 7977.
(246) Lin, Y. S.; Kumakiri, I.; Nair, B. N.; Alsyouri, H. Sep. Purif. Methods (295) Skoulidas, A. I.; et al. J. Membr. Sci. 2003, 227, 123.
2002, 31, 229. (296) Skoulidas, A. I.; Sholl, D. S. AIChE J. 2005, 51, 876.
(247) Benes, N. E.; Biesheuvel, P. M.; Verweij, H. AIChE J. 1999, 45, (297) Sholl, D. S. Acc. Chem. Res. 2006, 39, 403.
1322. (298) Cussler, E. L. Diffusion Mass Transfer in Fluid Systems, 2nd ed.;
(248) Verweij, H.; Lin, Y. S.; Dong, J. H. MRS Bull. 2006, 31, 756. Cambridge University Press: Cambridge, U.K., 1997.
(249) Duke, M. C.; Diniz da Costa, J. C.; Lua, G. Q.; Petch, M.; Gray, P. (299) Burggraaf, A. J. Transport and separation properties of membranes
J. Membr. Sci. 2004, 241, 325. with gases and Vapors; Elsevier: Amsterdam, 1996.
(300) Masuda, T.; Fukumoto, N.; Kitamura, M.; Mukai, S. R.; Hahimoto, (350) Koresh, J. E.; Soffer, A. J. Chem. Soc., Faraday Trans. 1 1980, 76,
K.; Tanaka, T.; Funabiki, T. Microporous Mesoporous Mater. 2001, 2507.
48, 239. (351) Koresh, J. E.; Soffer, A. J. Chem. Soc., Faraday Trans. 1 1981, 77,
(301) Park, D. H.; Nishiyama, N.; Egashira, Y.; Ueyama, K. Ind. Eng. 3005.
Chem. Res. 2001, 40, 6105. (352) Koresh, J. E.; Soffer, A. Sep. Sci. Technol. 1983, 18, 723.
(302) Sano, T.; Hasegawa, M.; Ejiri, S.; Kawakami, Y.; Yanagishita, H. (353) Koresh, J. E.; Soffer, A. J. Chem. Soc., Faraday Trans. 1 1986, 82,
Microporous Microporous Mater. 1995, 5, 179. 2057.
(303) Falconer, J. L.; George, S. M.; Ott, A. W.; Klaus, J. W.; Noble, R. (354) Hatori, H.; Takagi, H.; Yamada, Y. Carbon 2004, 42, 1169.
D.; Funke, H. H. Modification of zeolite or molecular sieve (355) Choudhary, T. V.; Sivadinarayana, C.; Goodman, D. W. Chem. Eng.
membranes using atomic layer controlled chemical vapor deposition. J. 2003, 93, 69.
U.S. Patent 6,043,177. (356) Iijima, S. Nature 1991, 354, 56.
(304) Yan, Y. S.; Davis, M. E.; Gavalas, G. R. J. Membr. Sci. 1997, 123, (357) Service, R. F. Science 2000, 290, 246.
95. (358) Odom, T. W.; Huang, J.-L.; Kim, P.; Lieber, C. M. Nature 1998,
(305) Nenoff, T. M.; Kartin, M.; Thoma, S. G. Enhanced Selectivity of 391, 62.
Zeolites by Controlled Carbon Deposition. U.S. Patent 7,041,616. (359) Wildöer, J. W. G.; Venema, L. C.; Rinzler, A. C.; Smalley, R. E.;
(306) Gu, X.; Dong, J. H.; Nenoff, T. M.; Ozokwelu, D. E. J. Membr. Sci. Dekker, C. Nature 1998, 391, 59.
2006, 280, 624. (360) O’Connell, M. J.; Bachilo, S. M.; Huffman, C. B.; Moore, V. C.;
(307) Hong, M.; Falconer, J. L.; Noble, R. D. Ind. Eng. Chem. Res. 2005, Strano, M. S.; Haroz, E. H.; Rialon, K. L.; Boul, P. J.; Noon, W. H.;
44, 4035. Kittrell, C.; Ma, J.; Hauge, R. H.; Weisman, R. B.; Smalley, R. E.
(308) Guan, G.; Kusakabe, K.; Morooka, S. Sep. Sci. Technol. 2001, 36, Science 2002, 391, 593.
2233. (361) Dyke, C. A.; Tour, J. M. Chem.sEur. J. 2004, 10, 812.
(309) Guan, G. Q.; Tanaka, T.; Kusakabe, K.; Sotowa, K. I.; Morooka, S. (362) Williams, K. A.; Eklund, P. C. Chem. Phys. Lett. 2000, 320, 352.
J. Membr. Sci. 2003, 214, 191. (363) Dresselhaus, M. S.; Williams, K. A.; Eklund, P. C. Mat. Res. Bull.
(310) Mitchell, M. C.; Autry, J. D.; Nenoff, T. M. Mol. Phys. 2001, 99, 1999, 24, 45.
1831. (364) Skoulidas, A. I.; Ackerman, D. M.; Johnson, J. K.; Sholl, D. S. Phys.
(311) Mitchell, M.; Gallo, M.; Nenoff, T. M. J. Chem. Phys. 2004, 121, ReV. Lett. 2002, 89, 185901.
1910. (365) Sokhan, V. P.; Nicholson, D.; Quirke, N. J. Chem. Phys. 2002, 117,
(312) Gallo, M.; Nenoff, T. M.; Mitchell, M. C. Fluid Phase Equilib. 2006, 8531.
247, 135. (366) Skoulidas, A. I.; Johnson, J. K.; Sholl, D. S. J. Chem. Phys. 2006,
(313) Adams, K. L.; Li, L.; Gu, X.; Dong, J. H.; Mitchell, M. C.; Nenoff, 124, 154708.
T. M. J. Membr. Sci., in preparation. (367) Chen, H.; Sholl, D. S. J. Membr. Sci. 2006, 269, 152.
(314) Seike, T.; Matsuda, M.; Miyake, M. J. Mater. Chem. 2002, 12, 366. (368) Sholl, D. S.; Johnson, K. J. Science 2006, 312, 1003.
(315) Bernal, M. P.; Coronas, J.; Menéndez, M.; Santamarı́a, J. AIChE J. (369) Holt, J. K.; Park, H. G.; Wang, Y.; Stadermann, M.; Artyukhin, A.
2004, 50, 127. B.; Grigoropoulos, C. P.; Noy, A.; Bakajin, O. Science 2006, 312,
(316) Gu, X.; Dong, J. H.; Nenoff, T. M. Ind. Eng. Chem. Res. 2005, 44, 1034.
937. (370) Hinds, B. J.; Chopra, N.; Rantell, T.; Andrews, R.; Gavalas, V.;
(317) Li, S.; Martinek, J. G.; Falconer, J. L.; Noble, R. D.; Gardner, T. Q. Bachas, L. G. Science 2004, 303, 62.
Ind. Eng. Chem. Res. 2005, 44, 3220. (371) Weller, S.; Steiner, W. A. Chem. Eng. Prog. 1950, 46, 585.
(318) DOE/H2 Multi-Year Research, Development and Demonstration Plan, (372) Zolandz, R. R.; Fleming, G. K. Gas Permeation Applications. In
available electronically at http://www.1.eere.energy.gov/hydroge- Membrane Handbook; Ho, W. S. W., Sirkar, K. K., Eds.; Chapman
nandfuelcells/. and Hall: New York, 1992; p 78.
(319) FutureGen coalition formed. Power Eng. 2005, 109, 16. (373) Gardner, R. J.; Crane, R. A.; Hannan, J. F. Chem. Eng. Prog. 1977,
(320) Peltier, R. Power 2003, 147, 52. 73, 76.
(321) Williams, M. C.; Strakey, J. P.; Surdoval, W. A. J. Power Sources (374) Koros, W. J.; Mahajan, R. J. Membr. Sci. 2000, 175, 181.
2005, 143, 191. (375) Henis, J. M. S.; Tripodi, M. K. Multicomponent Membranes for Gas
(322) Saufi, S. M.; Ismail, A. F. Carbon 2004, 42, 241. Separations. United States Patent 4,230,463, 1980.
(323) Fuertes, A. B.; Centeno, T. A. J. Membr. Sci. 1998, 144, 105. (376) Schell, W. J.; Houston, C. D. Chem. Eng. Prog. 1982, 78, 33.
(324) Ismail, A. F.; David, L. I. B. J. Membr. Sci. 2001, 193, 1. (377) Bollinger, W. A.; Long, S. P.; Metzger, T. R. Chem. Eng. Prog.
(325) Shiflett, M. B.; Foley, H. C. J. Membr. Sci. 2000, 179, 275. 1984, 80, 51.
(326) Shiflett, M. B.; Foley, H. C. Science 1999, 285, 1902. (378) Bollinger, W. A.; Maclean, D. L.; Narayan, R. S. Chem. Eng. Prog.
(327) Hayashi, J.; Mizuta, H.; Yamamoto, M.; Kusakabe, K.; Morooka, S. 1982, 78, 27.
J. Membr. Sci. 1997, 124, 243. (379) Farrauto, R.; et al. Annu. ReV. Mater. Res. 2003, 33, 1.
(328) Wang, H.; Zhang, L.; Gavalas, G. R. J. Membr. Sci. 2000, 177, 25. (380) Sato, S.; Nagai, K. Membranes 2005, 30, 20.
(329) Fuertes, A. B.; Centeno, T. A. Microporous Mesoporous Mater. 1998, (381) Knudsen, M. The Kinetic Theory of Gases; Some Modern Aspects;
26, 23. Methuen’s Monographs on Physical Subjects; Methuen: London,
(330) Acharya, M.; Foley, H. C. J. Membr. Sci. 1999, 161, 1. 1952.
(331) Liang, C.; Sha, G.; Guo, S. Carbon 1999, 37, 1391. (382) Hines, A. L.; Maddox, R. N. Mass Transfer; Prentice Hall PTR:
(332) Morthon-Jones, D. H. Polymer processing; Chapman and Hall: Upper Saddle River, NJ, 1985; p 553.
London, 1984; Chapter 2. (383) Hwang, S. T.; Kammerme, K. Can. J. Chem. Eng. 1966, 44, 82.
(333) Soffer, A.; Rosen, D.; Saguee, S.; Koresh, J. Carbon membranes. (384) Lee, K. H.; Hwang, S. T. J. Colloid Interface Sci. 1986, 110, 544.
GB patent 2207666. (385) Masaryk, J. S.; Fulrath, R. M. J. Chem. Phys. 1973, 59, 1198.
(334) Schindler, E.; Maier, F. Manufacture of porous carbon membranes. (386) Baker, R. W. Membrane Technology and Applications; McGraw-
U.S. patent 4919860. Hill: New York, 2000.
(335) Suda, H.; Haraya, K. J. Phys. Chem. B 1997, 101, 3988. (387) Rautenbach, R.; Albrecht, R. Membrane Processes; John Wiley &
(336) Geiszler, V. C.; Koros, W. J. Ind. Eng. Chem. Res. 1996, 35, 2999. Sons: Chichester, 1989.
(337) Strathmann, H. Membr. Technol. 1999, 113, 9. (388) Stern, S. A. J. Membr. Sci. 1994, 94, 1.
(338) Yoneyama, H.; Nishihara, Y. Carbon based porous hollow fiber (389) Vu, D. Q.; Koros, W. J.; Miller, S. J. J. Membr. Sci. 2003, 211, 311.
membrane and method for producing same. U.S. patent 5089135. (390) Singh, A.; Koros, W. J. Ind. Eng. Chem. Res. 1996, 35, 1231.
(339) Petersen, J.; Matsuda, M.; Haraya, K. J. Membr. Sci. 1997, 131, 85. (391) Lin, H.; Freeman, B. D. J. Membr. Sci. 2004, 239, 105.
(340) Pietrass, T. MRS Bull. 2006, 31, 765. (392) Reid, R. C.; Prausnitz, J. M.; Poling, B. E. The Properties of Gases
(341) Rao, A. M.; Sircar, S. Gas Sep. Purif. 1993, 7, 279. and Liquids, 4th ed.; McGraw Hill: Boston, MA, 1987; p 752.
(342) Rao, A. M.; Sircar, S. J. Membr. Sci. 1993, 85, 253. (393) Robeson, L. M. J. Membr. Sci. 1991, 62, 165.
(343) Viera-Linhares, A. M.; Seaton, N. A. Chem. Eng. Sci. 2003, 58, 4129. (394) Aitken, C. L.; Koros, W. J.; Paul, D. R. Macromolecules 1992, 25,
(344) Sircar, S.; Waldron, W. E.; Rao, M. B.; Anand, M. Sep. Purif. 3651.
Technol. 1999, 17, 11. (395) Aitken, C. L.; Koros, W. J.; Paul, D. R. Macromolecules 1992, 25,
(345) Viera-Linhares, A. M.; Seaton, N. A. Chem. Eng. Sci. 2003, 58, 5251. 3424.
(346) Koresh, J. E.; Soffer, A. Sep. Sci. Technol. 1987, 22, 973. (396) Aitken, C. L.; Paul, D. R.; Mohanty, D. K. J. Polym. Sci., Part B:
(347) Villar-Rodil, S.; Denoyel, R.; Rouquerol, J.; Martı́nez-Alonso, A.; Polym. Phys. 1993, 31, 983.
Tascón, J. M. D. Chem. Mater. 2002, 14, 4328. (397) Aitkin, C. L.; Paul, D. R. J. Polym. Sci., Part B: Polym. Phys. 1993,
(348) Koresh, J. E.; Soffer, A. J. Chem. Soc., Faraday Trans. 1 1980, 76, 31, 1061.
2457. (398) Bixler, H. J.; Sweeting, O. J. Barrier Properties of Polymer Films.
(349) Koresh, J. E.; Soffer, A. J. Chem. Soc., Faraday Trans. 1 1980, 76, In The Science and Technology of Polymer Films; Sweeting, O. J.,
2472. Ed.; Wiley-Interscience: New York, 1971; pp 1-130.
(399) Fritsch, D.; Peinemann, K. V. J. Membr. Sci. 1995, 99, 29. (451) Way, J. D.; Noble, R. D. In Membrane Handbook; Ho, W. S. W.,
(400) Hofman, D.; et al. Polymer 1996, 37, 4773. Sirkar, K. K., Eds.; Van Nostrand Reinhold: New York, 1992; p
(401) Kita, H.; Tabuchi, M.; Sakai, T. Polymer Separation Membrane. U.S. 833.
Patent 6,656,252 B2. (452) Way, J. D.; Noble, R. D.; Reed, D. L.; Ginley, G. M.; Jarr, L. A.
(402) McHattie, J. S.; Koros, W. J.; Paul, D. R. Polymer 1991, 32, 2618. AIChE J. 1987, 33, 480.
(403) McHattie, J. S.; Koros, W. J.; Paul, D. R. J. Polym. Sci., Part B: (453) Way, J. D.; Hapke, R. L. Prepr. Pap.sAm. Chem. Soc., DiV. Fuel
Polym. Phys. 1991, 29, 731. Chem. 1988, 33, 283.
(404) McHattie, J. S.; Koros, W. J.; Paul, D. R. Polymer 1991, 32, 840. (454) Pelligrino, J. J.; Nassimbene, R.; Ko, M.; Noble, R. D. Proc. Ninth
(405) McHattie, J. S.; Koros, W. J.; Paul, D. R. Polymer 1992, 33, 1701. Annu. Gasification Gas Stream Cleanup Syst. Contractors ReV.
(406) Min, K. E.; Paul, D. R. J. Polym. Sci., Part B: Polym. Phys. 1988, Meeting 1989, 211.
26, 1021. (455) Noble, R. D.; Pellegrino, J. J.; Grosgogeat, E.; Sperry, D.; Way, J.
(407) Mohr, J. M.; et al. J. Membr. Sci. 1991, 56, 77. D. Sep. Sci. Techmol. 1988, 23, 1595.
(408) Mohr, J. M.; et al. Polymer 1991, 32, 2387. (456) Langevin, D.; Pinoche, M.; Selegny, E.; Metayer, M.; Roux, R. J.
(409) Nagasaki, Y.; et al. Makromol. Chem., Rapid Commun. 1989, 10, Membr. Sci. 1993, 82, 51.
255. (457) Kim, M.; Park, Y.; Youm, K.; Lee, K. J. Membr. Sci. 2004, 245, 79.
(410) Srinivasan, R.; Auvil, S. R.; Burban, P. M. J. Membr. Sci. 1994, 86, (458) Matsuyama, H.; Teramoto, M.; Iwai, K. J. Membr. Sci. 1994, 93,
67. 237.
(411) Stern, S. A.; et al. J. Polym. Sci., Part B: Polym. Phys. 1989, 27, (459) Matsuyama, H.; Teramoto, M.; Sakakura, H.; Iwai, K. J. Membr.
1887. Sci. 1996, 117, 251.
(412) Takada, K.; et al. J. Appl. Polym. Sci. 1985, 30, 1605. (460) Matsuyama, H.; Teramoto, M.; Matsui, K.; Kitaura, Y. J. Appl. Polym.
(413) Tanaka, K.; et al. Polymer 1992, 33, 585. Sci. 2001, 81, 936.
(414) Yamamoto, H.; et al. J. Polym. Sci., Part B: Polym. Phys. 1990, 28, (461) Ho, W. S. W. U.S. Patent No. 5,611,843.
2291. (462) Ho, W. S. W. U.S. Patent No. 6,099,621.
(415) Freeman, B. D. Macromolecules 1999, 32, 375. (463) Ho, W. S. W. U.S. Patent No. 6,579,331.
(416) Kita, H.; et al. J. Membr. Sci. 1994, 87, 139. (464) Quinn, R.; Laciak, D. V. U.S. Patent 6,315,968.
(417) Liu, Y.; Ding, M. X.; Xu, J. P. J. Appl. Polym. Sci. 1995, 58, 485. (465) Quinn, R.; Laciak, D. V.; Appleby, J. B.; Pez, G. P. U.S. Patent
(418) Wright, C. T.; Paul, D. R. J. Membr. Sci. 1997, 129, 47. 5,336,298.
(419) Ding, Y.; Bikson, B.; Nelson, J. K. Polymide Gas Separation (466) Quinn, R.; Laciak, D. V. J. Membr. Sci. 1997, 131, 49.
Membranes. U.S. Patent 6,790,263. (467) Quinn, R. J. Membr. Sci. 1998, 139, 97.
(420) Kawakami, H.; et al. Gas Separation Membrane and Method of (468) Quinn, R.; Laciak, D. V.; Pez, G. P. J. Membr. Sci. 1997, 131, 61.
Producing the Same. U.S. Patent 6,709,491 B2.
(469) Namboodhiri, T. K. G. Trans. Indian Inst. Met. 1984, 37, 764.
(421) Baker, R. W.; et al. Gas Separation using Organic-Vapor-Resistant
Membranes in Conjunction with Organic-Vapor-Selective Mem- (470) Birnbaum, H. K. In Environmental Sensitive Fracture of Engineering
branes. U.S. Patent 6,572,697 B2. Materials; Proceedings of Symposium on EnVironmental Effects on
Fracture, Chicago, Illinois, 1977; Foroulis, Z. A., Ed.; Metallurgical
(422) Bikson, B.; et al. Hollow Fiber Membrane Gas Separation Cartridge
Society of AIME: Warendale, PA, pp 326-360.
and Gas Purification Assembly. U.S. Patent 6,814,780 B2.
(423) Engler, Y.; Fuentes, F. Plant for the Production of Hydrogen and of (471) Buschow, K. H. J.; Bouten, P. C. P.; Miedema, A. R. Rep. Prog.
Energy. U.S. Patent 5,989,501. Phys. 1982, 9, 937.
(424) Fuentes, F. Installation for the Production of Pure Hydrogen from a (472) Shkolnik, I. V.; Kulsartov, T. V.; Tazhibaeva, I. L.; Shestakov, V.
Gas Containing Helium. U.S. Patent 6,669,922 B1. P. Fusion Technol. 1998, 34, 868.
(425) Lokhandwala, K. A.; Baker, R. W. Hydrogen/Hydrocarbon Separation (473) Cermák, J.; Rothová, V. Intermetallics 2001, 9, 403.
Process, Including PSA and Membranes. U.S. Patent 6,592,749. (474) Takano, T.; Ishikawa, K.; Matsuda, T.; Aoki, K. Mater. Trans., JIM
(426) Siadous, N.; Engler, Y.; Monereau, C. Method for Separating a Gas 2004, 45, 3360.
Mixture with a Permeation Membrane Unit. U.S. Patent 6,977,007 (475) Komiya, K.; Shinzato, Y.; Yukawa, H.; Morinaga, M.; Yasuda, I.
B2. Fifth Pacific Rim Int. Conf. AdV. Mater. Process., Pts 1-5 2005,
(427) Wallace, P. S.; Kasbaum, J. L.; Johnson, K. A. Hydrogen Recycle 475-479, 2497.
and Acid Gas Removal Using a Membrane. U.S. Patent 6,416,568 (476) Cheng, X. Y.; Wu, Q. Y.; Sun, Y. K. J. Alloys Compd. 2005, 389,
B1. 198.
(428) Yamashita, N.; Yamamoto, T. Method and Apparatus for Recovering (477) Hashi, K.; Ishikawa, K.; Matsuda, T.; Aoki, K. Mater. Trans., JIM
a gas from a gas mixture. U.S. Patent 6,197,090 B1. 2005, 46, 1026.
(429) Ekiner, O. M. Gas Separation Membranes of Blends of Polyether- (478) Aoki, K.; Matsuda, T.; Ishikawa, K. U.S. Patent 0,217,480 A1.
sulfones with Aromatic Polyimides. U.S. Patent 5,917,137. (479) Jang, T. H.; Lee, J. Y. J. Non-Cryst. Solids 1990, 116, 73.
(430) MacKinnon, S. M., Process for Preparing Graft Copolymers and (480) Yan, S.; Maeda, H.; Kusakabe, K.; Morooka, S.; Akiyama, Y. Ind.
Membranes Formed Therefrom. U.S. Patent 6,828,386 B2. Eng. Chem. Res. 1994, 33, 2696.
(431) Nakanishi, S.; et al. Gas Separation Membrane and Method for its (481) Hwang, G. J.; Onuki, K.; Shimizu, S.; Ohya, H. J. Membr. Sci. 1999,
Use. U.S. Patent 6,464,755. 162, 83.
(432) Simmons, J. W. Block Polyurethane-ether and Polyurea-ether Gas (482) Koukou, M. K.; Papayannakos, N.; Markatos, N. C.; Bracht, M.;
Separation Membranes. U.S. Patent 6,843,829. Van Veen, H. M.; Roskam, A. J. Membr. Sci. 1999, 155, 241-259.
(433) Simmons, J. W. Block Polyester-ether Gas Separation Membranes. (483) Schafer, R.; Noack, M.; Kolsch, P.; Thomas, S.; Seidel-Morgenstern,
U.S. Patent 6,860,920 B2. A.; Caro, J. Sep. Purif. Technol. 2001, 25, 3.
(434) Perry, J. D.; Nagai, K.; Koros, W. J. MRS Bull. 2006, 31, 745. (484) Tsuru, T.; Yamaguchi, K.; Yoshioka, T.; Asaeda, M. AIChE J. 2004,
(435) Merkel, T. C.; et al. J. Membr. Sci. 2001, 191, 85. 50, 2794.
(436) Pinnau, I.; He, Z. J. J. Membr. Sci. 2004, 244, 227. (485) Makino, H.; et al. Process for Preparing Aromatic Polyimide
(437) Hirayama, Y.; et al. J. Membr. Sci. 1999, 160, 87. Semipermeable Membranes. U.S. Patent 4,378,324.
(438) Nagai, K.; et al. J. Membr. Sci. 2000, 172, 167. (486) Breck, D. W. Zeolite Molecular SieVes; John Wiley & Sons: New
(439) Nagai, K.; Nakagawa, T. Adv. Materials for Membrane Separations; York, 1974; p 783.
ACS Symposium Series; American Chemical Society: Washington, (487) Agrawal, R.; Offutt, M.; Ramage, M. P. AIChE J.2005, 51, 1582.
DC, 2004. (488) Ramage, M. P. The Hydrogen Economy: Opportunities, Costs,
(440) Suzuki, H.; et al. J. Membr. Sci. 1998, 146, 31. Barriers, and R&D Needs; National Research Council of the National
(441) Kazama, S.; et al. 7th International Conference on Greenhouse Gas Academies: 2004; pp ES-1 to ES-14.
Control Technologies, Vancouver, Canada, 2004. (489) Simbeck, D. R. Energy 2004, 29, 1633.
(442) Nagai, K. KOUBUSHIRONBUNSHU 2004, 61, 420. (490) Summers, W. A.; Gorensek, M. B. Chem. Eng. Prog. 2005, 101, 4.
(443) Nagai, K. Membrane 2004, 29, 42. (491) Winter, C. J. Int. J. Hydrogen Energy 2005, 30, 681.
(444) Anand, M.; et al. J. Membr. Sci. 1997, 123, 17. (492) Orme, C. J.; Stone, M. L.; Benson, M. T.; Peterson, E. S. Sep. Sci.
(445) Bondar, V. I.; Freeman, B. D.; Pinnau, I. J. Polym. Sci., Part B: Technol. 1995, 38, 3225-3283.
Polym. Phys. 2000, 38, 2051. (493) Zhang, Y.; Wang, Z.; Wang, S. J. Appl. Polym. Sci. 2002, 86, 2222.
(446) Pinnau, I.; et al. J. Polym. Sci., Part B: Polym. Phys. 1996, 34, 2613. (494) Matsuyama, H.; Terada, A.; Nakagawara, T.; Kitamura, Y.; Teramoto,
(447) Raharjo, R. D.; et al. Polymer 2005, 46, 6316. M. J. Membr. Sci. 1999, 163, 221.
(448) Robeson, L. M.; et al. Polymer 1994, 35, 4970. (495) Kim, T. J.; Li, B.; Hägg, M. B. J. Polym. Sci. B 2004, 42, 4326.
(449) LeBlanc, O. H.; Ward, W. J.; Matson, S. L.; Kimura, S. G. J. Membr. (496) Trachtenberg, M. C. Proceedings of EFC-conference on AdVanced
Sci. 1980, 6, 339. Membrane Technology, Barga, Italy, 2001.
(450) Kimura, S. G.; Ward, W. J.; Matson, S. L. U.S. Patent No. 4,318,-
714. CR0501792

Membranes For Hydrogen Separation 2007 PDF

Uploaded by

Copyright:

Available Formats

Membranes For Hydrogen Separation 2007 PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Membranes For Hydrogen Separation 2007 PDF

Uploaded by

Copyright:

Available Formats

4078 Chem. Rev.

2007, 107, 4078−4110

Membranes for Hydrogen Separation

Received November 27, 2006

Contents 6.3. H2 versus CO2 Selective Polymeric 4103

Many H2 membrane separation technologies are based on

Initial reforming reaction:

Water-gas shift (WGS) reaction:

Table 1. Current Status and Future H2 Membrane Property

combined result of these composition and performance issues

-DH2∂CH2 -DH2SH2∂PH1/22 -DH2SH2∆PH1/22

where JH2 is the hydrogen flux, DH2 is the concentration

The permeation and selectivity values associated with any

Scheme 1. Namboodhiri’s469 Taxonomy of H2 Damage in Solids

the three layers that comprise a silica membrane will be more

tion222 and centrifugal casting,236 resulting in high strength,

industrial applications. Currently, research is ongoing to

4.1. Membrane Growth Methods

The separation selectivity, R, is defined as the enrichment

4.3. Defect Site Diffusion/Nonzeolitic Pores

4.4. Thin Films

5.1. Carbon Membrane Preparations applications of membranes, it is preferable to fabricate a

carbonization. They used commercially available polyimide

5.7. Carbon Nanotubes

Table 10. Hydrogen Separation Ability of First-Generation,

Table 11. Hydrogen Purity Required in Industry380

Table 14. Permselective Properties of CO2-selective Polymeric Facilitated Transport Membranes

on poly(amido-amine) dendrimers. This membrane has a

Table 15. Comparison of Membrane Classes

You might also like

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.