-
Chemical Engineering Science 92 (2013) 4066
Contents lists available at SciVerse ScienceDirect
Chemical Engineering Science
0009-25
http://d
n Corr
E-m
journal homepage: www.elsevier.com/locate/ces
Review
Recent advances on membranes and membrane reactorsfor hydrogen
production
Fausto Gallucci a,n, Ekain Fernandez b, Pablo Corengia b, Martin
van Sint Annaland a
a Multiphase Reactors Group, Department of Chemical Engineering
and Chemistry, Eindhoven University of Technology, 5612 AZ
Eindhoven,
The Netherlandsb Tecnalia, Mikeletegi Pasealekua 2, 20009
Donostia, San Sebastian, Spain
H I G H L I G H T S
c Recent advances in hydrogen selective membranes are
presented.c Commercial membranes for hydrogen production are
discussed.c An overview of the membrane reactor concepts is
highlighted.c The application of membrane reactors to different
feedstock is reported.
a r t i c l e i n f o
Article history:
Received 22 October 2012
Received in revised form
17 December 2012
Accepted 6 January 2013Available online 23 January 2013
Keywords:
Membrane reactor
Hydrogen production
Packed bed
Fluidization
Separations
Membranes
09/$ - see front matter & 2013 Elsevier Ltd. A
x.doi.org/10.1016/j.ces.2013.01.008
esponding author. Tel.: 31 40 247 3675; faxail address:
[email protected] (F. Gallucci).
a b s t r a c t
Membranes and membrane reactors for pure hydrogen production are
widely investigated not only
because of the important application areas of hydrogen, but
especially because mechanically and
chemically stable membranes with high perm-selectivity towards
hydrogen are available and are
continuously further improved in terms of stability and hydrogen
flux. Membrane reactors are
(multiphase) reactors integrating catalytic reactions (generally
reforming and water gas shift reactions
for hydrogen production) and separation through membranes in a
single unit. This combination of
process steps results in a high degree of process
integration/intensification, with accompanying
benefits in terms of increased process or energy efficiencies
and reduced reactor or catalyst volume.
The aim of this review is to highlight recent advances in
hydrogen selective membranes (from
palladium-based to silica and proton conductors) along with the
advances for the different types of
membrane reactors available (from packed bed to fluidized bed,
from micro-reactors to bio-membrane
reactors). In addition, the application of membrane reactors for
hydrogen production from different
feedstock is also discussed.
& 2013 Elsevier Ltd. All rights reserved.
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . 41
2. Hydrogen separation membranes . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . 43
2.1. Membrane materials . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . 43
2.2. Membrane configurations . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
44
2.2.1. Geometry . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
44
2.2.2. Unsupported and supported membranes . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . 44
3. Dense metal membranes . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . 44
3.1. General comments . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . 44
3.2. Progresses in commercialization of dense metal membranes .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . 47
3.2.1. CRI/Criterion . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
47
3.2.2. ECN . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . 47
ll rights reserved.
: 31 40 247 5833.
www.elsevier.com/locate/ceswww.elsevier.com/locate/ceshttp://dx.doi.org/10.1016/j.ces.2013.01.008http://dx.doi.org/10.1016/j.ces.2013.01.008http://dx.doi.org/10.1016/j.ces.2013.01.008http://crossmark.dyndns.org/dialog/?doi=10.1016/j.ces.2013.01.008&domain=pdfhttp://crossmark.dyndns.org/dialog/?doi=10.1016/j.ces.2013.01.008&domain=pdfhttp://crossmark.dyndns.org/dialog/?doi=10.1016/j.ces.2013.01.008&domain=pdfmailto:[email protected]://dx.doi.org/10.1016/j.ces.2013.01.008
-
F. Gallucci et al. / Chemical Engineering Science 92 (2013) 4066
41
3.2.3. Eltron Research, Inc. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
47
3.2.4. Green Hydrotec . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
47
3.2.5. Hy9 . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . 47
3.2.6. M&P . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . 48
3.2.7. MRT . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. 48
3.2.8. Pall Corporation . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
48
3.2.9. REB . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . 48
3.2.10. Tokyo Gas . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. 49
3.2.11. UTRC . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . 49
4. Microporous membranes . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . 49
4.1. Zeolite membranes. . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . 49
4.2. Metalorganic framework membranes . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.3. Silica membranes . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . 50
4.4. Carbon membranes . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . 50
5. Proton conducting membranes . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . 51
5.1. Dense ceramic membranes . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
51
5.1.1. Perovskite-type membranes . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . 51
5.1.2. Non-perovskite-type membranes . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . 51
5.2. Cermet membranes . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . 52
6. Advances in membrane reactors for hydrogen production . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
6.1. Packed bed membrane reactors. . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
52
6.2. Fluidized bed membrane reactors . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
55
6.3. Membrane micro-reactors . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
59
6.4. Membrane bio-reactors . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. 59
7. Feedstock for hydrogen production. . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . 60
7.1. H2 production from methane in MR . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
60
7.2. H2 production from other hydrocarbons in MR . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . 61
7.3. H2 production from biological related feedstock in MR . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . 61
8. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . 62
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . 62
References . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . 62
Web references . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . 66
1. Introduction
It is widely accepted that the solution of the global
warmingproblem will be a combination of various contributions
rangingfrom carbon capture and sequestration (CCS) to improved
carbonefficiency of fossil fuels, to large use of renewable energy
sources(in the long term). Among the various strategies proposed,
CCSseems to be a good candidate for CO2 emissions
mitigation;however, CCS has to be regarded as a mid-term solution
as longas the energy economy is based on fossil fuels: CCS will be
used aslong as other technologies such as large-scale exploitation
ofrenewable sources and nuclear fusion are still under
develop-ment. CCS consists of two challenging processes, viz.
carbondioxide capture and its sequestration (mineralization or
storage).In this review only carbon capture is discussed, while
sequestra-tion (perhaps in off shore geological formations) is a
process withits own technological and societal challenges. Three
main tech-nological paths have been proposed for CCS from
fossil-fuel firedpower stations, namely the post-combustion
capture, the pre-combustion decarbonization and the oxyfuel
combustion route(see Fig. 1).
This review focuses on separation and production of hydrogenthus
can be applied to the decarbonization route. Hydrogen fromfossil
fuels can also be used to increase the efficiency of thesystem. An
example of increased efficiency is the conversion ofraw material
into hydrogen and its use in fuel cells. In fact, forexample in
automotive applications, electromotor combined withhydrogen powered
fuel cells shows an overall efficiency (4055%)significantly higher
than internal combustion engines (1330%).
Pt based catalyst are commonly used in polymer
electrolytemembrane (PEM) fuel cells, the performance of the
catalyst
decrease dramatically if carbon monoxide (a main product ofmost
of the conversion technologies) or hydrogen sulfide ispresent. For
this reason, the interest in the production of ultra-pure hydrogen
has strongly increased in the last years. It has beendemonstrated
that by using pure hydrogen produced by mem-brane reactors in
co-generation units the total efficiency will beincreased by
roughly 10% (Roses et al., 2010, 2011).
Traditionally, hydrogen is produced via steam reforming (SR)of
hydrocarbons such as methane, naphtha oil or methanol/ethanol. But
on industrial scale most of the hydrogen (more than80%) is
currently produced by SR of natural gas carried out inlarge
multi-tubular fixed-bed reactors. In small-scale applications,two
other main alternatives are generally considered along withSR:
partial oxidation reactions, with a significantly lower effi-ciency
than SR, and auto-thermal reforming, where the partialoxidation
(exothermic reaction) and SR (endothermic reaction)are carried out
in the same reactor.
The main drawbacks of conventional SR, partial oxidation
andauto-thermal conventional reactors are that all these
reactionsare equilibrium limited and (even in case of complete
fuelconversion) produce a hydrogen rich gas mixture
containingcarbon oxides and other by-products. Consequently, in
order toproduce pure hydrogen, these chemical processes are carried
outin a number of reaction units (typically high temperature
refor-mer, high and low temperature shift reactors) followed
byseparation units (mostly pressure swing adsorption). The
largenumber of different process steps decreases the system
efficiencyand makes scale-down uneconomical. A typical reaction
processscheme is reported in Fig. 2.
Using this process, high hydrogen yields are achieved, butcostly
high temperature heat exchangers and complex energy
-
F. Gallucci et al. / Chemical Engineering Science 92 (2013)
406642
integration among different process units are required to
obtainthe hydrogen at the desired high purity.
Among different technologies related to production, separa-tion
and purification of H2, membrane technologies seem to be
Fig. 2. Conventional steam reforming reaction scheme. HT shift
and L
Fig. 3. Membrane system and involv
Post-CombustionCapture
Pre-CombustionCapture
OxyfuelCombustion
Power & Heat
Air separatio
Gasification
Power &Heat
Air
Air/Steam
Air
O2
CoalGasBiomass
CoalBiomass
CoalGasBiomass
GasOil
Reforming+ CO2
separation
Fig. 1. The three main CO2 capture rout
the most promising and membrane separation is nowadays
increas-ingly considered as a good candidate for substituting
conventionalsystems. The specific thermodynamic constrains limiting
traditionalreactors can be circumvented by using innovative
integrated systems,
T shift are high and low temperature shift reactors,
respectively.
ed gas streams (Lu et al., 2007).
CO2separation
n
CO2compression &
Dehydration
Power &Heat
N2 , O2
CO2
CO2
CO2
Air
N2
es (adapted from Metz et al., 2005).
-
F. Gallucci et al. / Chemical Engineering Science 92 (2013) 4066
43
such as the so-called membrane reactors (MRs), engineering
systemsin which both reaction and separation are carried out in the
samedevice (see Fig. 3).
In comparison to a conventional configuration in which areactor
is combined with a downstream separation unit, the useof membrane
reactors can bring various potential advantagessuch as reduced
capital costs (due to the reduction in size of theprocess unit),
improved yields and selectivities (due to theequilibrium shift
effect) and reduced downstream separationcosts (separation is
integrated).
The success of membrane reactors for hydrogen productiondepends
crucially on: (i) the advances in the membrane produc-tion methods
for the production of thin membranes with highhydrogen fluxes and
high hydrogen perm-selectivities; (ii) thedesign of innovative
reactor concepts which allow the integrationof separation and
energy exchange, the reduction of mass andheat transfer resistances
and the simplification of the housing andsealing of the
membranes.
In this review we will first discuss recent advances in
(hightemperature) membranes to be applied in hydrogen
productionunits, categorized into dense metal membranes,
microporousmembranes and proton conducting membranes. An overview
ofthe most permeable and selective membranes is given togetherwith
a collection of commercial membranes for hydrogen produc-tion.
Subsequently, recent developments in membrane reactorsdesign will
be highlighted and the application of membranereactors for hydrogen
production from different feedstock willbe discussed.
2. Hydrogen separation membranes
Membranes are basically barriers that allow the flow of
somecomponents of a feed gas mixture stream. The stream
containingthe components that permeate through the membrane is
calledpermeate and the stream containing the retained components
iscalled retentate, as shown in Fig. 3. Membranes for
hydrogenseparation should have the following characteristics:
1.
TabCom
M
M
T
H
H
T
St
P
C
High selectivity towards hydrogen.
2. High flux.3. Low cost.4. High mechanical and chemical stability.
Fig. 4. Solutiondiffusion mechanism of hydrogen permeation
through a densemetal membrane (Yun and Oyama, 2011).
There have been many applications of catalytic inorganicmembrane
reactors for reactions involving hydrogen, such as
le 1parison of membrane types for hydrogen separation (Kluiters,
2004; Liu et al., 201
embrane type Polymeric Microporous ceramic
aterials Polymers: polyimide, cellulose
acetate, polysulfone, etc.
Silica, alumina, zirconia, titan
metal-organic frameworks (M
emperature (1C) o100 2006002 selectivity Low 5139
2 flux
(103 mol m2
s1) at DP1 bar
Low 60300
ransport
mechanism
Solution-diffusion Molecular sieving
ability issues Swelling, compaction,
mechanical strength
Stability in H2O
oisoning issues HCl, SOx, CO
ost Low Low
hydrogenation and dehydrogenation (Dittmeyer et al., 2001;Gimeno
et al., 2009), methane steam reforming (Matsumura andTong, 2008),
and watergas shift (Bi et al., 2009).
2.1. Membrane materials
Regarding the type of materials, hydrogen separation mem-branes
may be classified into the following categories:
polymericmembranes, porous membranes, dense metal membranes
andproton conducting membranes. Table 1 shows a comparisonbetween
the different membrane types for hydrogen separation.The most
important parameters when comparing membranes arethe
perm-selectivity, the flux and the temperature range at whichthe
membranes can be applied.
In order to obtain high purity hydrogen-permeate streams,dense
metal membranes (mainly palladium alloys) and denseceramic
membranes are currently the most suitable materialsdue to their
high hydrogen selectivity. Pdalloys (mainly PdAg,PdCu and PdXAu)
are used to decrease the embrittlementproblem and to decrease the
poisoning of the membrane when incontact with H2S and other
pollutants (such as CO). Microporousceramic membranes are promising
materials for high purityhydrogen production and separation.
However, these membranesseparate hydrogen by size exclusion and
thus selectivity is stilllimited compared to more expensive dense
inorganic membranes.Nowadays, the selective layers are thin (for
instance, Pd mem-branes are now produced with thicknesses in the
order of 1 mm)and therefore the cost of the selective layer is low
or moderatecompared to the entire cost of the membrane.
0)
Porous carbon Dense metallic Proton conducting
dense ceramic
ia, zeolites,
OF)
Carbon Palladium alloys Perovskites (mainly
SrCeO3d, BaCeO3d)
500900 300700 600900
420 41000 4100010200 60300 680
Surface diffusion,
molecular sieving
Solution-diffusion Solution-diffusion
Brittle, oxidizing Phase transition
(causes
embrittlement)
Stability in CO2
Strong adsorbing
vapors, organics
H2S, HCl, CO H2S
Low Moderate Low
-
F. Gallucci et al. / Chemical Engineering Science 92 (2013)
406644
2.2. Membrane configurations
2.2.1. Geometry
Membrane geometries may be planar, tubular (that includestubes,
capillaries and hollow fibers), plate and frame and spiralwound.
Currently the most commonly used geometries for gasseparation are
planar and tubular. The planar membranes are oftenused in earlier
laboratory research and development studies, whilefor medium scale
and industrial scale the tubular membranes arethe most preferred
option (due to their higher surface area-to-volume ratio in
comparison to planar membranes).
2.2.2. Unsupported and supported membranes
The membranes may be classified into unsupported andsupported
membranes.
Unsupported membranes need to be thick self-standing films(450
mm thick) in order to have a minimum mechanical stability.The main
drawback of these membranes is their low hydrogenpermeance.
Moreover, in the case of using an expensive mem-brane material the
cost of the whole membrane will be sharplyincreased by increasing
the membrane thickness. Thus there is athreshold between membrane
mechanical stability and mem-brane thickness (and thus flux and
costs). For this reason it isforeseen that the first industrially
available membranes will besupported membranes.
Supported membranes consist of a thin selective film depo-sited
onto a support that provides mechanical stability. Thus,
thehydrogen permeance of the membrane will be higher so that
lessmembrane area is required and the whole membrane cost will
belower than that for unsupported membranes. However, in thetotal
membrane cost, the cost of the support also becomesimportant.
Especially when very thin film membranes areselected, the support
pore size should be much lower and thesurface much smoother, so
that its cost will increase.
There are mainly two types of porous support materials:metallic
and ceramic.
Ceramic supports typically have better surface quality
provid-ing membranes with thinner selective layers. However, they
aremore fragile. Currently, the ceramic supports that are
commer-cially available are tubes and hollow fibers. As an example,
Inoporproduces and offers ceramic tubular supports of different
ceramicmaterials and pore sizes (http://www.inopor.com/). In
general,ceramic tubular supports consist of a porous ceramic
substrateprocessed by extrusion and a porous low roughness ceramic
layerdeposited onto the support by, for example, solgel
techniques.On the other hand, hollow fibers are usually prepared by
spinning(Li, 2007b).
Metallic supports are more robust than ceramic but thecommercial
ones, mainly tubular, have lower surface qualities.This is due to
the fact that they are not used as membranesupports but rather as
particle filters. GKN (http://www.gkn-filters.de/), Mott
(http://www.mottcorp.com/), and Pall (http://www.pall.com) are
known metallic support suppliers.
Some research groups are developing hollow fiber supports(both
ceramics and metallic) in order to increase the surface areato
volume ratio and also improve the surface quality (Luiten-Olieman
et al., 2012).
In the case in which both the selective layer and the supportare
metallic and the entire membrane is used at temperaturesabove the
Tamman temperature of one of the metallic parts, it isnecessary to
position an inter-metallic diffusion barrier layerbetween the
metallic support and the metallic selective layer.
This is a common situation in dense metal composite mem-branes
used for high temperature reforming reactions. In this casethe
materials used as barrier layer are: ZrO2 (Tarditi et al.,
2013;
Li et al., 2008a), YSZ (Sanz et al., 2011, Li et al., 2007a),
TiO2(Li et al., 2008b), CeO2 (Qiao et al., 2010) and Al2O3 (Dardas
et al.,2009). The typical deposition technologies used are:
atmosphericplasma spraying (APS) (Sanz et al., 2011; Huang and
Dittmeyer,2007), wet powder spraying (WPS) (Zhao et al., 2004) and
powdersuspension suction (Chi et al., 2010).
3. Dense metal membranes
3.1. General comments
Dense metal membranes are commonly used for high purityhydrogen
production. The mechanism of hydrogen permeationthrough dense metal
membranes has been extensively studied. Itis well known that it
generally follows a solutiondiffusionmechanism. The steps involved
in hydrogen transport from ahigh to a low pressure gas region are
the following (Lewis, 1967;Fig. 4): (a) diffusion of molecular
hydrogen to the surface of themetal membrane, (b) reversible
dissociative adsorption on themetal surface, (c) dissolution of
atomic hydrogen into the bulkmetal, (d) diffusion of atomic
hydrogen through the bulk metal,(e) association of hydrogen atom on
the metal surface, (f) desorptionof molecular hydrogen from the
surface, (g) diffusion of molecularhydrogen away from the
surface.
Common dense metal layer deposition technologies includephysical
vapor deposition (PVD, including magnetron sputtering,thermal
evaporation or pulsed laser evaporation), chemical vapordeposition
(CVD or MOCVD), electroless plating (ELP), electro-plating and
diffusion welding (Iniotakis et al., 1987). Each tech-nology has
its strengths and weaknesses; therefore, there is atrend of
tailoring the deposition technology to the features of thesupport
in order to obtain a suitable composite membrane.
The most widely used preparation technology for dense
metallayers is ELP due to its ability of covering supports with
complexgeometries, the simplicity of the required equipment and its
lowcost (absence of electrodes or electrical source) (Mallory
andHajdu, 1990).
On the other hand, PVD magnetron sputtering is a veryattractive
deposition technology because it could provide thinneruniform
layers (down to only few nanometers), much lower thanthe thickness
achieved with the ELP technique, with a controlledmicrostructure
and composition across of these coatings. Otherimportant advantage
of PVD versus ELP is its environmentalfriendly operation, without
producing waste liquids from chemi-cal baths (Klette, 2005).
Among dense metal materials used for hydrogen
purification,developments in Pd and Pd alloys have been carried out
for a longtime (Buxbaum, 1999; Grashoff et al., 1983; Holleck,
1970). Themost important problem associated with the use of pure
Pdmembranes is the hydrogen embrittlement phenomenon. Opera-tion
with hydrogen at a temperature below 300 1C and a pressurebelow 2
MPa, leads to the nucleation of the b-hydride phase fromthe a-phase
resulting in severe lattice strains. In this case a purepalladium
membrane becomes brittle after a few a2b cycles(Hsieh, 1989).
Another important problem is the palladium surface poison-ing,
which can be more significant for thin metal membranes, bysulfur
compounds (Edlund and Pledger, 1994), CO (Amandussonet al., 2000),
H2O (Li et al., 2000), chlorine, carbon, unsaturatedhydrocarbons,
etc.. In order to avoid hydrogen embrittlement,poisoning and reduce
membrane cost, palladium can be alloyedwith other metallic elements
such as Ag, Cu, Fe, Ni, Pt and Y(Bryden and Ying, 2002; Qiao et
al., 2010; Uemiya et al., 2007) orthe palladium grains can be used
in nanometer sized (PachecoTanaka et al., 2006).
http://www.inopor.com/http://www.gkn-filters.de/http://www.gkn-filters.de/http://www.mottcorp.com/http://www.pall.comhttp://www.pall.com
-
F. Gallucci et al. / Chemical Engineering Science 92 (2013) 4066
45
On the other hand, there is a group of metals called
refractorymetals (e.g., vanadium, niobium, and tantalum) that have
muchhigher permeability at these lower temperatures in comparison
topalladium. These metals are much cheaper and have
greatertolerances to high temperatures compared to palladium.
Never-theless, producing membranes with only these metals (as
purecomponents) is hindered by the high embrittlement produced
atroom temperature in the presence of 6.9 barhydrogen. On
thecontrary, several alloys of these metals are also
non-embrittling,making them good candidates for hydrogen
purification mem-branes (Dolan, 2010; Dolan et al., 2011; Ishikawa
et al., 2011;Paglieri et al., 2011; Wolden et al., 2013). These
metals are notnormally used alone for hydrogen permeation also
because ofpoor surface properties (particularly formation of
surface oxidelayers) that reduce hydrogen transport. When palladium
orpalladium alloys are applied over the surface of refractory
alloys(Makrides et al., 1967; Paglieri et al., 2009), forming
metalmetalmatrix membranes, the surface barriers are removed and
hydro-gen permeation follows the trend shown in Fig. 5.
As the number of possible combinations of these metals
isenormous, a modeling approach can help in identifying the
mostsuitable combination of metals for producing hydrogen
mem-branes. An interesting approach towards the definition of
suitablemodeling tools for this kind of study is given by the
application ofdensity functional theory (DFT) calculations to
predict the beha-vior of binary or ternary alloys with pure
hydrogen or with gasmixtures. For example the group of Sholl
published a series ofpapers showing the potentiality of DFT in
predicting the solubi-lity/diffusivity of hydrogen in different
alloys, resulting in inter-esting guidelines for the future
membrane production. In theirworks on PdCu based ternary alloys
(Semidey-Flecha et al., 2010,Kamakoti and Sholl, 2003, 2006) the
authors applied DFT tofind an effective ternary additive to the
PdCu alloy to retain thegood poisoning resistance of the binary
alloy while increasing thetotal permeability of the resulting
membrane. It was found thatamong different additives (such as Ni,
Au, Ag, Rh, Pt) an additionof Au or Ag can increase the
permeability of the alloy of up to5 times. They also found out that
for this PdCuAg alloys, theincrease of solubility due to the
increase of Ag is more dominanton the increase of permeability
compared to the effect ofdiffusivity. This means that for PdCu
ternary alloys an additivethat increases the solubility of hydrogen
is to be preferred.
1/T, 1/K0.0005 0.0010 0.0015 0.0020 0
Perm
eabi
lity,
mol
/m s
Pa1
/2
1e+01e-31e-61e-91e-121e-151e-181e-211e-241e-271e-301e-331e-361e-391e-421e-451e-48
Fig. 5. Hydrogen permeability through several metals
Solubility of hydrogen is also important for binary
alloys;Sonwane et al. (2006) have shown by using DFT calculations
thatthe maximum solubility for PdAg is found at 30% Ag while
forPdAu the maximum is found at 20% Au. For PdAg (30% Ag)
thehydrogen solubility has been found to be 10 times higher than
forpure Pd. Hao and Sholl (2011) also applied DFT to evaluate
thepermeability of amorphous metal membranes (such as ZrCubased
membranes). The results in this case show that thepermeability is
not directly related to the solubility of hydrogen,giving
guidelines also for the design of amorphous metal mem-branes. DFT
calculations have also been applied to evaluate theeffect of
contaminants in the gas mixture on the hydrogenpermeability (Ling
and Sholl, 2009; Ozdogan and Wilcox, 2010;Gallucci et al.,
2007a,b); it has been found that CO affects thepermeation of
hydrogen especially because there is a preferentialadsorption of CO
molecules on the same sites (hollow and bridge)used by H2; moreover
CO can jump between the two sites andthis is why a small amount of
CO is enough to deteriorate thehydrogen permeability (Gallucci et
al., 2007a,b). Ozdogan andWilcox showed that the adsorption of H2S
on the surface of themembrane depends on the metal used to alloy
palladium. Inparticular they found a H2S binding tendency such as
CuoPdoNb. This is why PdCu have are more resistant to H2S
poisoning.Ling and Sholl (2009) showed that any sulfide formation
on themembrane will decrease enormously the permeability of
themembrane especially due to the extremely slow diffusion ofH
through the sulfide layer.
Regarding the experimental works on membranes, some of themost
relevant current developments on dense metal membranesfor hydrogen
production are presented afterwards.
PdAg alloy membranes are usually prepared by electrolessplating
using a sequential deposition of the two metals. Thisprocedure
gives the deposition of bi-layered metal films andhence a high
temperature and longer time is required for thecomplete inter-phase
diffusion of two metals, in addition theprecise control of Pd/Ag
ratio in the membrane is difficult toobtain.
At AIST-Japan, Pacheco Tanaka et al. (2005) developed amethod
for the simultaneous plating of Pd-Ag with the desiredcomposition
of metals. This was achieved by the uniform deposi-tion of
nano-particles of Pd nuclei on the surface of the substrateand the
careful control of the composition of the plating solution.
.0025 0.0030 0.0035 0.0040
Al Be Co (eps) Co (alfa) Cu Ge Au Fe Mo Ni Nb Pd Pt Si Ta W V
Ti
as a function of temperature (Basile et al., 2008).
-
F. Gallucci et al. / Chemical Engineering Science 92 (2013)
406646
Alloying of Pd and Ag was possible under lower temperature
andshort heat treatment. The same group at AIST also prepared
thinPdAu alloy membranes by electroless plating of Au on a
Pdmembrane; the amount of Au in the alloy was controlled
byadjusting the concentration of gold in the plating
solution(Okazaki et al., 2008a). Okazaki and Pacheco Tanaka also
studiedthe effect of the substrate on the permeation of hydrogen at
hightemperature, they demonstrated that at temperatures higher
than650 1C alumina reacts with hydrogen and the PdAl alloy
formeddecrease the permeation of hydrogen in that extent that at
850 1Cno permeation occurs. On the other hand, YSZ is more stable
andcan be used until 750 1C. (Okazaki et al., 2008b, 2009). The
samegroup in Japan, developed hydrogen permeable membrane ofnovel
configuration (pore fill type membrane); in this configura-tion,
palladium particles are filled in the nano-size pores ofceramic
(g-Al2O3) layer located under the top surface (PachecoTanaka et
al., 2006). This pore fill configuration providesadvantages of
improved toughness on handling since palladiumlayer is not exposed
on the surface. Moreover nano-size palla-dium grains confined in
the nanopore substantially suppress theinternal stress associated
with ab phase transition. This Pdmembrane is tolerable below the
critical temperature, wherefatal damage occurs for usual. Almost
constant hydrogen permea-tion and selectivity were maintained for
more than 24 h at 50 1Cand 100 kPa pressure difference. At 300 1C
and 4 bar of pressuredifference, the hydrogen permeance and H2/N2
ideal selectivity ofthis membrane were 1.3106 mol m2 s1 Pa1 and
41000,respectively. Due to the difference of expansion
coefficientbetween Pd and alumina, nanoporous alumina filled with
Pdhas low stability at temperatures higher than 350400 1C.
Toovercome this problem, a membrane was prepared by
fillingpalladium into the nano pores of YSZg-Al2O3. The PdYSZAl2O3
composite membrane revealed excellent thermal stabilityallowing
long-term operation at elevated temperature (500 1C).This has been
attributed to improved fracture toughness ofYSZAl2O3 layer and
matching of thermal expansion coefficientbetween palladium and YSZ
(Pacheco Tanaka et al., 2008).At 425 1C and 4 bar of pressure
difference, the hydrogen per-meance and H2/N2 ideal selectivity of
this membrane were1.6106 mol m2 s1 Pa1 and 300, respectively.
Various Pd alloys have been developed and tested as mem-brane by
the Colorado School of Mines (CSM). Supported mem-branes have been
obtained with PdAu, PdCu and PdRu alloysupported on commercial
alumina and zirconia-coated porousstainless steel tubular
substrates using an ELP technique (Roaet al., 2009). The most
promising result has been obtained byusing an alumina supported
PdAu membrane. A H2 flux of482 ml min1 cm2 (STP) and a H2/N2 ideal
selectivity of 1000at a temperature of 400 1C and a partial
pressure difference ofaround 7 bar. On the other hand, a 2.3 mm
thick PdAu membranesupported on zirconia-coated porous stainless
steel achieved a H2flux of 1.01 mol m2 s1 and a H2/N2 ideal
selectivity of 82,000 at1.38 bar and 400 1C (Hatlevik et al.,
2010). At these experimentalconditions no flux reduction was
observed for this PdAu mem-brane for WGS mixture compared to a pure
H2 feed gas. Thismeans that no CO adsorption on the membrane
surface occurs atthese temperatures, which makes the membrane very
attractivefor WGS reactions or pre-combustion CO2 capture
plants.
A great research effort on novel membranes has been devotedin
China where the Dalian Institute of Chemical Physics (DICP)
isdeveloping supported composite PdAu alloy membranes onalumina
substrates using ELP technique (Goldbach and Xu,2011). Very thin
membranes with high fluxes have been producedby DICP. They reported
a remarkable H2 permeability of1.3108 mol m m2 s1 Pa0.5 at 400 1C
attained by a 5 mmPdAu composite membrane. The H2/N2 ideal
selectivity was
around 1100 at 500 1C. The H2 flux achieved with a 23 mm
thickPdAu composite membrane was 0.62 mol m2 s1 at 500 1C. TheH2/N2
ideal selectivity was around 1400. Moreover cycling inhydrogen
between 250 and 450 1C had no significant effect onhydrogen and
nitrogen permeation rates showing that mechan-ical stress caused by
the differing thermal expansion of the joinedmaterials is within
the tolerance of the metal/ceramic compositemembrane up to those
temperatures (Shi et al., 2010).
SINTEF (Norway) has developed a two-step technique
formanufacturing very thin defect free palladium-based
hydrogenseparation membranes. First, a defect-free Pdalloy thin
film isprepared by PVD magnetron sputtering onto a silicon
wafer(surface with minimum roughness). In a second step, the film
isremoved from the wafer and used as membrane. These films
mayeither be used as self-supported membrane or be integrated
withvarious supports of different pore sizes and geometries.
Thisallows the preparation of thin (12 mm), high-flux
membranessupported on macroporous substrates avoiding all the
masstransfer resistances given by microporous supports or
problemswith interdiffusion layers. It must be highlighted that
according tothe DFT calculations of Ling and Sholl, for membranes
of 1 mm, thedesorption resistance completely dominates the
permeation attemperatures below 230 1C and it has an important
contributionat temperatures below 450 1C. Thus, at these
temperaturessmaller thicknesses will not improve the permeation
flux of themembrane while will surely deteriorate the stability of
themembrane.
In pure hydrogen, and at 400 1C and a H2 pressure difference
of25 bar, the H2 permeance and H2/N2 ideal selectivity were1.46102
mol m2 s1 Pa0.5 and 2900. (Peters et al., 2011a).In WGS conditions
(57.5% H2, 18.7% CO2, 3.8% CO, 1.2% CH4 and18.7% steam) SINTEF
membranes have shown a H2 permeance of1.1103 mol m2 s1 Pa0.5 at 400
1C and 26 bar feed pres-sure. No membrane failure was detected
operating the membranefor more than 1 year using WGS and H2N2 feed
mixtures(Iaquaniello et al., 2011). SINTEF is also working in the
develop-ment of Pd-based binary and ternary alloy membranes for
H2separation (Peters et al., 2011b, 2012).
Research on self-supported membrane foils has been carriedout at
Southwest Research Institute (SwRI). The institute used amagnetron
sputtering technique to develop PdAu and PdAuPtfoils. SwRI tested
PdAuPt (702010%) foils at 400 1C, 12.7 barwith feed gas composition
of 50% H2, 30% CO2, 19% H2O, and 1%CO (Coulter et al., 2012), and a
H2 flux was approximately0.212 mol/m2/s. Currently, SwRI is
developing amorphousZr-based alloys, specially focusing on ZrCu
based ternary alloys.The reason for the development of these alloys
is that theyshowed good results (i.e., Zr30Cu60Ti10) in previous
computationalcalculations predicting H2 flux through ternary alloys
as functionsof operating temperature, H2 feed pressure, membrane
thickness,and trans-membrane pressure drop to advance testing
(CO2Handbook, 2011).
Important advances on membrane preparation have beenachieved at
the center for inorganic membranes of the WorcesterPolytechnic
Institute (WPI). The group directed by Prof. Ma isdeveloping
Pd-based membranes on tubular stainless steel oralloy supports
(such as, SS 316 L, Hastelloy and Inconel) by theELP technique. WPI
achieved a H2 flux of 0.098 mol m
2 s1 and aH2/He selectivity of around 4500 at 450 1C and 1.03
bar DP with a7 mm thick Pd membrane with an Inconel support shown
in Fig.
6(http://www.hydrogen.energy.gov/pdfs/progress09/ii_d_1_ma.pdf).
WPI is also developing supported molten metal membranes,consisting
of low-melting non-precious metals and its alloys (e.g.,Sn, In, Ga,
Bi) as supported thin films (CO2 Handbook, 2011).
Moreover, WPI has also built an engineering-scale
prototypemembrane with 8.8 mm thickness, 2 in. outer diameter, and
6 in.
http://www.hydrogen.energy.gov/pdfs/progress09/ii_d_1_ma.pdfhttp://www.hydrogen.energy.gov/pdfs/progress09/ii_d_1_ma.pdf
-
Fig. 6. Pd/Inconel hydrogen separation membranes manufactured by
WPI
(http://www.netl.doe.gov/publications/factsheets/project/Proj481.pdf).
Fig. 7. High temperature Pd-based composite commercial hydrogen
separationmembranes manufactured by CRI/Criterion
(http://www.cricatalyst.com/).
Fig. 8. Hysep 1308 hydrogen separation module manufactured by
ECN.
F. Gallucci et al. / Chemical Engineering Science 92 (2013) 4066
47
length. WPI has demonstrated long-term membrane testingwith
total test duration of 63 days at 450 1C, 1.03 bar DP,0.303 mol m2
s1H2 flux, 99.99% purity. This flux is equivalentto approximately
1.286 mol m2 s1 at 6.9 bar DP (Hydrogenfrom Coal Program,
2010).
Currently, WPI is performing long-term tests with
Pd-basedmembranes under mixed gas streams (Guazzone et al.,
2012;Augustine et al., 2012).
3.2. Progresses in commercialization of dense metal
membranes
There are different companies working on the commercializa-tion
of dense metal membranes for hydrogen separation/produc-tion. In
the following, the main progresses achieved are presented.Some of
these commercial membranes have been successfully usedin fluidized
membrane reactors for methane steam reforming andin membrane
separators in a pilot plant for 20 m3/h hydrogenproduction as
described below.
3.2.1. CRI/Criterion
CRI/Criterion (a company owned by Shell) is in the process
ofcommercializing Pd and Pdalloy membranes on sintered porousmetal
supports as published in a report (Shell Impact, 2010).
CRI/Criterion has produced membranes of this type as large as 2
in.OD by 48 in. L, by welding two separate 24 in. L sections.
Thesupport is polished with a robotized machine before the
Pddeposition. The H2 permeance of these membranes varies in arange
of 4070 Nm3 m2 h1 bar0.5. Both hydrogen flux andseparation
selectivity are stable at temperatures of 300500 1Cand differential
pressures of 2642 bar. H2 purity of 499% hasbeen demonstrated for
periods exceeding 4000 h in high tem-perature gas separations. In
Fig. 7 some manufactured mem-branes are shown.
3.2.2. ECN
The Energy research Centre of the Netherlands (ECN) producesand
offers a line of hydrogen separation modules (Hysep) on
apre-commercial basis for evaluation purposes as described in
itswebsite (http://www.hysep.com/). There are three modules
avail-able: Hysep 108 (area of 0.04 m2), Hysep 308 (0.1 m2) and
Hysep1308 (0.5 m2) (Fig. 8). The Hysep modules use palladium
compo-site membranes composed by a 39 mm thick palladium
layerdeposited by ELP onto a porous ceramic alumina support.
Thenominal capacity of the largest membrane module (Fig. 5)
equals3.56 Nm3 h1, based on the obtained hydrogen flux
applyingreformate with 33% H2, an inlet pressure of 25 bar and H2
outletpressure of 4 bar. Lifetimes of several thousands of hours
havebeen shown under different conditions and purities that can
reachthe range from 99.5% to 99.995% depending on the
initialcomposition. ECN is also developing novel Pd alloy
compositemembranes (e.g. PdAg, PdCu; Acha et al., 2011).
3.2.3. Eltron Research, Inc.
Eltron Research, Inc. has developed alloy-based membranesand has
developed a separator unit rated to produce 6.8 kg/day ofhydrogen
(Hydrogen from Coal Program, 2010; CO2 Handbook,2011). Eltrons best
alloy membrane has demonstrated a H2 fluxrate of 809 mL min1 cm2
(STP) at 400 1C and 6.9 bar DP withH2 pure gas feeding. Eltron has
tested the membranes under WGSfeed stream conditions; tubular
membranes were successfullytested for greater than 300 h with a
feed gas composition of 50%H2, 29% CO2, 19% H2O, 1% CO, and 1%
He.
3.2.4. Green Hydrotec
As shown on its website (www.grnhydrotec.com), GreenHydrotech
has developed Pd and PdCu membranes on porousstainless steel tubes
(Fig. 9) with an extremely high H2/N2 perm-selectivity (4100,000)
and these membranes provide in situpurification of hydrogen in a
steam reformer for high purityhydrogen (99.996%).
3.2.5. Hy9
Hy9 is commercializing HPSTM hydrogen purifiers based onplanar
palladium alloy metal membranes (Fig. 10), as describedon its
website (http://hy9.com/). Hy9s membrane purifiers
http://www.hysep.com/www.grnhydrotec.comhttp://hy9.com/http://www.netl.doe.gov/publications/factsheets/project/Proj481.pdfhttp://www.netl.doe.gov/publications/factsheets/project/Proj481.pdfhttp://www.cricatalyst.com/
-
Fig. 9. Pd-based composite membrane module manufactured by Green
Hydrotech.
Fig. 10. HPS 689 hydrogen purifier composed by PdCu alloy planar
membrane.
Fig. 11. Pd alloy composite tubular membrane manufactured by
Pall Corporation(active surface area: 15 cm2).
F. Gallucci et al. / Chemical Engineering Science 92 (2013)
406648
tolerate a wide variety of feed stocks and gas mixtures,
havingbeen successfully deployed with systems utilizing
methanol,natural gas, diesel, liquefied petroleum gas, ammonia
borane,and a variety of hydrides. The optimal operating
temperatureconditions range from 300 to 350 1C and at inlet
pressures from3.4 to 17.2 barg depending on specific
requirements.
3.2.6. M&P
Media and Process Technology, Inc. (M&P) (Liu et al., 2010)
hasdeveloped a Pd-based membrane module that is capable ofproducing
a H2 flow of 0.575 mol m
2 s1 and H2/N2 selectivityof over 1000. The Pd membrane
composite membranes consist ofa 2 mm thick Pd layer deposited onto
a porous ceramic supportwith a H2 flux of 0.575 mol m
2 s1 at 1.38 bar. Currently, testsinvolving a WGS membrane
reactor with this module are beingcarried out.
3.2.7. MRT
Membrane Reactor Technologies (MRT) is developing a rangeof
hydrogen purifiers to provide high-purity hydrogen and torecover
hydrogen from mixed gas streams, as described on itswebsite
(http://www.membranereactor.com/).
MRT produces membranes either as rolled foils or as
depositedthin films (815 mm).
While palladium-based foils of common compositions areavailable
commercially, MRT has developed its own alloy compo-sitions for
added performance and robustness. In addition, a
patent-pending bonding technique has been developed to
perma-nently attach membranes to support modules with a
perfect,hydrogen-tight seal.
For membranes thinner than 15 mm, MRT uses a proprietarycoating
technique (Iaquaniello et al., 2011). MRT achieved a H2flux of 95 N
m3 m2 h1 with a 5 mm thick Pd membranesupported on alumina-coated
porous stainless steel tubular sub-strate at 550 1C and hydrogen
pressure difference of 3.4 bar(Li et al., 2007c).
3.2.8. Pall Corporation
Pall Corporation produces and offers Pdalloy compositetubular
membranes on a pre-commercial basis (Fig. 11), aspublished on its
website (Pall.com). The support of this mem-brane is composed of
YSZ inter-diffusion barrier layer depositedonto a porous stainless
steel tube. Pall has a couple of techniquesto deposit its 15 mm
PdAu layer. Palls current membraneflux performance and H2/Ar ideal
selective are 295354 mL min1 cm2 (STP) and 10,00020,000 at a
temperatureof 400 1C and feed pressure of 1.38 barg (with permeate
atmo-spheric) (CO2 Handbook, 2011, Pall.com, Damle, A. 2010).
Devel-opment of a Pd-alloy composition tolerant up to 100 ppm H2S
iscurrently ongoing (Bredesen et al., 2011).
3.2.9. REB
REB Research and Consulting produces and offers metalmembranes
made of 1.51.6 mm thick Pd-coated refractory metaltubes (3/8 in. OD
0.007 in. wall) or palladiumsilver alloy tubes(1/8 in. OD 0.003
in.), as described on its website (REBre-search.com).
Palladiumsilver alloy membranes are availableplain, coated with
palladium-grey, or coated with palladiumcopper alloy.
The average membrane permeabilities of VTi based mem-branes are
0.15 mol m m2 s1 Pa0.5 at 600 1C and0.015 mol m m2 s1 Pa0.5 at 300
1C. For VNb membranes,the permeabilities are 0.2 mol m m2 s1 Pa0.5
at 429 1C and0.19 mol m m2 s1 Pa0.5 at 340 1C (REBresearch.com,
Buxbaum2008). On the other hand, REB is developing metalmetal
matrixmembranes which consist of a high permeable H2 metal
layercoated with Pdalloys on each side, with the aim of
reducingmembrane cost and increasing durability. A H2 flux of0.2
mol m2 s1 was achieved with a 0.5 mm PdCu coated oneach side of a
high permeable metal membrane (B2) at 400 1C and3.03 bar DP with
feed gas mixture containing H2, CO, CO2, CH4and H2S. These
membranes have resisted poisoning from 50 hwith 100 ppm H2S. These
novel membranes will be integratedin a disc membrane reactor to be
tested with coalgas(CO2 Handbook, 2011).
REB membranes have also been successfully tested in fluidizedbed
membrane reactors for hydrogen production as discussed inthe
following sections. The double membrane layer was able to
http://www.membranereactor.com/
-
F. Gallucci et al. / Chemical Engineering Science 92 (2013) 4066
49
assure a high selectivity for long time under bubbling
fluidizationconditions (Roses et al.). It should also be
highlighted that thesemembranes have been successfully tested at
temperatures up to650 1C without deteriorating the perm-selectivity
of the mem-branes which is probably the higher temperature used so
fare forPd-based membranes and it is also the temperature window
thatallow an effective integration of Pd membranes into
methanereformers (Gallucci et al., 2008a,b; Roses et al.).
3.2.10. Tokyo Gas
Tokyo Gas Co., Ltd. has been developing membrane reformers(MRF)
for hydrogen production with Mitsubishi Heavy Industries,Ltd. since
1992. Recently, Tokyo Gas developed a MRF test systemwith a
hydrogen production of 40 Nm3/h, hydrogen purity of99.99%, and
hydrogen production efficiency of 70% (Shirasakiet al., 2009). The
reformer has 112 reactor tubes, each of whichhas two planar-type
membrane modules composed of stainlesssteel supports and PdY(Gd)Ag
alloy films of less than 20 mmthick (Sakamoto et al., 1992). The
longest operation time for thismembrane reformer was 3000 h. An
interesting evolution of themembrane modules introduced by Tokio
Gas company is whatthey called membrane-on-catalyst (MOC) module.
This is a Pd-based membrane prepared on the porous surface of the
tubularstructured catalyst that has catalytic activity for steam
reformingreaction. The important aspect to be taken into account
whenpreparing MOCs is the thermal expansion of the catalyst that
hasto match with the membrane material, while also selecting
aproper porosity, mechanical strength and thermal conductivity
ofthe catalyst (Yasuda et al., 2006). This concept simplifies
enor-mously the membrane module design while also reducing themass
transfer resistances otherwise affecting the packed bedmembrane
modules used in their hydrogen generator.
3.2.11. UTRC
United Technologies Research Center (UTRC) is developing
apalladium copper (PdCu) trimetallic alloy hydrogen separator
forcentral H2 production from coal gasification-derived syngas
(CO2Handbook, 2011). UTRC has tested five separators using
PdCuTMalloy which showed increased surface stability in
bench-scaletests (Hydrogen from Coal Program, 2010). UTRCs current
mem-brane flux performance is approximately 0.230 mol m2 s1 at
atemperature of 400 1C and feed pressure of 6.9 bar). This
mem-brane has also shown pressure capability up to 27.6 bar,
sulfurtolerance of 20 ppmv, CO tolerance and the production of at
least99.5% pure H2.
Three types of commercial membranes (MRT, ECN and one froma not
specified Japanese company) are being tested it a 20 m3/hhydrogen
production plant at Chieti (Italy) by Technimont-KT(De Falco et
al., 2011). Preliminary experimental tests haveconfirmed the
potential of membrane technology with an overallconversion of
approximately 57.3% was achieved at 600 1C, 26%higher than what was
achieved in a conventional reformer. Thisconversion is expected to
be even increased up to 62.8% bydoubling the membrane surface by a
factor of two. Moreover, theeffect of operating temperature and gas
mixture space velocity havebeen evaluated over a period of 1000 h
without detrimental effectson the membranes performance.
4. Microporous membranes
Microporous membranes are referred to as those with a
porediameter smaller than 2 nm. Regarding the structure, the
micro-porous membranes for H2 separation may be classified
intocrystalline (zeolites and MOF) and amorphous (such as
silica,carbon, etc.).
4.1. Zeolite membranes
Zeolites are microporous crystalline aluminosilicates
withuniform molecular sized pores. The unique properties of
zeolitemembranes are: size and shape selective separation
behaviorand thermal and chemical stabilities. Due to their
crystallinity,zeolites have a well-defined pore size. The size of
the channels istypically 310 A, in the range of molecular
dimensions. Therefore,the hydrogen permeation through zeolite
membranes relies onmolecular sieve effect and/or competitive
diffusion mechanisms(Dong et al., 2000).
Regarding material and structure, the most extensively
studiedzeolite membranes for hydrogen separation are mainly MFI
(ZSM(Hong et al., 2005), silicate-1 (McLeary et al., 2006), LTA
(NaA(Xu et al., 2000), DDR (Lin and Kanezashi, 2007) and CHA
(SAPO-34(Hong et al., 2008).
The most commonly used technologies for zeolite processingare in
situ hydrothermal synthesis (Coronas and Santamaria,2004),
secondary (seeded) growth synthesis (Xomeritakis et al.,2000) and
vapor phase transport synthesis (Chiang and Chao,2001). Other
various techniques have been applied, such asconventional heating,
microwave heating (Li et al., 2006), synth-esis under centrifugal
field (Tiscareno-Lechuga et al., 2003),moved-synthesis (Richter et
al., 2003). Often, pre-treatment ofsupports (Berg et al., 2003) and
post-treatment of membranes(Yan et al., 1997) (e.g. for pore size
reduction and defects removal)are carried out to improve the
quality of the as-synthesizedmembranes.
Some of the most relevant current developments on
zeolitemembranes for hydrogen separation are presented below.
Lai et al. (Lai and Gavalas, 2000) developed ZSM-5
zeolitemembranes by hydrothermal synthesis using a template free
gelin order to avoid calcination step. The H2/N2 ideal selectivity
and H2permeance obtained were 109 and 1.2107 mol m2 s1
Pa1,respectively.
Welk et al. (2004) from Sandia National Laboratories (SNL)have
studied the potential applications of zeolite membranes forhydrogen
separation from reforming streams. ZSM-5 membraneswere prepared by
hydrothermal synthesis and the testing resultsshowed that the H2
purity was enriched from 76% to more than a98% after a single pass
through this membrane.
Tsapatsis and co-workers from University of Minnesota
haveprepared 1 mm thick MCM-22/silica layer onto porous
homemadealumina support discs. The H2/N2 ideal selectivity and H2
per-meance were 50 and 7108 mol m2 s1 Pa1 at 200 1C,respectively
(Choi and Tsapatsis, 2010). Besides, the same groupis developing
membranes consisting of ex-foliated MCM-22 layersonto commercial
stainless steel tubular support to their laterintegration in WGS
reactors and integrated gasification combinedcycle plants (CO2
Handbook, 2011).
Recently, Zhang et al. (2012) from Nanjing University
havedeveloped a MFI zeolite membrane on porous a-alumina
porousdiscs. The membrane was tested for the separation of
H2/CO2mixture containing 20 cm3 (STP)/min H2, 20 cm
3 (STP)/min CO2and 5 cm3 (STP)/min He under atmospheric
pressure. The perme-ate side was swept by He stream with the flow
rate of 30 cm3
(STP)/min. The H2/CO2 separation factor and the H2 permeancewere
42.6 and 2.82107 mol/m2/s/Pa at 500 1C.
4.2. Metalorganic framework membranes
Metalorganic frameworks are microporous crystalline
hybridmaterials consisting of metal cations or cationic oxide
clustersthat are linked by organic molecules. Pore size
tailorabilitycombined with tunable sorption behavior provides
promisingpossibilities for the applications of MOFs as membranes
for gas
-
F. Gallucci et al. / Chemical Engineering Science 92 (2013)
406650
separation applications. It is thought that MOF can solve
pro-blems that zeolites have for their industrial use, such as,
long-term stability, T-cycling, regeneration and difficult
housing(Gascon and Kapteijn, 2010).
MOF are mechanically less stiff and brittle and have lessenergy
intensive synthesis conditions compared to zeolites (Tanand
Cheetham, 2011; Shah et al., 2012).
The zeolitic imidazolate frameworks (ZIFs) form a subfamily
ofMOFs that are promising candidates as gas separation membranesdue
to their thermal and chemical stability in combination withtheir
small pores (generally less than 5 A) (Shah et al., 2012).Other MOF
materials that have been used for gas separation areMOF-type (Yoo
et al., 2009; Bux et al., 2009), HKUST-type(Guerrero et al., 2010;
Guo et al., 2009) and SIM-type (Aguadoet al., 2011a), among
others.
The most commonly used technologies for MOF preparationare in
situ growth synthesis (Liu et al., 2010; Huang et al., 2010;Guo et
al., 2009), secondary (seeded) growth synthesis (Liu et al.,2009;
Ranjan and Tsapatsis, 2009; Yoo et al., 2009) and liquidphase
epitaxy (Shekhah et al., 2011). Besides, postsyntheticmodifications
of MOF membranes have been reported, forinstance, introducing side
groups to give functionality to themembranes (Huang and Caro, 2011;
Aguado et al., 2011b).
Some of the most relevant current developments on MOF mem-branes
for hydrogen separation are presented afterwards. Li and co-workers
have prepared 2 mm thick ZIF-7 membrane deposited on anasymmetric
alumina disc by secondary (seeded) growth technique.The H2
permeance and H2/N2 ideal selectivity of the membrane at220 1C and
at 1 atm were 4.55108 mol m2 s1 Pa1 and 20.7,respectively (Li et
al., 2010).
Recently, the same group in Hannover (Huang and Caro, 2011)has
prepared a 20 mm thick ZIF-90 membrane on a-Al2O3 porousdiscs by
solvothermal reaction. After synthesis the membrane hasbeen
modified using ethanolamine and its H2 permeance hasslightly
decreased from 2.5 to 2.1107 mol m2 s1 Pa1 andthe H2/N2 ideal
selectivity has considerably increased fromaround 7 to 17.5 at 200
1C and 1 atm.
(McCarthy et al. (2010) from Texas A&M University
haveprepared a ZIF-8 membrane (20 mm thick) by in situ
solvothermalgrowth on a-Al2O3 support after its surface
modification. The H2permeance and H2/N2 ideal selectivity of the
membrane at 25 1Cand at 1 atm were 1.73107 mol m2 s1 Pa1 and
11.6,respectively.
The same group from Texas (Guerrero et al., 2010) hasprepared a
25 mm thick HKUST-1 membrane deposited on aporous alumina support
disc via secondary (i.e., seeded) growthmethod, using thermal
seeding in order to anchor HKUST-1 seedcrystals on the support. The
H2 permeance and H2/N2 idealselectivity of the membrane at 190 1C
and at 1 atm were1.1106 mol m2 s1 Pa1 and 7.3, respectively.
Permeation data about other reported MOF membranes can befound
in the review prepared by Shah et al. (2012).
4.3. Silica membranes
Silica membranes are the most important representatives
ofamorphous microporous membranes, because they can be moreeasily
prepared as ultra- or super-microporous thin layers incomparison to
other metal oxides (such as alumina, titania orzirconia) and these
can be used for molecular sieving applications.
The most widely used technologies for deposition of silicalayers
on porous substrates are solgel (Tsuru, 2008) and CVD(Nagano et
al., 2008). Silica membranes with pore diameters lessthan 1 nm can
be prepared by CVD which provides high hydrogenselectivity
(molecular sieving) but, consequently, a lower perme-ability.
Moreover, the CVD method requires substantial capital
investment and well defined and controlled deposition
condi-tions. On the contrary, solgel derived membranes
generallyachieve lower selectivities but higher permeabilities.
These mem-branes are much easier to produce with the possibility
ofcontrolling the pore size of the silica membranes, but
theprepared membranes still suffer of low reproducibility
whichmakes their industrial exploitation less attractive.
Akamatsu et al. (2008) prepared and tested silica membranesvia
CVD technique. The test results showed an excellent H2permeance at
600 1C of the order of 107 mol m2 s1 Pa1,and a high H2/N2
selectivity of over 1000.
Yoshino et al. (2005) reported silica membranes prepared
bysolgel technique with a permeance at 600 1C of7107 mol m2 s1 Pa1
and H2/N2 selectivity around 100.This group (Yoshino et al., 2006)
also fabricated a membranemodule with a membrane area of 0.05
m2.
The main problem of microporous silica membranes, as in thecase
of titania and alumina, is that they are not stable at
hightemperatures, especially in the presence of steam, leading to
lossof permeability. This is due to closure of pore channels
bydensification which is catalyzed by humidity, particularly at
hightemperatures (Lin, 2001). Furthermore, this phenomenon maycause
silica film embrittlement with the subsequent loss inseparation
properties.
In order to improve the stability of silica membranes,
differentapproaches have been proposed in the literature. The
firstapproach used is the doping of silica with inorganic oxides
(e.g.,titania, zirconia and alumina) (Kanezashi and Asaeda,
2006;Sekulic et al., 2002).
Kanezashi and Asaeda(2006) prepared Ni-doped silica mem-branes
with a permeance of 4.6106 m3 (STP) m2 s1 kPa1
for H2 with a H2/N2 selectivity of 400 even after being kept
insteam (steam pressure: 90 kPa) at 500 1C for about 6 days.
Tsuruet al. (2011) prepared Co-doped silica membranes (approx. 50
nmthick layer) with a H2 permeance of approximately 1.8107 mol m2
s1 Pa1 and a H2/N2 selectivity of 730 even after60 h of exposure to
steam (steam: 300 kPa) at 500 1C. Doping ofmetals into silica
membranes has been investigated in order toincrease the
hydrothermal stability at high temperatures and fortheir possible
application to membrane reactors, such as steamreforming of methane
(Battersby et al., 2009).
Another approach to obtain hydrothermal stable silica mem-branes
is to incorporate methyl groups in the silica microstruc-ture as
proposed by Campaniello et al. (2004). Recently,
hybrid(organicinorganic) silica membranes are widely being
studiedbecause the presence of organic groups in silica networks
couldimprove the hydrothermal stability of silica structures and
helpcontrolling the pore size of the membrane (Duke et al.,
2004).Castricum et al. (2008) prepared hybrid silica membranes
derivedby co-polymerization of methyltriethoxysilane and
bis(triethox-ysilyl)ethane with high hydrothermal stability.
Kanezashi et al.(2009) reported a hybrid silica membrane using
BTESE as pre-cursor with a high H2 permeance around 10
5 mol m2 s1 Pa1
but low H2/N2 selectivity of around 10 at 200 1C.
4.4. Carbon membranes
Carbon membranes are promising candidates for hydrogenseparation
due to their high separation performance and excellentthermal and
chemical resistance (Ismail and David, 2001; Korosand Mahajan,
2000). Based on the pore size, carbon membranesare generally
divided in carbon molecular sieve membranes(CMS) and selective
surface flow (SSF) membranes. CMS mem-branes could allow the
transport of small molecules through thepores, avoiding the passage
of larger molecules. Due to their smallpores, CMS membranes possess
a high selectivity for separation of
-
F. Gallucci et al. / Chemical Engineering Science 92 (2013) 4066
51
gas mixtures containing small gas species. In this case, a
precisecontrol of pore sizes near to molecular sieving size is
required. SSFmembranes have pores larger than the dimensions of the
mole-cules and thus the separation is based on the
preferentialadsorption of some components in gas mixture followed
bysurface diffusion in the carbon matrix (Seo et al., 2002).
Thissection will be focused on CMS membranes due to their
proper-ties for high H2 separation in comparison to SSF
membranes.
CMS membranes are the result of the pyrolysis of
polymericprecursors under an inert or vacuum atmosphere. The most
impor-tant parameters that would influence the final properties of
CSMmembranes are: (i) type of polymeric precursors such as
polyimides,polyfurfuryl alcohol, phenolic resins, polyvinylidene
chloride, poly-acronitrile, cellulose derivates, and polyetherimide
(Saufi and Ismail,2004); (ii) pyrolysis conditions (i.e., heating
rate, atmosphere andfinal temperature) (Geiszler and Koros, 1996;
Centeno et al., 2004);and (iii) modifications (i.e., pre- or
post-treatment such as stabiliza-tion, activation or oxidation, and
CVD) (Barsema et al., 2004; Leeet al., 2008; Li et al., 2008c; Wang
et al., 2002).
The most commonly used techniques for depositing CSMlayers onto
supports may be dip coating (Hayashi et al., 1997),ultrasonic
deposition (Shiflett and Foley, 2000), vapor deposition(Wang et
al., 2000), spin coating (Tseng et al., 2009), and spraycoating
(Acharya and Foley, 1999).
Some of the most relevant current developments on
carbonmembranes for hydrogen separation are presented
afterwards.
Hosseini and Chung (2009) from National University of Singa-pore
reported carbon membranes prepared from blends of PBIand polyimides
with a high H2/N2 selectivity of 460480 and H2permeability of 60180
barrer (1 barrer11010 cm3(STP)cm cm2 s1 cm Hg1). Campo et al.
(2010) have developedCMS membranes from cellophane paper with very
interestinghydrogen permeation properties: H2 permeability of 39.3
barrer,H2/N2 and H2/CO2 selectivities of 1310 and 58.7,
respectively(Celo600 Sample). Grainger and Hagg (2007) reported CMS
fromcellulose base based precursors with a H2/N2 selectivity of
740and H2 permeability of 1110 barrer at 90 1C and at a feed
pressureof 6 bar. Media and Process Technology, Inc. is
commercializingCSM composite membranes prepared by the pyrolysis of
a PEIprecursor deposited onto a ceramic porous substrate
(Abdollahiet al., 2010; Sedigh et al., 2000).
5. Proton conducting membranes
Proton conducting membranes can be classified into twogroups:
dense ceramic membranes and composite ceramic metal(cermet)
membranes. Both have their advantages and disadvan-tages in terms
of membrane flux, perm-selectivity and stability. Inthe following
section both types of membranes are discussed.
5.1. Dense ceramic membranes
Dense ceramic membranes can recover a very high purity H2stream
due to a proton transport mechanism, but they have tooperate at
temperatures as high as 900 1C. The hydrogen fluxthrough these
membranes is proportional to the ambipolarprotonicelectronic
conductivity. It is necessary to have highvalues for protonic and
electronic conductivities to obtain a highhydrogen flux. The dense
ceramic membranes may be classifiedinto two sub-categories:
perovskite-type and non-perovskite-type membranes.
5.1.1. Perovskite-type membranes
The protons are embedded in perovskites in the electron cloudof
an oxygen ion, forming hydroxide defects. They migrate by
hopping between oxide ions (Grotthuss mechanism). As for
theoxygen conductors, sufficiently high temperatures are
requiredfor high proton conductivity.
The general formula of perovskite-type oxides is AB1xMxO3d,
where the A element is taken from the group consistingof (Ca, Sr,
B); the B element is taken from the group consisting ofbe Ce, Tb,
Zr, Tl; M element is taken from the group consisting ofTi, Cr, Mn,
Co, Ni, Co, Al, Y, Ga; x is less than the upper limit ofsolid
solution formation range (usually less than 0.2) and d is theoxygen
deficiency per unit cell.
The most extensively studied high temperature perovskite-type
oxides are SrCeO3 (Higuchi et al., 2004; Song et al., 2002),BaCeO3
(Guan et al., 1997; Ma et al., 1998) and BaZrO3 (Tetsuoet al.,
2002). These perovskites have high protonic conductivitybut their
electronic conductivity is poor. In order to improve thelatter
conductivity it is common to dope these perovskites withcations,
such as Y (Sammes et al. 2004), Eu (Song et al., 2003) andGd (Shima
and Haile, 1998).
The common methods for perovskite power synthesis are solgel
(Selvaraj et al., 1991), spray drying (Varma et al.,
1994),hydrothermal synthesis (Zheng et al., 1997) and combustion
ofpolymerized complexes (Liu et al., 2002). On the other
hand,typical methods for perovskite film deposition onto ceramic
ormetal supports are CVD (Ngamou and Bahlawane, 2009), EVD(Pal and
Singhal, 1990) and PVD (Ma et al., 2008).
Some of the most relevant current developments onperovskite-type
membranes for hydrogen separation are pre-sented afterwards.
Yuan et al. (2010) have prepared
SrCe0.75Zr0.20Tm0.05O3dmembranes with a H2 permeation flux up to
0.042 mL min
1 cm2
at H2 partial pressure of 0.4 atm and at a temperature of 900
1C. TheZr doping can increase mechanical stability of the membrane
andthe resistance to reduction.
Yazdi et al. (2009) have developed BaCe1xYxO3d films by
DCmagnetron sputtering at room temperature. Ceramatec (Paglieriand
Way, 2002, CO2 Handbook, 2011) is producing a prototypemembrane for
hydrogen separation from coal-derived syngasusing perovskite-type
membranes. The pevoskite used is a mix-ture of barium cerate
(BaCeO3) and ceria, where the former isproton conductor and the
latter is electron conductor. The pre-liminary tests showed a
perfect H2 separation (100%) from H2/CO2mixture.
5.1.2. Non-perovskite-type membranes
The non-perovskite-type membranes used for hydrogenseparation
are mainly doped rare earth metal oxides andfluorite-structured
metal oxides. There are lots of different dopedrare earth metal
oxides that may be interesting for hydrogenseparation, such as,
Tb2O3, Ln2Ti2O7 and Er2Ti2O7.
Haugsrud and Norby (2006) prepared a 10 mm thick film
ofacceptor-doped LaNbO4 (i.e. La0.99Ca0.01NbO4) that has a H2 flux
of0.1 mL (STP) cm2 min1 at a temperature of 1000 1C and adifference
in pressure of 10 bar. This group also reported thatLn6WO12
membranes have sufficient mixed electronproton con-ductivity at
intermediate temperatures (Haugsrud, 2007).
Escolastico et al. (2011) have developed Nd5LaWO12 mem-branes
and the results showed that the addition of La to Nd6WO12increases
the hydrogen flux from 0.03 to 0.05 mL min1 cm2 at1000 1C and
difference pressure of 0.5 bar.
Among fluorite-structured metal oxides CeO2, YSZ and Y2O3are the
most used.
Nigara et al. (2003) have reported hydrogen permeation
throughdoped CeO2 and YSZ (Nigara et al., 2004) membranes, but
unfortu-nately H2 permeabilities of these membranes are very low.
Serraet al. (2005) studied H2 permeation through commercial
alumina
-
F. Gallucci et al. / Chemical Engineering Science 92 (2013)
406652
tubes obtaining as a result a H2 flux of 4108 mol s1 m2 at1400
1C.
5.2. Cermet membranes
Cermet membranes consist of a combination of a ceramicphase and
a metallic phase. The former is a pure proton con-ductor, while the
latter is a highly electron conductor. Combiningthese two phases
together may provide high H2 permeationbecause both proton and
electron conductivities become high,resulting in high hydrogen
permeation.
Due to the hydrogen transport by both metals and oxides,mainly
three different combinations with respect to functionalproperties
of both phases can be realized: (1) a metal with lowhydrogen
conductivity in combination with a highly protonconductive oxide,
(ii) a metal or an alloy with high hydrogenpermeability (i.e., Pd,
Pd/Ag, Pd/Cu, Nb, Ta, V) combined with aceramic of low hydrogen
permeability and (3) a combinationwhere both the metallic and
ceramic phases conduct hydrogen.
Some of the most relevant current developments on
cermetmembranes for hydrogen separation are presented in the
following.Balachandran et al. (2006) from Argonne National
Laboratory (ANL)developed Pd/YSZ composite membranes and the
highest H2 fluxwas 20.0 cm3 (STP) min1 cm2 for a 22 mm thick
membraneat 900 1C using 100% H2 as feed gas. Recently, Zhu et al.
(2011) havedeveloped a 30 mm thick NiBZCY (Ba(Zr0.1Ce0.7Y0.2)O3d)
cermetlayer onto ceramic support that exhibited a maximum H2 flux
of2.4107 mol cm2 s1 at 900 1C using 80% H2/N2 (with 3% ofH2O) as
feed gas and dry high purity argon as sweep gas.
Park et al. (2011) have developed a 0.5 mm thick Ta/YSZ
mem-brane and the highest H2 flux obtained was 1.2 ml min
1 cm2 at300 1C using 100% H2 as the feed gas and Ar as the sweep
gas.
Jeon et al. (2011a) have developed a 0.5 mm cermet mem-brane
composed of Pd embedded in proton conducting ceramicmatrix
(CaZr0.9Y0.1O3d) and its H2 flux was around 2.3 cm
3
(STP) min1 cm2 at 900 1C. This group also has developedPdGDC
(Ce0.8Gd0.2O2d) cermet membranes for high sulfurresistance and no
sharp drop in hydrogen permeation flux wasobserved using feed gases
with 220 ppm H2S (Jeon et al., 2011b).
The best performances of membranes investigated in thevarious
research organizations are summarized in Table 2.Besides, membranes
on the way to commercialization commen-ted before are shown in
Table 3. It is shown that relatively highfluxes with good
perm-selectivities can be now achieved withhydrogen membranes which
make this technology closer to thereal market.
6. Advances in membrane reactors for hydrogen production
The application of membrane reactors for
dehydrogenationreactions has been first proposed to the scientific
community byProf. Gryaznov in the late 60s (see for example
Gryaznov et al.,1970). Removing hydrogen through a thick membrane
resulted ina shift of the equilibrium reaction towards the product
of interest.Membrane reactors in dehydrogenation reactions were a
scien-tific curiosity until around 1996 with few papers published
peryear. With the increasing interest into hydrogen as a
possibleclean energy carrier, the scientific attention towards
membranereactors as high efficient hydrogen production systems
sharplyincreased in the last years (Fig. 12). Probably the echo
producedby the two books The Hydrogen Economy (Rifkin,
2002;National Academy of Engineering, 2004) was also reflected
inthe great increase of number of papers in the last 5
years.However, our research will focus on the progresses on
membranereactors in the last 10 years. Accordingly with the
increase of
number of papers, the patents awarded on hydrogen productionin
membrane reactors also increased with time and in the last
yearsthis number is increasing rapidly. In fact, most of the
patents havebeen awarded in the last 10 years (Gallucci et al.,
2009).
Different types of membrane reactors for hydrogen productionhave
been proposed in the literature. Most of the previous workhas been
performed in packed bed membrane reactors (PBMR);however, there is
an increasing interest in novel configurationssuch as fluidized bed
membrane reactors (FBMR) and micro-membrane reactors (MMR)
especially because better heat man-agement and decreased mass
transfer limitations can be obtainedin these reactor
configurations. In the following, these reactorconfigurations,
along with membrane bio-reactors (MBR) andcatalytic membrane
reactors (CMR) for hydrogen production willbe discussed in
detail.
6.1. Packed bed membrane reactors
The packed bed membrane reactor configuration is the firstand
most studied configuration for hydrogen production inmembrane
reactors. This is because, the first studies on mem-brane reactors
focussed on the effect of the hydrogen permeationthrough membranes
on the reaction system. Thus it was straight-forward to compare two
packed bed reactors (avoiding thecomplication of complex fluid
dynamics such as in fluidizedbed) in one of which a membrane was
used.
In the following table the main investigators working withPBMR
for hydrogen production are summarized (source Scopus):
The different authors showed that the concept of
Pd-basedmembrane reactor can be used to carry out and intensify
differentdehydrogenation reactions such as reforming of methane
anddifferent alcohols. In particular, packed bed membrane
reactorshave been used for producing hydrogen via reforming of
methane(Gallucci et al., 2006; Matsumura and Tong, 2008), reforming
ofalcohols (Kikuchi et al., 2008; Tosti et al., 2009),
autothermalreforming (Simakov and Sheintuch, 2009), partial
oxidation ofmethane (Tan and Li, 2009), etc. The results of these
studies areinteresting to show that indeed the Pd membranes are
notpoisoned by the different alcohols and the products of
thereactions such as CO2 or higher hydrocarbons. The only
poisoningcan be due to the presence of H2S as said above or CO (if
lowtemperature o300 1C is used).
In a packed bed membrane reactor the catalyst is confined
infixed bed configuration and in contact with a
perm-selectivemembrane. The most used packed bed configuration is
thetubular one where the catalyst may be packed either in
themembrane tube (Fig. 13a) or in the shell side (Fig. 13b), while
thepermeation stream is collected in the other side of the
membrane(in case of hydrogen selective membranes) or one reactant
is feedon the other side of the membrane (in case of oxygen
selectivemembrane Jin et al., 2000).
For multi-tubular membrane reactor configurations the cata-lyst
in tube configuration can be preferred especially for con-struction
reasons and for the extent of bed-to-wall mass and heattransfer
limitations which can be very detrimental when thecatalyst is
positioned in shell configuration.
Often, a sweep gas can be used in the permeation side of
themembrane in order to keep the permeation hydrogen
partialpressure as low as possible for minimizing the membrane
arearequired for the hydrogen separation. This practice is for
examplevery useful if hydrogen for ammonia plant is being produced,
inwhich case an amount of nitrogen can be used for sweeping
thepermeation side producing a synthesis stream (N2/H2 1/3)ready
for the final reaction step. If a sweep gas is used in
thepermeation side then a packed bed membrane reactor can beused in
both co-current and counter current modes. Using a
-
Table 2Permeation data of different hydrogen separation
membranes reported in the literature.
Institute Membrane material Preparation method Selective
layer
thickness
(mm)
T (1C) PermeanceH2 [10
8
(mol/m2 s Pa)]a
Permeability
H2 [1013
(mol m/m2 s Pa)]
Ideal
selectivity
Sulfur
tolerance
(ppm)
Durability
(h)
AIST (Pacheco Tanaka et al., 2006) Pd/Al2O3 Pore filling 5 300
170 41,000 (H2/N2) AIST (Pacheco Tanaka et al., 2008) Pd/YSZ Pore
filling 5 425 210 300 (H2/N2)
CSM (Hatlevik et al., 2010) PdAu/YSZ/PSS ELP 2.3 400 710 160
82,000 (H2/N2)
DICP (Goldbach and Xu, 2011) PdAu/Al2O3 ELP 23 500 620 160 1400
(H2/N2)
SINTEF (Peters et al., 2011a)b PdAg/PSS PVD-MS 2.8 400 1,500 420
2900 (H2/N2) 2000
SwRI (Coulter et al., 2012)b PdAuPt PVD-MS 25 400 54 130
WPI (Ma, 2009) Pd/Inconel ELP 7 450 96 67 4,500 (H2/He)
2,200CALTECH (Li et al., 2000) ZSM-5/Al2O3 Hydrothermal synthesis
67 150 12 7.8 109 (H2/N2)
SNL (Welk et al., 2004) ZSM-5/Al2O3 Hydrothermal synthesis 10 25
68 68 16.6 (H2/CO2, CO,CH4)
Nanjing University (Zhang et al., 2012) MFI/Al2O3 Hydrothermal
synthesis 500 28 42.6 (H2/CO2)
Liebniz University Hannover (Li et al., 2010) ZIF-7/Al2O3
Secondary (seeded) growth 2 220 4.5 0.9 20.7 (H2/N2)
Liebniz University Hannover (Huang and
Caro, 2011)
ZIF-90/Al2O3 In situ growth (modified after synthesis
with ethanoldiamine)
20 200 21 42 17.5 (H2/N2)
Texas A&M University (McCarthy et al.,
2010)
ZIF-8/Al2O3 In situ growth (support surface modified) 20 25 17
42.5 11.6 (H2/N2)
Texas A&M University (Guerrero et al.,
2010)
HKUST-1/Al2O3 Secondary (seeded) growth 25 190 110 275 7.3
(H2/N2)
University of Minnesota (Choi and
Tsapatsis, 2010)
(MCM-22)-SiO2/Al2O3 Layer-by-layer deposition 1 200 7 0.7 50
(H2/N2)
The University of Tokyo (Akamatsu et al.,
2008)
SiO2/Al2O3 CVD 600 1014 4 1,000 (H2/N2) Total
Noritake Company (Yoshino et al., 2005) SiO2/Al2O3 Sol-gel 0.02
600 70 0.14 100 (H2/N2)
Hiroshima University (Kanezashi and
Asaeda, 2006)
NiSiO2/Al2O3 Solgel 0.3 500 20 0.6 400 (H2/N2) 144
Hiroshima University (Tsuru et al., 2011) CoSiO2 Solgel 0.05 500
18 0.09 730 (H2/N2) 60
Hiroshima University (Kanezashi et al.,
2009)
Hybrid silica (BTESE as
precursor)
Solgel 0.5 200 1,000 50 10 (H2/N2)
National University of Singapore (Hosseini
and Chung, 2009)
Carbon (from PBI and
polyimide blends)
Pyrolisis 35 0.20.27 460480 (H2/N2)
University of Porto (Campo et al., 2010) Carbon (from
cellophane) Pyrolisis 9.22 29.5 0.14 0.13 1,310 (H2/N2)
NTNU (Grainger and Hagg, 2007) Carbon (from cellulose-
based precursors)
Pyrolisis 90 3.7 740 (H2/N2) 350
South China University of Technology (Yuan
et al., 2010)
SrCe0.75Zr0.20Tm0.05O3-d Liquid citrate method 900 0.64
University of Oslo (Haugsrud and Norby,
2006)
La0.99Ca0.01NbO4 Pressing 10 1000 0.22 0.22
ITQ- UPV (Escolastico et al., 2011) Nd5LaWO12 Solgel 900
1000
0.63
ANL (Balachandran et al., 2006) PdYSZ Pressing 22 900 150 330
University of Science and Technology of
China (Zhu et al., 2011)
Ni-BZCY/ceramic Pressing 30 900 2.9 8.7
Korea Institute of Energy Research (Park
et al., 2011)
TaYSZ Pressing 500 300 8.9 450
Chonnam National University (Jeon et al.,
2011a)
PdCZY Pressing 500 900 7.8 390
Chonnam National University (Jeon et al.,
2011b)
PdGDC Pressing 282 900 19 540 220
a Permeance values have been calculated for a H2 partial
pressure of 1 bar.b Self-supported membranes.
F.G
allu
cciet
al.
/C
hem
ical
En
gin
eering
Science
92
(20
13
)4
0
66
53
-
Table 3Major investigators on packed bed membrane reactors
Investigator
name
Institution Number
of papers
Basile A. Institute on Membrane Technology (Italy) 35
Tosti S. ENEA (Italy) 29
Rahimpour,
M.R
Sharaz University (Iran) 27
Itoh N. Utsunomiya University (Japan) 12
Lombardo
E.A.
Instituto de Investigaciones en Catalisis y
Petroqumica (Argentina)
11
Nomura N. University of Tokyo (Japan) 7
Fig. 12. Number of papers on hydrogen production in membrane
reactors peryear. Database Scopus (www.scopus.com). Keywords
membrane reactor and
hydrogen production11 December 2012
Fig. 13. Membrane reactor catalyst in tube (A) and catalyst in
shell(B) configurations.
F. Gallucci et al. / Chemical Engineering Science 92 (2013)
406654
counter-current mode leads to completely different partial
pres-sure profiles in reaction and permeation sides with respect to
theco-current mode (independently on the reaction system
consid-ered) (Gallucci et al., 2008c).
Although the tube in tube configuration is quite useful to
workin lab scale and for proof of principle of membrane reactors,
forindustrial scale some other configurations need to be used
inorder to increase the membrane area per volume of vessel used.In
fact, the amount of hydrogen produced is directly related to
theamount of membrane area installed in the reactor. Starting for
thetube in tube configuration, a straightforward way to increase
themembrane area in packed bed is the tube in shell
configuration(Buxbaum, 2002; Tosti et al., 2008). An example of
multi-tubemembrane housing has been patented by Buxbaum (2002)
andreported in Fig. 14. In this case the catalyst is loaded in the
shellside of the reactor while the membrane tubes are connected to
acollector for the pure hydrogen. In particular, in the figure
thepossibility to use a catalyst in a separate chamber is shown.
Incase of reforming reactions, this chamber acts as a
pre-reformingzone where the greatest temperature profiles are
concentrated. Inthis way the membranes will work at an almost
constanttemperature.
The second way to increase the membrane area per volume
ofreactor is adopting the hollow fiber configuration. For example,
incase of perovskite membranes the membrane flux is generallyquite
low and the hollow fiber configuration is quite interesting.The
main investigators of hollow fiber membrane reactors aresummarized
in Table 4.
These studies confirmed that the membrane preparation proce-dure
can be also intensified to produce hollow fiber membraneswith
similar selectivities of the planar or tubular membranes.
Kleinert et al. (2006) studied for example POM in a hollowfiber
membrane reactor. The perovskite membranes used by theauthors were
produced from Ba(Co,Fe,Zr)O3d (BCFZ) powder viaphase inversion
spinning technique. A tube in tube configurationhas been used while
the catalyst was packed in the shell side ofthe reactor.
In their paper the authors show that the membrane was ableto
give quite interesting results with a methane conversion of 82%and
a CO selectivity of 83%. Moreover the membrane was quitestable
under the reactive conditions investigated. Finally, thecombination
of steam reforming and POM was studied by feedingsteam along with
methane in order to suppress the carbonformation. Even in these
conditions the membrane reactorshowed good stability.
The membrane area required for the separation can be reducedby
incr