University of Southern California Dissertations and Theses

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

(USC Thesis Other)

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

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content
University of Southern California
Mork Family Department of Chemical Engineering and
Materials Science

The Use of Carbon Molecule Sieve and Pd
Membranes for Conventional and
Reactive Applications

A Dissertation Presented to the USC Graduate School
In Partial Fulfillment of the Requirements for the Degree
Doctor of Philosophy
Chemical Engineering

Jiang Yu
Table of Contents

Abstract ........................................................................................................................ 1

Chapter 1: Introduction .............................................................................................. 3
1.1 General Information ................................................................................................ 3
1.2 Inorganic Membranes .............................................................................................. 5
1.3 CMSM-Based Membrane Reactors and IGCC Plants ............................................. 9
1.4 Pd Membranes and Fuel Cells ............................................................................... 14
1.5 Thesis Outline ........................................................................................................ 19
1.6 References ............................................................................................................. 21

Chapter 2: The Use of CMSM-Based MR for the Production of Hydrogen from
Coal-Derived Syngas ................................................................................................. 31
2.1 Introduction ........................................................................................................... 31
2.2 Experimental Set-up and Procedures ..................................................................... 32
2.3 Mathematical Model .............................................................................................. 35
2.4 Kinetics Studies ..................................................................................................... 36
2.5 Membrane Reactor Experiments ........................................................................... 40
2.6 Reactor Design and Scale-up ................................................................................. 52
2.7 Summary and Conclusions .................................................................................... 61
2.8 Appendix ............................................................................................................... 62
2.9 References ............................................................................................................. 65

Chapter 3: The Use of CMSM-Based MR for the Production of Hydrogen from
Biomass-Derived Syngas ........................................................................................... 68
3.1 Introduction ........................................................................................................... 68
3.2 Experimental Set-up and Procedures ..................................................................... 69
3.3 Membrane Stability ............................................................................................... 71
3.4 Membrane Reactor Studies .................................................................................... 73
3.5 Summary and Conclusions .................................................................................... 83
3.6 References ............................................................................................................. 85

Chapter 4: The Use of CMSM-Based MR for the Production of Hydrogen from
Biomass-Derived Syngas. The Impact of Tars and Organic Vapors .................... 87
4.1 Introduction ........................................................................................................... 87
4.2 Experimental Set-up and Procedure ...................................................................... 89
4.3 Results and Discussion .......................................................................................... 91
4.4 Conclusions ......................................................................................................... 101
4.5 References ........................................................................................................... 103

Chapter 5: Ultra-Pure Hydrogen Production from a Reformate Mixture Using a
Palladium Membrane Reactor ............................................................................... 105
5.1 Introduction ......................................................................................................... 105
5.2 Experimental Set-up and Procedure .................................................................... 111
5.3 Membrane Characterization ................................................................................. 114
5.3.1 Membrane Permeation Studies ................................................................. 114
5.3.2 CO Effect on the Pd Membrane ............................................................... 116
5.4 Reactor Studies .................................................................................................... 120
5.5 Scale-Up .............................................................................................................. 125
5.6 Summary .............................................................................................................. 128
5.7 References ........................................................................................................... 130

Chapter 6: Preliminary Studies of the Transport Characteristics of CMS
Membranes at High Pressures and Temperatures ............................................... 133
6.1 General Introduction ............................................................................................ 133
6.2 Experimental System ........................................................................................... 134
6.3 Experimental Results ........................................................................................... 135
6.4 References ........................................................................................................... 148

Chapter 7: Concluding Remarks and Suggestions for Future Work ................. 150

1
Abstract
High-temperature gas separations using inorganic membranes have attracted
increased attention in recent years. In particular, the use of such membranes in
high-temperature and high-pressure membrane reactors has the potential to enhance
process intensification and to increase energy savings and/or product yield. Though the
potential benefits of high-temperature conventional as well as reactive gas separations are
substantial, commercialization still remains elusive. A major technical barrier is the lack
of robust inorganic membranes and full-scale modules which are suitable for use at the
high-temperature and high-pressure conditions required.
In this thesis, we focus on the application of two different types of such inorganic
membranes, namely carbon molecular sieve membranes (CMSM) and Pd membranes in
high-temperature and high-pressure reactive processes related to power generation, in
particular, in the context of the Integrated Gas Combined Cycle (IGCC). Specifically, we
study a “one-box” process in which the coal/biomass-derived syngas is fed directly into a
water gas shift (WGS) reactor, which efficiently converts the CO into hydrogen in the
presence of a number of troublesome impurities found in such syngas (e.g., H
2
S, tars,
organic vapors, etc.), and delivers a contaminant-free hydrogen product. This reactor
makes use of a hydrogen-selective carbon molecular sieve membrane and a
sulfur-tolerant Co/Mo/Al
2
O
3
catalyst. In our studies, we investigate the membrane
stability, catalytic kinetics, and the reactor’s overall behavior for different experimental
conditions. The results are also used to validate a mathematical model for the membrane
reactor, which is then used to further discuss the potential scale-up of the proposed
process.
We also study in this Thesis a realistic size, ultra-permeable Pd membrane for pure
hydrogen production for fuel cell applications from a feed with a simulated reformate
2
composition through the water gas shift reaction. Prior to its use in the reactor
experiments, the membrane is characterized through single-gas permeation measurements.
The effect of different conditions during the WGS experiments is experimentally studied,
and the results are again compared with those of a mathematical model. The model is
then further used to study the design aspects of the proposed process. It is shown that the
Pd membrane reactor system under study is capable of attaining almost complete CO
conversion and full hydrogen recovery at realistic experimental conditions akin to those
utilized in industrial applications.

3
Chapter 1: Introduction
1.1 General Information
During the past few decades, as the pressure of a looming environmental crisis has
mounted, research on alternative clean energy has attracted increased attention, with
emphasis on the production of clean fuels such as hydrogen (H
2
), and the increased use of
renewable forms of energy like solar, wind, and tidal waves. This is because one of the
main culprits blamed for environmental pollution today are traditional fossil fuels, which,
when combusted, generate pollutants such as CO
2
, nitrogen oxides (NO
x
), various
unburned hydrocarbons (UHC), and SO
2
, etc. As noted above, an alternative fuel that is
attracting attention, as a result, today is H
2
, because during power generation it burns
cleanly without any pollution (no UHC and CO
2
), according to reaction R1 below.
2H
2
+ O
2
= 2 H
2
O (R1)
Another benefit of adopting hydrogen as a future energy source, in addition to reducing
UHC and CO
2
emissions during its utilization, is that it can be produced cleanly from
readily available resources like coal and renewable biomass; this then diminishes the
need to use the world’s dwindling crude-oil resources.
Hydrogen is used for power and electricity generation primarily in fuel cells for both
mobile and stationary applications. Due to its increased use in these systems, the interest
in its production, delivery, and utilization is growing rapidly as well. However, in order
to make the so-called H
2
energy economy a reality, there are still four major challenges
that remain [1], as shown in Table 1 below. Conventional as well as reactive membrane
separations are playing an increasingly important role in helping overcome some of these
challenges. A number of novel membranes and membrane-based technologies are being
4
investigated. For example, membrane reactors (MR) have already attracted significant
attention to date for use in the two key reactions utilized presently for hydrogen
production, namely steam reforming of methane (SRM) and the water gas shift (WGS)
reaction, since when using highly hydrogen-permselective (e.g., Pd and Pd-alloy)
membranes they can provide higher yields and a highly purified hydrogen product.
Table 1.1 Critical technologies needed for the hydrogen economy [1]
(1) Cost-effective production of hydrogen in a carbon-constrained global
energy system.
(2) Hydrogen purification and storage technologies that will be able to
separate, and purify the hydrogen streams to satisfy the requirements of the
subsequent storage and utilization systems.
(3) An efficient, widely available and well-managed hydrogen delivery and
distribution infrastructure.
(4) Efficient fuel cells and other energy conversion technologies that utilize
hydrogen.

For example, Basile et al. [2] achieved 99% CO conversion for the WGS reaction in a
membrane reactor operating at 595 K, which is higher that the equilibrium CO
conversion (~85%) under such conditions as well as the conversion attained in a
conventional packed-bed reactor. Conventional membrane separations show promise as
well. For example, using lab-scale carbon molecular sieve (CMS) membranes having an
extremely high H
2
/CH
4
separation factor of ~2500 at 298 K, Grainger et al. [3] performed
the separation of hydrogen from a binary H
2
/CH
4
mixture containing 30% H
2
and feed
pressure of 40 bar, obtaining over 90% H
2
recovery and 99.6% H
2
purity. However, there
5
are no commercial CMS membranes (CMSM) currently attaining such high separation
factors, and this is, as result, is a very active area of research today and a key focus in this
Thesis (as is the use of such membranes in reactive separations as well). Though
polymeric membranes have received some attention as well, in this area it is primarily
inorganic membranes that are attracting the greater share of attention due to their ability
to work under high temperatures and pressures.
1.2 Inorganic Membranes
There are different types of inorganic membranes which can be, generally, separated
into two distinct groups, namely dense-phase and porous membranes. As shown in Table
1.2, each type encompasses a variety of membranes made from different materials.
Table 1.2 Different types of inorganic membranes
Dense-phase membranes
Dense-phase metallic membranes, such as Pd
membranes
Dense-phase metallic alloy membranes, such as
Pd/Ag membranes
Porous inorganic
membranes
CMS membranes
Alumina, silica, silica functionalized ceramic
membranes, and SiC membranes
Zeolite membranes

For the dense-phase membranes, the permeation of hydrogen through the metallic
(such as Pd or Pd-alloy) film follows a rather complicated mechanism. The first step is
sorption of hydrogen molecules onto the film surface, followed by hydrogen dissociation,
6
and transport of the hydrogen atoms through the metal lattice, and finally desorption from
the other side of the metallic film (and, potentially, diffusion through a ceramic support
substrate, for the case of supported metal membranes). For thick metallic films, the
diffusion of the hydrogen atoms through the metal bulk is, typically, rate-limiting.
However, for thinner membranes some of the other steps may also become rate
controlling. For ideal membranes (with no pinholes or imperfections), based on the above
mechanism, only H
2
can permeate through the metallic film, resulting in very high
separation factors for the H
2
over other gases. Often, however, these membranes develop
cracks during use and their separation factor significantly diminishes. Pd membranes are
the most common example of dense-phase membranes. Comparing to other metals, Pd
and its alloys have high membrane permeabilities towards hydrogen, which is the reason
that they are used commercially today to produce hydrogen-selective membrane
separation units [4]. The interaction of hydrogen with Pd and its alloys, as a result, has
been widely studied for many years [5].
Transport phenomena through mesoporous and microporous inorganic membranes
can be described by four different mechanisms, as shown schematically in Figure 1.1.
These four different transport mechanisms vary significantly in the resulting membrane
selectivity. Knudsen diffusion gives relatively low separation factors (inversely
proportional to the square root of the ratio of the molecular weights), and is dominant in
mesoporous membranes. Surface diffusion is governed by the selective adsorption of
some of the components of the gas mixture onto the pore surface and the effective
transport of these compounds on these surfaces. For membranes with smaller pores,
multi-layer adsorption of the most condensable compounds takes place (capillary
condensation) resulting in pore blockage and preventing some of the lighter components
from permeating through. For membranes with even smaller pores, molecular sieving
becomes the main separation mechanism, whereby the smaller molecules like hydrogen
7
are allowed to permeate through the membrane while the larger molecules like methane
are excluded [6]. Transport through a given membrane is, generally, via more that one of
the aforementioned mechanisms, though usually one of the four mechanisms (e.g.,
molecular sieving for CMS membranes) may dominate.

Figure 1.1. Transport mechanism of porous membranes: (a) Knudsen, (b) surface
diffusion, (c) capillary condensation, (d) molecular sieving

Like Pd membranes, microporous inorganic membranes also show, generally, a high
permeation rate for H
2
, but their separation factors are, typically, lower than those for the
metallic membranes. CMS membranes are among the most popular microporous
inorganic membranes today, due to their relative ease of preparation and their ability to
operate under conditions which are detrimental for other membranes (e.g., in the presence
of H
2
S that destroys metallic membranes, and steam that severely impacts silica
membranes). CMS membranes have been utilized by our group and others in both gas
8
separation [7] as well as in membrane reactor application [8], and are the primary focus
of this Thesis.
Carbon membranes have been known for more than 30 years. However, it is only
during the recent 15 years that their potential for difficult gas separations has been fully
recognized and acknowledged [9]. Ismail and David have written a comprehensive
review on the development of various types of carbon membranes [9]. As noted above,
CMS membranes have shown high-temperature resistance and excellent chemical
resistance to acids, hot organic solvents, and alkaline solutions. Based on these unique
properties, CMS membranes are reported to have a distinct advantage over other types of
inorganic membranes for a variety of applications. Given the growing importance of
CMS membranes, a number of studies through the years (starting in the early 1980’s)
have been devoted to the study of how the preparation (typically, via the controlled
atmosphere pyrolysis of a polymeric precursor) conditions affect their separation
properties [10-14], and have led to a better understanding of their preparation methods.
For example, it has been discovered that by changing the type of polymeric precursor and
the pyrolysis conditions, one can prepare CMS membrane with the desired gas
permeances and separation factors.
As previously mentioned, the advantage of Pd membranes is that they provide high
H
2
permeance and almost infinite separation factor for H
2
over other gases, which makes
them uniquely qualified for a number of niche applications (one of which will be
discussed in this Thesis – see Chapter 5). However, there are a number of disadvantages
of the Pd membranes, particularly in comparison to the CMS membranes. They include:
1: Other components in the gas mixture (e.g., H
2
O and CO) can also adsorb on the Pd
surface and block dissociation sites for hydrogen, especially when operating at
temperatures below 623 K [15], hence, decreasing the H
2
permeance.
9
2. As noted above, H
2
S is a well-known toxic impurity for Pd membranes, even at
single-digit ppm level. Exposure of the Pd surface to H
2
S has been shown to result into
the formation of an irreversible grey surface scale of palladium sulfide [16] or in the
dramatic pitting of the membrane surface [17]. This is a key challenge for the commercial
use of such membranes, because a number of the hydrogen containing mixtures (e.g.,
syngas from coal and biomass gasification) also contain substantial concentrations of
H
2
S.
3. Pd is a highly-priced precious metal, in limited supply, and its use in large-scale
applications (e.g., coal gasification and power generation) is likely to further exacerbate
problems with current availability and prices [18].
4. The only use of Pd membranes is for H
2
separation, which limits their broad
applicability.

1.3 CMSM-Based Membrane Reactors and IGCC Plants
As noted above, a number of raw materials including fossil fuels, coal and renewable
biomass can be used to produce hydrogen. Coal is an attractive and relatively inexpensive
raw material to produce hydrogen as well as a key energy source for power generation in
a number of the largest economies of the world including the USA and China, where it is
found as an abundant resource [19]. To produce hydrogen, coal must be first gasified via
the aid of an oxidant like air or pure oxygen at high temperatures.

This then produces a
gas mixture, known as coal-gasifier off-gas or syngas, that contains as key species H
2
,
CO, CO
2
, H
2
O, CH
4
, and other gaseous by-products such as various organic vapors, tars,
H
2
S, NH
3
, etc. [19]. The exact syngas composition depends on the operating conditions
10
of the gasifier (e.g., pressure, temperature, coal and oxidant flow rates, gasifier
configuration, etc.), and the type of coal and oxidant used [19]. To produce hydrogen, the
syngas must be first cleaned of its undesired components that include, in addition to the
gaseous by-products noted above, various particulates like coke, and inorganic matter
known as ash [19] (for example, H
2
S can be removed using granular solid sorbents
[20-22], via catalytic conversion [23], or through a solvent-based [24] absorption process).
The clean syngas is then processed in a reactor which converts the CO via its catalytic
reaction with steam, known as the water gas shift (WGS) reaction, into additional
hydrogen and carbon dioxide; the WGS reactor effluent must be further processed to
produce a pure hydrogen product [25]. When coal (and/or biomass – see further
discussion below) gasification and hydrogen generation are combined with electricity
production (e.g., via the use of the hydrogen in a turbine or in a fuel cell) this is known as
the integrated gas combined (power generation) cycle or IGCC [26]. IGCC power plants
are attracting renewed interest today because they are ideally suited for carbon dioxide
capture for its storage and sequestration (CCS) [26].
Another abundant energy source, and also a promising raw material for hydrogen
production, as well as for environmentally-benign power generation via IGCC, is biomass
[27]. The term biomass, as an energy source, encompasses a broad range of materials,
including ligno-cellulosic products such as wood and wood waste, agricultural products
and by-products, food processing and municipal waste, algae and various other aquatic
plants, etc. [28] Moderately-dried biomass can be directly used as a fuel (e.g., via
co-firing in coal power plants) for electricity generation. Ligno-cellulosic materials can
be used to produce liquid fuels (e.g., ethanol) via hydrolysis followed by fermentation of
the resulting sugars [28]. A broader range of liquid fuels can be produced via biomass
pyrolysis [28], though the process is still not commercial.
11
For environmentally-benign power generation from biomass, IGCC offers
potentially the best option (biomass has an added environmental advantage over coal, as
it is a renewable energy source, and thus offers a much greater potential for CO
2
emissions reductions). Here, as with coal-based IGCC, biomass is first gasified in the
presence of oxygen and steam to produce a gasifier off-gas containing similar main
components as with coal-derived syngas (albeit at different concentrations) together with
additional impurities, including H
2
S, NH
3
, and various high molecular weight compounds
known as tars [29,30].

For IGCC power generation, this syngas must be further reacted with steam in a
WGS reactor to enrich its hydrogen content. The WGS reaction is exothermic and is
favored by low temperatures, with equilibrium conversion decreasing as temperature
increases. Therefore, typically two reactors are needed to overcome both equilibrium and
kinetic limitations, and to increase the H
2
conversion [31]. One of these, known as the
high-temperature shift (HTS) reactor, operates at high temperatures using Fe/Cr-based
catalysts, and the other, known as the low-temperature shift (LTS) reactor, operates at
lower temperatures using a Cu/Zn-based catalyst [32]. As is the case with coal-derived
syngas, the WGS reactor product must be further processed in order to produce pure
hydrogen and a CO
2
stream for further storage and sequestration. The whole process is
complicated and highly energy-intensive. In its place, our Group at USC in collaboration
with Media and Process Technology, Inc. (MPT) has recently proposed – see further
discussion to follow and Chapters 2, 3, 4 of this Thesis-- a novel process [33], termed the
“one-box” process that substitutes the conventional dual-bed WGS reactor with a
catalytic membrane reactor. By using a hydrogen-permselective CMSM and a sulfided
Co/Mo catalyst, which are both resistant to the impurities found in biomass-derived
syngas, this novel process avoids the need to use a separate syngas pre-treatment step as
well as a hydrogen purification step, thus significantly simplifying process design (further
12
details about the lab-scale efforts for the development of the one-box process are
provided in this Thesis, see Chapters 2, 3, 4).
As noted above, WGS membrane reactors have received prior attention for the
production of hydrogen. Uemiya et al. [34] were the first to study an atmospheric
pressure Pd-membrane WGS reactor treating pure CO and using Ar as a sweep. At
400 °C (with a stoichiometric H
2
O/CO ratio) they achieved a maximum CO conversion
of 96%. Since the WGS reaction requires higher temperatures, most studies today (with a
few notable exceptions [35]) utilize inorganic membranes, as noted previously, including
dense Pd [2, 36-40], microporous silica [41-44], microporous zeolite [45], and carbon
molecular sieve membranes. [8, 33, 46,47].
Pd or Pd-alloy membrane reactors have attracted most of the attention, starting with
the earliest studies by Uemiya et al. [34] noted above. Bi et al. [39] studied a MR
operating at 375 °C and a pressure of 1.2 bar with a simulated syngas feed (H
2
, 7%; CO,
25%; CO
2
, 15%; N
2
, 53%) and a N
2
sweep gas rate of 28.3 cm
3
/min using a conventional
Co/Cr catalyst and a porous glass supported Pd membrane. They obtained a maximum
CO conversion of 98%. Augustine et al. [48]

used a WGS-membrane reactor to treat a
simulated syngas mixture (H
2
, 22.0%; H
2
O, 45.4%; CO, 22.7%; CO
2
, 9.9%) with a dense
Pd membrane supported on porous stainless steel. They obtained a maximum CO
conversion of 98% and H
2
recovery of 88% with a H
2
O/CO ratio of 2.6 at the reactor
temperature of 450 °C, feed-side pressure of 14.4 bar for GHSV= 2900 h
-1
. Ma and
coworkers [49]

also studied the WGS reaction at temperatures ranging from 420 to
440 °C and pressures ranging from 7 to 20 bar using a larger size composite Pd
membrane, a simulated syngas mixture (H
2
, 40%; CO, 42.2%; CO
2
17.8%), and H
2
O/CO
ratios ranging from 2.5 to 3.5. They obtained a maximum CO conversion of 98.1% and a
H
2
recovery of 85.1% at 440
o
C, and a GHSV = 1130 h
-1
.
13
As the above studies indicate, Pd membranes when used for the WGS reaction (for a
more recent concise review see [50]) can deliver high CO conversions, and a high-purity
hydrogen product, and our own efforts in this area are detailed in this Thesis, see
Chapters 5, 6; their main drawback for the WGS application (other than the limited
availability of the metal, which may ultimately, however, hinder their widespread use for
this large-scale application) is that they are sensitive to the syngas impurities, particularly
H
2
S which adversely affects their characteristics, even at single-digit ppm levels.
Exposure of Pd to H
2
S has been shown to reduce its permeability, and to result into the
formation of a surface scale of Pd sulfide [16], as well as in dramatic pitting of the
membrane surface [17], as noted previously. An exhaustive clean-up step for the syngas,
therefore, becomes necessary [51,52].
WGS membrane reactors making use of silica membranes have been studied and
shown excellent performance as well [41-44]. The drawback with silica membranes for
such a reaction is well known, in that silica undergoes condensation in the presence of
steam. Efforts to improve the hydrothermal stability include functionalization of the
membranes using surfactants to form a hydrophobic silica surface [53], and incorporation
into the silica structure of various metals [54,55] and carbon [56] during the preparation
step, but all these efforts have, so far, found limited success (microporous silica
membranes, on the other hand, show good potential for application to other reactions
where the presence of steam is not required [57,58]).
CMS membranes as a result of the challenges other types of membranes face with
the WGS reaction, have attracted recent attention [8,33,46,52,59]. These membranes, as
noted above, are made via the pyrolysis of polymeric precursors in various atmospheres,
important conditions influencing their properties being: (1) the type of precursor utilized
[60]; (2) the pyrolysis conditions such as the atmosphere and heating protocols [61]; and
14
(3) post-pyrolysis modifications such as activation, oxidation and stabilization [62].
Initial efforts by our group in using CMS membranes for the WGS reaction [8,46] were
important to prove that these membranes show good performance and good stability in
the presence of high-temperature steam. However, these studies were performed with
feeds that did not contain the impurities typically encountered in coal-derived and
biomass-derived syngas. In two most recent studies [33, 59] by our group (which are also
detailed in this Thesis, see Chapters 2, 3,4), CMS membranes were utilized to treat
simulated coal-derived [59] and biomass-derived [33] syngas containing realistic
concentrations of H
2
S and NH
3
. In these studies use was also made of the so-called
sour-shift commercial catalysts [63,64], and during the lab-scale investigations lasting
more than a month, both membranes and catalysts exhibited good and stable performance
in the temperature range of 250-350
o
C (recently Dong and coworkers [45] used silylated
zeolite membranes to carry out the HTS reaction in the presence of H
2
S; though the
reported steam stability of the silica surfaces contrasts prior studies by other groups, see
above, the results are nevertheless promising and will, hopefully, provide impetus for the
further study and development of these membranes).
1.4 Pd Membranes and Fuel Cells
Fuel cells are widely considered as a clean and energy-efficient technology that
shows promise as a future potential replacement for the internal-combustion (IC) engines
in vehicles, IC engines and gas turbines in stationary power generation, and for batteries
in portable power applications [65]. Proton-exchange membrane (also known as polymer
electrolyte membrane or PEM) fuel cells are particularly promising as they are compact
and lightweight units that work at comparatively lower temperatures and pressures, which
makes them good candidates for use in mobile and small-scale distributed stationary
power generation applications. PEM fuel cells make use of a thin, permeable polymeric
15
membrane as the electrolyte, and noble metal (e.g., Pt) electrodes are used on either side
of the membrane in order to catalyze the reactions. Hydrogen is, typically, utilized as the
fuel and is fed to the anode, where it catalytically splits into protons (which transport
through the membrane to the cathode) and electrons that travel along an external load
circuit to the cathode, thus producing electricity. Oxygen (as air) is simultaneously
supplied to the cathode, where it combines with the hydrogen ions to produce water [66].
As noted already in this Introduction, the hydrogen, that PEM fuel cells need to
operate, can be produced from different renewable (e.g., water, biomass) and
non-renewable (e.g., fossil fuels) feed-stocks through various processes [67,68]. Due to
its higher efficiency, currently steam methane reforming is the primary industrial process
that is used for H
2
production. In SMR, steam reacts with methane at high temperatures
and pressures in the presence of a catalyst to produce hydrogen according to the
following reaction:
CH
4
+ H
2
O → CO + 3H
2
(R2)
CO that is simultaneously produced as an undesirable by-product of SMR is known to be
particularly detrimental for PEM fuel cell operation. Even when present at trace
concentrations in the H
2
stream, CO is reported to degrade PEM fuel cell performance as
it severely poisons the electro-active Pt surface in the anode, thus preventing H
2
oxidation.
The tolerance of PEM fuel cells to CO depends on the material used for preparing the
anode, with most studies reporting an acceptable level of 10 ppm or less. Substantially
higher tolerances have been reported as well. Pt–Ni/C, for example, is reported to be
tolerant to up to 50 ppm of CO [69], while Pt–Ru supported on defect-free carbon
nano-tubes has been reported capable to tolerate up to 100 ppm of CO [70].
For the hydrogen produced from SMR to be utilized in PEM fuel cells, care must,
16
therefore, be paid towards decreasing its CO content prior to entering the fuel cell stack.
Typically, the first step involves utilizing the water gas shift (WGS) reaction (reaction R1
above) in tandem with the SMR in order to react the CO in the reformate mixture and to
produce additional H
2
. Conventional WGS reactors can attain high CO conversions, but
not to the degree required to prevent CO-induced PEM fuel cell stack performance
degradation. This, then, means that following the H
2
/CO
2
separation step (e.g., via
pressure swing adsorption (PSA)) an additional CO removal step must be implemented
involving either the catalytic oxidation of CO (this is known as preferential oxidation or
PROX, reaction R3) or the methanation (reaction R4) in order to reduce the CO level into
the acceptable fuel cell operational range.
2CO + O
2
→ 2CO
2
(R3)
CO + 3H
2
→ CH
4
+ H
2
O (R4)
Both approaches, unfortunately, end-up consuming part of the H
2
(since in the
presence of oxygen some catalytic oxidation of the hydrogen is unavoidable).
Methanation offers more simplicity than PROX, since no oxygen needs to be added to the
reactor. On the other hand, it typically results in higher hydrogen consumption for the
same CO conversion (because of its reaction with the CO
2
present in the syngas mixture).
For either process, the lower the CO content is, the less hydrogen is consumed and the
more efficient the process is. Therefore, it is highly desirable to be able to produce either
pure H
2
or a H
2
product with an acceptable CO content, which does not require
sacrificing part of hydrogen, during further purification, and increasing process
complexity.
The above conventional SMR + WGS + PSA + PROX (or methanation) system is
one of significant complexity. To add to this complexity, since WGS is an
17
equilibrium-limited reaction, and as previously noted, usually two reactors are utilized:
one operating at high temperatures, and the other at low temperatures in order to
maximize the CO conversion. In order to improve process efficiency and to minimize
system complexity, membrane reactors are employed for the WGS reaction, as previously
noted, which combine reaction and separation in one unit, and in the context of their
application to pure hydrogen production provide many advantages. By removing
hydrogen from the reactor, for example, conversion increases and less steam is required.
CO
2
is separated in the reject side at high pressure, thus eliminating the need for using a
separate PSA unit. With highly permselective membranes the need for using a separate
PROX or methanation step may be completely eliminated, but even for less permselective
membranes the overall system efficiency for the PROX (or the methanation) step
improves, and hydrogen loss is minimized. Even the use of inexpensive, low-pressure
steam as a sweep for the permeate side of these WGS-MR may provide an advantage, as
it generates a pre-humidified H
2
product as a feed, which is a requirement for many PEM
fuel cell stack applications. Our own efforts in this area are detailed in this Thesis, see
Chapters 5, 6.
Membrane reactors have been used for both the SMR [71-72] and the WGS reactions
[7, 33, 37, 38, 39, 59, 73, 74] for H
2
production. Due to the high process temperatures
required, metal membranes like Pd and its alloys [37, 73, 74], inorganic membranes like
microporous silica [41-44], and carbon molecular sieve (CMS) membranes [7, 46, 59]
have all been utilized. As discussed above, CMSM provide a great advantage for the
WGS application in the context of IGCC applications for coal and biomass, as they are
very robust to the various impurities (e.g., H
2
S, NH
3
, tar, etc.) one, typically, encounters
in such systems, which are highly detrimental to the mechanical stability of the metal
membranes. Pd and Pd-alloy membranes are, however, well-suited for high-purity
hydrogen production via the use of SMR for fuel cell applications, due to their very high
18
(in principle, infinite, as noted above) selectivity towards hydrogen. As a result, a number
of groups have investigated their application for hydrogen production. Jansen et al. [75],
for example, performed a techno-economic analysis of two process schemes involving
the use of Pd membrane reactors integrated in natural gas-based combined power
generation cycles. These include (a) a process utilizing a WGS, and (b) one utilizing an
SMR membrane reactor. The cycle involving the use of the WGS membrane reactor
proved to provide higher efficiency and lower capital cost. Jansen et al. [75] attributed
this to the fact that the typical feed to an SMR unit contains methane and steam, with
little or no H
2
being present. The WGS feed, on the other hand, contains considerable
amounts of H
2
, which are produced during the SMR reaction. Thus, when using the Pd
membrane for the WGS step, since the driving force for H
2
transport across the
membrane is higher, less membrane area is required. Jansen et al. [75] also note that
SMR reactors, typically, operate at much higher temperatures when compared to their
WGS counterparts. High temperature has a negative impact on Pd membrane stability due
to accelerated membrane degradation; to counteract this impact one may have to utilize
thicker membranes, thus increasing the membrane cost, but more importantly also
reducing the H
2
flux.
A number of researchers have studied the use of Pd membranes for the WGS reaction,
as already noted. Basile et al. [73] used Pd and Pd/Ag membranes for the WGS reaction,
and reported conversion values higher than thermodynamic equilibrium. They used a
model WGS feed to their reactor which contained no inerts or other reaction products one
would normally expect to find in a typical reformate mixture, and for simplicity they
utilized nitrogen as sweep. Iyoha et al. [37] performed the WGS reaction in the absence
of any catalyst at a temperature of 900 °C and an operating pressure of 2.41 bar making
use of Pd and Pd-alloy (80 wt% Pd–20 wt% Cu) tubular membranes. The high rate of
hydrogen extraction through the Pd-based membranes resulted in increasing the carbon
19
monoxide conversion to 93%. The conversion decreased from 93% to 66% and hydrogen
recovery from 90% to 85% when the Pd membranes were replaced with the Pd–Cu ones,
which Iyoha et al. [37] attributed to the lower permeance of the Pd–Cu alloy membranes.
Though the study of Iyoha et al. [37] provides the advantage of not needing to use a
catalyst, the long-term mechanical stability of these materials at these extremely high
temperatures raises concerns.
Mendes et al. [74] recently studied the performance of a Pd–Ag membrane reactor for
the WGS reaction using a Cu–Zn LTS catalyst. A dense metallic permeator tube (5 cm in
length, and 1 cm in internal diameter) was assembled from a commercial flat sheet of
Pd–Ag. The catalyst was packed in the membrane lumen. A simulated reformate feed
composition (4.7% CO, 34.78% H
2
O, 28.70% H
2
, 10.16% CO
2
, balanced with N
2
) was
utilized. The MR performance was evaluated in terms of CO conversion and H
2
recovery
at different experimental conditions, e.g., different temperatures (200–300 °C), pressures
(1–4 bar), and using either vacuum (30 mbar) or N
2
as a sweep stream. The MR delivered
CO conversions above the thermodynamic equilibrium in most of the cases, but hydrogen
recoveries were rather low (except in the case of using vacuum). No data were presented
for the permselectivity of the membrane or for the purity of the hydrogen product.

1.5 Thesis Outline
This Thesis is organized as follows. Chapter 1 contains a general introduction about the
production methods for hydrogen generation, and the use of both conventional and
reactive membranes separations to improve reactor yield and product purity. Chapter 2
describes our experimental and modeling efforts in using CMSM-based WGS/MR for
producing hydrogen from coal-derived syngas. Chapter 3 describes our complimentary
20
efforts in in using CMSM-based WGS/MR for producing hydrogen from
biomass-derived syngas. Chapter 4 describes our further efforts in applying
biomass-derived syngas to produce hydrogen using CMSM-based WGS/MR, with
particular focus on the impact of tars and organic vapors, commonly found in the
biomass-derived syngas. Chapter 5 describes our experimental and modeling efforts in
using Pd-based WGS/MR for producing high purity hydrogen from a reformate mixture.
Chapter 6 describes our preliminary experimental and modeling efforts on the transport
characteristics of CMSM, with a particular focus on their eventual use in high-pressure
WGS/MR applications.

21
1.6 References
[1] G.Q. Lu, J.C. Diniz da Costa, M. Duke, S. Giessler, R. Socolow, R.H. Williams, T.
Kreutz, Inorganic membranes for hydrogen production and purification: A critical review
and perspective. J. of Colloid and Interface Sci. 314 (2007) 589–603.
[2] A. Basile, A. Criscuoli, F. Santella, E. Drioli, Membrane reactor for water gas shift
reaction, Gas. Sep. Purif. 10., (1996) 243-254.
[3] D. Grainger, M.B. Ha¨gg, The recovery by carbon molecular sieve membranes of
hydrogen transmitted in natural gas networks, J. of Colloid and Interface Sci. 314 (2007)
589–603.
[4] S. Uemiya, State-of-the-art of supported metal membranes for gas separation. Sep.
Purif. Methods 28(1) (1999) 51-85.
[5] FA. Lewis, The palladium hydrogen system. London: Academic Press (1967).
[6] H.L. Fleming. In: 1986 Membrane Technology/Planning Conference Proceeding,
Business Communications Co., Cambridge, MA (1986).
[7] A. Harale, H. T. Hwang, P. K.T. Liu, M. Sahimi, T. T. Tsotsis, A membrane-based
reactive separation system for CO
2
removal in a life support system, J. Membr. Sci. 315
(2008) 116–124.
[8] A. Harale, H. T. Hwang, P. K.T. Liu, M. Sahimi, T. T. Tsotsis, Experimental studies
of a hybrid adsorbent-membrane reactor (HAMR) system for hydrogen production, Chem.
Eng. Sci. 62 (2007) 4126 – 4137.
22
[9] A.F. Ismail, L.I.B. David, A review of the latest development of carbon membranes
for gas separation, J. Membr. Sci. 193 (2001) 1–18.
[10] R. F. P. M. Moreira, H. J. José, A. E. Rodrigues, Modification of pore size in
activated carbon by polymer deposition and its effect on molecular sieve selectivity,
Carbon 39 (2001) 2269–2276.
[11] J. Koresh, A. Soffer, Study of molecular sieve carbons. Part 1. Pore structure,
gradual pore opening and mechanism of molecular sieving, J. Chem. Soc., Faraday Trans.
76 (1980) 2457-2471.
[12] J. Koresh, A. Soffer, Molecular sieving range of pore diameters of adsorbents, J.
Chem. Soc., Faraday Trans. 76 (1980) 2507-2509
[13] A. B. Fuertes, T. A. Centeno, Carbon molecular sieve membranes from
polyetherimide, Micropor. Mesopor. Mater. 26 (1998) 23-26.
[14] V. Geizler, W.J. Koros, Effects of polyimide pyrolysis conditions on carbon
molecular sieve membrane properties, Ind. Eng. Chem. Res. 35 (1996) 2999–3003.
[15] H. Li, A. Goldbach, W. Z. Li, H. Y. Xu, PdC formation in ultra-thin Pd membranes
during separation of H2/CO mixtures, J. Membr. Sci. 299 (2007) 130-137.
[16] R. C. Hurlbert, J. O. Konecny, Diffusion of hydrogen through palladium, J. Chem.
Phys. 34 (1961) 655-658
[17] A. Kulprathipanja, G. O. Alptekin, J. L. Falconer, J. Douglas Way, Pd and Pd–Cu
membranes: inhibition of H
2
permeation by H
2
S, J. Membr. Sci. 254 (2005) 49-62.
[18] J.N. Armor, Applications of catalytic inorganic membrane reactors to refinery
23
products, J. Membr. Sci. 147 (1998) 217–233.
[19] N. V. Gnanapragasam, B. V. Reddy, M. A. Rosen, Hydrogen production from coal
gasification for effective downstream CO
2
capture, Int. J. Hydrogen Energy 35 (2010)
4933−4943.
[20] X. Bu , Y. Ying, X. Ji, C. Zhang, W. Peng, New development of zinc-based sorbents
for hot gas desulfurization, Fuel Process Technol. 88 (2007) 14−147.
[21] E. R. Monazam, L. J. Shadle, D. A. Berry, Modeling and analysis of S-sorption with
ZnO in a transport reactor, Chem. Eng. Sci. 63 (2008) 2614−23.
[22] J. M. Sa´nchez-Herva´, J. Otero, E. Ruiz, A study on sulphidation and regeneration
of Z-Sorb III sorbent for H
2
S removal from simulated ELCOGAS IGCC syngas, Chem.
Eng. Sci. 60 (2005) 2977−89.
[23] J. S. Chung, S. C. Paik, H. S. Kim, D. S. Lee, I. S. Nam, Removal of H
2
S and/or SO
2

by catalytic conversion technologies, Catal. Today 35 (1997) 37-43.
[24] F. P. Nilsen, I. S. L. Nilsen, H. Lidal, Novel contacting technology selectively
removes H
2
S, Oil & Gas J. 100 (2002) 56−62.
[25] G. Weber, Q. Fu, H. Wu, Energy efficiency of an integrated process based on
gasification for hydrogen production from biomass, Developments in Chemical
Engineering and Mineral Processing 14 (2006) 33–49.
[26] M. J. Beer, High efficiency electric power generation: the environmental role,
Progress in Energy and Combustion Science 33 (2007) 107–34.
24
[27] Biomass Basics: The Facts About Bioenergy, U. S. Department of Energy, Feb.
2010
[28] K. Nath, D. Das, Hydrogen from biomass, Current Science 85 (2003) 265-271.
[29] A. V. Bridgwater, Renewable fuels and chemicals by thermal processing of biomass,
Chem. Eng. J. 91 (2003) 87−102.
[30] G. F. Schiefelbein, Biomass thermal gasification research: recent results from the
United States DOE’s research program, Biomass 19 (1989) 145−159.
[31] J. R. Ladebeck, J.P. Wagner, Handbook of fuel cells-fundamentals, Technology and
Applications, Wiley, Chichester, 2003.
[32] M. P. Aznar, M. A. Caballero, J. Corella, G. Molina, J. M. Toledo, Hydrogen
production by biomass gasification with steam-O
2
mixtures followed by a catalytic steam
reformer and a CO-shift system, Energy Fuels 20 (2006) 1305−1309.
[33] M. Abdollahi, J. Yu, H. T. Hwang, P. K. Liu, R. Ciora, M. Sahimi, T. T. Tsotsis,
Process intensification in hydrogen production from biomass-derived syngas, Ind. Eng.
Chem. Res. 49 (2010) 10986−10993.
[34] S. Uemiya, N. Sato, H. Ando, E. Kikuchi, The water gas shift reaction assisted by a
palladium membrane reactor, Ind. Eng. Chem. Res. 30 (1991) 585–589.
[35] J. Zou, J. Huang, W. S. W. Ho, CO
2
-selective water gas shift membrane reactor for
fuel cell hydrogen processing, Ind. Eng. Chem. Res. 46 (2007) 2272−2279.
[36] S. Tosti, A. Basile, G. Chiappetta, C. Rizzello, V. Violante, Pd–Ag membrane
25
reactors for water gas shift reaction, Chem. Eng. J. 93 (2003) 23–30.
[37] O. Iyoha, R. Enick, R. Killmeyer, B. Howard, B. Morreale, M. Ciocco,
Wall-catalyzed water-gas shift reaction in multi-tubular Pd and 80 wt% Pd–20 wt% Cu
membrane reactors at 1173 K, J. Membr. Sci. 298 (2007) 14–23.
[38] G. Barbieri, A. Brunetti, G. Tricoli, E. Drioli, An innovative configuration of a Pd
based membrane reactor for the production of pure hydrogen: experimental analysis of
water gas shift, J. Power Sources 182 (2008) 160–167.
[39] Y. Bi, H. Xu, W. Li, A. Goldbach, Water-gas shift reaction in a Pd-membrane
reactor over Pt/Ce
0.6
Zr
0
.
4
O
2
catalyst, Int. J. Hydrogen Energy 34 (2009) 2965–2971.
[40] A. Brunetti, G. Barbieri, E. Drioli, Upgrading of a syngas mixture for pure hydrogen
production in a Pd–Ag membrane reactor, Chem. Eng. Sci. 64 (2009) 3448–3454.
[41] S. Giessler, L. Jordan, J.C. Diniz da Costa, G.Q.M. Lu, Performance of hydrophobic
and hydrophilic silica membrane reactors for the water gas shift reaction, Sep. Purif.
Technol. 32 (2003) 255–264.
[42] A. Brunetti, G. Barbieri, E. Drioli, T. Granato, K.H. Lee, A porous stainless steel
supported silica membrane for WGS reaction in a catalytic membrane reactor, Chem. Eng.
Sci. 62 (2007) 5621–5626.
[43] S. Battersby, M.C. Duke, S. Liu, V. Rudolph, J.C. Diniz da Costa, Metal-doped
silica membrane reactor: operational effects of reaction and permeation for the water gas
shift reaction, J. Membr. Sci. 316 (2008) 46–52.
[44] S. Battersby, S. Smart, B. Ladewig, S. Liu, M.C. Duke, V. Rudolph, J.C. Diniz da
26
Costa, Hydrothermal stability of cobalt silica membranes in a water gas shift membrane
reactor, Sep. Purif. Technol. 66 (2009) 299–305.
[45] SJ. Kim, Z. Xu, G. K. Reddy, P. Smirbiotis, JH. Dong, Effect of pressure on
high-temperature water gas shift reaction in microporous zeolite membrane reactor, Ind.
Eng. Chem. Res. 51 (2012) 1364−1375.
[46] A. Harale, H.T. Hwang, P.K.T. Liu, M. Sahimi, T. T. Tsotsis, Design aspects of the
cyclic hybrid adsorbent-membrane reactor (HAMR) system for hydrogen production,
Chem. Eng. Sci. 65 (2010) 427−435.
[47] S. Sá, H. Silva, J.M. Sousa, A. Mendes, Hydrogen production by methanol steam
reforming in a membrane reactor: palladium vs. carbon molecular sieve membranes, J.
Membr. Sci. 339 (2009) 160–170.
[48] A. S. Augustine, Y. H. Ma, N. K. Kazantzis, High pressure palladium membrane
reactor for the high temperature water gas shift reaction, Int. J. Hydrogen Energy 36
(2011) 5350−5360.
[49] J. Catalano, F. Guazone, I. P. Mardilovich, N. K. Kazantzis, Y. H. Ma, Hydrogen
production in a large scale water gas shift Pd-based catalytic membrane reactor, Ind. Eng.
Chem. Res. 52 (2013) 1042–1055.
[50] F. Galluci, E. Fernandez, P. Corengia, M. van Sint Annaland, Recent advances in
membranes and membrane reactors for hydrogen production, Chem. Eng. Sci. 92 (2013)
40−66.
[51] M. Abdollahi, J. Yu, P. K. T. Liu, R. Ciora, M. Sahimi, T. T. Tsotsis, Ultra-pure
hydrogen production from reformate mixtures using a palladium membrane reactor
27
system, J. Membr. Sci. 390–391 (2012) 32–42.
[52] P. K. T. Liu, M. Sahimi, T. T. Tsotsis, Process intensification in hydrogen
production via membrane-based reactive separations, Current Opinion in Chemical
Engineering 1 (2012) 342−351.
[53] H. L. Castricum, A. Sah, R. Kreiter, D. H. A. Blank, J. F. Vente, Hydrothermally
stable molecular separation membranes from organically linked silica, J. Mater. Chem.18
(2008) 2150–2158.
[54] M. Kanezashi, M. Asaeda, Hydrogen permeation characteristics and stability of
Ni-doped silica membranes in steam at high temperature, J. Membr. Sci. 271 (2006)
86–93.
[55] T. Tsuru, R. Igi, M. Kanezashi, T. Yoshioka, S. Fujisak, Y. Iwamoto, Permeation
properties of hydrogen and water vapor through porous silica membranes at high
temperatures, AIChE. J. 57 (2011) 618–629.
[56] M. Kanezashi, K. Yada, T. Yoshioka, T. Tsuru, Organic–inorganic hybrid silica
membranes with controlled silica network size: preparation and gas permeation
characteristics, J. Membr. Sci. 348 (2010) 310–318.
[57] S Nakao, K. Akamatsu, Development of membrane reactors for dehydrogenating
organic chemical hydrides to supply high-purity hydrogen, J. Jpn. Pet. Inst. 54 (2011)
287−297.
[58] K. Akamatsu, T. Sugawara, M. Nakane, T. Hattori, S. Nakao, Development of a
membrane reactor for decomposing hydrogen sulfide into hydrogen using a
high-performance amorphous silica membrane, J. Membr. Sci. 325 (2008) 16–19.
28
[59] M. Abdollahi, J. Yu, P. K. T. Liu, R. Ciora, M. Sahimi, T. T. Tsotsis , Hydrogen
production from coal-derived syngas using a catalytic membrane reactor based process, J.
Membr. Sci.363 (2010) 160–169.
[60] S. M. Saufi, A. F. Ismail, Fabrication of carbon membranes for gas separation—a
review, Carbon 42 (2004) 241–259.
[61] V. C. Geiszler, W. J. Koros, Effects of polyimide pyrolysis conditions on carbon
molecular sieve membrane properties, Ind. Eng. Chem. Res. 35 (1994) 2999–3003.
[62] J. N. Barsema, S. D. Klijnstra, J. H. Balster, N. F. A. Van der Vegt, G. H. Koops, M.
Wessling, Intermediate polymer to carbon gas separation membranes based on matrimid
PI, J. Membr. Sci. 238 (2004) 93–102.
[63] E. Knozinger, J. Weitkamp, Handbook of heterogeneous catalysis, Wiley, Hoboken,
1997.
[64] H. F. Rase, Handbook of commercial catalysts: heterogeneous catalysts, CRC Press,
Boca Raton, 2000.
[65] Loyselle, Patricia; Prokopius, Kevin, Teledyne Energy Systems, Inc., Proton
Exchange Member (PEM) Fuel Cell Engineering Model Powerplant. Test Report: Initial
Benchmark Tests in the Original Orientation, NASA, Glenn Research Center, Retrieved
15 Sep. 2011
[66] Fuel Cells, Office of Energy Efficiency & Renewable Energy, U. S. Department of
Energy
29
[67] M. N. Islam, M. Dixon, Review of various process routes for hydrogen production,
Fuel Energy Abstr. 45 (2004) 254.
[68] J. D. Holladay, J. Hu, D.L. King, Y. Wang, An overview of hydrogen production
technologies, Catal. Today 139 (2009) 244–260.
[69] T. Okada, H. Yano, C. Ono, Novel CO tolerant anode catalysts for PEFC based on
platinum and organic metal complexes, J. New Mater. Electrochem. Syst. 10 (2007)
129–134.
[70] E. Yoo, T. Okada, T. Kizuka, J. Nakamura, Effect of various carbon substrate
materials on the CO tolerance of anode catalysts in polymer electrolyte fuel cells,
Electrochemistry 75 (2007) 146–148.
[71] J. Shu, B.P.A. Grandjean, S.Kaliaguine, Methane steam reforming in a symmetric Pd
and Pd Ag/porous SS membrane reactors, Appl. Catal. A: Gen. 119 (1994) 305–325.
[72] Y. Chen, Y. Wang, H. Xu, G. Xiong, Efficient production of hydrogen from natural
gas steam reforming in palladium membrane reactor, Appl. Catal. B: Environ. 81 (2008)
283–294.
[73] A. Basile, G. Chiappetta, S. Tosti, V. Violante, Experiments and simulation of both
Pd and Pd/Ag for a water gas shift membrane reactor, Sep. Purif. Technol. 25 (2001)
549–571.
[75] D. Mendes, V. Chibante, J.M. Zheng, S. Tosti, F. Borgognoni, A. Mendes, L.M.
Madeira, Enhancing the production of hydrogen via water gas shift reaction using
Pd-based membrane reactors, Int. J. Hydrogen Energy 35 (2010) 12596–12608.
30
[77] D. Jansen, J.W. Dijkstra, R.W. van den Brink, T.A. Peters, M. Stange, R. Bredesen,
A. Goldbach, H.Y. Xu, A. Gottschalk, A. Doukelis, Hydrogen membrane reactors for
CO
2
capture, Energy Procedia 1 (2009) 253–260.

31
Chapter 2: The Use of CMSM-Based MR for the Production of Hydrogen from
Coal-Derived Syngas
*

2.1 Introduction
In this Chapter, we study a WGS-MR treating a feed, which contains substantial
quantities of H
2
S (several thousand ppm) typical of what may be encountered in the
off-gas of a coal gasifier. The idea is to make this MR the “heart” of a “one-box” process
in which the gasifier syngas is fed directly into the WGS reactor, which then effectively
converts the CO into hydrogen in the presence of H
2
S and other impurities, and delivers a
substantially contaminant-free hydrogen product. For the MR we have chosen to use
CMS membranes, which are prepared by the deposition of polymeric precursors on
tubular alumina substrates commercially available by Media and Process Technology, Inc.
(for further details about the preparation technique, see [1-3]). In ongoing field studies by
our team these membranes have already proven stable in the treatment of commercial
refinery streams containing high levels of contaminants, such as H
2
S and NH
3
. CMS
membranes have also been shown previously [4-5] to be highly stable in the presence of
steam in the WGS reaction environment. Since H
2
S in the syngas poisons the common
WGS catalysts [6], in our study we are making use of a so-called sour-shift catalyst [7],

*
The work described in this Chapter was carried out collaboratively with my USC student
colleague Dr. Mitra Abdollahi, and is described in the following publication: M. Abdollahi, J. Yu,
P. K. T. Liu, R. Ciora, M. Sahimi, T. T. Tsotsis, Hydrogen production from coal-derived syngas
using a catalytic membrane reactor based process, Journal of Membrane Science 363 (2010)
160–169.

32
in order to overcome the problem of catalyst poisoning. These sulfur-resistant WGS
catalysts, containing sulfided Co-Mo or Ni-Mo supported on alumina [6], and on various
other supports, such as zeolites [8], titania and zirconia [9], were first proposed almost 20
years ago; they have since been shown to exhibit good performance (at relatively low
temperatures, 250–350 °C) with syngas feeds containing high H
2
S concentrations (if
adequately pre-sulfided prior to use, they perform satisfactorily even in feed streams that
contain low level of H
2
S).
In what follows, we will first discuss the experimental studies to determine the
reaction kinetics and rate parameters of the commercial sulfur-tolerant Co/Mo/Al
2
O
3

catalyst that we utilize. Then, we will discuss the experimental membrane reactor
performance for a range of pressures and sweep ratios, and compare it with results from a
simple isothermal MR model. Finally, the same model is used to further investigate the
design features of the proposed process.

2.2 Experimental Set-up and Procedures
A schematic of the CMS-MR system used in this thesis is shown in Figure 2.1. The
tubular CMS membrane is sealed inside the tubular stainless steel (SS) reactor using
graphite o-rings and compression fittings. The catalyst particles are first thoroughly
mixed with ground quartz particles, and are then loaded into the annular space in between
the membrane and the reactor body. We dilute the catalyst with inert quartz particles in
order to completely fill the annular reactor volume, and to be able to operate the reactor
bed under isothermal conditions. The experimental system consists of three sections: (i)
The feed section, which consists of gas cylinders, mass-flow controllers (MFC), syringe
pumps, and the steam-generating units; (ii) the reactor section, which consists of the MR,
33
a furnace for heating the reactor, pressure gauges for measuring the pressure, two
condensers and two moisture-traps to remove the water from the reject- and
permeate-side streams of the reactor, and two traps to remove the H
2
S from the same
streams; (iii) the analysis section, that consists of a gas chromatograph to analyze the
concentration of the exit gas streams, two bubble flow-meters for measuring the total
flow rates, and Drager tubes to measure the H
2
S concentration.

Figure 2.1. Experimental set-up used in the membrane reactor experiments.
For the experiments, the reactor is maintained isothermal by placing it in a six-zone
furnace, the temperature in each zone controlled with the aid of six temperature
controllers and thermocouples installed in six different locations in the bed. An additional
thermo-well is installed in the bed in order to monitor the temperature along the length of
the bed using a sliding thermocouple. The feed and sweep gas streams flow at specified
rates controlled by mass-flow controllers. Pressures are controlled by adjusting the needle
valves at the exit of the reactor and sweep sides. Pressure gauges are installed at the inlet
and outlet of both the feed and permeate sides in order to monitor the pressure.
34
Two syringe pumps are used in order to supply a controlled flow of water into two
steam-generating units (one connected to the feed line, the other to the sweep stream line).
The steam generators are well-insulated SS vessels which are packed with quartz beads,
in order to accelerate the water evaporation and to dampen out any fluctuations in the
flow of the steam that is generated. They are heated by heating tapes wrapped around
them, and their temperature is controlled with the aid of temperature controllers. All the
stream lines, including feed, sweep, permeate, and reject lines are insulated and
heat-traced using heating tapes. Their temperature is also controlled with temperature
controllers. In particular, the feed and sweep gas flows are preheated to the reaction
temperature before entering the reactor.
The above experimental system is also utilized to carry out permeation studies for
characterizing the membrane properties. For such experiments, the sweep gas
(permeate-side) inlet is closed, gas flows into the feed side and the flow rates and the
compositions of the permeate and reject streams are measured. For calculating the water
permeance, the permeated stream by-passes the condenser and goes directly into the
adsorbent bed where the water is captured. The amount of water that permeates is
calculated by measuring the weight of the adsorbent before and after water permeation.
During the MR experiments, the gas streams exiting the reject and permeate sides flow
first through condensers and then through moisture-traps in order to capture the water.
The flow rates of the water-free stream are then measured by a bubble flow-meter. A
small slip-stream from both the reject and permeate sides is intermittently removed to
measure the H
2
S content through the use of Drager tubes. These are graduated tubes that
contain a Cu compound that reacts with the H
2
S and produces CuS, which results into a
color change from blue to black. The degree of color change, read on a linear scale on the
colorimetric detection tube, is translated into an accurate measurement of the level of H
2
S
(as low as 0.2 ppm) present in the gas stream. Another small slip-stream from both the
35
reject and permeate sides is allowed to pass through an adsorption bed (in order to
remove its H
2
S content), and is then used to measure the composition with an online gas
chromatograph.
To carry out the packed-bed reactor experiments (to compare its performance with
that of the MR) and for measuring the catalytic reaction kinetics, the same procedure is
followed, except that the inlet and exit valves for the sweep gas are closed.
2.3 Mathematical Model
In several previous papers published by our group [10,11], we developed an
isothermal co-current membrane reactor model to describe the behavior of our CMS –MR
membrane reactor and also the kinetics for the WGSR catalyst. Here is a brief description
about this model.
Empirical equation for mass transport through the CMS membrane:
(2.1)
Mass balances for components in the feed and permeate sides:
(2.2)
(2.3)
The pressure drop in the packed-bed is calculated using the Ergun equation:
(2.4)
36
(2.5)
(2.6)
Boundary conditions:
At V = 0: , , (2.7)
CO conversion is defined by Eq. 2.8.
(2.8)
Hydrogen recovery ( ) is given by Eq. 2.9.
(2.9)
To analyze the data, we use an isothermal co-current flow (feed to permeate) MR
model previously utilized by our group for describing such reactors [12]. Several
assumptions are made in order to simplify the mathematical analysis. Briefly, it is
assumed that the reactor operates isothermally (this has been validated experimentally)
under ideal gas law conditions, and that the external mass-transfer resistances are
negligible for the catalyst and the membrane.

2.4 Kinetics Studies
The kinetics of the WGS reaction has received substantial attention in recent years
37
(e.g., [13-15]). Several researchers, in particular, have also studied the WGS kinetics over
sour WGS catalysts (e.g., [16-20]). Various rate expressions have been reported and
different mechanisms [13] have been proposed to explain the observed reaction rate
equations. Most researchers, however, make use of an empirical, power-law rate
expression, without reference to any specific reaction mechanism. In this study, a
commercial Co-Mo/Al
2
O
3
sour-shift catalyst is utilized to perform the WGS reaction (the
physical properties of the catalyst are shown in Table 2.1). The catalyst and quartz
particles are crushed separately into smaller particles and their sizes are sorted with the
aid of mesh-screens in the range of 600–800 µm. Prior to loading into the reactor, the
catalyst is mixed and diluted with the quartz particles in order to completely fill the
reactor space and to be able to conveniently operate the reactor bed under isothermal
conditions. The catalyst particles are irregular in shape, but are roughly considered
spherical for the estimation of the bed properties. Since the Co and Mo metal components
of the fresh catalyst, as received, are in the oxidized form, they must be sulfided prior to
the reaction. The activation procedure involves the in situ reduction of the metals using a
gas mixture containing H
2
and H
2
S using the temperature and pressure protocol as
specified by the catalyst manufacturer.
Table 2.1. Physical and chemical properties of the sour WGS catalyst.
Catalyst form Extrudates
Catalyst size 0.003 m
Chemical
composition
CoO: 3–4 wt%; MoO
3
: 13–15 wt%; Al
2
O
3
: 80–85 wt%
Bulk density 592.68 × 10
3
g/m
3

Surface area 160–220 m
2
/g
38
Catalyst form Extrudates
Pore volume 0.55–0.65 × 10
−6
m
3
/g

Since the catalyst manufacturer did not provide any reaction rate information on the
catalyst, a series of kinetic experiments have been carried out in a PBR using 12 g of the
catalyst intermixed with 90 g of the quartz at temperatures in the range of (220–300 °C),
pressures in the range of (1–5 atm) and W
c
/F
CO
(weight of undiluted catalyst (g) over the
molar flow rate of CO (mol/h)) in the range of (70–320). Results are shown in Figure 2.2
and Figure 2.3 in terms of CO conversion vs. W
c
/F
CO
for a feed composition of
H
2
:CO:CO
2
:CH
4
:H
2
S = 2.6:1:2.13:0.8:0.05 (corresponding to a fractional composition of
39.5% H
2
, 15.2% CO, 32.4% CO
2
, 12.2% CH
4
and 0.7% H
2
S) and a near stoichiometric
H
2
O/CO ratio in the feed of 1.2. To validate the reaction rate, the CO conversion data
were fitted using nonlinear regression analysis and the following empirical rate
expression was found to provide the best fit for all the experimental data generated
(including the PBR experiments carried out in tandem with the MR experiments):
(2.10)
where (bar) is the partial pressure for component j, and K
eq
is the
overall reaction equilibrium constant [21].
39

Figure 2.2. CO conversion as a function of Wc/F
CO
for various packed-bed reactor
pressures at T = 250 °C

Figure 2.3. CO conversion as a function of Wc/F
CO
for various packed-bed reactor
temperatures at P = 3 atm.
40
2.5 Membrane Reactor Experiments
The purpose of these experiments was to show that both the membrane and catalyst
perform stably under the WGS reaction environment. In all the MR experiments reported
here, we used a feed with composition (on a dry basis) of H
2
:CO:CO
2
:CH
4
:H
2
S =
2.6:1:2.13:0.8:0.05 (intended to simulate a coal gasifier's exit composition [22]) and a
near stoichiometric H
2
O/CO ratio in the feed of 1.2. Commercial WGS reactors generally
operate in the presence of substantial excess steam. One potential advantage of WGS-MR
is that they give the same conversion, at or near stoichiometric H
2
O/CO ratios, thus the
choice of the low H
2
O/CO ratio in our experiments. Two different CMS membranes of
the exact same dimensions (L = 254 mm, ID = 3.5 mm, OD = 5.7 mm) were utilized in
the isothermal experiments reported here. The first membrane (CMS#1) with relatively
high flux but moderate selectivity was used for a preliminary series of membrane reactor
experiments, and also for an extended series of membrane characterization studies, with
the entire series of experiments lasting for over 1 month. Table 2.2 presents three sets of
mixed-gas permeation data. The first composition is that of the dry feed for the MR
experiments. The second composition is the exit (on a dry basis) reactor composition
corresponding to 70% CO conversion, while the third composition is the same as the
second one, but with water being present (the lower H
2
S content in the second and third
compositions is because H
2
S is purchased premixed in the CO gas cylinder). As can be
seen in Table 2.2, varying the mixed-gas composition (including the H
2
S concentration)
has little effect on the permeance of most of the gases (other than hydrogen for which the
permeance varies by <15%).
Figure 2.4 shows the CO conversion and H
2
recovery at three different W
c
/F
CO
for the
MR experiments at P = 5 and permeate steam sweep gas ratio =0.1
41
(the error bars reflect the carbon and hydrogen loss or gain due to the experimental errors
in measuring the flow rates and compositions).
Table 2.2. Mixed-gas permeation data for CMS#1.
Gas mixture composition
Gas Permeance m
3
/(m
2
h bar) Separation Factor (S.F.)
(1) H
2
:CO:CO
2
:CH
4
:H
2
S = 39.5%:15.2%:32.4%:12.2%:0.7%
H
2
1.37 1.0
CO 0.02 68.5
CO
2
0.05 27.4
CH
4
0.01 137.0
H
2
S 0.01 137.0

(2) H
2
:CO:CO
2
:CH
4
:H
2
S = 45.36%:4.52%:38.91%:11%:0.21%
H
2
1.40 1.0
CO 0.02 70.0
CO
2
0.04 35.0
CH
4
0.01 140.0
H
2
S 0.01 140.0

(3)
H
2
:CO:CO
2
:CH
4
:H
2
O:H
2
S = 42.45%:4.233%:36.4%:10.29%:6.43%:0.197%
H
2
1.56 1.0
CO 0.02 78.0
42
Gas mixture composition
Gas Permeance m
3
/(m
2
h bar) Separation Factor (S.F.)
CO
2
0.05 31.2
CH
4
0.01 156.0
H
2
O 1.1 1.4
H
2
S 0.01 156.0
In the experiments we used 15 g of the catalyst diluted with 80 g of ground quartz
glass, packed inside the MR in the annular space between the reactor wall and the
membrane (membrane shell-side). Shown on the same figure are the simulated
conversion and recovery lines using the model, utilizing the experimental rate expression,
and the last set of mixed-gas permeances in Table 2.2. Figure 2.5 shows the conversion
and hydrogen recovery at 250 °C for a different set of experimental conditions as shown
in the figure caption.
43

Figure 2.4. Comparison of the experimental (a) conversion and (b) recovery with the
model predictions atT = 250 °C, P = 5 atm and sweep ratio = 0.1 using CMS#1.
In order to further explore the range of appropriate operating conditions for the
CMS-MR, and to further validate the ability of membranes and catalysts to function
stably, a second series of MR experiments for the WGS reaction were carried out at
300 °C, a temperature which pushes the limits of application for both the sour-shift
catalyst and the CMS membranes.
44

Figure 2.5. Comparison of the experimental (a) conversion and (b) recovery with the
model predictions atT = 250 °C, P = 3 atm and sweep ratio = 0.3 using CMS#1.

In these experiments, we again used the same feed gas composition and H
2
O/CO
ratio, with the experiments being carried out at two different feed pressures (3 and 5 atm,
with the permeate-side pressure under atmospheric conditions), and two different
45
permeate steam sweep gas ratios of 0.1 and 0.3. In this series of experiments, in tandem
with the MR experiments, we also carried out under identical conditions PB experiments
(every MR experiment was followed by a PB experiment during which the permeate side
is kept closed, as previously described). A different membrane (CMS#2), this time with
high selectivity but relatively low permeability was utilized, and 10 g of the catalyst
diluted with 80 g of ground quartz glass was packed in the membrane shell-side (the
reactor shell-side volume was slightly smaller in these series of experiments, thus the use
of a smaller amount of catalyst).
Since the emphasis in these experiments was on the MR performance, we carried out
only a limited number of permeation studies using single gases before and after the MR
experiments. Prior studies by our group with several of these membranes indicate that the
mixed-gas permeances of the various gases generally remain relatively close to the values
measured during the single-gas experiments [16]. Table 2.3 indicates the single-gas
permeances for this membrane measured prior to the initiation of the reactor experiments.
For H
2
S, the MR experiments indicated that it does not permeate through the membrane
(within the detection limit of the Drager tube utilized) and, hence, its permeance was
taken to be zero, since it has no impact on the modeling results. (In extensive studies in
which both the surface of the membrane module and the plumping were specifically
coated to avoid potential wall adsorption, the H
2
S permeance was always found to lie in
between the permeance of CO/N
2
and CH
4
). The permeance of water was deduced by
fitting all the compositional data available for both the reject and permeate sides of the
MR (e.g., see Figure 2.6 for the fit for one set of such data).

46

Figure 2.6. Compositions of (a) reject and (b) permeate side at P = 3 atm and sweep
ratio = 0.3.

The MR experiments lasted more than 1 month during which period membrane gas
permeances changed less than 7% before and after the MR experiments, indicative of the
47
good stability of the membranes under the WGS-MR environment. The model discussed
earlier was again used to simulate the experimental results, together with the
experimental empirical power-law rate expression and the experimental single-gas
permeances (Table 2.3) as discussed above.

Table 2.3. Single-gas permeation data for CMS#2.
Pure Gas

Permeance m
3
/(m
2
h bar) Separation factor (S.F)
H
2
0.5354 1
CO 0.0037 145.88
CO
2
0.0107 50.03
CH
4
0.0014 385.18
H
2
O
a
0.0922 5.8
a: Fitted value.

Figure 2.7 shows the CO conversion and H
2
recovery at three different W
c
/F
CO
for
the MR and PB experiments at 3 atm and a steam sweep ratio of 0.1. Figure 2.8 presents
the CO conversion and H
2
recovery at the same conditions mentioned above, but at a
steam sweep ratio equal to 0.3. Figure 2.9 and Figure 2.10 show the CO conversion and
H
2
recovery for the MR and the PB experiments at 5 atm and steam sweep ratios equal to
48
0.1 and 0.3, respectively (the solid lines in the figures represent the modeling results).

Figure 2.7. Comparison of the experimental (a) conversion and (b) recovery with the
model predictions atT = 300 °C, P = 3 atm and sweep ratio = 0.1 using CMS#2.
49

Figure 2.8. Comparison of the experimental (a) conversion and (b) recovery with the
model predictions atT = 300 °C, P = 3 atm and sweep ratio = 0.3 using CMS#2.
50

Figure 2.9. Comparison of the experimental (a) conversion and (b) recovery with the
model predictions atT = 300 °C, P = 5 atm and sweep ratio = 0.1 using CMS#2.
51

Figure 2.10. Comparison of the experimental (a) conversion and (b) recovery with the
model predictions atT = 300 °C, P = 5 atm and sweep ratio = 0.3 using CMS#2.

These figures make it clear that the model does, generally, a good job in predicting
the experimental MR as well as the PB behavior. The MR attains conversions which are
52
higher than those for the packed-bed. The relatively low conversions are due to
limitations with the size of our laboratory system, which accommodates only one small
CMS membrane, and the limited amount of catalyst that can be filled inside the small
reactor. The pressure drop measured in the laboratory (and also the one calculated using
the Ergun equation) was also negligible due to the same reasons. Higher conversions can
be attained for higher W
c
/F
CO
, as the simulation results also indicate (see below). We are
limited, however, in our laboratory system by the amount of catalyst that can be utilized.
Once the amount of catalyst is fixed, the maximum value of W
c
/F
CO
attained is
determined by the minimum flow rate one is able to provide, reflecting the lower limits of
the MFC in our experimental system (it should be noted, however, that
high Wc/F
CO
values are also not interesting for practical applications).
A key conclusion from this series of experiments is that the membrane and the
catalyst exhibited robust behavior and remained stable throughout the series of
experiments which lasted almost 1-month in the presence of hydrogen sulfide under the
harsh WGS environment.

2.6 Reactor Design and Scale-up
Since the model performs reasonably well in describing the experimental results, it
can be used to further study the effect of various parameters on WGS-MR performance,
in terms of reactor conversion, hydrogen recovery, and purity. The target here is to
choose appropriate conditions which maximize both the CO conversion and H
2
recovery,
and minimize the CO content of the hydrogen product. In the simulations that follow the
experimental power-law reaction rate expression together with the experimental
single-gas permeances for the CMS#2 membrane were used. The membrane length is
53
increased to 1 m and it is assumed that the catalyst and quartz are packed along the entire
length of the membrane. The conditions utilized, unless otherwise noted in the figures’
captions, are listed in Table 2.4.
Table 2.4. The base-case and the range of the experimental conditions used in the
simulations.
Parameter Base-case Applied range
Feed side pressure (PF) 5 atm 5–30 atm
Permeate-side pressure (PP) 1 atm –
Reactor temperature (TR) 300 °C –
Steam sweep to feed ratio (SR) 0.3 0.3–2
Number of membranes 1 1–4
Length of the membrane (L) 1 m –
Inner diameter of the membrane (ID) 0.0035 m –
Outer diameter of the membrane (OD) 0.0057 m
Inner diameter of the reactor 0.0318 m
Weight of the catalyst (Wc) 10 g –
Hydrogen permeance 0.54 m
3
/(m
2
h bar) 0.5–3
H
2
/CO separation factor (S.F.) 146 100–300
Figure 2.11 shows the effect of pressure. Since the coal gasifier typically operates in
the pressure range of 20–30 atm [23], the advantages of operating the CMS-MR at high
pressures is obvious. Increasing the pressure helps increase both the conversion and the
hydrogen recovery by increasing the partial pressure difference of hydrogen across the
membrane. The pressure effect is more prominent at lower W
c
/F
CO
, and for a constant
54
weight of catalyst that means higher feed flow rates. As expected, even at the highest
pressures the reactor does not attain complete conversion due to the omnipresent loss of
CO, which indicates that a more appropriate type of reactor may be a hybrid system
consisting of a packed-bed, followed by an MR [24].

Figure 2.11. Effect of pressure on (a) conversion and (b) recovery, L = 1 m, T = 300 °C
and sweep ratio = 0.3.
Figure 2.12 shows the effect of the sweep ratio. As the figure indicates, increasing
55
the SR increases both the conversion and recovery, as expected, since sweeping helps
maintain the permeate-side partial pressures low. Increasing the sweep ratio does not
affect the CO transport through the membrane as much as increasing the reactor side
pressure, and as a result the impact of CO loss is not as severe and CO conversion
continues to increase as the sweep ratio increases.
Figure 2.12. Effect of sweep ratio on (a) conversion and (b)
recovery, L = 1 m, T = 300 °C and P = 5 atm.
56
Figure 2.13 shows the effect of the membrane area on the MR performance. In these
simulations, we have kept the amount of catalyst constant and have increased the number
of membranes that are packed into the reactor. The amount of quartz utilized has been
adjusted to fill the annular space between the membrane and the reactor wall (we estimate
that we can comfortably fit up to four membranes inside the reactor). As Figure 2.13
indicates, increasing the number of membranes (n), which translates into increasing the
membrane area per unit catalyst weight, increases both CO conversion and H
2
recovery,
due to the more rapid transfer of products to the permeate side. This favorably shifts the
reaction equilibrium towards the product side. Similarly to the pressure effect, the effect
of increasing the membrane area is stronger at lower W
c
/F
CO
, which at constant weight of
the catalyst corresponds to higher feed flow rates.

57

Figure 2.13. Effect of the membrane area on (a) conversion and (b)
recovery, L = 1 m, T = 300 °C, P = 5 atm and sweep ratio = 0.3.

58

Figure 2.14. Effect of the H
2
permeance on (a) conversion and (b)
recovery, L = 1 m, T = 300 °C, P = 5 atm and sweep ratio = 0.3.

Figure 2.14 shows the effect of varying the H
2
permeance on the MR performance
(while keeping the separation factors towards the other species constant, as indicated in
Table 2.3). As expected, increasing the H
2
permeance helps increasing both CO
conversion and H
2
recovery, the effect being more pronounced for the hydrogen recovery.
Figure 2.15 shows the effect of varying hydrogen permeance (while maintaining the
59
separation factors constant) on CO concentration (on a dry basis) in the product stream.
Note that while higher permeances have a positive impact on CO conversion and
hydrogen recovery, they have, on the other hand, a negative impact on hydrogen purity.
Finally, Figure 2.16 shows the effect of varying the H
2
/CO separation factor by varying
the CO permeance, while maintaining the permeances of all other species constant (see
Table 2.3), on the CO concentration (on a dry basis) in the product stream. Decreasing
the CO permeance impacts the conversion (not shown here) by decreasing the inadvertent
CO loss to the permeate side, but as expected, its most significant impact is lowering the
CO concentration in the product stream.

Figure 2.15. Effect of the H
2
permeance on CO concentration in the product
stream, L = 1 m, T = 300 °C,P = 5 atm, sweep ratio = 0.3 and H
2
/CO separation factor
(S.F.) = 146.

60

Figure 2.16. Effect of the CO permeance on CO concentration in the product
stream, L = 1 m, T = 300 °C,P = 5 atm, sweep ratio = 0.3 and
H
2
permeance = 0.54 m
3
/(m
2
h bar).

As noted previously, the proposed “one-box” approach is being studied in the context
of IGCC power plants, where the goal is to carry out the WGS step without the need to
cool the gasifier off-gas stream to remove its various contaminants (e.g., H
2
S) and then
having to reheat it back to the WGS reaction temperature. Turbines, internal combustion
(IC) engines and proton exchange membrane (PEM) fuel cells have been studied for
power generation using the hydrogen-enriched syngas. Turbines and IC engines are
significantly more tolerant to low levels of sulfur, CO and other contaminant than PEM
fuel cells. If PEM fuel cells are the option of choice for power generation, however, then
as the simulations above indicate an additional polishing step (absorption/adsorption for
the H
2
S and other contaminants and preferential oxidation for CO) may be required.
However, the energy needed for such a step, when treating the permeate stream of the
61
proposed “one-step” process, is likely to be a fraction of what would be needed to treat
the off-gas of a conventional sour-shift reactor.
2.7 Summary and Conclusions
In this Chapter, the “one-box” process which combines reaction and membrane
separation in the same unit was experimentally evaluated for the WGS reaction. The
kinetics of the same reaction over sulfided Co-Mo/Al
2
O
3
catalyst was investigated and a
data-validated rate expression and kinetic parameters were obtained. Nanoporous carbon
molecular sieve membranes were used for the in situ hydrogen separation. The
membranes’ performance was investigated under the operating conditions and their
transport properties were used for the model predictions. The modeling studies indicated
good agreement with the experimental data. The MR performance was investigated for a
range of pressures and sweep ratios, and showed higher CO conversions and H
2
purity
compared with those of the traditional packed-bed reactor. The effect of the membrane
properties and experimental conditions on the performance of the system was also
investigated. The “one-box” process proved to possess several advantages over the
traditional systems including increasing CO conversion, decreasing the amount of steam
required for the reaction, and being able to deliver a product with significantly lower CO
content. Using impurity-resistant catalyst adds another advantage to this system by
allowing one to perform the reaction in the presence of hydrogen sulfide; for the IGCC
power plants this would result in considerable energy savings. The catalyst and the CMS
membranes have demonstrated good stability in the presence of hydrogen sulfide in
continuous reactor experiments lasting over a month.

62
2.8 Appendix
Table 1 Nomenclature
A
F
cross-sectional area for the feed-side (m
2
)
d
p
particle diameter in the feed side (m)
f

friction factor
F
j
molar flux for component j (mol/m
2
. h)

molar flux for hydrogen (mol/m
2
. h)
g
c
gravity conversion factor
G
F
superficial mass flow velocity in the feed side (g/m
2
·h)
ID inner diameter of the membrane (m)
k reaction rate constant (mol/g.h.bar
2
)
K
eq
equilibrium constant
L length of membrane (m)
n pressure exponent
n
j
F
molar flow rate for component j in the feed side (mol/h)

molar flow rate for hydrogen in the feed side (mol/h)
n
j
P
molar flow rate for component j in the permeate side (mol/h)

molar flow rate for hydrogen in the permeate side(mol/h)
Re
F
N

Reynolds number for the feed-side
OD outer diameter of the membrane (m)
P
j
partial pressure for component j (bar)

partial pressure for hydrogen(bar)
P
F
feed-side pressure (bar)
63
P
j
F
partial pressure for component j in the feed-side (bar)

partial pressure for hydrogen in the feed side (bar)
P
P
permeate side pressure (bar)
P
j
P
partial pressure for component j in the permeate side (bar)

partial pressure for hydrogen in the permeate side (bar)
r overall reaction rate expression (mol/g.h)

hydrogen recovery, defined by Eq. (16)
SR
steam sweep gas ratio

n
j 0
P
∑
n
j 0
F
∑ ( )

T temperature
u
F
superficial flow velocity on the feed side (m/h)
U
j
membrane permeance for component j (mol/m
2
.h.bar)

H
2
permeance (mol/m
2
.h.bar
n
)
V reactor volume variable (m
3
)
W
C
weight of the catalyst (g)

carbon monoxide conversion, defined by Eq. (15)
m
α
membrane area per feed side reactor volume (m
2
/m
3
)
β

equilibrium coefficient
c
β
fraction of solid volume occupied by catalyst
v
ε

bed porosity in the feed side
F
µ
viscosity of the fluid (g/m·h)
ρ
F
average density of the fluid (g/m
3
)
ρ
c
catalyst density (g/m
3
)
64
ν
j
stoichiometric coefficient for component j

65
2.9 References
[1] M.G. Sedigh, W.J. Onstot, L. Xu, W.L. Peng, T.T. Tsotsis, M. Sahimi, Experiments
and simulation of transport and separation of gas mixtures in carbon molecular sieve
membranes, J. Phys. Chem. 102 (1998) 8580–8589.
[2] M.G. Sedigh, L. Xu, T.T. Tsotsis, M. Sahimi, Transport and morphological char-
acteristics of polyetherimide-based carbon molecular sieve membranes, Ind. Eng. Chem.
Res. 38 (1999) 3367–3380.
[3] M.G. Sedigh, M. Jahangiri, P.K.T. Liu, M. Sahimi, T.T. Tsotsis, Structural charac-
terization of polyetherimide-based carbon molecular sieve membranes, AIChE J. 46
(2000) 2245–2255.
[4] A. Harale, H.T. Hwang, P.K.T. Liu, M. Sahimi, T.T. Tsotsis, Design aspects of the
cyclic hybrid adsorbent-membrane reactor (HAMR) system for hydrogen production,
Chem. Eng. Sci. 65 (2010) 427–435.
[5] A. Harale, H.T. Hwang, P.K.T. Liu, M. Sahimi, T.T. Tsotsis, Experimental stud- ies
of a hybrid adsorbent-membrane reactor (HAMR) system for hydrogen production,
Chem. Eng. Sci. 62 (2007) 4126–4137.
[6] D.S. Newsome, The water-gas shift reaction, Catal. Rev. 21 (1980) 275–318.
[7] K. Kochloefl, G. Ertl, H. Knözinger, J. Weitkamp (Eds.), Handbook of Heteroge-
neous Catalysis, vol. 4, Wiley, Weinheim, 1997.
[8] M. Laniecki, W. Zmierczak, Acid–base properties of sulfided Ni–Mo–Y zeo- lite
catalysts for water-gas shift reaction, Stud. Surf. Sci. Catal. 75 (1993) 2569–2572.
66
[9] M. Laniecki, M. Malecka Grycz, F. Domka, Water gas shift reaction over sulfided
molybdenum catalysts. I. Alumina, titania and zirconia-supported catalysts, Appl. Catal.
A: Gen. 196 (2000) 293–303.
[10] M. Abdollahi, J. Yu, H. T. Hwang, P. K. Liu, R. Ciora, M. Sahimi, T. T. Tsotsis,
Process intensification in hydrogen production from biomass-derived syngas, Ind. Eng.
Chem. Res. 49 (2010) 10986−10993.
[11] M. Abdollahi, J. Yu, P. K. T. Liu, R. Ciora, M. Sahimi, T. T. Tsotsis , Hydrogen
production from coal-derived syngas using a catalytic membrane reactor based process, J.
Membr. Sci.363 (2010) 160–169.
[12] H.T. Hwang, A. Harale, P.K.T. Liu, M. Sahimi, T.T. Tsotsis, A membrane-based
reactive separation system for CO2 removal in a life support system, J. Membr. Sci. 315
(2008) 116–124.
[13] M.V. Twigg, Catalyst Handbook, Wolfe Publishing Ltd., London, 1990.
[14] E. Knozinger, J. Weitkamp, Handbook of Heterogeneous Catalysis, Wiley, Hoboken,
1997.
[15] H.F. Rase, Handbook of Commercial Catalysts: Heterogeneous Catalysts, CRC
Press, Boca Raton, 2000.
[16] A. Cimino, B. A. DeAngelis, The application of X-ray photo electron spectroscopy
to the study of molybdenum oxides and supported molybdenum oxide catalysts, J. Catal.
36 (1975) 11–22.
[17] P. Hou, D. Meeker, H. Wise, Kinetic studies with a sulfur-tolerant water gas shift
catalyst, J. Catal. 80 (1983) 280–285.
67
[18] N.R. Srivatsa, S.W. Weller, Water-gas shift kinetics over sulfided catalyst: elevated
pressure, in: 9th International Congress on Catalysis, Calgary, Canada, 1988.
[19] R. Hakkarainen, T. Salmi, R.L. Keiski, Water-gas shift reaction on a
cobalt–molybdenum oxide catalyst, Appl. Catal. A: Gen. 99 (1993) 195–215.
[20] M. Nagai, K. Matsuda, Low-temperature water-gas shift reaction over
cobalt–molybdenum carbide catalyst, J. Catal. 238 (2006) 489–496.
[21] Y. Choi, H. G. Stenger, Water gas shift reaction kinetics and reactor modeling for
fuel cell grade hydrogen, J. Power Sources 124 (2003) 432–439.
[22] Gasification Introduction, National Energy Technology Laboratory, U. S.
Department of Energy
[23] F. Starr, (IGCC) Integrated gasification combined circle for carbon capture &
storage, A Claverton Group Paper, 11 Feb. 2009
[24] G. Barbieri, A. Brunetti, G. Tricoli, E. Drioli, An innovative configuration of a
Pd-based membrane reactor for the production of pure hydrogen: experimental analysis
of water gas shift, J. Power Sources 182 (2008) 160–167.

68
Chapter 3: The Use of CMSM-Based MR for the Production of Hydrogen from
Biomass-Derived Syngas
*

3.1 Introduction
In this Chapter, we study a WGS-MR treating a feed which contains H
2
S and NH
3
,
typical of what may be encountered in the off-gas of a biomass gasifier. The idea here, as
in Chapter2, is to make this MR the “heart” of a “one-box” process in which the gasifier
syngas is fed directly into the WGS reactor (containing the CMS membrane and the
impurity-resistant Co/Mo/Al
2
O
3
catalyst), which then effectively converts the CO into
hydrogen in the presence of impurities, and delivers a contaminant-free hydrogen product.
The primary objective of the project is to demonstrate the technical feasibility of the
proposed “one-box” process, specifically focusing on the unique issues associated with
biomass-derived syngas.
In what follows, we first present results of the stability tests of the CMS membranes
under the expected operating conditions in the presence of simulated biomass gasifier
off-gas containing model organic vapor and tar compounds. We then present WGS
packed-bed and MR experiments with feeds containing H
2
S and NH
3
at levels typical of
those encountered in biomass-derived syngas. The experimental MR performance for a
range of pressures and sweep ratios is discussed and compared with the results obtained
from a simple isothermal MR model. Lastly, the same model is used to further investigate
the design features of the proposed process.

*
The work described in this Chapter was carried out collaboratively with my USC colleague Dr.
Mitra Abdollahi, and is described in the following publication: Process intensification in
hydrogen production from biomass-derived syngas, Mitra Abdollahi, Jiang Yu, Hyun T. Hwang,
Paul K. T. Liu, Richard Ciora, Muhammad Sahimi, Theodore T. Tsotsis, Industrial and
Engineering Chemistry Research 2010, 49, 10986-10993

69
3.2 Experimental Set-up and Procedures
The experiment set-up and procedure is the same as described earlier in Chapter 2. A
schematic of the MR system used in this study is shown in Figure 2.1. In order to prepare
the simulated biomass feed composition, a new feed gas tanks was added and also an
additional NH
3
adsorbents following the original H
2
S adsorbent in the set-up of Figure
2.1 (for further experimental details, please consult the original publication).
In this Chapter we also report preliminary stability tests of the membrane under
non-reactive conditions (these experiments were carried out in collaboration with our
industrial colleagues at MPT). The experimental system used in these tests (housed at
MPT) is shown in Figure 3.1. It consists of two parts, namely (i) the feed subsystem, and
(ii) the membrane subsystem. The feed subsystem is used to prepare and deliver
simulated tar-laden biomass gasifier off-gas to the membrane. It consists of an HPLC
pump and check-valves and shut-off valves used to control the liquid flow to the unit.
During the tests, a solution of naphthalene (serving as a model tar) in toluene (serving as
a model organic vapor) is first prepared and then delivered by the HPLC pump into a
constant temperature vaporization chamber at a constant flow rate. Gas is also delivered
to the unit using a separate line, its flow controlled via a mass flow controller. The liquid
solution is fed to the top of the mixing chamber and vaporized in a hot zone. This design
assures that a uniform gas/vapor stream is delivered to the membrane and prevents the
naphthalene from crystallizing in the feed lines prior to vaporization. The membrane
subsystem consists of a stainless steel (SS) module placed in a temperature-controlled
furnace, in which the CMS membrane (with dimensions of 5.7 mm OD, 3.5 mm ID, and
254 mm length) is sealed with the aid of graphite packing. During the experiments, the
retentate stream in the membrane module passes through a carbon-trap to remove the
naphthalene and toluene, and its flow rate is then measured by a bubble-flow meter, as is
70
the flow rate of the permeate-side stream. (There is no need to use a carbon trap for the
permeate stream, as for these highly selective membranes very little naphthalene or
toluene escapes through). The pressure of the retentate chamber was controlled by a
back-pressure regulator, while the permeate-side pressure was always kept atmospheric.

Figure 3.1. Schematic of the membrane testing unit used in the membrane stability tests
in the presence of simulated biomass derived syngas with model tar compounds.

71
3.3 Membrane Stability
During the testing of membrane performance in the presence of the tar (naphthalene)
and organic vapor (toluene) surrogate compounds, the membrane performed well. Figure
3.2, for example, shows the permeance and selectivity for one of the CMS membranes we
tested as a function of time at an operating temperature of 250 °C and a pressure of ~2.0
atm. After an initial decay in the membrane permeance, the performance was stable
throughout the remainder of the run. The initial 10−15% decline in membrane permeance
is attributed to naphthalene/toluene molecules blocking the largest pores in the membrane.
However, pores below a certain size are inaccessible to these large molecules, and remain
free for He (and H
2
) permeation. Very little impact was observed on the membrane
selectivity during the test.

Figure 3.2. He permeance and He/N
2
selectivity of the CMS membrane in the presence
of naphthalene/toluene as model tar and organic vapors, respectively, in the biomass
gasifier off-gas at T = 250 °C and P = 2 atm.
72

Condensation and accumulation of tar-like species on the membrane surface leading
to significant or even complete blocking of the membrane surface was, prior to the
initiation of our study, a key concern for potential fouling of the membrane during
biomass gasifier off-gas conditioning and H
2
recovery. Though the membrane tests above
indicated that the membrane is stable under temperature and pressure conditions akin to
the WGS environment, one cannot exclude situations where, due to various process
upsets, the membrane surface will foul due to tar deposition. It is important then that the
CMS membrane can be regenerated under field implementable conditions consisting of
temperatures below 300 °C, and in the presence of inert purge gases such as steam, N
2
,
etc. During this project, fouling and subsequent regeneration were tested by exposing the
membrane to naphthalene and toluene at low temperatures to promote surface fouling,
followed by regeneration by He purging at 250 °C. Figure 3.3 shows that considerable
permeation loss occurs when a CMS membrane is operated in the presence of these
model tar and organic vapor compounds at 150 °C. However, regeneration is readily
achieved by purging of the membrane in the presence of inert gas at 250 °C. This
indicates that at low temperatures only surface coverage through condensation occurs
with little or no irreversible pore-plugging of the membrane. These results are also
consistent with our previous experience with these membranes in pilot-plant tests
conducted in the presence of VGO hydrocracker and coal gasifier off-gas.
73

Figure 3.3. Testing of the fouling and regeneration of a CMS membrane in the presence
of naphthalene/toluene as model tar and organic vapors, respectively, in the biomass
gasifier off-gas.

3.4 Membrane Reactor Studies
As noted above, the key goal in this study is to validate the ability of the WGS-MR,
(which is to function as the “heart” of the proposed “one-box” process) to effectively
convert CO in the biomass-derived syngas in the presence of its common impurities, and
to deliver a contaminant-free hydrogen product. All MR experiments reported here were
carried out with a simulated syngas feed with the following composition
(H
2
/CO/CO
2
/N
2
/CH
4
/NH
3
/H
2
S = 0.67:1.00:1.00:2.67:0.2:0.00067:0.0006) which is
typical of the composition of an air-blown biomass gasifier off-gas [1]. A near
stoichiometric H
2
O/CO ratio in the feed of 1.1 is used. To study the H
2
S/NH
3
effect on
the MR performance, in this series of preliminary experiments we did not add any tar or
74
organic vapors into the syngas. (Experiments with syngas compositions containing these
additional contaminants are reported in Chapter 4).
Prior to initiation of the WGS-MR experiments, packed-bed reactor experiments were
carried out to derive a rate expression for the WGS reaction to be used in the MR models.
Though the WGS reaction has been studied by a number of groups in recent years, [2, 3]
and several of these studies use sour-shift catalysts, [4-10] there is still no general
agreement on the WGS rate expression. Weller and co-workers [7-10] have used the
following empirical power law rate expression for the WGS reaction over a sour-shift
catalyst:
, where (3.1)
where P
j
is the partial pressure for component j, and K
eq
is the overall reaction
equilibrium constant [11]. They studied the reaction over a sulfided Mo/Al
2
O
3
they
prepared in the laboratory, [7] as well as a Co/Mo/Al
2
O
3
catalyst they prepared in the
laboratory and a commercial Co/Mo/Al
2
O
3
catalyst [8-10]. For the WGS reaction over
the Mo/Al
2
O
3
catalyst they reported two sets of reaction rate parameters, corresponding
to the experiments at two different pressures, namely 5 and 15 atm (these parameters are
shown in Table 3.1). For the data with the Co/Mo/Al
2
O
3
catalysts they reported one set of
rate parameters also shown in Table 3.1. Lund [12,13] analyzed the same data using a
microkinetic model, which provides a better insight into the behavior of the catalyst
(including the presence of a maximum in the rate as a function of the CO/H
2
O ratio, and
the effect of H
2
S), but which also makes use of a much larger set of rate parameters that
must be determined experimentally.

75
Table 3.1. Kinetic parameters for the power-law rate model
tions, using the same feed composition to compare the results
with those of the MR experiments.
Prior to the initiation of the MR experiments, the membrane
was characterized through single-gas permeation experiments
(other than for water, for which the permeance was measured
as described above). The permeances of the various species are
shown in Table 2. Our previous studies with CMS membranes
indicate that the mixed-gas permeances of the various gases
generally remain fairly close to the values measured during the
single-gas experiments.
23
For the H
2
S and NH
3
syngas con-
taminants, the MR experiments indicate that they do not
substantially permeate through the membrane, and their per-
meance was set equal to zero for the simulations. (In extensive
studies in which both the surface of the membrane module and
the plumping were specifically coated to avoid potential wall
adsorption,theH
2
Spermeancewasalwaysfoundtoliebetween
the permeances of N
2
and CH
4
).
Figure 6 shows the effect of the feed-side pressure on the
CO conversion at differentW
C
/F
CO
for the MR and the packed-
bed experiments at 300 °C and an SR equal to 0.3. Figure 7
shows the effect of the steam SR on the CO conversion for the
MR experiments at different W
C
/F
CO
at 300 °C and a feed
pressureof3atm(theerrorbarsinbothfiguresreflectthecarbon
loss or gain due to experimental errors in measuring the flow
rates and compositions). The lines represent the simulated
conversions using our power-law rate expression (Table 1) and
the single gas permeances shown in Table 2. In general, the
Table 1. Kinetic Parameters for the Power-Law Rate Model
k
0
mol/(atm
(a+b+c+d)
·h·g) E (J/mol) ab c d
ref 36 219 38826 0.52 0.21 -0.1 -
ref 35
P ) 5 atm
6.3 22064 0.7 0.14 --
ref 3535
P ) 15 atm
6.0 24911 0.8 0.29 -0.07 -
this work 4.9 28145 0.61 ((0.0605) 0.37 ((0.0752) -0.38 ((0.0103) -0.54 ((0.0079)
Figure 5. Experimental vs. fitted CO conversion using the power law rate
expression at various packed-bed reactor experimental conditions.
Table 2. Single-Gas Permeation Data for the CMS Membrane
before and after the MR Experiments
pure gas
(before MR experiments)
pure gas
(after MR experiments)
gas
permeance
m
3
/(m
2
·h·bar)
separation
factor S. F.
permeance
m
3
/(m
2
·h·bar)
separation
factor S. F.
H
2
1.121 1.00 0.828 1.00
CO 0.024 46.71 0.026 31.85
CO
2
0.062 18.08 0.072 11.5
N
2
0.014 80.07 0.015 55.2
CH
4
0.009 124.56
H
2
O 0.374 3.00
Figure 6. Effect of the MR feed-side pressure on CO conversion at T )
300 °C and a steam sweep ratio) 0.3.
Figure7. Effect of the steam sweep ratio on CO conversion atT) 300 °C
and P) 3 atm.
Ind. Eng. Chem. Res., Vol. 49, No. 21, 2010 10991

In our study, we also utilized the above empirical power law rate expression to fit our
experimental packed-bed reactor data (including the packed-bed experiments carried out
in tandem with the MR experiments) using a commercial Co/Mo/Al
2
O
3
catalyst (further
details about the catalyst’s physical properties and characteristics and its activation
procedure can be found elsewhere [14]). The rate parameters (including their 95%
confidence limits) are also shown in Table 3.1. The goodness of fit can be seen in
Figure 3.4, in which we compare the experimental and fitted CO conversions for various
packed-bed temperatures (220−300 °C) and pressures (1−5 atm) using two different feed
compositions, namely H
2
/CO/CO
2
/N
2
/CH
4
/NH
3
/H
2
S =
0.67:1.00:1.00:2.67:0.2:0.00067:0.0006, typical of the composition of an air-blown
biomass gasifier off-gas, and H
2
/CO/CO
2
/CH
4
/H
2
S = 2.6:1:2.1:0.8:0.05), typical of the
composition of a coal/oxygen-blown gasifier off-gas. We utilize in these
experiments W
C
/F
CO
values in the range of 70−320, where W
C
(10−12 g) is the weight of
undiluted catalyst, and F
CO
is the molar flow rate of CO (mol/h).
76

Figure 3.4. Experimental vs. fitted CO conversion using the power law rate expression at
various packed-bed reactor experimental conditions.
77
We also tested all four sets of rate parameters in Table 1 in simulating the conversion
from an isothermal packed-bed reactor operating at 250 °C with feeds similar to those
used by Weller and co-workers [7-10] (the studies were carried out using feeds not
containing H
2
or CO
2
), and there was satisfactory agreement between the values obtained.
However, as the concentration of the H
2
/CO
2
in the feed increased differences in the
calculated conversions were amplified, indicative of the common problem with such
empirical rate expressions in that they are often unable to predict data far removed from
the set of experimental conditions utilized to derive them. Our empirical rate expression,
however, adequately describes all the experimental packed-bed and MR data generated in
our study.
MR experiments were carried out using a CMS membrane (with dimensions of 5.7
mm OD, 3.5 mm ID, and 254 mm length). For these experiments 10 g of the commercial
Co/Mo/Al
2
O
3
catalyst were diluted with 80 g of ground quartz glass and packed in the
membrane shell-side. The experiments were carried out at different feed pressures, with
the permeate side pressure under atmospheric conditions, and with different permeate
steam sweep gas ratios (SR), defined as the ratio of inlet molar flow rate in the permeate
side to the inlet molar flow rate in the feed side. Packed-bed reactor experiments (as
previously described, in these experiments the permeate side is kept closed) have also
been carried out under identical conditions, using the same feed composition to compare
the results with those of the MR experiments.
Prior to the initiation of the MR experiments, the membrane was characterized
through single-gas permeation experiments (other than for water, for which the
permeance was measured as described above). The permeances of the various species are
shown in Table 3.2. Our previous studies with CMS membranes indicate that the
mixed-gas permeances of the various gases generally remain fairly close to the values
78
measured during the single-gas experiments [15]. For the H
2
S and NH
3
syngas
contaminants, the MR experiments indicate that they do not substantially permeate
through the membrane, and their permeance was set equal to zero for the simulations. (In
extensive studies in which both the surface of the membrane module and the plumping
were specifically coated to avoid potential wall adsorption, the H
2
S permeance was
always found to lie between the permeances of N
2
and CH
4
).
Table 3.2. Single-gas permeation data for the CMS membrane before and after the MR
experiments
tions, using the same feed composition to compare the results
with those of the MR experiments.
Prior to the initiation of the MR experiments, the membrane
was characterized through single-gas permeation experiments
(other than for water, for which the permeance was measured
as described above). The permeances of the various species are
shown in Table 2. Our previous studies with CMS membranes
indicate that the mixed-gas permeances of the various gases
generally remain fairly close to the values measured during the
single-gas experiments.
23
For the H
2
S and NH
3
syngas con-
taminants, the MR experiments indicate that they do not
substantially permeate through the membrane, and their per-
meance was set equal to zero for the simulations. (In extensive
studies in which both the surface of the membrane module and
the plumping were specifically coated to avoid potential wall
adsorption,theH
2
Spermeancewasalwaysfoundtoliebetween
the permeances of N
2
and CH
4
).
Figure 6 shows the effect of the feed-side pressure on the
CO conversion at differentW
C
/F
CO
for the MR and the packed-
bed experiments at 300 °C and an SR equal to 0.3. Figure 7
shows the effect of the steam SR on the CO conversion for the
MR experiments at different W
C
/F
CO
at 300 °C and a feed
pressureof3atm(theerrorbarsinbothfiguresreflectthecarbon
loss or gain due to experimental errors in measuring the flow
rates and compositions). The lines represent the simulated
conversions using our power-law rate expression (Table 1) and
the single gas permeances shown in Table 2. In general, the
Table 1. Kinetic Parameters for the Power-Law Rate Model
k
0
mol/(atm
(a+b+c+d)
·h·g) E (J/mol) ab c d
ref 36 219 38826 0.52 0.21 -0.1 -
ref 35
P ) 5 atm
6.3 22064 0.7 0.14 --
ref 3535
P ) 15 atm
6.0 24911 0.8 0.29 -0.07 -
this work 4.9 28145 0.61 ((0.0605) 0.37 ((0.0752) -0.38 ((0.0103) -0.54 ((0.0079)
Figure 5. Experimental vs. fitted CO conversion using the power law rate
expression at various packed-bed reactor experimental conditions.
Table 2. Single-Gas Permeation Data for the CMS Membrane
before and after the MR Experiments
pure gas
(before MR experiments)
pure gas
(after MR experiments)
gas
permeance
m
3
/(m
2
·h·bar)
separation
factor S. F.
permeance
m
3
/(m
2
·h·bar)
separation
factor S. F.
H
2
1.121 1.00 0.828 1.00
CO 0.024 46.71 0.026 31.85
CO
2
0.062 18.08 0.072 11.5
N
2
0.014 80.07 0.015 55.2
CH
4
0.009 124.56
H
2
O 0.374 3.00
Figure 6. Effect of the MR feed-side pressure on CO conversion at T )
300 °C and a steam sweep ratio) 0.3.
Figure7. Effect of the steam sweep ratio on CO conversion atT) 300 °C
and P) 3 atm.
Ind. Eng. Chem. Res., Vol. 49, No. 21, 2010 10991

Figure 3.5 shows the effect of the feed-side pressure on the CO conversion at
different W
C
/F
CO
for the MR and the packed-bed experiments at 300 °C and an SR equal
to 0.3. Figure 3.6 shows the effect of the steam SR on the CO conversion for the MR
experiments at different W
C
/F
CO
at 300 °C and a feed pressure of 3 atm (the error bars in
both figures reflect the carbon loss or gain due to experimental errors in measuring the
flow rates and compositions). The lines represent the simulated conversions using our
power-law rate expression (Table 3.1) and the single gas permeances shown in Table 3.2.
In general, the CO conversion increases with increasing W
C
/F
CO
, feed pressure, and
permeate side steam SR. This is consistent with the general expectation, given the higher
79
driving force for H
2
permeation and removal from the reaction zone, and the resultant
shift in the production of additional H
2
product.

Figure 3.5. Effect of the MR feed-side pressure on CO conversion at T = 300 °C and a
steam sweep ratio = 0.3.

Figure 3.6. Effect of the steam sweep ratio on CO conversion at T = 300 °C and P = 3
atm.
80
From these figures, it is clear that there is generally good agreement between the
model predictions and the experimental results. The MR always delivers higher CO
conversions than those of the packed-bed reactor, ranging from 5% to 12% depending on
the experimental conditions used. A key conclusion from these experiments, lasting for
more than two months, is that the catalyst and membrane exhibited fairly robust behavior.
We observed, for example, no notable changes in catalyst activity. The single gas
permeances of various species were also measured after all the experiments were
completed (see Table 3.2). The permeance of the less permeable species (CO and N
2
)
changed very little, but the permeance of H
2
decreased by 25%
,
(though it is still high for
this type of membrane). The reason for the moderate decline in hydrogen permeance is
under investigation, as are methods for in situ regeneration of these membranes.
The good agreement between the simulations and experiments makes it possible to
use the model to further study the effect of various parameters on system performance in
terms of CO conversion and H
2
recovery. Doing so allows one to identify the appropriate
conditions that maximize both conversion and product recovery. Figure 3.7 shows, for
example, the effect of feed pressure on MR performance. Increasing pressure helps by
increasing both CO conversion and H
2
recovery by increasing the partial pressure
difference of H
2
across the membrane. The pressure effect is more prominent at
lower W
C
/F
CO
, where it compensates for the decrease in contact time between the gas
species and the catalyst due to the high flow rate (in these simulations W
C
is kept
constant). As shown in the figure, increasing pressure increases H
2
recovery to close to
100%. Though these pressures are significantly higher than the operating range of our
bench-scale experimental system, they are very much in line with the range of pressures
under which biomass gasifiers usually operate, and the advantage of using the “one-box”
process to produce a contaminant-free H
2
is, therefore, obvious.
81
Figure 3.7. Effect of the MR feed-side pressure on (a) CO conversion and (b)
H
2
recovery at T = 300 °C, sweep ratio = 0.3, and W
C
= 10 g.

Figure 3.8 shows the effect of steam SR on MR performance at P = 20 atm. With our
laboratory system we were limited to feed pressures up to 6 atm, and thus we used steam
as a sweep stream. Under commercial conditions, however, the reactor will be operated at
high pressures (>20 atm), and as Figure 3.8 indicates, the use of sweep is not at all
82
necessary to obtain the desired conversion and recovery (it should be noted, however, that
low-pressure steam, to be used as sweep, is plentifully available in the chemical industry,
and it does not entail the expense of high-pressure steam to be used as a reactant in
high-pressure WGS reactors).
Figu
re 3.8 Effect of the sweep ratio on (a) CO conversion and (b) H
2
recovery at T = 300 °C,
pressure = 20 atm, and W
C
= 10 g.

83
3.5 Summary and Conclusions
A novel MR system termed as “one-box” process, in which reaction and separation
are combined in the same unit, was successfully utilized for producing hydrogen from a
feed with a simulated biomass-derived syngas composition containing common
impurities such as H
2
S and NH
3
. A CMS membrane was used for the in situ hydrogen
separation. The membrane was characterized in terms of its gas permeances which were
used for the model predictions. The CMS membrane stability was also investigated in the
presence of naphthalene and toluene as model tar and organic vapor compounds, and the
membrane proved to be stable at experimental conditions akin to the WGS reaction
environment. At lower temperatures (e.g., 150 °C), permeation loss occurred; however,
regeneration was readily achieved by purging of the membrane in the presence of inert
gas at 250 °C. This indicates that at low temperatures only surface coverage through
condensation occurs, with little or no irreversible pore plugging of the membrane.
The kinetics of the WGS reaction over a contaminant-tolerant Co/Mo/Al
2
O
3
catalyst
was also investigated as part of our study, and an empirical power-law rate expression
was obtained. The performance of the MR (the “one-box” process) using such
membranes and catalysts was investigated experimentally for a range of pressures and
sweep ratios; the MR showed higher conversions compared with those of the traditional
packed-bed reactor. Parallel modeling investigations indicated good agreement with the
experimental data. The performance of the system under different experimental
conditions was further investigated using the model.
The “one-box” process shows several advantages over the traditional packed-bed
system, including improvements in CO conversion and H
2
purity, while allowing one to
perform the reaction in the presence of hydrogen sulfide and ammonia and being able to
84
deliver a contaminant-free hydrogen product. Use of the process in hydrogen production
from biomass-derived syngas should, therefore, result in considerable energy savings.

85
3.6 References
[1] J. Corella, A. Orı′o, P. Aznar, Biomass Gasification with Air in Fluidized Bed:
Reforming of the Gas Composition with Commercial Steam Reforming Catalysts. Ind.
Eng. Chem. Res. 37 (1998) 4617-4624
[2] E. Knozinger, J. Weitkamp, Handbook of Heterogeneous Catalysis; Wiley: Hoboken,
NJ, (1997).
[3] R. Howard, Handbook of Commercial Catalysts: Heterogeneous Catalysis; CRC Press:
Boca Raton, FL, (2000).
[4] A. Cimino, B. A. De Angelis, The Application of X-ray Photo- electron Spectroscopy
to the Study of Molybdenum Oxides and Supported Molybdenum Oxide Catalysts. J.
Catal. 36 (1975) 11-22.
[5] P. Hou, D. Meeker, H. Wise, Kinetic Studies with a Sulfur-Tolerant Water Gas Shift
Catalyst. J. Catal. 80 (1983) 280-285.
[6] M. Nagai, K. Matsuda, Low-Temperature Water-Gas Shift Reaction over
Cobalt-Molybdenum Carbide Catalyst. J. Catal. 238 (2006) 489-496.
[7] D. M. Spillman, An Investigation of the High Pressure Kinetics of the Water-Gas
Shift Reaction over a Sulfided Molybdenum Oxide-Alumina Catalyst Promoted by
Cobalt Oxide and an Alkali Metal or Rare Earth Oxide. M.S. Thesis, SUNY-Buffalo,
Buffalo, NY, (1988).
[8] N. R. Srivatsa, Kinetic Studies of the Water-Gas Shift Reaction on a Sulfided
Cobalt-Molybdena-Alumina Catalyst. Ph. D. Dissertation, SUNY -Buffalo, Buffalo, NY,
(1987).
86
[9] A. N. Ramaswamy, Investigations on the Water-Gas Shift Reaction, Effect of
Hydrogen and of Catalyst Acidity. M.S. Thesis, SUNY-Buffalo, Buffalo, NY, (1990).
[10] N. R. Srivatsa, S. W. Weller, Water-Gas Shift Kinetics over Sulfided Catalyst:
Elevated Pressure. 9th International Congress on Catalysis 1988, Calgary, Canada.
[11] Y. Choi, H. G. Stenger, Water Gas Shift Reaction Kinetics and Reactor Modeling for
Fuel Cell Grade Hydrogen. J. Power Sources 124 (2003) 432-439.
[12] C. R. F. Lund, Microkinetics of Water-Gas Shift over Sulfided Mo/ Al2O3 Catalysts.
Ind. Eng. Chem. Res. 35 (1996) 2531-2538.
[13] C. R. F. Lund, Effect of Adding Co to MoS2/Al2O3 upon the Kinetics of the
Water-Gas Shift. Ind. Eng. Chem. Res. 35 (1996) 3067-3073.
[14] M. Abdollahi, J. Yu, P. K. T. Liu, R. Ciora, M. Sahimi, T. T. Tsotsis, Hydrogen
Production from Coal using a Membrane Reactor Based Process. J. Membr. Sci. 363
(2010) 160-169.
[15] A. Harale, H. T. Hwang , P. K. T. Liu, M. Sahimi, T. T. Tsotsis, Experimental
Studies of a Hybrid Adsorbent-Membrane Reactor (HAMR) System for Hydrogen
Production. Chem. Eng. Sci. 62 (2007) 4126-4137.

87
Chapter 4: The Use of CMSM-Based MR for the Production of Hydrogen from
Biomass-Derived Syngas. The Impact of Tars and Organic Vapors
＊

4.1 Introduction
In addition to H
2
S and NH
3
, the presence of organic vapors and of tar-like species is a
concern for the operation of WGS membrane reactors processing biomass-derived syngas.
Appropriate modifications in gasifier design and operating conditions along with using
catalysts and additives, as well as novel technologies such as supercritical water
gasification [1] help to minimize tar formation, and also to convert organic vapors in the
syngas into H
2
and CO via steam reforming. However, their presence in the syngas
cannot be completely eliminated, and they must, therefore, be taken into account during
the design of downstream systems for further syngas clean-up and processing. High
concentrations of organic vapors were previously shown, for example, to be highly
detrimental to the performance of glassy polymeric membranes used in the separation of
syngas-relevant mixtures. For example, White et al. [2] performed the separation of a
(CO
2
/CH
4
) binary mixture saturated with toluene using polyimide membranes. The
presence of the organic vapor reduced the selectivity by about 50%. Tanihara et al. [3]

also reported similar losses in selectivity (~85%) when separating an equimolar (H
2
/CH
4
)
mixture in the presence of (1600-7600) ppm of toluene vapor.
Accumulation of the tar-like species on the membrane surface and condensation of
organic vapors within the membrane pore structure, thus blocking transport and leading

＊
The work described in this Chapter was carried out collaboratively with my USC student
colleague Mingyang Tan, and is described in the following publication: Hydrogen Production
from Biomass-Derived Syngas Using a Catalytic Membrane Reactor Based Process, Jiang Yu,
Mingyang Tan, Paul K.T. Liu, Muhammad Sahimi, and Theodore T. Tsotsis, submitted to Ind.
Eng. Chem. Res. 2013

88
to severe reductions in performance, remained prior to recent studies by our group [1,4] a
key concern for the use of CMS membranes in biomass-derived syngas environments. An
earlier study by Vu et al. [5] had hinted, however, that CMS membranes may be
significantly more robust to the presence of organic vapors than their polymeric
counterparts. They studied the influence of the presence of toluene vapor (70 ppm) on the
selectivity of a CMS membrane treating a (10% CO
2
/90% CH
4
) mixture at a pressure of
3448 KPa and at 35 °C. In tests, lasting as long as 60 h, the membrane showed stable
performance, which Vu et al. [5] attributed to the inability of the toluene vapor to enter
the membrane nanopores.
During our group’s lab-scale testing [1] -- see also Chapter 3 -- the CMS
membranes were exposed to simulated biomass-derived syngas mixtures containing
realistic amounts of naphthalene, as a model tar-like species, and toluene as a model
organic vapor (these model compounds were selected based on prior literature work [6]).
The membranes performed well [1] with very little impact on throughput and selectivity
observed. However, temperature is a key consideration with stable performance observed
as long as the operating temperature stayed above 250
o
C, with losses in performance
observed for lower temperatures, becoming particularly severe below 200
o
C. These
results were also validated in field tests [4] by our group with a slip-stream of real
coal-derived and/or biomass-derived syngas at DOE’s NCCC facility with a pilot-scale
CMSM module containing 86 membrane tubes. Again, the membranes performed stably
as long as the testing temperature remained above 250
o
C, with progressively increasing
losses in membrane performance observed upon lowering the test temperatures.
In this Chapter, we report the results of a lab-scale investigation using a single
CMSM tube with properties similar to the membranes used to prepare the pilot-scale
module. In these studies the membrane and catalyst were challenged with a simulated
89
biomass-derived syngas that contained in addition to H
2
S and NH
3
realistic
concentrations of a model organic vapor (toluene) and a model tar-like species
(naphthalene). The results of these studies are detailed in this Chapter, and provide
impetus and further motivation for undertaking the field-testing of the technology with
real biomass-derived syngas on the way to the eventual technology commercialization.

4.2 Experimental Set-up and Procedure
The experiment set-up and procedure is similar to the one described earlier in Chapter
2. A schematic of the MR system used in this study is shown in Figure 4.1. In order to
make the simulated biomass feed composition, a few changes have been made. First,
different gas tanks are used. In order to add toluene and naphthalene to the feed
composition, one additional syringe pump is added in the front at the feed line, which is
parallel to the water syringe. Additional adsorbent beds for the NH
3
adsorbents and
organic vapors are added after the H
2
S adsorbent bed.
90

Figure 4.1. Experimental set-up used in the membrane reactor experiments

For the reactor experiments reported here we use a simulated syngas with a
composition typical of an air-blown biomass gasifier off-gas
(H
2
/CO/CO
2
/N
2
/CH
4
/NH
3
/H
2
S)=(0.67:1.00:1.00:2.67:0.2:0.00067:0.0006) [1,7]

along
with 0.8 vol% of naphthalene and 6.4 vol% of toluene added to the syngas feed. For the
reactor experiments we use a near stoichiometric H
2
O/CO ratio in the feed of 1.1.
The same procedure is followed in performing the packed-bed reactor (PBR)
experiments (to compare its performance with that of the MR), and for measuring the
catalytic reaction kinetics, with the only difference being that the inlet and exit valves for
the sweep gas are closed. The same system has also been used for single-gas permeation
91
studies in order to characterize the membrane properties. For such experiments, the
sweep gas (permeate side) inlet is closed, gas flows into the feed-side and the flow rates
of permeate and reject streams are measured.

4.3 Results and Discussion
As noted above, the experimental system in Figure 4.1 has also been used for
single-gas permeation studies in order to characterize the membrane properties. For such
experiments, the sweep gas (permeate side) inlet is closed, gas flows into the feed side
and the flow rates of permeate and reject streams are measured. For measuring the water
permeance, an Ar gas stream containing a predetermined concentration of water is fed
into the system. The permeate stream then passes directly through an adsorbent bed
where the water is captured. The amount of water that permeates is then calculated by
measuring the weight of the adsorbent before and after water permeation.
As noted previosuly, these membranes were previously tested in the laboratory in
the presence of a model organic vapor (toluene) and a model tar (naphthalene) under
non-reactive conditions [1] as well as in field-scale studies where they were exposed to
real coal-derived and/or biomass-derived syngas [4] and they were shown to perform well.
The main goal in this study, therefore, was to test their ability to perform satisfactorily as
well under reactive conditions in the presence of the aforementioned impurities. Prior to
the initiation of the MR experiments, the membrane was characterized through single-gas
permeation experiments. The permeances of the various species (calculated with respect
to the inside membrane area) for the fresh membrane are shown in Table 4.1. (Our
previous studies with CMS membranes indicate [1, 8] that the mixed-gas permeance of
the various gases generally remain fairly close to the values measured during the
92
single-gas experiments, so no such experiments were performed here). For the H
2
S and
NH
3
syngas contaminants, the MR experiments indicate that they do not permeate
through the membrane, and their permeance was, therefore, set equal to zero for the
simulations shown below (In extensive studies in which both the surface of the membrane
module and the plumping were specifically coated to avoid potential wall adsorption, the
H
2
S permeance was always found to lie in between the permeance of N
2
and CH
4
). A key
conclusion from these experiments, lasting for more than 8 weeks, is that the catalyst and
membrane exhibited fairly robust behavior. We observed, for example, no notable
changes in catalyst activity, as manifested by rate measurements spread over the 8-week
period – see discussion below. The single-gas permeances (other than that for water, for
which the permeance was measured as described above) of the various species were also
measured after all the reactor experiments were completed (see Table 4.1). The
permeance of the less permeable species (CO, N
2
, CH
4
) changed very little, and the
permeance of H
2
decreased by ~ 7%, which is very much in line with our field-scale
observations with these membranes [4].
Another goal of the study was to investigate the impact the presence of an organic
vapor and of a model tar-like species may have on catalyst performance. The reaction
kinetics of this particular catalyst (Süd-Chemie Co/Mo/Al
2
O
3
C-25 WGS catalyst) was
extensively investigated previously by our group, and they are detailed in our previous
papers [1, 8]. Specifically, an empirical rate law first proposed by Weller and coworkers
[9-12] and discussed previously in Chapters 2 and 3, was used to fit an extensive reactor
data set that included studying syngas compositions corresponding to biomass and coal
gasifiers, under a broad range of pressure, temperatures, and reactor residence times.

93
Table 4.1. Single-gas permeation data for the CMSM before and after the MR
experiments
Gas
Permeance (m
3
/m
2
*hr*bar)
Before After
H
2
2.21 2.04
CO 0.036 0.037
H
2
O Not measured 0.73
CO
2
0.073 0.079
CH
4
0.007 0.007
N
2
0.016 0.017

As reported in our previous papers

[1, 8], this empirical rate law performed well in
fitting both our packed-bed reactor as well as our membranes reactor data. However, the
simulated syngas used in the present investigation contains a substantial amount of two
model impurities (toluene and naphthalene) and their impact on the reaction kinetics of
the catalyst was not known prior to this study (such impurities, for example, could have
undergone catalytic cracking to produce coke that deactivates the catalyst or could have
interfered, in some other way, with its WGS activity). For further investigating this issue,
additional reactor data were generated and compared with the predictions of the empirical
rate law (all these experiments were carried out at 300
o
C and a CO/H
2
O=1.1, which are
the conditions utilized in the MR experiments reported here). The agreement between the
data and the predictions from the empirical rate model are shown in Figure 4.2, which
compares the experimentally measured conversions with the calculated conversions from
the model. The experimental data agree well with the model indicating that under the
94
conditions studied in this paper the organic vapor and tar-like model impurities have no
impact on catalytic activity. Furthermore, as noted previously, the catalyst activity
remained stable through the 8 weeks of experiments. Since the empirical reaction rate
model previously developed was shown to perform adequately, it was also utilized for all
the reactor simulations reported in this paper, as discussed below.

Figure 4.2. Experimental vs. calculated CO conversions using the power-law rate
expression [1, 8]

under various packed-bed reactor experimental conditions

One of the major potential advantages of the proposed “one-box” process is that it is
a multi-functional system that combines the WGS reaction, the separation of the
hydrogen product, and the removal of the various syngas impurities (H
2
S, NH
3
, organic
vapors and tars, etc.) into a single step. In particular, the ability of the CMSM to separate
in situ the various syngas impurities offers great potential benefit in that it eliminates the
need for using a warm-gas clean-up unit (WGCU) for removing these impurities from the
syngas prior to its being processed in the WGS reactor. In our previous studies we
95
reported on the ability of the CMSM to remove H
2
S and NH
3
[1, 4, 8, 13]. However, due
to experimental limitations, we were not able to check their separation characteristics
towards the organic vapors and the tar-like species. In this study, we have carried out a
series of experiments in order to specifically check for the ability of such membranes to
remove these types of impurities as well. Since these membranes are permeable to water
(and also steam is used as the sweep gas) and toluene and naphthalene are somewhat
soluble in water (see below), it is not straightforward to check the separation
characteristics under reactive conditions. (In principle, through the use of condensers on
the permeate side, one could condense both water and the organic vapor and tar-like
species. However, because the CMSM does not allow such impurities to go through – see
further discussion below -- no separate organic phase can be detected). Instead, an
experiment was carried out in which the CMS membrane was exposed to a flowing gas
mixture of 6.4 vol% toluene in hydrogen at 300
o
C and at 5 bar, and the gas phase exiting
the permeate side (virtually pure hydrogen) was sampled via GC/MS for the presence of
toluene. In these experiments we were unable to detect any toluene in the permeate
stream (analytical instrument detection limit <1 ppm). Thus, we conclude, based on these
experiments, that toluene does not permeate through this tight-pore CMS membrane.
Since naphthalene is a bigger molecule than toluene, hence one may also conclude
(barring a set of unknown circumstances) that naphthalene (and the tar-like species in real
syngas) will be unlikely to penetrate through the CMS membrane either.
During the membrane reactor experiments, we collected the liquids from the
condenser on the reject side and studied their volume and composition. The key reason
for doing that, is so that we are able to investigate whether toluene and naphthalene react
to a substantial extent in the WGS-MR, and also whether they leak-through to the
permeate side during the process. These experiments, typically, involved carrying out the
MR experiments for a certain period of time (while collecting all the condensable liquids
96
produced), and then switching the feed flow into pure He gas, depressurizing the reactor
and under the same temperature flushing the reactor system with He for an additional 8 hr.
Since the solution of naphthalene in toluene (a 1:8 molar ratio) has a lower density than
water, the liquid phase collected consists of two phases, with the organic phase residing
on the top. The organic phase is then carefully separated from the water phase and
weighed. The total amount of organic phase collected corresponds to more than 95% of
the amount of organic impurities (toluene + naphthalene) fed to the reactor during the
period for which the liquid phase was collected. The composition of the organic phase
was measured using a Bruker GC450 - MS300, making use of a 30-m DB-5
non-polarized column. (The analysis procedure starts at 50 °C and keeping the oven at
that temperature for 2 min. Then the oven temperature increased to 250 °C, at 50 °C /min.
After reaching 250 °C, the column was kept at that temperature for 4 min). The GC/MS
was calibrated using toluene/naphthalene mixtures in methanol with a molar ratio ranging
from 1:1 to 16:1. The ratio of (toluene/naphthalene) in the organic phase collected from
the reactor was (7.7:1). Given that toluene has a small but finite solubility in water (the
toluene’s solubility in water is 490 mg/L, and the naphthalene’s solubility is 30 mg/L)
and it is also volatile, it is not surprising that both the total amount of liquids but also the
amount of toluene collected is somewhat smaller than the corresponding feed amounts
(albeit less than 5%). Nevertheless, one can conclude from these results than neither one
of these compounds gets substantially converted in the membrane reactor during the
WGS reaction experiments (which is in line with the experimental findings above that
they do not impact the reaction kinetics), nor do they leak-though the membrane to the
permeate side (which is a finding consistent with the permeation studies with the toluene
in hydrogen mixtures noted above).
Figure 4.3 shows the CO conversion as a function of W
cat
/F
co
(the weight of catalyst,
which for these experiments is 10 g, over the molar feed flow rate of CO) during the
97
membrane reactor experiments for a feed pressure of 3 bar and two different sweep ratios
(SR=0.1 and 0.3). Shown on the same Figure are the simulations based on the MR model
using the independently measured reaction rate expression and the membranes
permeances (the average values measured for the fresh membrane and the membrane
after 8 weeks of experiments, see Table 4.1). For comparison purposes, packed-bed
experimental data and simulations are also shown on the same Figure. A key observation
from Figure 4.3 is that the model does an adequate job in describing both the MR as well
as the packed-bed reactor experiments. The membrane reactor’s conversion is higher than
the packed-bed reactor’s conversion, and both conversions increase with W
cat
/F
co
as
expected. In addition, increasing the sweep ratio also improves the reactor conversion.

Figure 4.3. CO conversion under 3 bar of reactor pressure and at 300
o
C for different
sweep ratios

98
Figure 4.4 shows the CO conversion as a function W
cat
/F
co
during the membrane
reactor experiments for a different feed pressure of 5 bar and two different sweep ratios
(SR=0.1 and 0.3). Shown on the same Figure, in addition, are the simulations based on
the MR model. Once more, as Figure 4.4 indicates, the model does an adequate job in
describing the MR. The membrane reactor’s conversion again increases with W
cat
/F
co
and
the sweep ratio. When comparing the results between Figures 4.3 and 4.4, it is clear that
running the reactor at higher pressures benefits performance, an important result in terms
of the eventual commercialization of the technology, since typical gasifiers are run at
much higher pressures (20 bar and above).

Figure 4.4. CO conversion under 5 bar of reactor pressure and at 300
o
C for different
sweep ratios

Figure 4.5 shows the hydrogen recovery, defined as the fraction of total hydrogen
that ends-up as part of the permeate stream [1, 13] as a function of W
cat
/F
co
and the
99
reactor pressure and sweep ratio. Shown on the same Figure are also the recoveries
calculated using the data-validated MR model. As can be seen in Figure 4.5, the model
does an adequate job, again, in describing the hydrogen recoveries for all conditions
studied. As expected higher sweep ratios and reactor pressure increase H
2
recovery, and
for a sweep ratio=0.3 and a pressure of 5 bar, a H
2
recovery higher than 70% is achieved.
(The recoveries shown in Figure 4.5 are rather low due to limitations with the size of our
laboratory system, which accommodates only one small size CMS membrane – under
optimized conditions, see discussion below, recoveries in excess of 90% can be attained
under realistic IGCC conditions).

Figure 4.5. Hydrogen recovery under various experimental conditions

Finally, Figure 4.6 shows the hydrogen purities (dry-basis) in the permeate side as a
function of W
cat
/F
co
and the reactor pressure and sweep ratio. Shown on the same Figure
100
are also the hydrogen purities calculated using the data-validated MR model, which does
a decent job in predicting the experimental values (as a result of the un-optimized
lab-scale reactor operation these purities are rather low -- under optimized conditions [4]
purities in excess of 90% with a CO content of a few hundred ppm can be attained under
realistic IGCC conditions). The results in Figures 4.5 and 4.6 manifest the challenge one
faces in optimizing such reactors, whereby conditions maximizing recovery may lead to
diminished hydrogen purity. Thus, to optimize the reactor operating conditions, one
needs a fine balance between W
cat
/F
co
, the sweep ratio and the reactor pressure in order to
reach high CO conversion with acceptable hydrogen recovery and purity.

Figure 4.6. Permeate-side H
2
purity (dry-basis) under various experimental conditions

Since the model performs reasonably well in describing the experimental results, it
can be used to further study the effect of various parameters on WGS-MR performance,
101
in terms of reactor conversion, hydrogen recovery, and purity. The target here is to
choose appropriate conditions which maximize both the CO conversion and H
2
recovery,
and minimize the CO content of the hydrogen product. Results of such process design
and scale-up simulations have been presented elsewhere [1, 4, 8, 13]. For example,
simulations, based upon membrane properties measured during field tests demonstrate [4]
that the “one-box” process, operating on a typical oxygen-blown gasifier off-gas, can
deliver more than 90% hydrogen recovery at more than 90% purity (dry-basis), and thus
shows good promise for commercial application.
4.4 Conclusions
A novel MR system termed as the “one-box” process, in which syngas clean-up, and
product separation are combined in the same unit was successfully utilized for producing
hydrogen from a feed with a simulated biomass-derived syngas containing common
impurities such as H
2
S and NH
3
, a model organic vapor (toluene) and a model tar-like
species (naphthalene). A single CMS membrane was used for the in-situ hydrogen
separation. The membrane was characterized in terms of its single-gas permeances, which
were used for the model predictions. The CMS membrane stability was also investigated
in the presence of these impurities, and the membrane proved to be stable under the
experimental WGS reaction conditions.
The kinetics of the WGS reaction over a commercial Co/Mo/Al
2
O
3
sour-shift
catalyst was also investigated in the presence of the organic vapor and of the model
tar-like species as part of our study, and no impact of these impurities was observed. The
performance of the MR (the “one-box” process) using such membranes and catalysts was
investigated experimentally for a range of pressures and sweep ratios; the MR showed
higher conversions compared with those of the traditional packed-bed reactor. Parallel
modeling investigations indicated good agreement with the experimental data.
102
The proposed “one-box” process shows several advantages over the traditional
packed-bed reactor system, including improvements in CO conversion and H
2
purity,
while allowing one to perform the reaction in the presence of common impurities such as
H
2
S, NH
3
, organic vapor and tar-like species, and being able to deliver a
contaminant-free hydrogen product. Use of the process in hydrogen production from
biomass-derived syngas should, therefore, result in considerable energy savings.

103
4.5 References
[1] M. Abdollahi, J. Yu, H. T. Hwang, P. K. Liu, R. Ciora, M. Sahimi, T. T. Tsotsis,
Process intensification in hydrogen production from biomass-derived syngas, Ind. Eng.
Chem. Res. 49 (2010) 10986−10993.
[2] L. S. White, T. A. Blinka, H. A. Kloczewski, I. F. Wang, Properties of a polyimide
gas separation membrane in natural gas streams, J. Membr. Sci. 103 (1995) 73-82.
[3] N. Tanihara, H. Shimazaki, Y. Hirayama, S. Nakanishi, T. Yoshinaga, Y. Kusuki, Gas
permeation properties of asymmetric carbon hollow fiber membranes prepared from
asymmetric polyimide hollow fiber, J. Membr. Sci. 160 (1999) 179−186.
[4] D. Parsley, R. J. Ciora, Jr. D. L. Flowers, J. Laukaitaus, A. Chen, P. K. T. Liu, J. Yu,
M. Sahimi, A. Bonsu, T. T. Tsotsis, Field evaluation of CMS membranes for the
separation and purification of hydrogen from coal and biomass derived syngas, Submitted
to J. Membr. Sci.
[5] De Q. Vu, W. J. Koros, S. J. Miller, Effect of condensable impurity in CO
2
/CH
4
gas
feeds on performance of mixed matrix membranes using carbon molecular sieves, J.
Membr. Sci. 221 (2003) 233−239.
[6] R. Coll, J. Salvadó, X. Farriol, D. Montané, Steam Reforming of Model Compounds
of Biomass Gasification Tars: Conversion at Different Operating Conditions and
Tendency Towards Coke Formation, Fuel Process. Technol. 2001, 26, 19.
[7] J. Corella, A. Orı´o, P. Aznar, Biomass Gasification with Air in Fluidized Bed:
Reforming of the Gas Composition with Commercial Steam Reforming Catalysts. Ind.
Eng. Chem. Res. 1998, 37, 4617
104
[8] M. Abdollahi, J. Yu, P. K. T. Liu, R. Ciora, M. Sahimi, T. T. Tsotsis , Hydrogen
production from coal-derived syngas using a catalytic membrane reactor based process, J.
Membr. Sci.363 (2010) 160–169.
[9] D. M. Spillman, An investigation of the high pressure kinetics of the water-gas shift
reaction over a sulfided molybdenum oxide-alumina catalyst promoted by cobalt oxide
and an alkali metal or rare earth oxide, M.S. Thesis, SUNY-Buffalo, Buffalo, NY, 1988.
[10] N. R. Srivatsa, Kinetic studies of the water-gas shift reaction on a sulfided
cobalt-molybdena-alumina catalyst, Ph.D. Dissertation, SUNY-Buffalo, Buffalo, NY,
1987.
[11] A. N. Ramaswamy, Investigations on the water-gas shift reaction, effect of hydrogen
and of catalyst acidity, M.S. Thesis, SUNY-Buffalo, Buffalo, NY, 1990.
[12] N. R. Srivatsa, S. W. Weller, Water-gas shift kinetics over sulfided catalyst: elevated
pressure, 9th International Congress on Catalysis1988, Calgary, Canada.
[13] P. K. T. Liu, M. Sahimi, T. T. Tsotsis, Process intensification in hydrogen
production via membrane-based reactive separations, Current Opinion in Chemical
Engineering 1 (2012) 342−351.

105
Chapter 5: Ultra-Pure Hydrogen Production from a Reformate Mixture Using a
Palladium Membrane Reactor
*

5.1 Introduction
In this study, we investigate the use of commercial-scale Pd membranes, prepared via
electroless plating (ELP) on 0.05 µm pore size asymmetric tubular Media and Process
Technology, Inc. (MPT) commercial ceramic substrates (Figure 5.1 shows a typical SEM
cross-section of the MPT Pd membranes tested in this study). Realistic feeds with a
simulated reformate composition are being used in our study, together with steam as a
sweep. The membrane reactor performance is investigated at different feed pressures and
flow rates, and steam sweep ratios. The experimental results are compared with the
simulation results from a mathematical model. The model is further used to study the
effect of experimental conditions on CO conversion, H
2
recovery and purity of the
hydrogen product.
Hydrogen transport through Pd and Pd-alloy membranes is commonly described by
Eq. 5.1 (for Pd membranes other gas species are thought to transport only through cracks
and imperfections, and their transport in this Chapter is described by Eq. 5.2). The
mechanism of hydrogen transport through Pd membranes, as also noted in Chapter1, is
thought to involve the dissociative adsorption of hydrogen on the Pd surface on the
feed-side, its diffusion through the metallic lattice, and recombination of the hydrogen
atoms and desorption of hydrogen from the membrane surface on the permeate-side. In

*
The work described in this Chapter was carried out collaboratively with my USC colleague Dr.
Mitra Abdollahi, and is described in the following publication: Ultra-pure hydrogen production
from reformate mixtures using a palladium membrane reactor system, Mitra Abdollahi, Jiang Yu,
Paul K. T. Liu, Richard Ciora, Muhammad Sahimi, Theodore T. Tsotsis, Journal of Membrane
Science 2012, 390-391, 32-42

106
practical situations, external gas-phase mass transfer limitations as well transport
limitations through the porous support, and diffusion and Poiseuille flow through
membrane pinholes and cracks may also substantially affect H
2
permeation through
composite Pd membranes [2]; the presence of various surface impurities that affect
hydrogen adsorption/desorption on the Pd surface may also impact hydrogen permeation.

Figure 5.1. SEM cross-section of the Pd layer deposited on the alumina support using the
ELP method.

107
Table 5.1. Pressure-dependence of hydrogen transfer through palladium and
palladium-alloy membranes.
Membrane Temperature
(°C)
Synthesis method Pd layer
thickness
(µm)
Pressure
exponent
(n)
Ref.
Pd 350 – 24–154 0.68 [3]
Pd 350–900 Punching circular
disks out of a
1-mm thick
palladium sheet
1000 0.5–0.62 [4]
Pd/Al
2
O
3
300–500 Metal-organic
chemical vapor
deposition
2 1 [5]
Pd/Al
2
O
3
450 Electroless
plating
5–8 1 [6]
Pd/Al
2
O
3
250–450 Electroless
plating
5–8 0.75 [7]
Pd–Ag 350 Electron beam
evaporation of
Ag on
commercial Pd
foil
25 1 [8]
Pd–Ag/Al
2
O
3

400 Electroless
plating
20 0.76 [9]
108
Membrane Temperature
(°C)
Synthesis method Pd layer
thickness
(µm)
Pressure
exponent
(n)
Ref.
Pd–Ag/Al
2
O
3

350–500 Electroless
plating
8.6 0.5 [10]
Pd–Ag/glas
s
200–400 Magnetron
sputtering of
Pd–Ag on glass
support
6.8 0.5 [11]
Pd–Cu/zirco
nia coated
alumina
350–450 Electroless
plating
1–11 0.52–1 [12]

For a clean, defect-free, thick Pd membrane at high temperatures (>300 °C),
diffusion through the bulk metal is likely to be the rate limiting step, and the pressure
exponent n = 0.5 (Eq. 5.1 is then known as Sievert's law). However, for composite
membranes prepared by depositing thin Pd films (<10 µm thick) on porous supports (as
the membranes in this study), external mass transfer through the support layer can
provide substantial added resistance to H
2
transport. The presence of impurities on the
membrane surface and at the grain boundaries (as well mass transport through
unavoidable defects) may also have substantial effects. In such instances, the n value
ranges from 0.5 to 1, and should be obtained experimentally. Table 5.1, summarizes
selected literature data on the pressure-dependence of hydrogen transport through Pd and
Pd-alloy membranes [3-12]. As seen in the Table, some studies report n values >0.5 even
for membranes with thicknesses larger than 10 µm [3, 4, 6, 7] and in one study for
109
membranes as thick as 1 mm [4]. On the other end of the spectrum, other studies report n
values equal to 0.5 for membranes, which are thinner than 10 µm [10,11]. Clearly, the
synthesis method which affects the physical and chemical structure of the final Pd
membrane film has also a strong effect on the value of the exponent n.
(5.1)
(5.2)
Also, CO is known to affect H
2
permeation through Pd and Pd-alloy membranes
through its competitive adsorption on the membrane surface, especially at lower
temperatures, which favor its adsorption [8]. For example, for a feed consisting of CO
and H
2
, almost no H
2
permeation was reported in Pd membranes at temperatures below
150 °C [8]. For temperatures of 300 °C and above, however, the literature data are not in
complete agreement. Some of the studies have shown that CO has almost no effect on the
hydrogen permeation [8], while other studies report a decline in the hydrogen flux due to
the presence of CO [13]. A direct comparison is unfortunately difficult to attain, as the
membrane synthesis method and experimental conditions used (e.g., temperature,
pressure, and feed composition) vary widely among the various studies. For a perfect,
defect-free Pd membrane, only atomic hydrogen permeates through the Pd metal lattice,
with all other gases found in the reformate mixture having negligible solubility and
diffusivity; perfect Pd membranes, therefore, should in principle have nearly infinite
selectivity towards H
2
. That they do not, is indicative of the fact that other gases permeate
through defects and metal lattice imperfections, that are often unavoidable during the
preparation of ultra-thin layer Pd membranes. The hydrogen permeance and selectivity
are then two important indicators of the quality of the resulting Pd membranes. Yun and
Oyama [14] recently reviewed the performance of Pd and Pd-alloy membranes
110
(prepared via different fabrication methods such as ELP, chemical vapor deposition
(CVD), and conventional electroplating (EPD)) in hydrogen separation. They
concluded [14] that Pd membranes prepared by the ELP technique are the most promising
for practical applications, since ELP allows for easy preparation using simple equipment
and various types of supports. Table 5.2 summarizes the recently published gas
separation properties of Pd and Pd-alloy composite membranes prepared by the ELP
method, as are the supported membranes used in this study.
Table 5.2. Hydrogen permeance and selectivity of alumina supported palladium or
palladium-alloy membranes prepared using the ELP method.
Membrane Thickness
(µm)
T(°C) H
2
permeance
(mol/m
2
s bar)
Separation
factor
Ref.
Pd/γ-Al
2
O
3
6 480 0.260 2100
(H
2
/Ar)
[15]
Pd/γ-Al
2
O
3
packed
with
yitria-stabilized
zirconia
5 450 0.125 – [16]
Pd/γ-Al
2
O
3
5 507 0.126 600
(H
2
/N
2
)
[16]
Pd
88
Ag
12
/γ-Al
2
O
3
11 550 0.121 2000
(H
2
/N
2
)
[17]
Pd/γ-Al
2
O
3
2.6 370 0.048 3000
(H
2
/N
2
)
[18]
Pd/silicalite-1 5 500 0.180 1300 [19]
111
Membrane Thickness
(µm)
T(°C) H
2
permeance
(mol/m
2
s bar)
Separation
factor
Ref.
zeolite (H
2
/N
2
)

5.2 Experimental Set-up and Procedure
The experiment set-up and procedure is the same as described earlier in Chapter 2. A
schematic of the MR system used in this study is shown in Figure 5.2. Instead of the
CMSM used in previous chapters, a tubular Pd membrane (L = 762 mm, ID = 3.5 mm,
OD = 5.7 mm), is utilized which is prepared by depositing a thin Pd layer via the ELP
technique on a 0.05 µm pore size asymmetric, tubular mesoporous commercial MPT
alumina support. Also a different commercial, low-temperature Cu–Zn/Al
2
O
3
(44% CuO,
44% ZnO, balance Al
2
O
3
) catalyst is utilized instead of the Co-Mo/Al
2
O
3
sour-shift
catalyst. This type of catalyst has been extensively utilized for the-low temperature WGS
reaction.
112

Figure 5.2. The experimental set-up and the membrane module.

5.3 Reaction Kinetics
Since a different commercial catalyst has been used in this study, the rate expression
shall be different. Extensive kinetic investigations for an expanded range of pressures,
temperatures, and feed compositions with a Cu–Zn supported catalyst of a very similar
113
composition were previously carried out by our group [1]. A reaction rate expression
consistent with a Hougen–Watson type surface mechanism was developed (Eqs. 5.3 and
5.4, below), where the various rate parameters are summarized in Table 5.3.
(5.3)
(5.4)

Table 5.3. Hougen–Watson rate parameters [1].
E 30.387 (kJ/mol)
k 48.16 (mol/g s bar
2
)
ΔH (kJ/mol) k
0
(bar
−1
)
K H
2
−12 0.0178
K H
2
−2.42 0.005
KCO
2
−28 0.0410
KCO −0.86 0.0303

This rate expression also satisfactorily describes the packed-bed experimental data
derived with the catalyst used in this study, as Figure 5.3 indicates.
114

Figure 5.3. Measured vs. fitted CO conversion data using the Hougen–Watson rate
expression [1].

5.3 Membrane Characterization
5.3.1 Membrane Permeation Studies
Our group has extensively investigated the hydrogen transport through the ultra-long,
ultra-thin Pd membranes, prepared by the electroless plating technique that are used in
this study [6] and [7]. Our results indicate that, depending on the preparation conditions,
H
2
transport through our Pd membranes obeys Eq. 5.1, with the exponent n ranging from
0.75 to 1. Figure 5.4 shows single-gas H
2
permeation data at T = 300 °C with one of the
supported Pd membranes, prepared under identical conditions with the membrane used in
115
the membrane reactor experiments in this study. The membrane permeation data are
described well by Eq. 5.1, with the pressure exponent n = 0.96. Transport of gases, other
than hydrogen, through our Pd membranes were found experimentally to obey Eq. 5.2
(i.e., the permeability is independent of pressure). During membrane synthesis, large
defects and cracks sometimes develop that may result in convective flows; however, we
have found their impact on performance to be so severe that such membranes are not
selected for further study. The experimentally measured single-gas permeances of the
membrane utilized in the MR experiments are shown in Table 5.4.

Figure 5.4. Flux of hydrogen as a function of at T = 300 °C.

116
Table 5.4. Single-gas permeances measured at 300 °C and calculated separation factors
for the Pd membrane before and after the MR experiments.
Gas Before MR exp.

After MR exp.

Permeance
(mol/m
2
s bar)
Separation
factor
Permeance
(mol/m
2
s bar)
Separation
factor
H
2
0.222
a
1 0.208
a
1
CO 8.805E−05 2526 8.681E−05 2398
CO
2
7.068E−05 3147 7.192E−05 2894
Ar 8.433E−05 2638 7.937E−05 2623
N
2
8.185E−05 2718
CH
4
– – 8.557E−05 2433
a: The unit for H
2
permeance is (mol/m
2
s bar
0.96
).

5.3.2 CO Effect on the Pd Membrane
We have, ourselves, studied experimentally the effect of CO on the H
2
permeation
rate through our Pd membranes at the conditions utilized in this study. In a series of
experiments (with the membrane used in the MR experiments), the hydrogen flux through
the membrane was first measured (at 2.02 bar and 300 °C) in a flowing pure hydrogen
stream (6 mol/h), prior to exposing the membrane to CO. Then, the membrane was
exposed to a flowing pure CO stream (at P = 1.01 bar and T = 300 °C) for different
periods of time, ranging from 30 min to 5 h. Following exposure to CO, the feed was then
117
switched back to pure hydrogen (same flow rate of 6 mol/h, at 2.02 bar), with the
membrane temperature staying at 300 °C; the hydrogen permeation rate was then
immediately measured (within 5 min from the gas atmosphere switch). Figure 5.5 shows
the experimental results in terms of the normalized H
2
flux (defined as the ratio of the
hydrogen flux immediately measured after membrane exposure to CO, to the starting
membrane H
2
flux). As shown in Figure 5.5, there are no systematic trends in the data
pointing to inhibition due to exposure to CO. The maximum difference between the
hydrogen flux of the fresh membrane and the flux measured after the membrane had been
exposed to CO is 7.1%. Note, however, that for longer periods of exposure to CO (3 and
5 h) there are virtually no changes in the hydrogen flux upon exposure to CO. One can
conclude from these observations that CO adsorbs rather weakly at 300 °C on the
particular Pd membrane used in this study, consistent with prior literature data.

Figure 5.5. Effect of CO on the membrane H2 permeation rate.
118
We also include here additional experimental data (at 300 °C and 3.08 bar) with a Pd
membrane prepared under identical conditions with that in Figure 5.5 (this membrane is
shorter, 25.4 cm long, and its hydrogen permeance is 20% less than the membrane in
Figure 5.5, still however, within an acceptable quality tolerance for the synthesis of such
membranes). In these experiments, the membrane is exposed to different mixed-gas,
reformate-type feed compositions under sufficiently high feed flow conditions for which
the partial hydrogen pressure difference between the feed and permeate sides at the exit
of the membrane module is substantially non-zero (note that this is important for these
high-flux membranes to avoid data falsification due to the fact that otherwise a large
fraction of the membrane area may not be active in transport). Three different feed
compositions were tested. The first feed composition (hereinafter referred to as Feed 1)
corresponds to the reformate gas mixture used in the MR experiments (see further
discussion to follow), namely H
2
:CO:CO
2
:CH
4
:H
2
O = 5.22:1:0.48:0.1:2.8. The second
composition (Feed 2) is the same composition but on a “dry basis”, that is without any
water being present, namely H
2
:CO:CO
2
:CH
4
= 5.22:1:0.48:0.1. The third feed
composition (Feed 3) is H
2
:CO:CO
2
:CH
4
:H
2
O = 6.12:0.1:1.38:0.1:1.9 corresponding to
90% CO conversion of the original composition (Feed 1). For each series of experiments
the membrane was kept in the corresponding feed composition for at least 4 h before
measuring the flow rates and compositions of the reject and permeate sides. The
mixed-gas hydrogen permeance was then calculated by fitting the experimental data
using Eqs. 2.7 and 2.8 from the mathematical model described in Chapter 2 (without the
reaction term), and compared with the single-gas permeance for hydrogen measured
before and after the experiments. Figure 5.6 shows the experimental results in terms of
the normalized H
2
permeance (defined as the ratio of the mixed-gas hydrogen permeance
divided by the starting single-gas H
2
permeance). Again, no systematic trends are
observed pointing out any detrimental effects (at these relatively high temperatures and
119
low total pressures) that relate to the presence of CO (or any other reformate mixture
component) on the hydrogen permeance. The maximum difference between the
single-gas and mixed-gas hydrogen permeance is less than 8%, and to repeat we observe
no systematic trends.

Figure 5.6. Normalized H
2
permeance for various feeds; Bar 1 for Feed 1, Bar 2 for Feed
2, Bar 3 for Feed 3, and Bar 4 for single-gas H
2
permeance measured after the mixed-gas
experiments were completed.

Table 5.4 shows the measured single-gas permeances and calculated separation
factors at a temperature of 300 °C for the Pd membrane utilized in the MR experiments,
measured first before the experiments were started and then again after the experiments
were completed, during which time the membrane had stayed for more than one month (5
weeks) on stream. As Table 5.4 indicates, the membrane under study possesses a very
high hydrogen permeance (see Table 5.3 for comparison) and selectivity towards the
120
other gases found in the reformate mixture, which is the feed for the WGS reactor.
Comparing the hydrogen permeation rates measured before and after the MR experiments,
there is a small difference between the hydrogen permeance values measured ( ∼6%
decrease), indicative perhaps of small changes occurring on the surface or in the bulk of
the metal, most likely due to accumulation of carbon (and/or some yet to be identified
impurity) on the membrane surface. The permeation rates of the other gases remained
virtually unchanged, indicating that no additional cracks or pinholes were formed as a
result of the exposure of the Pd membranes to the WGS environment. These results,
therefore, look very promising in terms of long term membrane usage for WGS reaction.

5.4 Reactor Studies
Membrane reactor experiments were carried out using the experimental system
shown in Figure 5.2, utilizing a “commercial size” supported Pd membrane (L = 762 mm,
ID = 3.5 mm, OD = 5.7 mm), whose transport characteristics are shown in Table 5.4. As
noted above, the majority of the past studies utilizing Pd membrane reactors for the WGS
reaction have used feed compositions very different from what are used industrially. In
this study, the feed gas consists of all the species potentially present in the stream exiting
the SMR reactor upstream of the WGS reactors. For the membrane reactor experimental
data presented here, in particular, the feed had the following composition (in molar ratios)
H
2
:CO:CO
2
:CH
4
:H
2
O = 5.22:1:0.48:0.1:2.8, which is chosen to match the calculated
equilibrium composition from an SMR reactor operating at a H
2
O/CH
4
= 3, T = 850 °C
and P = 10 bar. For the membrane reactor experiments, 30 g of the commercial
Cu–Zn/Al
2
O
3
catalyst were diluted with ground quartz glass (enough to fill the reactor
volume) and packed in the membrane shell-side (the annular volume in between the
membrane and the reactor shell). The reactor was maintained under isothermal conditions,
121
and the permeate side was maintained under atmospheric pressure conditions. Steam was
utilized as the sweep gas stream.
In the experiments, the effect of varying the feed-side (reactor) pressure and
permeate-side sweep ratio (SR = ratio of inlet steam molar flow rate in the permeate side
to the inlet molar feed flow rate in the feed-side) on reactor conversion and recovery are
studied. Figure 5.7a and b shows the CO conversion and H
2
recovery as a function of
W
cat
/F
CO
(weight of undiluted catalyst (g) over the feed molar flow rate of CO (mol/h)), at
T = 300 °C and SR = 0.3 for two different operating pressures namely 3.08 and 4.46 bar
(the error bars reflect carbon and hydrogen loss or gain due to experimental errors in
measuring the flow rates and compositions, and/or possible carbon deposition). Shown in
Figure 5.7a is also the equilibrium conversion based on the feed composition. Clearly the
removal of hydrogen from the reaction mixture has a substantially beneficial effect on
reactor conversion. Increasing the and the reactor pressure improves conversion and
hydrogen recovery, as expected, with almost complete conversion and hydrogen recovery
in excess of 90% being attained. Figures 5.8a and b shows the CO conversion and H
2

recovery as a function of W
cat
/F
CO
for a temperature of 300 °C, a reactor pressure of 4.46
bar and three different sweep ratios, namely SR = 0 (no sweep), SR = 0.1, SR = 0.3
(shown in Figure 5.8a is also the calculated equilibrium conversion). As shown in these
two figures, increasing the sweep ratio increases both conversion (up to almost complete
conversion) and recovery (up to almost 90%), but the effect is stronger for the recovery.
Increasing the sweep ratio (as is the case with increasing the feed pressure—see Figure
5.7a and b) increases the hydrogen partial pressure difference across the membrane, thus
providing a higher driving force for H
2
to diffuse through the membrane. The use of
sweep is of particular value for the smaller, and industrially more relevant W
cat
/F
CO
.
122

Figure 5.7. Effect of pressure on (a) CO conversion and (b) H
2
recovery, T = 300 °C and
SR = 0.3.
123

Figure 5.8. Effect of sweep ratio (SR) on (a) CO conversion and (b)
H
2
recovery, T = 300 °C and P = 4.46 bar.

124
Another important indicator of good MR performance is the purity of the H
2
product,
since as noted above fuel cells are very vulnerable to the presence of CO even in small
quantities (ppm level). Table 5.5 shows the CO content of the H
2
product from the
membrane reactor (the stream exiting the permeate side) for a number of different
experiments. The CO concentration in the hydrogen product for all experiments reported
in Table 5.5 remains below 75 ppm, which may be potentially acceptable for few types of
PEM fuel cells, as noted previously [20]. For the majority of the PEM fuel cell
applications, however, for which CO concentrations below 50 ppm are required, the CO
content can be decreased or completely eliminated efficiently with a conventional method
such as methanation efficiently, since the low (<75 ppm) CO content in the permeate side
stream implies that very little H
2
is going to get consumed during the CO methanation.
Furthermore, the presence of trace amounts of CO
2
as well implies that conventional
(rather than highly selective) methanation catalysts need to be utilized.

Table 5.5. CO content of the MR H
2
product (permeate side) at two different sets of the
experimental conditions.
T = 300 °C, P = 4.46 bar no sweep
T = 300 °C, P = 4.46
(g-cat h/mol-CO)

(g-cat h/mol-CO)
CO concentration
(ppm)
Sweep ratio (SR) CO concentration
(ppm)
466 73 0.3 59
266 55 0.1 68
186 41 None 73
125
5.5 Scale-Up
Due to limitations in terms of the membrane area and the range of pressures that
could be employed in our experimental system it was not possible to completely explore,
via experimentation, the full range of capabilities of Pd membrane reactors for the WGS
reaction. Instead, in this study we have made use of the reaction rate expression in Eq. 5.3,
and the measured single-gas permeances (this based on the experimental observations
that mixed-gas permeances do not differ substantially form their single-gas
counterparts—see Figure 5.5 and Figure 5.6 and discussion therein) in order to
investigate, through simulations utilizing the mathematical model described in Chapter 2
(other than the fact that hydrogen permeance is described by Eq. 1), the behavior of the
WGS-MR system for conditions more akin to the industrial ones (e.g., in terms of
pressure and feed flow rates). The model provides satisfactory agreement with the
experimental data, with satisfactory agreement (maximum difference of 6% between the
two), see Table 5.6, for example, which presents measured and calculated CO
conversions and H
2
recoveries for two sets of experimental conditions.

Table 5.6. Measured and calculated CO conversion and H
2
recovery for two sets of the
experimental conditions.

Conv. (exp.) Conv. (sim.) Rec. (exp.) Rec. (sim.)
T = 300 °C, P = 4.46 bar, SR = 0.3
266 99.7 98.6 88.9 90.3
T = 300 °C, P = 4.46 bar, SR = 0.1
266 99.6 98.2 87.1 86.9
126
Figure 5.9 shows the effect of feed pressure on membrane reactor conversion.
Increasing reactor pressure has a significant effect on conversion, especially at lower
W
cat
/F
CO
, where it compensates for the decrease in contact time between the gas
species and the catalyst due to the high flow rates (in these simulations W
c
is kept
constant). These results are very promising, considering the fact that high pressures
( ∼10 bar and higher) and low W
cat
/F
CO
(<10) are the conditions which are currently
applied industrially. Increasing pressure has also the additional beneficial effect in
that it increases H
2
recovery (Figure 5.10) by increasing the partial pressure
difference of H
2
across the membrane.

Figure 5.9. Effect of pressure on CO conversion, T = 300 °C and SR = 0.1.
127

Figure 5.10. Effect of pressure on H2 recovery, T = 300 °C and SR = 0.1.
Another very important factor determining good MR performance is the H
2
purity, as
previously noted. As Figure 5.11 shows, increasing pressure decreases the H
2
purity due
to the fact that it provides a higher driving force for other species, in addition to H
2
, to
diffuse through the flaws of the membrane. The effect becomes stronger at higher
W
cat
/F
CO
where the flow rate is lower and the contact time between the gas and the
membrane is higher allowing for a relatively larger fraction of CO to transport through
the membrane to the permeate side without undergoing reaction. The purity of the
hydrogen product for realistic pressures and W
cat
/F
CO
is very high (above 99.9%),
indicating that the use of Pd-MR for industrially relevant conditions is, indeed,
promising.
128

Figure 5.11. Effect of pressure on H2 purity, T = 300 °C and SR = 0.1.

5.6 Summary
In this chapter, an ultra-thin, long, high-performance (in terms of its H
2
permeance
and selectivity) supported palladium membrane is used in a membrane reactor system to
produce pure hydrogen through the use of the water-gas shift reaction from a gas stream
with a simulated reformate composition. The membrane is characterized using
single-gas permeation measurements. A Cu–Zn/Al
2
O
3
catalyst is utilized for the WGS
reaction. The system performance is investigated under various experimental conditions,
namely, different pressures, feed flow rates and sweep ratios. The best performance is
obtained at T = 300 °C, P = 4.46 bar and the permeate sweep gas ratio = 0.3 with almost
complete CO conversion and 90% hydrogen recovery. The product hydrogen purity is
always at more than 99.9% with CO concentration of less than 100 ppm. A model is
used for further study of the design aspects of the system. It is shown that the Pd
129
membrane reactor system under study is capable of delivering almost complete CO
conversion and H
2
recovery at experimental conditions akin to the industrial
applications. The membrane exhibits good stability with only a 6% change in the
H
2
permeance, and almost no change in the permeation rates of the other gases after
being used in the reactor for more than a month under the WGS environment. Hence, it
is concluded that the Pd-based WGS-MR is, potentially, a promising system for
hydrogen production for fuel cell applications.

130
5.7 References
[1] A. Harale, H.T. Hwang, P.K.T. Liu, M. Sahimi, T.T. Tsotsis, Design aspects of the
cyclic hybrid adsorbent-membrane reactor (HAMR) system for hydrogen production,
Chem. Eng. Sci. 65 (2010) 427–435.
[2] T.L. Ward, T. Dao, Model of hydrogen permeation behavior in palladium mem-
branes, J. Membr. Sci. 153 (1999) 211–231.
[3] R.C. Hurlbert, J.O. Konecny, Diffusion of hydrogen through palladium, J. Chem.
Phys. 34 (1961) 655–658.
[4] B.D. Morreale, M.V. Ciocco, R.M. Enick, B.I. Morsi, B.H. Howard, A.V. Cugini, K.S.
Rothenberger, The permeability of hydrogen in bulk palladium at elevated temperatures
and pressures, J. Membr. Sci. 212 (2003) 87–97.
[5] S.Yan, H.Maeda, K.Kusakabe, S.Morooka, Thin palladium membrane formed in
support pores by metal-organic chemical-vapor-deposition method and application to
hydrogen separation, Ind. Eng. Chem. Res. 33 (1994) 616–622.
[6] A. Harale, A hybrid adsorbent membrane reactor (HAMR) system for hydrogen
production, Ph.D. dissertation, University of Southern California, Los Angeles, CA,
2008.
[7] H.T. Hwang, A study of the application of membrane-based reactive separations to
the carbon dioxide methanation, Ph.D. dissertation, University of Southern California,
Los Angeles, CA, 2009.
[8] H. Amandusson, L.G.Ekedahl, H.Dannetun, Hydrogen permeation through surface
modified Pd and Pd–Ag membranes, J. Membr. Sci. 193 (2001) 35–47.
131
[9] S. Uemiya, N. Sato, H. Ando, E. Kikuchi, The water gas shift reaction assisted by a
palladium membrane reactor, Ind. Eng. Chem. Res. 30 (1991) 585–589.
[10] Y. Guo, G. Lu, Y. Wang, R. Wang, Preparation and characterization of
Pd–Ag/ceramic composite membrane and application to enhancement of catalytic
dehydrogenation of isobutene, Sep. Purif. Technol. 32 (2003) 271– 279.
[11] E. Gobina, R. Hughes, Ethane dehydrogenation using a high-temperature catalytic
membrane reactor, J. Membr. Sci. 90 (1994) 11–19.
[12] F. Roa, J.D. Way, Influence of alloy composition and membrane fabrication on the
pressure dependence of the hydrogen flux of palladium-copper membranes, Ind. Eng.
Chem. Res. 42 (2003) 5827–5835.
[13] S.H. Israni, M.P. Harold, Methanol steam reforming in Pd–Ag membrane reactors:
effects of reaction system species on transmembrane hydrogen flux, Ind. Eng. Chem. Res.
49 (2010) 10242–10250.
[14] S. Yun, S.T. Oyama, Correlations in palladium membranes for hydrogen separation:
a review, J. Membr. Sci. (2011), doi:10.1016/j.memsci.2011.03.057.
[15] X. Zhang, G. Xiong, W. Yang, A modified electroless plating technique for thin
dense palladium composite membranes with enhanced stability, J. Membr. Sci. 314 (2008)
226–237.
[16] D. Tanaka, M. Tanco, T. Nagase, J. Okazaki, Y. Yakui, F. Mizukami, T. Suzuki,
Preparation of pore-fill type Pd-YSZ-gamma-Al2 O3 composite membrane supported on
alpha-Al2 O3 tube for hydrogen separation, J. Membr. Sci. 320 (2008) 436–441.
132
[17] B. Nair, J. Choi, M.P. Harold, Electroless plating and permeation features of Pd and
Pd/Ag hollow fiber composite membranes, J. Membr. Sci. 288 (2007) 67–84.
[18] B. Nair, M.P. Harold, Pd encapsulated and nanopore hollow fiber membranes:
synthesis and permeation studies, J. Membr. Sci. 290 (2007) 182–195.
[19] Y. Guo, X. Zhan, H. Deng, X. Wang, Y. Wang, J. Qiu, J. Wang, K.L. Yeung, A
novel approach for the preparation of highly stable Pd membrane on macroporous α-Al
2

O
3
tube, J. Membr. Sci. 362 (2010) 241–248.
[20] E. Yoo, T. Okada, T. Kizuka, J. Nakamura, Effect of various carbon substrate
materials on the CO tolerance of anode catalysts in polymer electrolyte fuel cells,
Electrochemistry 75 (2007) 146–148.

133
Chapter 6: Preliminary Studies of the Transport Characteristics of CMS
Membranes at High Pressures and Temperatures
6.1 General Introduction
Though the WGS-MR experiments in the laboratory, described in Chapters 2, 3, 4 of
this Thesis, were carried out under relatively low pressure conditions of ~ 5-6 atm, the
WGS reaction in the IGCC application will likely occur at much higher pressures, >20
atm, as noted above. Relatively, little is known about the transport characteristics of
complex mixtures of gases at these conditions, like the quaternary WGS gas mixture (CO,
H
2
, CO
2
and H
2
O) or the steam reforming mixture that contains in addition CH
4
. One key
focus of our future efforts is of our Group at USC is to study the transport of gas mixtures
relevant to these two reactions under realistic pressure and temperature conditions. Such
studies will also, hopefully, lead to better understanding of the transport of other
important industrial gas mixtures, particularly those that contain both condensable (e.g.,
CO
2
and H
2
O for the WGS mixture) and non-condensable gases (a current parallel effort
is undergoing in our Group in studying the separation of chemical warfare agents from
contaminated air streams, see Lee et al. [1] for a recent paper by our Group).
Under high-pressure conditions, CO
2
and H
2
O behave as non-ideal gases, and in the
modeling of their transport this must be taken under consideration. For microporous
systems, like the CMS membranes under study here, the effect of capillary condensation
which is prevalent under such conditions should also be considered. Capillary
condensation, as a phenomenon that is active in determining the transport characteristics
of microporous membranes, has attracted scientific interest for quite a long time. Barrer
and co-workers [2-4] were among the first groups to study the effect of capillary
condensation during their experimental study of various gas mixtures in microporous
carbon membranes. They described a “blocking” effect, whereby the condensed phase in
the membrane eliminated or “blocked” the transport of the non-condensed species. Tin et
al. [5], on the other hand, who studied the separation of binary CO
2
/CH
4
mixture at a
pressure of 20 atm with a CMS membrane (derived from the pyrolysis of a polyimide
precursor), reported that the permeance of CO
2
and CH
4
in the gas mixture was not
134
significantly different from the single-gas permeances under the same conditions. Hong
et al. [6], who studied the H
2
/CO
2
binary mixture reported that the H
2
permeance was
inhibited by strong CO
2
adsorption, especially under low temperatures, and that the CO
2

feed concentration in the mixture affected the CO
2
/H
2
selectivity. For example, at 253 K
and 11 atm of total pressure, and 43% CO
2
in the feed, the separation factor reached its
maximum value of 140. An earlier publication of our group [7] at 293 K and 30 psig,
reports that the membrane permeance of a quaternary mixture (CO
2
/CO/H
2
/CH
4
)
decreased slowly during continuous testing with this mixture. This decline in
performance was found, however, to be reversible. The initial membrane permeance was
recovered by heating the membrane in an inert atmosphere of Ar. This behavior is
consistent with a slowly developing CO
2
capillary condensation in the membrane
structure. Adams et al. [8] observed a separation factor for CO
2
/CH
4
of 34 at 40 psia,
while when the pressure increased to 440 psia, the separation factor dropped to ~25. For a
binary 90 mol% CO
2
and 10 mol% CH
4
mixture, Kosuri and Koros observed [9] that the
selectivity of CO
2
/CH
4
decreased from 50.4 to 39.6 as the pressure increased from 240
psi to 1220 psi.
It is clear from the above discussion that relatively little is known about the transport
of gas mixtures under high temperature and pressure conditions through microporous
membranes, in general, and supported CMSM, in particular. In what follows we first
describe some of our own preliminary work in this area, and then describe some proposed
future work in this area.

6.2 Experimental System
The experimental system utilized is shown in Figure 6.1. Mass flow controllers are
used to adjust the feed gas flow rates and mixture concentration. The membrane module
is wrapped-up with a heating tape, which is used to maintain the membrane to the desired
experimental temperature with the aid of a thermocouple installed very close to the
membrane surface. A back-pressure regulator (BPR) installed on the reject side (the
135
high-pressure gas flows in contact with the outside membrane surface) is used to maintain
the feed-side pressure; gauges installed at the feed, reject, and permeate lines, monitor
their pressures. The flow rate of the gas streams are measured with bubble flow-meters,
while their composition is measured with a Gas Chromatograph equipped with a capillary
column and a TCD detector. Only a small slip-stream from the exit gases is directed to
the GC, in order to avoid potential pressure drops and interference with the membrane
permeation measurements, and the rest is directed to the hood.

1: on/off valve; 2: pressure gauge (0-2500 psig); 3: back pressure regulator; 4: pressure
gauge (0-15 psig)
Figure 6.1. Experimental set-up for the high-pressure, mixed-gas separation experiments

6.3 Experimental Results
We report here some of our preliminary results of gas-mixture transport using a 10 in
long CMSM provided by industrial collaborator MPT. To characterize the properties of
the membrane, first single gas (H
2
or CH
4
) permeances were measured for different
pressures at 200
o
C. Note, that the CMS membrane is hydrogen-selective with a
single-gas separation fact of 164 at the pressure of ~ 102 psig. The single-gas hydrogen
136
permeance varies little with pressure (less than the experimental error, see Table 6.1 and
Figure 6.3), however, the methane permeance increases linearly with average pressure
(see Table 6.1 and Figure 6.2), which is indicative that these two gases transport through
different parts of the membrane micropore structure. Hydrogen diffuses by a molecular
sieving mechanism, likely, through membrane pore channels that do not allow methane to
permeate through, while methane transports through the cracks and pinholes via both
Knudsen and bulk (convective) flow.
Subsequently, mixed-gas separation experiments were carried out with the binary
(H
2
/CH
4
) mixture for a constant feed composition of H
2
:CH
4
equal to 60.1%:39.9% under
different feed pressures at the temperature 200
o
C. Upon the completion of this series of
mixed-gas experiments, another series of single-gas permeation measurements were
carried out in order to check the status of the CMS membrane, and the impact that the
permeation experiments may have.
Table 6.1. Single-gas permeance at 200
o
C for H
2
and CH
4
performed before the
mixed-gas separation experiment (calculated with respect to the outside membrane area).
CH
4

Pressure (psig) permeance m
3
/(m
2
*hr*bar)
102.3 0.0138
204.7 0.0147
302.5 0.0155
399.6 0.0166
H
2

Pressure (psig) permeance m
3
/(m
2
*hr*bar)
102.7 2.2630
199.5 2.2845
300.5 2.2509
137
Figure 6.2 shows the single-gas CH
4
permeance together with the mixed-gas
permeance in the binary (H
2
/CH
4
) mixture (see discussion below on how the mixed-gas
permeance is calculated). From Figure 6.2, the single-gas and the mixed-gas CH
4

permeance appear to be close to each other (less than 5% difference, which may also be
due to a slight concentration polarization effect, see further discussion below), and both
depend linearly on the average pressure. Also note, that the single-gas permeance stays
the same before and after the mixed-gas separation experiments, indicative that no
changes have occurred in the part of the membrane structure through which methane
transports. In contrast, the H
2
permeance in the mixed-gas experiments decreases as the
system pressure increases, as can be seen in Figure 6.3. This, we believe, can be
explained by a concentration polarization effect of H
2
, which is a rather unusual
observation for microporous membranes [10-14]; however, these CMS membranes are
highly permeable, and in the presence of large transmembrane gradients, as those utilized
here, it may be possible that external mass transfer limitations may come into play.

Figure 6.2. CH
4
permeance at 200
o
C
138

Figure 6.3. H
2
permeance at 200
o
C
To verify that this may be, indeed, the case, we modified the membrane separator
model to account for concentration polarization effects. A membrane separator operating
under isothermal conditions in the absence of concentration polarization is described by
the following equations:
(6.1)
(6.2)
where
F
j
n is the molar flow rate (mol/h) for component j in the feed side,
P
j
n the
corresponding molar flow rate (mol/h) in the permeate side, V the reactor volume
variable (m
3
),
m
α the surface area of the membrane per unit reactor volume (m
2
/m
3
) and
U
j
(mol/m
2
. h. bar) the permeance for component j, described by the following equation.
(6.3)
where F
j
is the molar flux (mol/m
2
. h), through the membrane, P
j
F
(bar) the partial
139
pressure for component j in the feed side, and P
j
P
(bar) the partial pressure for component
j in the permeate side. The mixed-gas permeances reported in Figures 6.2 and 6.3 are
calculated by fitting the experimental data, via the use of the above equations, and by
assuming that for a constant total membrane separator pressure the permeance does not
vary along the length of the separator.
In order to account for concentration polarization effects the assumption that external
mass transport limitations are negligible must be relaxed and a revised model must be
used for the membrane separator operating under such conditions. In this model, instead
of using U
j
in the above equations we use instead U
j
overall
as an overall effective
permeance, which incorporates both the true membrane permeance as well as the external
mass transfer coefficient for component j, k
j
[14]. The equations describing the
membrane separator in the presence of concentration polarization are as follows:
(6.4)
(6.5)
(6.6)
(6.7)
(6.8)
For an annulus, hydraulic diameter [15], (6.9)
(6.10)
140
(6.11)
(6.12)
n: molar flow rate of component, mol/s;
V: reactor volume, cm
3
;
m
α
:
membrane area per feed side reactor volume, cm
2
/cm
3
;
d
e
: the hydraulic diameter, cm;
D: overall diffusion coefficient, cm
2
s
-1
;
D
j
: diffusion coefficient of each feed gas component, cm
2
s
-1
;
u
j
: the velocity of pure component j, cm s
-1
;
u: the velocity of the feed gas and here this is the average velocity on the feed side,
cm s
-1
;
d
i
: the inside diameter of the reactor, cm;
d
0
: the outside diameter of the membrane, cm;
υ: is the overall kinematic viscosity of the feed gas, cm
2
s
-1
; [14]
V
mol,j
: is the molar volume of component j;
P
j
: partial pressure for component j, bar, P
P
is for permeate side and P
F
is for feed
side;
141
In the above equations we assume that only hydrogen is subjected to concentration
polarization, as its flux is two orders of magnitude higher than that of methane. For the
hydrogen mixed-gas permeance we assume that it is the same as the single-gas
permeance, which makes sense as only hydrogen permeates through these small pores
and its partial pressure should not have any impact on its permeation rate.
By first assuming that the hydrogen mixed-gas permeance is equal to the single-gas
permeance and not taking into account concentration polarization, we use the membrane
separator Eqns. 6.1 and 6.2 above and calculate the hydrogen flow to the permeate side
and compare it with the experimental results at 200
o
C and 245
o
C. As Figure 6.4 shows,
the simulated H
2
flows are significantly higher than the experimental H
2
flow rates,
indicating the presence of an additional resistance to transport which effectively reduces
the overall permeability of the membrane (see Eqn. 6.8).

Figure 6.4. Simulation of the H
2
flow on the permeate side without taking into account
the external mass transfer limitations, and comparison with the experiments
142
Using Eqns. 6.4 – 6.8 and assuming again that the mixed-gas permeance is the same
with the single-gas permeance provides a much better fit between simulations and
experiments, as Figures 6.5 and 6.6 indicate, lending credence to the assumption that
concentration polarization is present under high-temperature and high-pressure conditions
for these highly permeable CMS membranes.

Figure 6.5. Simulation of the H
2
flow in the permeate side in the presence of external
mass transfer limitations, and comparison with the experiments at 200
o
C.

143
Figure 6.6. Simulation of the H
2
flow in the permeate side in the presence of external
mass transfer limitations, and comparison with the experiments at 245
o
C.

The net effect of the concentration polarization effect is a decrease in the H
2

permeance in the mixed-gas as the total system pressure goes-up, and since the CH
4

permeance stays almost the same with the single gas values, the real separation factor
between H
2
and CH
4
is significantly smaller than the ideal factor based on single-gas
permeance values. The net result is a decrease in the hydrogen purity of the permeate
product, as Table 6.2 indicates.
Table 6.2. Experimentally measured H
2
purity of permeating gas at 200
o
C in the
mixed-gas separation experiment
Pressure (psig) H
2
purity (%)
106.4 99.39
161 99.30
224.7 99.18
280.3 99.05
355.1 98.85
To further investigate the impact of concentration polarization on membrane
separator performance (for the H
2
/CH
4
binary mixture), we have used the model above to
calculate the effective membrane permeability while varying the overall feed flow-rate
and membrane separator feed pressure. The results are presented in Figures 6.7 and 6.8.
As expected, as the feed flow increases, the effective H
2
permeance comes closer to the
single-gas H
2
permeance, which means that the impact of concentration polarization
diminishes as the feed flow rate increases. Also, as the system pressure increases, the
effective H
2
permeance decreases, as we have observed experimentally.
144

Figure 6.8. Effective H
2
permeance as a function of pressure
145

Figure 6.9. Effective H
2
permeance as a function of H
2
feed flow rate.

The above sequence of experiments was repeated at a different temperature of 245
o
C.
Table 6.3 below shows the methane and hydrogen single-gas permeances. As with the
data at the lower temperature, the methane permeance increases with pressure (though the
change is somewhat smaller, ~ 10%). On the other hand, the hydrogen permeance
remains unchanged. The methane single gas permeance measured before and after the
mixed-gas experiments as well as the mixed-gas methane permeance are shown in Figure
6.10. The mixed gas permeance also increases with increasing to total pressure, but there
seem to be larger differences between the single-gas and mixed-gas permeances for the
higher temperature than the data generated at lower temperatures. One of the reasons for
the discrepancy between the mixed-gas and single-gas permeance values is also apparent
in the Figure as the methane single-gas permeance after the mixed gas experiments has
increased by ~ 10%, which seems to indicate a widening of the pores through which the
methane goes through. The increase in temperature does seem to affect the pores that are
146
available for hydrogen transport, however, as the hydrogen single-gas permeance before
and after the experiments is virtually unchanged, as Figure 6.11 indicates. Note once
more in Figure 6.11 the presence of the concentration polarization effect.
Table 6.3. Single-gas permeances at 245
o
C for H
2
and CH
4
before the mixed-gas
separation experiments
CH
4

Pressure (psig) permeance m
3
/(m
2
*hr*bar)
107.8 0.0200
198.2 0.0209
319.5 0.0217
408.3 0.0223
H
2

Pressure (psig) permeance m
3
/(m
2
*hr*bar)
102.9 3.1825
197.1 3.1711
239.2 3.1707

Figure 6.10. CH
4
permeance at 245
o
C.
147

Figure 6.11. H
2
permeance at 245
o
C.
For future work, in this area we recommend the study of the quaternary
(CO/H
2
/CH
4
/CO
2
) mixture under both high-pressure and temperature conditions. As in
our previous study by our group under the lower pressure and temperature conditions, the
study needs to be systematic involving the study of the individual binary as well as
ternary subsystems. One additional effect that needs to be investigated is the effect of
water vapor on the transport characteristics of the quaternary gas mixture. As with the
preliminary experiments, so far, single-gas permeation tests must be used as a sensitive
indicator of the state of the membrane stability.

148
6.4 References
[1] H. C. Lee, M. Monji, D. Parsley, M. Sahimi, P. Liu, F. Egolfopoulos, and T. Tsotsis,
Use of Steam Activation as a Post-Treatment Technique in the Preparation of Carbon
Molecular-Sieve Membranes, Ind. Eng. Chem., Res., 53 (2013) 1122-1132.
[2] R. Ash, R. M. Barrer, C. G. Pope, Flow of adsorbable gases and vapors in a
microporous medium I Single sorbates, Proc. Roy Sot. A, 271 (1963) 1-18.
[3] R. Ash, R. M. Barrer, C. G. Pope, Flow of adsorbable gases and vapors m a
microporous medium II Binary mixtures, Proc. Roy Sot. A, 271 (1963) 19-33.
[4] R. Ash, R. M. Barrer, R. T. Lowson, Transport of single gas and binary gas mixtures
m a microporous carbon membrane, J. Chem. Soc. , Faraday Trans I, 69 (1973)
2166-2178.
[5] P. S. Tin, T. S. Chung, Y. Liu, R. Wang, Separation of CO
2
/CH
4
through carbon
molecular sieve membranes derived from P84 polyimide, Carbon 42 (2004) 3123–3131.
[6] M. Hong, S. Li, J. L. Falconer, R. D. Noble, Hydrogen purification using a SAPO-34
membrane, J. of Membrane Sci. 307 (2008) 277–283.
[7] M. G. Sedigh, W. J. Onstot, L. Xu, W. L. Peng, T. T. Tsotsis, M. Sahimi,
Experiments and Simulation of Transport and Separation of Gas Mixtures in Carbon
Molecular Sieve Membranes, J. Phys. Chem. A, 102 (1998) 8580-8589.
[8] R. T. Adams, J. S. Lee, T. H. Bae, J. K. Ward, J.R. Johnson, C. W. Jones, S. Nair, W.
J. Koros, CO
2
–CH
4
permeation in high zeolite 4A loading mixed matrix membranes, J.
Membrane Sci. 367 (2011) 197-203.
[9] M. R. Kosuri, W. J. Koros, Defect-free asymmetric hollow fiber membranes from
Torlon®, a polyamide–imide polymer, for high-pressure CO
2
separations, J. Membrane
Sci. 320 (2008) 65-72.
[10] R. G. Nemmani, S. V. Suggala, An Explicit Solution for Concentration Polarization
for Gas Separation in a Hollow Fiber Membrane, Sep. Sci. Tech., 45 (2010) 581–591.
[11] S. Hara, K. Sakaki, N. Itoh, Decline in Hydrogen Permeation Due to Concentration
149
Polarization and CO Hindrance in a Palladium Membrane Reactor, Ind. Eng. Chem. Res.
38 (1999) 4913-4918.
[12] J. Zhang, D. Liu, M. He, H. Xu, W. Li, Experimental and simulation studies on
concentration polarization in H
2
enrichment by highly permeable and selective Pd
membranes, J. Membrane Sci. 274 (2006) 83–91.
[13] M. E. Ayturk, N. K. Kazantzis, Y. H. Ma, Modeling and performance assessment of
Pd- and Pd/Au-based catalytic membrane reactors for hydrogen production, Energy
Environ. Sci., 2 (2009) 430–438.
[14] G. He, Y. Mi, P. L. Yue, G. Chen, Theoretical study on concentration polarization in
gas separation membrane processes, J. Membrane Sci. 153 (1999) 243-258.
[15] H. Chanson, The hydraulics of open channel flow, Butterworth-Heinemann, Oxford,
UK, 2
nd
Edition, 2004

150
Chapter 7: Concluding Remarks and Suggestions for Future Work
Air pollution is today a serious problem throughout the world, especially in the
industrialized and developing regions, with power-plant, factory, and motor vehicle
emissions being major contributors. The use of hydrogen, as an alternative energy source
to substitute for fossil fuels, is generally acknowledged as a potentially effective way for
reducing such air emissions. One of the benefits of adopting hydrogen as a future energy
source, in addition to reducing carbon dioxide emissions, is that it can be produced from
readily available resources like coal and renewable biomass; this then diminishes the
need to use the world’s dwindling crude-oil resources.
Hydrogen production integrated within IGCC power plants, that show promise for
environmentally-benign power generation, is a key focus of this Thesis. In these plants
coal and/or biomass are first gasified into syngas, which is then processed in a water
gas-shift (WGS) reactor to further enhance its hydrogen content for power generation.
However, impurities in the syngas, primarily H
2
S, NH
3
, various organic vapors and
tar-like species are detrimental to catalyst life and must be removed before the gas enters
the WGS reactor. This, then, means cooling the syngas for clean-up and then reheating it
to the WGS reaction temperature. For use in various industrial applications, and
potentially for CO
2
capture/sequestration, hydrogen purification is required. In this
research we have investigated, instead, a novel CMSM-based MR system termed as the
“one-box” process, in which syngas clean-up, and product separation are combined in the
same unit. It has proven successful for producing hydrogen from a feed with a simulated
syngas composition containing common impurities such as H
2
S, NH
3
, and model organic
vapor (toluene) and model tar-like (naphthalene) species. The lab-scale MR we have
studied employs a single CMS membrane used for the in-situ hydrogen separation. In our
study, the membrane was characterized in terms of its single-gas permeances, which were
151
also used for model predictions. The CMS membrane stability was also investigated in
the presence of the aforementioned impurities, and the membrane proved to be stable
under the experimental WGS reaction conditions.
The kinetics of the WGS reaction over a commercial Co/Mo/Al
2
O
3
sour-shift
catalyst was also investigated in the presence of H
2
S, NH
3
, the organic vapor and the
model tar-like species as part of this Thesis, and no impact of these impurities was
observed. The performance of the MR (the “one-box” process) using such membranes
and catalysts was investigated experimentally for a range of pressures and sweep ratios;
the MR showed higher conversions compared with those of the traditional packed-bed
reactor. Parallel modeling investigations indicated good agreement with the experimental
data. The proposed “one-box” process shows several advantages over the traditional
packed-bed reactor system, including improvements in CO conversion and H
2
purity,
while allowing one to perform the reaction in the presence of common impurities such as
H
2
S, NH
3
, organic vapor and tar-like species, and being able to deliver a
contaminant-free hydrogen product. Use of the process in hydrogen production from
coal/biomass-derived syngas should, therefore, result in considerable energy savings.
Another important and promising application for hydrogen are PEM fuel cells.
However, their operating temperature is typically low and, as a result, common impurities
found in conventional hydrogen production like CO, can adsorb onto and poison the
catalysts utilized. Therefore, high-purity hydrogen is required for the operation of such
systems. In this Thesis, an ultra-thin, long, high-performance (in terms of its H
2

permeance and selectivity) supported palladium membrane is used in a membrane reactor
system to produce pure hydrogen through the use of the water-gas shift reaction from a
gas stream with a simulated reformate composition. The system performance is
investigated under various experimental conditions, namely, different pressures, feed
152
flow rates and sweep ratios. At best performance, the system is able to reach almost
complete CO conversion and 90% hydrogen recovery. The product hydrogen purity is
always at more than 99.9% with CO concentration of less than 100 ppm. A model is used
for further study of the design aspects of the system. It is shown that the Pd membrane
reactor system under study is capable of delivering almost complete CO conversion and
H
2
recovery at experimental conditions akin to the industrial applications. The membrane
exhibits good stability in the reactor for more than a month under the WGS environment.
Hence, it is concluded that the Pd-based WGS-MR is, potentially, a promising system for
hydrogen production for fuel cell applications. Furthermore, we develop a hybrid Pd-MR
+ Methanation PBR System to produce ultra-high purity, ‘CO-free’ (less than 50 ppm of
CO) hydrogen for a potential application on fuel cells and more. The hybrid experiments
we performed indicate that by using this system, we can produce hydrogen with less than
3 ppm of CO and 30 ppm of CO
2
. Reaction kinetics experiments of the methanation
catalyst for low-concentration CO and CO
2
were also performed. These are incorporated
in a mathematical model of this hybrid reactor system, which is used to study its design
and scale-up aspects.
For the future, additional efforts should be made to further move the ‘one-box’
membrane reactor system towards field-scale evaluation and industry application. For
that, the lab-scale test module must be modified to be able to accommodate multiple
tubular membranes. The membrane reactor behavior must then be studied at different
experimental conditions and modeled for further scale-up and process design. The next
step will be testing this MR system under real industry production conditions to obtain
more information for its commercial application. Also, additional effort must be made for
improving the characteristics of the CMSM as they are important in making this process
more economical.
153
For fuel cell applications making use of Pd membranes, additional effort must be
made to make these Pd membranes less costly to prepare. Also, a multi-tubular
membrane reactor system must be prepared for use for further field-testing and
evaluation.

Abstract (if available)

Abstract High-temperature gas separations using inorganic membranes have attracted increased attention in recent years. In particular, the use of such membranes in high‐temperature and high‐pressure membrane reactors has the potential to enhance process intensification and to increase energy savings and/or product yield. Though the potential benefits of high‐temperature conventional as well as reactive gas separations are substantial, commercialization still remains elusive. A major technical barrier is the lack of robust inorganic membranes and full-scale modules which are suitable for use at the high‐temperature and high‐pressure conditions required. ❧ In this thesis, we focus on the application of two different types of such inorganic membranes, namely carbon molecular sieve membranes (CMSM) and Pd membranes in high‐temperature and high‐pressure reactive processes related to power generation, in particular, in the context of the Integrated Gas Combined Cycle (IGCC). Specifically, we study a “one-box” process in which the coal/biomass‐derived syngas is fed directly into a water gas shift (WGS) reactor, which efficiently converts the CO into hydrogen in the presence of a number of troublesome impurities found in such syngas (e.g., H₂S, tars, organic vapors, etc.), and delivers a contaminant‐free hydrogen product. This reactor makes use of a hydrogen‐selective carbon molecular sieve membrane and a sulfur‐tolerant Co/Mo/Al₂O₃ catalyst. In our studies, we investigate the membrane stability, catalytic kinetics, and the reactor’s overall behavior for different experimental conditions. The results are also used to validate a mathematical model for the membrane reactor, which is then used to further discuss the potential scale‐up of the proposed process. ❧ We also study in this Thesis a realistic size, ultra‐permeable Pd membrane for pure hydrogen production for fuel cell applications from a feed with a simulated reformate composition through the water gas shift reaction. Prior to its use in the reactor experiments, the membrane is characterized through single‐gas permeation measurements. The effect of different conditions during the WGS experiments is experimentally studied, and the results are again compared with those of a mathematical model. The model is then further used to study the design aspects of the proposed process. It is shown that the Pd membrane reactor system under study is capable of attaining almost complete CO conversion and full hydrogen recovery at realistic experimental conditions akin to those utilized in industrial applications.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

On the use of membrane reactors in biomass utilization

PDF

Process intensification in hydrogen production via membrane-based reactive separations

PDF

An integrated 'one-box' process for hydrogen production

PDF

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

PDF

A high efficiency, ultra-compact process for pre-combustion CO₂ capture

PDF

A hybrid adsorbent-membrane reactor (HAMR) system for hydrogen production

PDF

Methanol synthesis in the membrane reactor

PDF

Methanol synthesis in a membrane reactor

PDF

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

PDF

Fabrication of silicon carbide sintered supports and silicon carbide membranes

PDF

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

PDF

A flow-through membrane reactor for destruction of a chemical warfare simulant

PDF

Lab-scale and field-scale study of siloxane contaminants removal from landfill gas

PDF

Fabrication of nanoporous silicon carbide membranes for gas separation applications

PDF

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

PDF

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

PDF

The purification of contaminated air streams via recative separation techniques

PDF

Hydrogen storage in carbon and silicon carbide nanotubes

PDF

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

PDF

Novel methods for landfill gas and biogas clean-up

Asset Metadata

Creator Yu, Jiang (author)

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

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Chemical Engineering

Publication Date 05/12/2014

Defense Date 09/09/2013

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

Tag carbon molecular seive(CMS) membrane reactor,hydrogen production,IGCC power plant,kinetics,OAI-PMH Harvest,Pd membrane reactor,process intensification,water gas shift (WGS) reaction

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Tsotsis, Theodore T. (committee chair), Egolfopoulos, Fokion N. (committee member), Sahimi, Muhammad (committee member)

Creator Email jiangyu@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-412920

Unique identifier UC11296544

Identifier etd-YuJiang-2506.pdf (filename),usctheses-c3-412920 (legacy record id)

Legacy Identifier etd-YuJiang-2506-0.pdf

Dmrecord 412920

Document Type Dissertation

Format application/pdf (imt)

Rights Yu, Jiang

Type texts

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

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA