Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
A hybrid adsorbent-membrane reactor (HAMR) system for hydrogen production
(USC Thesis Other)
A hybrid adsorbent-membrane reactor (HAMR) system for hydrogen production
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
A HYBRID ADSORBENT-MEMBRANE REACTOR (HAMR)
SYSTEM FOR HYDROGEN PRODUCTION
by
Aadesh Harale
A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(CHEMICAL ENGINEERING)
December 2008
Copyright 2008 Aadesh Harale
ii
Acknowledgments
I would like to thank my advisors, Professors Theodore Tsotsis and
Muhammad Sahimi, for their support and encouragement throughout
the course of this work, as well as their insightful comments and
suggestions, and for the great degree of latitude they allowed me in
conducting the research reported in this thesis.
I also wish to thank Professor Hai Wang for serving on my dissertation
committee and for his helpful advice during my Ph.D. studies.
I sincerely appreciate the help of Ms. Karen Woo, Mr. Brendan Char,
and Mrs. Tina Silva of the Mork Family Department of Chemical
Engineering and Materials Science, and thank them for their kindness.
In addition, I must thank my fellow graduate students for their help and
collaboration. In particular, I want to thank Hyun Hwang, Babak
Fayyaz, and Yongman Kim for their assistance at various stages of this
research. I also want to thank Mr. Don Wiggins of the USC machine
shop for his kind help on numerous occasions.
I am indebted to all my friends for their great friendship and
encouragement during the course of my life.
iii
Above all, I would like to thank my parents for their patience,
encouragement and support all through during my Ph.D. studies. I am
also grateful to my sister and brother for their constant support and
encouragement throughout my seemingly endless years at school.
iv
Table of Contents
Acknowledgements ii
List of Tables vii
List Figures x
Abstract
xvi
Chapter I. Introduction & Overview
1
1.1 Motivation & background
1
1.2 Hydrogen Production Routes
3
1.2.1 Steam Reforming
4
1.2.2 Partial oxidation & autothermal reforming
4
1.2.3 Water electrolysis
6
1.2.4 Gasification and woody biomass
conversion
7
1.2.5 Biological hydrogen production
9
1.3 Scope of work
10
Chapter II. HAMR Overview & Model 15
2.1 Introduction 15
2.2 HAMR Concept 15
2.3 Proposed hydrogen production HAMR process 20
2.4 The mathematical model of the HAMR system 29
Chapter III. Membrane Studies 44
3.1 Introduction 44
v
3.2 Overview on membrane separation 46
3.3 Membrane performance evaluation 48
3.3.1 Membrane module design 49
3.3.2 Membrane permeance calculations 50
3.3.2.1 CMS membrane data 54
3.3.2.2 Palladium membrane data 66
Chapter IV. Adsorbent Studies 73
4.1 Introduction 73
4.2 CO2 Adsorbent Studies 74
4.3 Effect of Aluminum content in the hydrotalcite on
the adsorption capacity
78
4.4 Characterization of Hydrotalcite 80
4.5 Adsorption Equilibria 82
4.3 Adsorption Kinetic Studies 86
4.3 Effect of cycle number on adsorption capacity 93
4.3 Adsorbent regeneration/desorption studies 94
Chapter V. Reaction Kinetic Studies 97
5.1 Introduction 97
PART (A): WGS REACTION KINETICS
5.2 Introduction 98
5.3 Thermodynamic Aspects 99
5.4 Mechanism and Kinetics 100
PART (B): REFORMING KINETICS
5.5 Introduction 111
5.6 Thermodynamic Aspect 112
5.7 Reaction Kinetic and Mechanism 115
Chapter VI. Results & Conclusions 125
vi
6.1 Introduction 125
6.2 Experimental Set-up 125
6.3 Experimental Result 128
6.4 Sensitivity analysis 141
6.5 Summary & Conclusion 173
Bibliography 176
Appendix A 194
Appendix B 214
vii
List of Tables
2.1 Comparison of conventional and HAMR operating
conditions
18
2.2 Boundary conditions for the regeneration steps 43
3.1 Physical properties and experimental conditions
investigated
54
3.2 Single and mixed-gas permeance data for CMS membrane
# 1
56
3.3 Single and mixed-gas permeance data for CMS membrane
# 2
56
3.4 H2-CO2 binary gas permeance data for CMS membrane #
3
57
3.5 Comparison of permeance data for the feed side vs. the
tube side
58
3.6 Single and mixed-gas permeance data for palladium
membrane # 1
67
3.7 mixed-gas permeance data for palladium membrane # 1 67
3.8 Summary of the literature data for the palladium
membranes
69
3.9 Effect of steam on the palladium membrane performance 72
4.1 The physical and chemical properties of hydrotalcite
investigated
79
4.2 Sips isotherm parameters 84
viii
4.3 Diffusivity Data for Hydrotalcite 89
4.4 Mass transfer coefficient values for various temperatures
using uptake data from TGA
90
4.5 Mass transfer coefficient values for various temperatures
using packed bed data
93
5.1 WGS equilibrium constant as a function of temperature 99
5.2 Physical and chemical properties of LK-821-2 catalyst 104
5.3 Experimental conditions investigated 106
5.4 Langmuir-Hinshelwood rate model and kinetic parameters 110
5.5 Equilibrium constant as a function of temperature 113
5.6 Reforming catalysts and supports overview 114
5.7 Various reforming catalyst and their operating range 115
5.8 Summary of reforming kinetic mechanism 116
5.9 Physical and chemical properties of reforming catalyst 118
5.10 Rate expressions and thermodynamic properties for the
methane-steam reforming reaction
120
5.11 Comparison of reforming kinetic parameters 124
ix
6.1 Parameters used in the simulations 149
6.2 Parameters used in the cyclic HAMR operation
simulations
158
x
List of Figures
2.1 PSA and TSA operation diagram 20
2.2 HAMR-hydrogen production process 21
2.3 Schematic illustrations of the HAMR process steps 26
2.4 4-bed/4-step HAMR hydrogen production process 28
2.5 Schematic of hybrid adsorbent-membrane reactor 29
3.1 Membrane unit nomenclatures 47
3.2 Schematic of membrane reactor module 50
3.3 Hydrogen permeance as a function of H2/CO2 ratio 58
3.4 H2/CO2 (single gas) vs. H2 permeance 59
3.5 H2/CO2 (mixed gas) vs. H2 permeance 60
3.6 H2/CO (single gas) vs. H2 permeance 61
3.7 H2/CO (mixed gas) vs. H2 permeance 62
3.8 H2/ CH4 (single gas) vs. H2 permeance 63
3.9 H2/ CH4 (mixed gas) vs. H2 permeance 64
xi
3.10 H2/ H2O (mixed gas) vs. H2 permeance 65
3.11 Hydrogen flux as a function of trans-membrane pressure 70
3.12 Pressure-dependence of hydrogen permeance 71
4.1 Structural models for hydrotalcite-like compounds 77
4.2 The structure of hydrotalcite materials 78
4.3 Adsorption capacity as a function of Al content of the
HTC at 250° C and 1 bar
80
4.4 Nitrogen adsorption/desorption isotherm in the
hydrotalcite at 77K
81
4.5 H-K plot for the hydrotalcite sample 82
4.6 CO2 isotherms and the Sips fit 84
4.7 Adsorption capacities as a function of dry and wet
conditions for MG50
86
4.8.a Short-time behavior of CO2 uptake Mt/M∞ at 523 K 88
4.8.b Long-time behavior of CO2 uptake Mt/M∞ at 523 K 89
4.9 Mass transfer coefficient calculation using packed bed
data
92
4.10 Effect of cycle number on adsorption capacity of
hydrotalcite at 250°C, Pressure = 1 atm
94
xii
4.11 CO2 desorption profiles using Argon as a purge gas for
temperatures 250-450°C
96
5.1 Arrhenius plot for the Haugon-Watson reaction kinetics 108
5.2 Experimental vs. fitted CO conversion for various
temperatures, pressures and compositions
109
5.3 CH4 conversion measured as a function of W/FCH4 122
5.4 Experimental vs. fitted CH4 conversion for various
temperatures
123
6.1 Experimental set-up 126
6.2 Comparison of the experimental data and model
predictions for = 282 at 250°C
130
6.3 Comparison of the experimental data and model
predictions for = 186 at 250°C
132
6.4 Comparison of the experimental data and model
predictions for = 350 at 250°C
133
6.5 Comparison of the experimental simulated CO2
breakthrough for = 350 at 250°C
134
6.6 Comparison of the experimental data and model
predictions for = 300 at 250°C
135
xiii
6.7 Comparison of experimental and simulated CO2
breakthrough for = 300 at 250°C
136
6.8
Cyclic behavior of HAMR for = 466 at 250°C
137
6.9
Cyclic behavior of HAMR for = 287 at 250°C
138
6.10 Comparison of effect of Sweep Ratio on the CO
conversion for = 466 at 250°C
139
6.11 Comparison of the experimental and simulated CH4
conversion for = 187 at 450°C
141
6.12 H2 permeance sensitivity analysis 142
6.13 CO permeance sensitivity analysis 143
6.14 CO2 permeance sensitivity analysis 144
6.15 H2O permeance sensitivity analysis 145
6.16 mCO2 sensitivity analysis 146
6.17 Mass transfer coefficient sensitivity analysis 147
xiv
6.18 Performance of the HAMR and AR systems for
different
151
6.19 Effect of membrane properties on the HAMR and AR
systems
153
6.20 Effect of adsorbent properties on the HAMR and AR
systems
155
6.21 Effect of initial condition for step 1 for AR systems 156
6.22 Schematic illustration of the four-bed HAMR process 159
6.23 Conversion and permeate-side CO concentration (ppm)
profile during the reaction step
160
6.24-a Effect of varying the DaPe number on CO conversion for
Wc/Fco= 200
162
6.24-b Effect of varying the DaPe number on the time-averaged
CO concentration (ppm) for Wc/Fco= 200
163
6.25 Effect of varying the DaPe number on H2 recovery for
Wc/Fco= 200
163
6.26-a Effect of varying the H2/CO separation factor on CO
conversion for Wc/Fco= 200
165
6.26-b Effect of varying the H2/CO separation factor on the
time-averaged CO concentration in the permeate-side for
Wc/Fco= 200
165
6.27-a CO conversion profile as a function of the reaction step
time
166
xv
6.27-b CO concentration profile in the permeate side (ppm) as a
function of the reaction step time
167
6.28-a Feed-side H2 mole fraction profiles during the blow-down
step
168
6.28-b Feed-side CO2 mole fraction profiles during the blow-
down step
169
6.29
Feed-side CO2 mole fraction profiles during the purge
step
170
6.30 Feed-side H2, H2O and CO2 mole fraction profiles during
the pressurization step
171
6.31 Pressure history during the cyclic HAMR operation 172
xvi
Abstract
As a result of stricter environmental regulations worldwide, hydrogen
is progressively becoming an important clean energy source. For H2 to
replace fossil fuels in mobile applications, it will require the creation of
a production and delivery infrastructure equivalent to that currently
existing for fossil fuels, which is an immense task. As an alternative,
and as an interim step towards the new hydrogen economy, various
groups are currently studying steam reforming of methane (SRM) for the
on-board generation of hydrogen, or for on site production, in order to
alleviate the need for compressed or liquid hydrogen gas storage.
Conventional technologies are, however, neither convenient nor
economical to apply for small-scale (on site or on-board) hydrogen
generation. Reactive separation processes have, as a result, been
attracting renewed interest for application in H2 production through
SRM. One such technology is the hybrid adsorbent-membrane reactor
(HAMR) system, which couples reaction and membrane separation
steps with adsorption on the reactor and/or membrane permeate side.
The HAMR concept was originally proposed by USC group
for
esterification reactions, and it was adapted recently for on-board or on-
site hydrogen production applications. Our early studies involved the
development of a mathematical model for the HAMR system (applied to
hydrogen production through SRM); recently experimental
xvii
investigations with the water-gas shift reaction, using microporous
membranes and CO2 hydrotalcite-type adsorbents, were carried out in
order to validate the HAMR design models. Experimental data were
compared with the model predictions, and found to be consistent. The
practical process design aspects of the HAMR hydrogen production
process was also investigated both experimentally and through
modeling studies.
Chapter 1
Introduction and Overview
1.1 Motivation and background
With the start of the twenty-first century global civilization is
threatened by the prospect of economic disarray and incitement to war,
caused by the local depletion and global misdistribution of high quality
fossil fuels, especially oil, as well as by the prospect of almost
unimaginable environmental, economic, and cultural disruption caused
by climate volatility and triggered, primarily, by our energy system’s
carbon dioxide effluent (Ogden et al. 1999; Padró and Putsche 1999;
Ogden 2001). The growing technological advances are transforming the
manner that we live in, and, as a result, the demand for energy will
continue to increase considerably in the future.
Fossil fuels are of finite resource, and if we remain tethered to
them for our existence, it is merely a question of when, not if, that one
or both of the catastrophic specters, outlined above, becomes a reality
(Ogden et al. 1999; Padró and Putsche 1999; Ogden 2001). Today, we
are likely seeing the precursors of both. From the viewpoint of Earth’s
future, climate volatility is much more critical. This suggests that the
global environment, not energy resource depletion, is the absolute cap
2
on fossil fuel use (Ogden et al. 1999; Padró and Putsche 1999; Ogden
2001; Hemmes K. 2003; Babu 2004; McHugh 2005).
With the growing global population, the number of vehicles and
transportation energy demand are both projected to grow rapidly
(Ogden 2001). Continued reliance on fossil fuels and vehicle
technologies, however, poses significant challenges with respect to air
pollution, greenhouse gas emissions and energy supply security.
Combustion of liquid fuels for transportation and heating contributes to
about two thirds of all greenhouse gas emissions. Even with gains in
energy efficiency, it is likely that low or zero carbon fuels will be needed
to meet future carbon emission reduction goals, since the
transportation sector accounts for a large fraction of air pollutant
emissions (Gregory and Pangborn 1976; Philcox and Fenner; Yacobucci
2004; McHugh 2005; Shoko et al. 2006).
Today virtually all transportation fuels are derived from oil, but oil
production is projected to peak worldwide within a decade or so. We
must, therefore, rapidly develop sustainable energy sources, which, in
addition, do not emit carbon dioxide. Such alternative energy sources
include solar, wind-derived power, and safe, economic, next–generation
nuclear power technologies.
Almost a century ago, hydrogen was recognized as a potential
energy carrier (Ogden et al. 1999; Trimm and Onsan 2001). Hydrogen
3
contains no carbon that can form pollutants or greenhouse gases. The
carbon dioxide effluents from fossil fuel consumption are pushing our
planet towards climate destabilization that, if unabated, will be
catastrophic. To eliminate greenhouse gas emissions, we need to
develop non-carbon energy sources. Hydrogen, as fuel, is an important
example of a non-carbon energy source which, together with other
renewable sources of energy, including hydro-power, solar, wind-based
energy, and nuclear fission, and perhaps fusion (Gregory and Pangborn
1976; Philcox and Fenner; Padró and Putsche 1999; Ogden 2001;
Trimm and Onsan 2001) will, if developed, potentially diminish or
completely eliminate the prospect of global warming. Various hydrogen
production routes are available today, and are discussed in the
following section
1.2 Hydrogen production routes
Hydrogen makes up about 75% of the known baryonic universe,
but is not a commonly utilized energy source like oil, coal, wind, or solar
power. Hydrogen is, potentially, an important versatile energy carrier
because like oil and gas, it can be stored in large amounts (albeit
currently at a higher storage cost), and can be conveniently made from
many of the current energy resources, and used to service almost any
energy need. Hydrogen is the lightest element and molecule. The reason
hydrogen is not a common energy source is that it is almost never
4
found by itself, the way coal, oil, and natural gas are. Instead, it must
be first released from chemical compounds in which it is bound to
(Gregory and Pangborn 1976; Philcox and Fenner; Padró and Putsche
1999; Ogden 2001; Trimm and Onsan 2001).
1.2.1 Steam reforming
Catalytic steam reforming of methane is a well-known,
commercially available process for hydrogen production (Twigg 1989;
Rostrup-Nielsen and Alstrup 1999). In the United States, most
hydrogen (over 90%) is currently manufactured via steam reforming of
natural gas (Twigg 1989; Yacobucci 2004; McHugh 2005).
Conventionally, hydrogen production is accomplished in several steps:
steam reforming, water-gas-shift (WGS) reaction, and hydrogen
purification.
The steam-reforming reaction,
CH
4
+ H
2
O ! CO + 3H
2
"H = +206.16 kJ /molCH
4
(1)
is endothermic and requires an external heat input. Economics favor
reactor operation at pressures of 3-25 atmospheres and temperatures of
700-850 °C (Twigg 1989; Hufton et al. 1999; Sircar et al. 1999; Sircar
2002). This reaction is discussed in detail in Chapter 5.
1.2.2 Partial oxidation and autothermal reforming
Another commercially available method for generating hydrogen
from hydrocarbons is partial oxidation (POX). Here, methane (or some
5
other hydrocarbon feedstock, such as crude oil) is oxidized to produce
carbon monoxide and hydrogen according to the following reaction
CH
4
+
1
2
O
2
! CO + 2H
2
"H =#36 MJ /kmol CH
4
(2)
The reaction is exothermic and no heat exchanger is needed. Catalysts
are not required because of the high temperature. However, the
hydrogen yield per mole of methane input can be significantly enhanced
by use of catalysts. A hydrogen plant based on POX includes a POX
reactor, followed by a shift reactor, and the associated hydrogen
purification equipment. Large-scale POX systems have been used
commercially to produce hydrogen from hydrocarbons, such as residual
oil, for such applications as refineries.
The POX reactor is more compact than a steam reformer, where
heat must be added indirectly via a heat exchanger. The efficiency of the
POX unit is relatively high (70-80%). However, POX systems are
typically less energy efficient than steam reforming because of the
higher temperatures involved (which exacerbates heat losses) and the
problem of heat recovery. Although the POX reactor is likely to be less
expensive than a steam reformer vessel, the downstream shift and
purification stages are likely to be more expensive (Twigg 1989; Hufton
et al. 1999; Ogden et al. 1999; Sircar et al. 1999; Ogden 2001; Sircar
2002; McHugh 2005).
6
Autothermal reforming combines POX and steam reforming. The
name refers to the heat exchange between the endothermic steam
reforming process and the exothermic POX. Autothermal reformers
combine some of the best features of the steam reforming and POX
systems. In autothermal reforming a hydrocarbon feed (methane or a
liquid fuel) is reacted with both steam and air to produce a hydrogen-
rich gas. Both the steam-reforming and POX reactions take place, for
example, with methane according to reactions (1) and (2). With the right
mixture of input fuel, air and steam, the POX reaction supplies all the
heat needed to drive the catalytic steam reforming reaction. Unlike a
steam reformer, the autothermal reformer requires no external heat and
the use of heat exchangers. This makes autothermal reformers simpler
and more compact than steam reformers, which is why often
autothermal reformers have a lower capital cost. Since all the heat
generated by the POX reaction is fully utilized to drive the steam
reforming reaction, autothermal reformers typically offer higher system
efficiency than POX systems, where excess heat is not easily recovered
(Twigg 1989; Hufton et al. 1999).
1.2.3 Water electrolysis
The conversion of electric energy into hydrogen (and oxygen) by
water electrolysis has been known for a long time (first demonstrated by
Faraday in 1820, and widely used since about 1890). If the electricity is
7
produced by the use of fossils fuels, then the cost of hydrogen obtained
this way is higher than the one associated with steam reforming of
natural gas. On the other hand, the high hydrogen purity needed in
some applications is easier to achieve this way, and for this reason
electrolysis currently has almost 5% market share of hydrogen
production. Though the cost of electricity dominates the economics of
hydrogen production from electrolysis, the situation would be quite
different if the electricity used is surplus electricity from various
renewable resources, such as wind or solar power. In cases where such
electricity is produced at times of no local demand and no evident
option for export to other regions exists, using it to produce hydrogen
fuel by electrolysis may be a valid option (Twigg 1989; Hufton et al.
1999; Chen and Elnashaie 2005; McHugh 2005).
1.2.4 Gasification and wood biomass conversion
The gasification of wood biomass is a century-old technology that
flourished before and during World War II. The interest in gasification
technology has undergone many ups and downs in the intervening
years. Just like the entire subject area of biomass energy, interest in
biomass gasification has peaked and receded with the availability and
price of oil and other fossil fuels. While biomass gasification is an old
technology, it is also a developing technology, since it was never fully
embraced on a large commercial scale.
8
Gasification is a form of pyrolysis in which the solid biomass is
converted into gas through a thermochemical reaction. Through
pyrolysis, solid biomass can be converted to liquid fuel by heating the
biomass in the absence of oxygen, or partially combusting it in a limited
oxygen supply. Normally a hydrocarbon-rich gas mixture, an oil-like
liquid, and a carbon-rich solid residue are produced. The carbon-rich
solid is charcoal, the liquid is bio-oil, and the gas is called producer gas,
synthetic gas, or syngas. In the gasification process, pyrolysis is carried
out with the introduction of excess air and at high temperatures,
generally in the range of 600-1000 °C.
The gas produced from biomss gasification (syngas) is a mixture
of carbon monoxide, hydrogen, and methane, together with carbon
dioxide and nitrogen. The energy content of the syngas depends
significantly on the approach used to supply heat to drive the
gasification reactions. Most designs use oxygen as an oxidizing agent,
either in air or in its pure form. Heat is generated by partially
combusting the biomass feedstock. In air-blown gasification, nitrogen in
the air dilutes the syngas product, which results in a low-energy gas.
The low-energy syngas can be used in applications where the heat
content of the gas is not critical (Gregory and Pangborn 1976;
Dresselhaus 2003; Hemmes K. 2003; Babu 2004; McHugh 2005).
9
Gasification coupled with WGS is the most widely practiced
process route for biomass to hydrogen. Thermal, steam and partial
oxidation gasification technologies are under development around the
world. Feedstocks include agricultural and forest product residues of
hard wood, soft wood and herbaceous species. Thermal gasification is
essentially high-rate pyrolysis carried out in the temperature range of
600–1000 °C in fluidized bed gasifiers. Other relevant gasifier types are
bubbling fluid beds and the high-pressure high-temperature slurry-fed,
entrained flow gasifier. However, all these gasifiers need to include
significant gas conditioning along with the removal of tars and inorganic
impurities and the subsequent conversion of CO to H2 by the WGS
reaction. At present, there are no commercial biomass gasification
processes for hydrogen production.
1.2.5 Biological hydrogen production
Production of hydrogen from biomass may also be achieved by
biological fermentation, or by other bacterial or algae decomposition of
water, or another suitable substrate. The conversion processes may be
carried out in the dark or with the assistance of light. Growing the
biological substance in the first place requires energy input, normally
from sunlight and, thus, there are several conversion efficiencies
involved: from the primary energy source to the biological material, from
the energy in the biological material to the energy contained in the
10
hydrogen produced, with all of them combined constituting the overall
conversion efficiency from solar radiation to hydrogen. Because
production of molecular hydrogen is rarely the end-purpose of natural
biological systems, they have to be modified accordingly in order to
serve this purpose, e.g., by genetic engineering. Photosynthesis, for
which light is required, and fermentation, which occurs in the dark, are
the two major biological ways today of producing hydrogen (Hemmes K.
2003; Babu 2004).
1.3 Scope of the work
Fuel cells are considered to be a viable alternative to the
conventional technologies for future clean and efficient power
generation. It is anticipated, therefore, that the fuel cell market will
expand significantly, if a number of the key technical barriers that
currently exist can be overcome. Two such key barriers are fuel
processing and purification; both are critical in determining the fuel cell
system’s size, and process complexity. The existing fuel processing is
based on conventional steam reforming of natural gas (NG) and of other
hydrocarbons. The conventional process, as noted previously, is
practiced commonly at large-scale for H2 production, for NH3 and
methanol synthesis; it is, however, a complicated and capital- and
energy-intensive operation, which makes it difficult to scale-down for
fuel cell applications. As mentioned above, hydrogen is a valuable
11
supplemental fuel in power generation, as it is known to significantly
improve the burning characteristics of common fuels like natural gas. A
key barrier to its more widespread use, however, is its production cost.
Reactive separation processes have been attracting renewed
interest for application in catalytic steam reforming and the water-gas
shift reaction. From the standpoint of hydrogen production with
concomitant removal of CO2, reactive separation could offer a unique
opportunity to reduce the hydrogen production cost via the separation
of CO2 from the reactor, which can then be ready for sequestration.
Commonly discussed reactive separation processes include packed-bed
catalytic membrane reactors (MR)
(Park et al. 1998; Nam et al. 2000;
Hwang 2001; Lim et al. 2002; Sanchez and Tsotsis 2002) and, more
recently, adsorptive reactor (AR) processes (Balasubramanian et al.
1999; Hufton et al. 1999; Waldron and Sircar 2000; Ding and Alpay
2000a; Ding and Alpay 2000b; Ding and Alpay 2001; Ortiz and
Harrison 2001; Xiu et al. 2002; Xiu et al. 2002; Xiu et al. 2003a; Xiu et
al. 2003b; Xiu et al. 2004). Their potential advantages over the more
conventional reactors have been widely discussed. They include, (i)
increasing the reactant conversion and product yield through shifting of
the equilibrium towards the products, potentially allowing one to
operate under milder operating conditions (e.g., lower temperatures,
pressures and reduced steam consumption), and (ii) reducing the
12
downstream purification requirements by in-situ separation from the
reaction mixture the desired hydrogen product (in the case of MR) or
the undesired CO2 (in the case of AR). The MR show substantial
promise in this area and, typically, utilize nanoporous inorganic or
metallic Pd or Pd-alloy membranes (Sanchez and Tsotsis 2002). The
latter are better suited for pure hydrogen production. However, metallic
membranes are very expensive and become brittle during reactor
operation (Nam et al. 2000) or deactivate in the presence of sulfur or
coke. Nanoporous membranes are better suited for the steam-reforming
environment. They are difficult to manufacture, however, without
cracks and pinholes and, as a result, often have inferior product yield.
In addition, the hydrogen in the permeate side contains other by-
products, and may require further treatment for use in fuel-cell-
powered vehicles
The essence of this study is the employment of a unique
membrane- and adsorption-enhanced reactor, termed the hybrid
adsorbent-membrane reactor (HAMR), which allows for in-situ
preferential H2 permeation and simultaneous CO2 adsorption. Thus, the
reactor can produce an ultra-pure H2 product (without the need for
using POX or methanizer reactors) continuously until the adsorbent is
saturated for regeneration via a pressure-swing adsorption (PSA)
operation. This unique reactor configuration can be viewed as a
13
simplified MR under PSA operation, suitable for using in a scaled-down
version of the steam-reforming process. The MR- and sorption-
enhanced PSA-based reactor technologies proposed in the literature
allow only one of the two ultimate reaction products from steam
reforming (H2 or CO2), to be removed. The reaction rate enhancement
that results is not however, sufficient, to achieve significant reforming
at such low temperatures.
The outline of this dissertation is as follows. Chapter 1 discusses
the motivation for the research. In Chapter 2, the HAMR concept and
the model developed for the HAMR hydrogen production process are
discussed. In Chapter 3, the hydrogen-selective membranes used in the
HAMR-hydrogen production process are studied. In this study,
hydrogen-selective porous carbon-molecular-sieve and dense palladium
membranes are used, and are characterized through single and mixed-
gas permeation measurements. The effects of temperature, pressure,
and composition on the permeation characteristics of these membranes
are also investigated. In Chapter 4, the transport characteristics and
adsorption of carbon dioxide in Mg-Al-CO3 hydrotalcite materials are
studied with the aid of gravimetric experiments in the temperature
range, 250–450 °C. For the diffusion coefficient, the transient CO2
uptake data are measured gravimetrically at each elevated temperature,
and then the diffusion coefficients are estimated by fitting the
14
experimental data to the solution of the relevant diffusion equation. The
desorption characteristic of the adsorbent are also discussed. In
Chapter 5, the kinetics of the WGS reaction on commercial Cu/Zn
catalyst and steam methane reforming (SMR) on commercial Ni catalyst
under the respective operating conditions are investigated. Rate
expressions exhibiting the most accurate fit to the experimental data
are chosen. In Chapter 6, the experimental results of the WGS and SRM
reaction using the HAMR concept are discussed and compared with the
HAMR model predictions. The sensitivity of the HAMR model to the
membrane and adsorbent properties is also discussed.
15
Chapter 2
HAMR Overview and Model
2.1 Introduction
In this chapter, the HAMR concept and its application towards
the development of a novel hydrogen production process through the
equilibrium-limited SRM and WGS reaction processes are discussed.
The mathematical model describing the HAMR hydrogen production
process is described and discussed in detail.
2.2 The HAMR
Le Chatelier’s principle states that for the equilibrium-limited
reactions, the conversion of the reactants to the products and the rate
of the overall reaction can be enhanced by selectively removing the
reaction products from the reaction zone (Gluud et al. 1931).
Applications of this concept have been advanced in fixed-bed,
continuous countercurrent moving bed, and simulated countercurrent
moving bed chromatographic reactors. A detailed review of the literature
is given by Carvill et al. (1996).
Recently, the application of the Le Chatelier’s principle for the
hydrogen production process has been investigated by various groups
(Balasubramanian et al. 1999; Hufton et al. 1999; Waldron and Sircar
2000; Ding and Alpay 2000a; Ding and Alpay 2000b; Ding and Alpay
16
2001; Ortiz and Harrison 2001; Xiu et al. 2002; Xiu et al. 2002 ; Xiu et
al. 2003a; Xiu et al. 2003b; Xiu et al. 2004).
In this work, a novel reactor system, HAMR is proposed and
investigated for hydrogen production. The HAMR concept, originally
proposed by Park and Tsotsis et al. (1998, 2004), couples the reaction
and membrane separation steps with adsorption on the reactor and/or
membrane permeate side. The HAMR system investigated previously
involved a hybrid pervaporation MR system, and studied the reaction
and pervaporation steps through a membrane with water adsorption.
Coupling reaction, pervaporation, and adsorption significantly improved
the performance. Most recently, Elnashaie and co-workers (Prasad and
Elnashaie 2002; Prasad and Elnashaie 2003; Prasad and Elnashaie
2004; Chen and Elnashaie 2004b)
analyzed mathematically the
behavior of a circulating fluidized-bed system utilizing Pd membranes.
Their reactor was assumed to operate at steady state by re-circulating
the catalyst and adsorbent through a second reactor for regeneration.
The ability of Pd membranes to withstand the extreme conditions of the
fluidized-bed steam-reforming environment and of the adsorbents to
undergo continuous recirculation and regeneration still remains the key
challenges.
The HAMR system can be potentially used with equilibrium- or
selectivity-limited reactions, in which one of the products can be
17
adsorbed while another (or the same) product can be simultaneously
removed via a membrane. What limits the application of the concept is
the availability of efficient adsorbents that are stable at the reaction
conditions. Esterification reactions (such as the ethanol reaction with
acetic acid to produce ethyl acetate, previously studied by Park and
Tsotsis 2004), through the use of water adsorbents, and the production
of hydrogen (through steam-reforming or the WGS reactions) are two
key potential applications. The HAMR system promises to exhibit
behavior that is more advantageous than either the MR or AR, in terms
of the attained yields and selectivities.
In addition, the HAMR system potentially allows for significantly
greater process flexibility than either the MR or AR system. The
membrane, for example, can potentially be used to separate the catalyst
from the adsorbent phase, thus allowing for in-situ continuous
regeneration of the adsorbent (though this is not the configuration
studied in this Thesis). This offers, potentially, a significant advantage
over the AR, which is, by definition, batch system and requires the
presence of multiple beds (one being in operation while the other is
being regenerated) to simulate continuous operation. The HAMR system
shows, furthermore, significant potential advantages with respect to the
conventional MR system. Beyond the improved yields and selectivities,
the HAMR system has the potential for producing a CO-free fuel-cell-
18
grade hydrogen product, which is of significance for the proposed fuel-
cell-based mobile applications of such systems. The additional benefits
include cost reduction by replacing the high temperature, expensive
alloyed steel construction for the SRM reactor with less expensive
stainless steel construction. They may also involve reduced costs
associated with the downstream separation processes, such as PSA for
hydrogen purification, due to the significant reduction on the CO and
CO2 content of the SMR process effluent. The milder operating
conditions may also minimize or eliminate the coke formation in the
reactors. Table 2.1 shows the comparison of the operating
characteristics of the conventional SMR process and the proposed
HAMR process.
Conventional HAMR
Temperature 800-1000 °C 400-500 °C
CH4 Conversion 80-85 % 99+ %
Product Pressure 200-400 psig 20-100 psig
H2O/CH4 - in feed 5-10 3-4
Table 2.1 Comparison of conventional (Twigg and Spencer 2001) and
HAMR operating conditions
The spent/used adsorbent can be regenerated by two ways:
19
(1) By heating the adsorbent to high temperatures using hot gas.
At high temperatures the adsorbent’s capacity is reduced, and
impurities are desorbed and removed from the adsorption unit. This is
an efficient way of regenerating the adsorbent. It is, however, limited by
the number of cycles that can be run over a certain period of time, since
heating and cooling are relatively slow processes. For this reason, to
date, the temperature-swing processes are limited to the removal of
small quantities of strongly adsorbed impurities.
(2) The second potential way of regenerating the spent adsorbent
involves the pressure-swing process. Here, simply by reducing the
operating pressure at constant temperature, the impurities are
desorbed, and the adsorbent is regenerated. Since pressure-swing is
more convenient to operate, a much more rapid cycling can be achieved
and larger quantities of impurities can be removed. Figure 2.1
illustrates both the pressure- and temperature-swing processes.
20
Figure 2.1 PSA and TSA operation diagram (Stöcker et al. 2005)
2.3 Proposed hydrogen production by the HAMR process
Hydrogen production can be conducted at 400-500 ºC with the
proposed HAMR process, as opposed to 800-1000 ºC in the
conventional hydrocarbon reforming. Figure 2.2 shows a schematic of
the HAMR hydrogen production process through reforming of
hydrocarbons on commercial Ni catalyst with a hydrogen selective
membrane, and hydrotalcite material as the CO2 adsorbent.
21
Figure 2.2 HAMR-hydrogen production process
Membrane reactor technology and the sorption-enhanced- PSA-
type processes proposed in the literature (Estrin and Roizenma.R. 1972;
Nagamoto and Inoue 1985; Kikuchi et al. 1989; Shu et al. 1994b; Noble
and Falconer 1995; Basile et al. 1996; Carvill et al. 1996; Hsieh 1996;
Hufton et al. 1999; Sircar et al. 1999; Ward and Dao 1999; Kikuchi
2000; Nam et al. 2000; Ding and Alpay 2000b; Itoh et al. 2001; Ortiz
and Harrison 2001; Xiu et al. 2002; Karagiannakis et al. 2003; Tosti et
al. 2003; Basile et al. 2003a; Gallucci et al. 2004; Gallucci et al. 2004;
22
Jordal et al. 2004; Chen and Elnashaie 2004a; Ferreira-Aparicio and
Benito 2005; Kim et al. 2005; Kulprathipanja et al. 2005; Basile et al.
2005a; Basile et al. 2005b; Adhikari and Fernando 2006; Chen et al.
2006; Chiappetta et al. 2006; Kusakabe et al. 2006; Reijers et al. 2006;
Tsuru et al. 2006; Brunetti et al. 2007) allow only one of the reaction
products, such as H2 and CO2, respectively, to be removed. The
conversion enhancement that results is not sufficient. Use of dual
adsorbents can technically achieve an effect similar to that in the
HAMR-process; however, adsorbent regeneration and collection of the
products separately are way more complex than proposed
membrane/adsorbent hybrid processes. Additionally, the proposed
HAMR-reforming process offers two unique features that are of value for
the fuel cell applications of the technology,
The reverse reforming reaction (i.e., methanation) is minimized;
thus, the methane content in the final product is very low (<2%).
The WGS reaction is highly promoted; thus, extremely low levels
of CO (<3 ppm) may be attainable in the final product.
Both features address the key concerns about the specific fuel quality
requirement for the existing fuel cells (PEM), i.e., hydrogen purity and
CO contaminant levels.
Additionally, the proposed system offers potentially many process
advantages over the conventional reforming process. They include:
23
No separate WGS reactor, as an additional step for further
conversion of CO to H2, is required. The proposed ultra-low-
temperature steam reformer offers the right environment for
WGS in the same reactor, hence, completely eliminating the
need for a separate WGS reactor.
No post-treatment for CO clean-up is necessary. Under the
conditions of low-temperature reforming and the in-situ
removal of H2 and CO2, the CO level is expected to meet the
required specifics for fuel cell use.
All heating and cooling requirements involve the temperature
range of up to 400 to 500 °C vs. 600 to 800 °C in the
conventional catalytic reforming. Also, coke formation is,
potentially, inhibited at the low-temperature range of
operation.
In the HAMR process, the reactants, steam, methane (for the
reforming reaction) or carbon monoxide (for the WGS reaction), are fed
into a reactor containing a catalyst, an adsorbent for removing CO2, and
a hydrogen-permselective nanoporous membrane. Hydrogen-rich
product is obtained in the membrane permeate side during this step.
Once the adsorbent is saturated with CO2, it is regenerated in situ via
the principle of the PSA. The process is, therefore, cyclic, and at least
two reactors must be utilized for commercial application, each of which
24
is undergoing repetitive reaction/regeneration steps. Thus, a
continuous hydrogen-rich product is obtained from the system when
two or more beds are utilized.
The proposed steps for the direct production of essentially pure
H2 by the cyclic HAMR process are shown in Figure 2.3, and described
below:
1. Adsorption-reaction-membrane-separation step: The reactor is
initially pre-saturated with a portion of the product H2 and steam at the
desired reaction temperature and pressure. A mixture of steam and CO
(or methane for the steam reforming reaction) at a prescribed ratio is
then fed into the reactor, and an essentially pure H2 product is collected
at the permeate side. The reaction step is continued up to the time
period needed for CO2 breakthrough to occur through the adsorbent
bed. After this period, the product purity, reactor conversion, and
hydrogen recovery start to decline, and the HAMR’s full capabilities are
not utilized. When the H2 purity and recovery decrease to preset levels-
the time for this to happen depends upon the adsorbent’s diffusivity,
adsorption characteristics, and other reactor parameters- the feed is
diverted into a second identical reactor.
2. Blow-down step: During this step, the reactor is depressurized to
a lower pressure level PL, countercurrently to the feed flow direction.
The effluent gas stream contains all the components in the reactor,
25
while the stream is recycled as a feed to another reactor or to be used
as fuel.
3. Purge step: The reactor is countercurrently purged with a weakly
adsorbing gas, such as steam or hydrogen, to desorb the CO2. The
desorption step operates at atmospheric pressure (PL). The desorbed gas
consists of CO, CH4, CO2, H2, and H2O, and is either separated for
recycling of purge gas or used as fuel.
4. Pressurization step: The reactor is countercurrently pressurized
to the reaction pressure with a mixture of steam, and H2. At this point,
regeneration of the reactor is completed, so that it is ready to undergo a
new cycle.
26
Figure 2.3 Schematic illustrations of the HAMR process steps
Figure 2.4 shows schematic of the HAMR process for hydrogen
production process scheme for a 4-bed/4-step process. While one bed is
undergoing reaction-separation, the other three beds are undergoing
27
one of the regeneration steps. One needs to point out here that, for the
adsorptive reactors during the steps 3 and 4 of the regeneration cycle, it
is very important to purge and pressurize the reactor using the
reaction-product hydrogen. Application of non-product gases for the
purge and the pressurization steps reduces the purity of hydrogen
during the next reaction step. For the HAMR process, on the other hand,
the purge and pressurization gases can be gases other than the product
hydrogen, since the product is collected on the permeate side of the
reactor.
28
Figure 2.4 4-bed/4-step HAMR hydrogen production process
29
2.4 The mathematical model of the HAMR system
Figure 2.5 Schematic of the hybrid adsorbent-membrane reactor
30
In this part of the Thesis, the mathematical model developed for
the HAMR for the hydrogen production via the WGS reaction is
presented. A schematic of the HAMR system is shown in Figure 2.5. In
this figure the catalyst and adsorbent are packed in the exterior of the
membrane (signified by the superscript F, for the feed side), with
additional adsorbent also packed in the interior of the membrane
volume (signified by the superscript P, for the permeate side). There are,
of course, a number of other potential reactor configurations, as
previously noted. For example, the catalyst may be loaded in the feed
side, while the adsorbent may also be loaded on the permeate side, or
the catalyst and adsorbent may only be loaded in the feed side, with no
adsorbent or catalyst being present in the permeate side, which is the
configuration that is analyzed here. To simplify matters, we assume
that the reactor operates isothermally, that external mass transfer
resistances are negligible for the transport through the membrane as
well as for the catalysts, and that internal diffusion limitations for the
catalyst, and internal or external transport limitations for the adsorbent,
are accounted for by the overall rate coefficients. Moreover, plug-flow
conditions are assumed to prevail for the interior and exterior
membrane volumes, as well as the ideal gas law conditions.
In the simulations performed, the experimentally measured
transport characteristics of a carbon molecular-sieve (CMS) membrane
31
are utilized. These membranes have been shown to be thermally and
hydrothermally stable under conditions akin to the WGS reaction. Mass
transfer through the porous membrane is described by the following
empirical equation:
F
j
= U
j
(P
j
F
! P
j
P
) (1)
where Fj is the molar flux (mol/m
2
.s), the partial pressure of
component j on the membrane feed side (bar), the partial pressure of
component j on the membrane permeate side (bar), and Uj the
membrane permeance for component j (mol/m
2
.bar.s). Equation 1 is, of
course, a simplified empirical expression for describing flux through a
nanoporous membrane for which the size of the pores approaches that
of the diffusing molecules. Simple analytical expressions for describing
transport through such membranes are currently lacking, however;
hence, the choice of the commonly utilized empirical Equation 1 in this
preliminary reactor modeling investigation.
The mass balance on the feed-side of the reactor packed with
WGS catalyst and, potentially, an adsorbent is described by the
following equations for CO2, CO, H2, H2O, and an inert species
(potentially used as a sweep gas or a blanketing agent - for catalytic
SMR or WGS a practical sweep gas would be either steam or hydrogen,
however):
32
!
F
"C
j
F
"t
+
"n
j
F
"V
=#$
m
U
j
(P
j
F
# P
j
P
)+ 1#!
b
F
( )
%
c
&
c
R
j
F
# 1#!
b
F
( )
(1#%
c
)&
a
G
j
F
+!
b
F
A
F
( )
2
"
"V
D
L
F
"C
j
F
"V
#
$
%
&
'
(
; j=1,2, ..., n (2)
In the above equation is the molar flow rate (mol/s) for species j, and
is the gas phase concentration (kmol/m
3
) equal to
(n
j
F
/Q
F
), where
is the volumetric flowrate (m
3
/s). V is the feed-side reactor volume
variable (m
3
), A
F
the cross-sectional area for the reactor feed side (m
2
),
m the membrane area per feed-side reactor volume (m
2
/m
3
), the bed
porosity on the feed side, the total feed side bed porosity (which
includes the bed and the catalyst porosities), the fraction of the solid
volume occupied by the catalysts (
!
c
= 1 , when no adsorbent is
present ), and the catalyst and the adsorbent densities (kg/m
3
),
and the reaction rate expression, which is either described by
individual component rate expression, or is equal to zero if j is an inert
species. Assuming that the adsorbent adsorbs only CO2, is zero for
all other components except CO2. (m
2
/s) is the axial dispersion
coefficient, given by the following equation generally applicable for
describing dispersion phenomena through packed beds (Edwards and
Richardson 1968):
33
D
L
F
= 0.73D
m
F
+
0.5u
F
d
P
F
1+ 9.49
D
m
F
u
F
d
P
F
(3)
where (m
2
/s) is the molecular diffusivity , (m/s) is the flow
velocity on the feed side, and (m) is the particle diameter on feed
side.
One finds a number of approaches in the literature for describing
. Ideally, one would like to account explicitly for both external and
internal mass transport, and finite rates of adsorption. Such an
approach goes beyond the scope of this work, however, in addition to
the fact that there are currently no experimental high-temperature
transport/adsorption CO2 data to justify such a level of mathematical
detail. Traditionally, in the modeling of adsorptive reactors, simpler
models have been utilized, instead (Ding and Alpay 2000a; Ding and
Alpay 2000b). Two such models have received the most attention. They
are, (i) the model based on the assumption of instantaneous local
adsorption equilibrium (ILE) between the gas and the adsorbent phases
(Ding and Alpay 2000a; Ding and Alpay 2000b), and (ii) the linear
driving force models (LDF), according to which (Karger 1992) is
described by the following expression:
dC
s
dt
= G
CO
2
F
= k
a
(C
s
eq
! C
s
) (4)
34
where is the adsorption equilibrium CO2 concentration (mol/kg) on
the adsorbent corresponding to the prevailing gas-phase concentration,
is the existing adsorbed CO2 concentration (mol/kg), and ka (s
-1
) is a
parameter which “lumps” together the effects of external and
intraparticle mass transport and the sorption processes, and which, as
a result, is often a strong function of temperature and pressure –
though, typically, in modeling it is taken as temperature/pressure
independent. To calculate , we utilize the data obtained for CO2
adsorption on layered-double hydroxide (LDH). The CO2 adsorption data
has been found to follow a Sips adsorption isotherm under both dry and
wet conditions, described by the following equation (Tien 1994):
C
s
eq
=
m
CO
2
(b
CO
2
P
CO
2
)
1/ n
1+ (b
CO
2
P
CO
2
)
1/ n
(5)
where (mol/kg) is the total adsorbent capacity, and (bar
-1
) the
adsorption equilibrium constant, which is described by the van’t Hoff
equation:
b
CO
2
(T )= b
CO
2
(T
0
)exp !
Q
R
"
#
$
%
&
'
1
T
!
1
T
0
"
#
$
%
&
'
(
)
*
*
+
,
-
-
1
n
=
1
n(T
0
)
+. 1!
T
0
T
"
#
$
%
&
'
(6)
35
Here,
b
CO
2
(T
0
) and
n(T
0
) are the affinity constant and the exponent at
some reference temperature , ! is a constant, and is a measure of
the heat of adsorption. Equations (2) and (4) must be complemented by
the initial and boundary conditions. For simplicity, we assume that the
reactor, prior to the initiation of the reaction/adsorption step, has
undergone a start-up procedure as described by Ding and Alpay (2000),
which involves, (i) heating up the reactor to the desired temperature
under atmospheric pressures by feeding H2 in the reactor feed side and
the chosen sweep gas on the permeate side; (ii) supplying water to the
system so that the feed H2O:Ar ratio is the same with the H2O:CO ratio
to be used during the reaction step; (iii) pressurizing the feed and
permeate side to the desired pressure conditions, and (iv) switching
from Ar to CO to initiate the reaction/adsorption step. In the
simulations the conditions prevailing at the start of step (iv) are those
prevailing at steady state during step (iii). In addition, during step (iv)
the following conventional boundary conditions prevail (Xiu et al. 2002;
Xiu et al. 2002; Xiu et al. 2003a; Xiu et al. 2003b; Xiu et al. 2004):
@ V = 0;
!x
j
F
!V
= -
u
0
F
(x
j0
F
" x
j
F
)
A
F
#
b
F
D
L
F
(7)
@ V = V
R
;
!x
j
F
!V
= 0 (8)
36
where (m/s) is the inlet superficial velocity, (m
3
) the total reactor
volume, the mole fraction and the inlet mole fraction for species
j.
Assuming that the catalyst and adsorbent particles have the
same size, the pressure drop in a packed bed can be calculated using
the Ergun equation:
!
dP
F
dV
= 10
!6
f
F
G
m
F
( )
2
A
F
g
c
d
P
F
"
F
F
(9)
at V=0, P
F
= P
0
F
(10)
f
F
=
1!"
b
F
"
b
F
( )
3
#
$
%
%
%
&
'
(
(
(
1.75+
150 1!"
b
F
( )
N
Re
F
#
$
%
%
&
'
(
(
(11)
N
Re
F
< 500 1!"
b
F
( )
(12)
N
Re
F
=
d
P
F
G
m
F
µ
F
(13)
where P
F
(bar) is
the feed side pressure, the inlet feed side pressure,
µ
F
(Pa.s)
the viscosity, (m) the particle diameter on the feed side,
G
m
F
= !
F
F
u
F
(kg/m
2
.s) the superficial mass flow velocity on the feed side,
(kg/m
3
)the density of the fluid, and the gravity conversion factor
equal to one in the SI units.
Since the CMS and composite palladium membranes do not
exhibit substantial CO2 permeation, we assume that no adsorbent or
37
catalyst is present on the permeate side, for which the following
equation is utilized:
!C
j
P
!t
+ k
!n
j
P
!V
="
m
kU
j
(P
j
F
# P
j
P
)+ A
F
( )
2
!
!V
D
L
P
!C
j
P
!V
$
%
&
'
(
)
, j=1,2,...,n (14)
where, , (m
2
) being the cross-sectional area on the permeate
side, and (m
2
/s) is the axial Taylor-Aris dispersion coefficient
(Levenspiel 1998) on the permeate side for empty tubes given as:
D
L
P
= D
m
P
+
u
P
( )
2
d
t
P
( )
2
192D
m
P
(15)
where (m
2
/s) is the molecular diffusivity, (m/s) is the velocity at
the permeate side and (m) is membrane inside diameter. In the
simulations, the conditions prevailing on the permeate side at the start
of step (iv) are those prevailing at steady state during step (iii). In
addition, during step (iv) the following conditions prevail on the
permeate side,
@ V = 0;
!x
j
P
!V
= -
u
0
P
(x
j0
P
" x
j
P
)
A
F
D
L
P
(16)
@ V = V
R
;
!x
j
P
!V
= 0 (17)
where is the mole fraction, the inlet mole fraction for species j on
the permeate side, and (m/s) the superficial flow velocity at the inlet.
38
Since no adsorbent or catalyst is present on the permeate side, we
ignore any potential pressure drops.
The reactor conversion (based on CO, which is typically the
limiting reagent fot the WGS reaction) is defined by:
X
CO
=
n
CO
0
F
! n
CO,ex
F
+ n
CO,ex
P
( )
n
CO
0
F
(18)
where is the inlet molar flow rate of CO, and , and are the
CO molar flow rates (mol/s) at the exit of the reactor feed and permeate
side, correspondingly. The yield of hydrogen product, defined as the
fraction of moles of CO fed into the reactor that have reacted to produce
hydrogen, is given by the following equation:
Y
H
2
=
(n
H
2
,ex
F
! n
H
2 0
F
)+ (n
H
2
,ex
P
! n
H
2 0
P
)
n
CO
0
F
(19)
where and are, respectively, the hydrogen molar flow rates at
the exit of the reactor feed and permeate side, and and (mol/s)
the H2 molar flow rates potentially present at the inlet of the reactor
feed and permeate sides. When all the CO has reacted completely
reacted to produce CO2
and H2, then .
Equations (1) - (19) are now written in dimensionless form by
defining the following variables and groups:
39
!
"
= (k
a
)
#1
;
!
F
=
"
F
V
R
A
F
u
0
F
;
! =
"
F
"
#
;
! =
V
V
R
;
u
F
=
Q
F
A
F
; ;
!
F
=
u
F
u
0
F
;
!
P
=
u
P
u
0
P
;
; ;
! =
P
0
P
P
0
F
;
!
j
=
MW
j
MW
H
2
;
x
j
F
=
P
j
F
P
F
;
x
j
P
=
P
j
P
P
P
;
! = k
a
t ;
!
j
=
U
j
U
H
2
;
K
eq1
'
=
K
eq1
(P
0
F
)
2
;
K
CO
'
= K
CO
P
0
F
;
K
H
2
'
= K
H
2
P
0
F
;
Da =
!
c
1"#
b
F
( )
$
c
k
1
(T
0
) V
R
RT
A
F
u
0
F
(P
0
F
)
1.5
;
!
CO
2
= (b
CO
2
P
0
F
)
1/n
;
Pe =
A
F
u
0
F
U
H
2
V
R
!
m
RT
;
!
F
=
"
b
F
A
F
D
L
F
u
0
F
V
R
;
!
P
=
A
F
D
L
P
u
0
P
V
R
Ha =
1! "
c
( )
1!#
b
F
( )
V
R
$
a
k
a
RTm
CO
2
A
F
u
0
F
P
0
F
;
! =
Ha
Da
;
! = (Da)(Pe) ;
! = 10
"6
f
F
(u
0
F
)
2
MW
H
2
V
R
A
F
g
c
d
p
F
RT
;
! =
A
P
u
0
P
A
F
u
0
F
;
!
seq
F
=
C
seq
F
m
CO
2
;
!
s
F
=
C
s
F
m
CO
2
;
The dimensionless equations equivalent to Eqs. (2) – (19) are
!
"x
j
F
"#
+
! x
j
F
$
F
"$
F
"#
= % &
F
"x
j
F
"'
+ x
j
F
"&
F
"'
+
x
j
F
&
F
$
F
"$
F
"'
(
)
*
*
+
,
-
-
%
Da.
j
/
x
j
F
% x
j
P
0
$
P
$
F
(
)
*
+
,
-
+Da
1
!
F
R'
j
F
" Da#
1
!
F
G '
j
F
+$
F
%
2
x
j
F
%&
2
+ 2$
F
1
!
F
%x
j
F
%&
'
(
)
*
+
,
%!
F
%&
'
(
)
*
+
,
, j=1,2,...,n-1(20)
!"
F
!#
= $
"
F
%
F
!%
F
!#
$
Da
&
'
j
x
j
F
$ x
j
P
(
%
P
%
F
)
*
+
,
-
.
j
/
+ Da
1
%
F
R'
j
F
j
/
$0Da
1
%
F
G '
CO
2
F
$
1
t
k
a
V
R
A
F
u
0
F
%
F
!%
F
!2
(21)
40
!
"
F
#k
$x
j
P
$%
= & '
P
$x
j
P
$(
+ x
j
P
$'
P
$(
+
x
j
P
'
p
)
P
$)
P
$(
*
+
,
,
-
.
/
/
+
Da0
j
#1
x
j
F
)
F
2)
P
& x
j
P
*
+
,
-
.
/
+
!
P
"
2
x
j
P
"#
2
+ 2!
P
1
$
P
"x
j
P
"#
%
&
'
(
)
*
"$
P
"#
%
&
'
(
)
*
, j= 1,2,...,n+ 1 (22)
!"
P
!#
=$
"
P
%
P
!%
P
!#
+
Da
&'
(
j
x
j
F
%
F
)%
P
$ x
j
P
*
+
,
-
.
/
j
0
(23)
!"
F
!#
=$%(&
F
)
2
"
F
x
j
F
'
j
(
(24)
d!
S
F
d"
=!
S
eq
F
#!
S
F
(25)
X
CO
=
x
CO
0
F
! x
CO
F
"
F
#
F
)
ex
+ (x
CO
P
$"
P
%#
P
( )
ex
x
CO
0
F
(26)
Y
H
2
=
(x
H
2
F
!
F
"
F
)
ex
# x
H
2 0
F
+ (x
H
2
P
$!
P
%"
P
)
ex
# x
H
2 0
P
$%
x
CO
0
F
(27)
where, in dimensionless form
G '
CO
2
F
= (!
seq
F
"!
s
F
) (28)
!
seq
F
=
"
CO
2
(x
CO
2
F
#
F
)
1/ n
1+"
CO
2
(x
CO
2
F
#
F
)
1/ n
(29)
and R’j are dimensionless forms of Ri. Equations (21) and (23) that
express the dimensionless velocity distributions are obtained by the
overall mass balances in the feed and the permeate side. In the
absence of substantial pressure drop in the permeate side in Equation
41
(22),
!
P
= 1 , and
!"
P
!#
= 0 . The initial condition at the start of the
adsorption/reaction step is that prevailing during step 3 (start-up
procedure as described by Ding and Alpay (2000a))
previously described.
The final distributions of concentrations and pressure along the reactor
column for one step are the initial conditions for the next step.
Equation (30)-(35) express the boundary conditions used for the
reaction-separation step:
! > 0 ; @"= 0; #
F
= 1 ; #
P
= 1 (30)
!
F
= 1 ; !
P
= 1 (31)
!x
j
F
!"
= -
1
#
F
(x
j0
F
$ x
j
F
); i= 1, 2, ...,n (32)
!x
j
P
!"
= -
1
#
P
(x
j0
P
$ x
j
P
); i= 1, 2, ..., n (33)
! > 0; @ "= 1;
#x
j
F
#"
= 0 (34)
!x
j
P
!"
= 0 (35)
Table 2.2 shows the corresponding boundary conditions for the
regeneration steps involved. The interbed pressure dynamics for the
variable pressure steps are estimated using a linear dynamic pressure
(Sereno and Rodrigues 1993; Malek and Farooq 1997).
42
!P
F
!t
= 0 for step (1)
!P
F
!t
= " P
H
" P
L
( )
t
2
for step (2)
!P
F
!t
= 0 for step (3)
!P
F
!t
= P
H
" P
L
( )
t
4
for step (4)
(36)
The system of coupled non-linear partial differential Equations
(20)- (25) and the associated boundary conditions have been solved in
MATLAB using the method of lines (MOL) (Schiesser 1991; Wouwer et al.
2004). The system of partial differential equations was converted to a
set of ordinary differential equations by discretizing the spatial
derivative in the !- direction using a five-point biased upwind finite-
differences scheme to approximate the convective term. A fourth-order
central differences scheme was used to approximate the diffusive term.
For the finite difference discretization, the reactor volume was divided
into n sections with (n+1) nodes. The initial-value ordinary differential
equations and other explicit algebraic equations at a time were
simultaneously solved using ’ode45.m’, a MATLAB built-in solver for
initial-value problems for ordinary differential equations
43
Blow-down Purge Pressurization
@ V = 0
!x
j
F
!V
= 0
P= P(t)
*
!x
j
F
!V
= 0
P= P
L
!x
j
F
!V
= 0
u= 0
@ V = V
R
!x
j
F
!V
= 0
u= 0
!x
j
F
!V
=
u
3,0
F
(x
j0
F
" x
j
F
)
A
F
#
b
F
D
L
F
u=!u
3,0
!x
j
F
!V
=
u
4,0
F
(x
j0
F
" x
j
F
)
A
F
#
b
F
D
L
F
P= P(t)
*
* Equation 36 is used
Table 2.2 Boundary conditions for the regeneration steps
44
Chapter 3
Membrane Studies
3.1 Introduction
CMS based membranes represent a “next generation” membrane
material, because they vastly improve on the limits of the selectivity vs.
permeability relationship set by the polymeric membranes for gas
separations. The CMS membranes have been studied since the early
1980’s, with the current emphasis being on the scale-up and their use
at the field scale.
Palladium has been known for its permeance to H2 for well over a
century. However, its embrittlement due to the !"# phase
transformation, poisoning by sulfur compounds (Adhikari and
Fernando 2006; Guazzone et al. 2006), and the cost have inhibited its
wide industrial use as a membrane material for hydrogen separations.
Furthermore, degradation of the pure metal due to exposure to O2 and
H2 cycling was reported recently (Roa and Way 2003a). Recent
development has focused on the use of Pd-Cu alloy to alleviate such
problems (Pavlova et al. 1972; Fort and Harris 1975; Suzuki and
Kimura 1984; Ermilova et al. 1988; Shu et al. 1991; Uemiya et al. 1991;
Berseneva et al. 1993; Shu et al. 1993; Suzuki and Kimura 1993;
Timofeev et al. 1994; Shu et al. 1995; Aoki et al. 1996; Basile et al.
45
1996; Paglieri et al. 1999; Jung et al. 2000; Tosti et al. 2000; Hollein et
al. 2001; Chen et al. 2002; Paglieri and Way 2002; Roa et al. 2002;
Tosti et al. 2002; Ma et al. 2003; Kulprathipanja et al. 2004; Sholl and
Ma 2006; Thoen et al. 2006; Iyoha et al. 2007; Ma and Guazzone 2007).
Currently, the use of Pd alloys offers an attractive and viable alternative
for the practical implementation of the palladium membrane for
hydrogen separations. Dense Pd (or Pd alloy) foil (≥25 µm) membrane
separators are too costly, due to the low permeate flux and high
material cost. Most recent research has focused on the development of
thin dense Pd films supported on, (i) Group V metals, such as tantalum,
vanadium, etc., and (ii) porous inorganic substrates, such as stainless
steel, metal oxides, etc. The former configuration has demonstrated
excellent permeance and selectivity at 300 ºC (Shu et al. 1993; Shu et
al. 1994b; Mardilovich et al. 2002; Rothenberger et al. 2004; Tong and
Matsumura 2006). However, at higher temperatures, i.e., 400 ºC,
interlayer diffusion between the Pd and the metallic substrate led to
deterioration of the H2 permeance, particularly for the alloys. The latter
phenomenon is not a serious concern with porous ceramic substrates.
Membranes that are based upon Pd-Cu/Al2O3 for H2 separations up to
450 ºC have been reported recently. However, the thermal mismatch
and the lack of chemical bonding between the palladium and the
ceramic sublayer remain major concerns (Li et al. 1999; Li et al. 1999a;
46
Li et al. 1999b; Li et al. 2000; O'Brien et al. 2001; Sun et al. 2006). The
test results reported in the literature mostly involve a consistent H2 feed
at a constant temperature, and require a very meticulous start-up and
shutdown procedure.
In this chapter, a brief overview of the membrane separation
terminology and various hydrogen selective membranes is provided. In
our HAMR-hydrogen production studies we have considered and
investigated the CMS and palladium membranes, prepared by Media &
Process Technology, Inc. These membranes are characterized through
single and mixed-gas permeation tests under the WGS and SRM
conditions. Part of the work in this Chapter was done in collaboration
with my colleague Hyun Hwang.
3.2 Overview on membrane separation
A membrane is a barrier that permits selective mass transport
between two phases (Baker 2004). It is selective typically because some
components can pass through the membrane more easily than others.
This makes membranes a suitable means to separate a mixture of
components. The phases on either side of the membrane can be liquid
or gas. Although we may not be aware of their presence, membranes
play an important role in life. Probably the best-known example of a
membrane is the human skin (Baker 2004; Kluiters 2004). The skin
permits selective transport of both gases and liquids (e.g., water can not
47
flow in but it can flow out when sweating). Figure 3.1 illustrates the
nomenclature used for membrane processes. The two sides of the
membrane are called the feed (or upstream side) and permeate sides (or
downstream). In practice, permeation can take place in both directions.
Generally speaking, the feed and permeate sides are chosen consistent
with the rule that the permeation of the (most) relevant species takes
place from the feed to the permeate side. The feed side flow is initially
called the feed flow. The flow resulting after permeation is called
retentate (or residue) flow. On the permeate side the inlet flow is called
sweep flow and the exit flow the permeate flow.
Figure 3.1 Membrane unit nomenclatures
Performance and efficiency of membranes are usually measured
in terms of flow (or flux) through the membrane and membrane
selectivity towards the mixtures. The flow can be measured in volume or
mass per unit time. The selectivity is a measure of the differences in the
permeabilities (the relative ease with which species can permeate
through the membrane) of different components. In other words, it is a
Membrane
Feed Side
Permeate Side
Sweep
Feed
Retentate
Permeate
48
measure of the membrane’s separation effectiveness. The selectivity
factor of two components A and B in a mixture is defined by:
!
A B
=
y
A
y
B
x
A
x
B
(1)
where yA and yB are the fractions of components A and B in the
permeate, and xA and xB are the fractions of the same components in
the feed. A, and B are usually chosen in such a way that the selectivity
factor is greater than unity. If the selectivity factor is equal to one, there
is no separation. The higher the selectivity factor, the more selective the
membrane is to a certain species (which is usually a desirable
membrane property). Two other important quantities to describe
membrane performance are the recovery and volume reduction. The
recovery or yield (S) is the fraction of the feed flow passing through the
membrane:
S =
q
p
q
f
(2)
where qp is the permeated flow and qf is the feed flow. The volume
reduction (VR) is the ratio of the initial feed flow rate and the retentate
flow rate:
VR=
q
f
q
r
(3)
49
3.3 Membrane performance evaluation
In what follows, we described how the performance of a membrane is
evaluated.
3.3.1 Membrane module design
The building block of a membrane system is called the module.
All module types applied in the industry, so far, are based on two types
of membrane configurations: flat and tubular. Module types based on
flat membranes are the plate-and-frame and spiral-wound modules.
Tubular membranes typically consist of a thin selective membrane layer
deposited on the inner- or outer side of a tubular support with a
diameter generally larger than 10 mm. The number of tubes put
together in the module may vary from 1 to 10, but is not limited to this
number.
Figure 3.2 shows the schematic of the module used in our studies.
A thermowell, bored through the module, is used to install the
thermocouples. The feed flows through the outer side of the membrane
tubes. Three thermocouples are installed to control the temperature in
the reactor. Two thermocouples are installed using the thermowell. A
third thermocouple is installed on the top, as shown in the figure. This
hole is also used to fill in the catalyst and adsorbent in the reactor.
50
Figure 3.2 Schematic of membrane reactor module
3.3.2 Membrane permeance calculations
The two most important characteristics of membranes are the
permeance and separation factor. Permeance is a measure of the gas
flow rate per unit area per unit pressure difference. A more
fundamental unit is the permeability, which is the permeance
multiplied by the thickness of the membrane. In most cases, the
thickness of the membrane is not known very accurately and, therefore,
so the permeance is a more practical quantity. The separation factor is
meaningful only with respect to a mixture of two gases. The transport of
51
gases through membranes behaves differently, as the pore diameter is
reduced. Gas transport can also be affected by temperature, and a
change in temperature can affect diffusion of gases differently in pores
of different diameters. However, measuring pore diameters that are
smaller than 2 nm is difficult. Therefore, it is critically important to be
able to follow the changes in the transport mechanisms of different
gases during pore diameter reduction, in order to help determine the
extent to which pores have been reduced.
The details of the experimental set-up used in the membrane
characterization studies are discussed in Chapter 6. Two types of
hydrogen selective membranes have been investigated, the CMS and
palladium membranes. In this study, tubular-type membranes were
used and characterized through single and mixed-gas permeation
measurements. The membrane was installed in the tubular stainless
steel reactor module, shown in Figure 3.2. Graphite seals along with
compression fittings were used to seal the area between the membrane
and reactor module.
The reactor module was installed in the furnace, and isothermal
conditions were maintained in the system with the help of three
temperature controllers and heating coils. The exit and inlet flow lines
were insulated to avoid water condensation in the lines. Pressure was
measured using two transducers located at the inlet and outlet for the
52
outer cell, which is called the feed side and another pressure
transducer, is located at the exit of the inner cell, called as membrane
side. Feed gas flow rates were controlled using mass flow controllers
and gas cylinder pressure regulators. A back-pressure regulator is used
to control the pressure on the feed/outer side. Atmospheric pressure is
maintained on the membrane/permeate side. The flow rates of the
retentate and permeated gases were measured using a soap bubble-flow
meter.
The compositions of the separated gases in mixed gas
measurements were analyzed using two on-line mass spectrometers,
installed each on the permeate and retentate sides. In order to prevent
the system from being contaminated from other gases, the feed and
permeate sides are swept using argon or nitrogen for more than 20 min.
The temperature in the system is raised under an argon or nitrogen
environment.
To determine the permeability and the separation factors of the
membranes used, a series of experiments were carried out for single,
binary, and quaternary gases. It has been already shown that the
separation of species through membranes can be different depending on
the way of feeding: single-gas or mixed-gas feed. In particular, when the
gases are fed singularly, the selectivities are higher than in the other
case. This is due to the fact that, by using mixed-gas feeds the flux
53
values are influenced by the multicomponent interactions, which
strongly impact the overall behavior. On the basis of these results, to
realistically simulate the process, the permeabilities and separation
factors of the membranes have been determined by feeding to the
reactor a mixture of gases containing all the species present during the
reaction (H2, CO, CO2, CH4, and H2O). The effect of temperature and
pressure on the membrane performance was also investigated.
The mixed-gas permeabilities (Uj) were obtained by
simultaneously solving the governing equations for flow through porous
membrane equation (1) from Chapter 2 for each component. The
permeabilities were fitted as parameters using either feed- or permeate-
side flow rates measured using the bubble flow meter and the partial
pressures, obtained using the exit stream composition analyzed with
the help of online mass spectrometers.
Table 3.1 presents the physical dimensions of the CMS and
palladium membranes tested, as well the operating conditions used in
this study.
54
Carbon molecular
sieve membranes
Palladium
membranes
Inner diameter (mm) 3.5 3.5, 6.75
Outer diameter (mm) 5.7 5.7, 9.0
Length (mm) 254 254, 100
Temperature (°C) 120-250 450
Pressure (Psi) 30-75 35-75
Gases Investigated H2, CO, CO2, CH4, H2O, Ar, N2
Table 3.1 Physical properties and experimental conditions investigated
3.3.2.1 CMS membrane data
Table 3.2 presents the single (H2, CH4, CO, CO2 ) and mixed-gas
(H2 : CO : CO2 : H2O) permeances and ideal separation factors (defined
as the permeance of hydrogen divided by the permeance of the
corresponding gas) obtained for the CMS membrane #1, which is also
used in the HAMR-WGS-reaction experiments reported in chapter 6.
The mixed-gas composition (H2: CO: CO2: H2O = 4:1:1:1) studied is also
shown in Table 3.2. Table 3.3 shows the corresponding data for another
membrane, CMS membrane #2. Note that the permeances of the
various gases measured during the mixed-gas experiments generally
remain close to the values measured during the single-gas experiments.
Table 3.4 shows the H2-CO2 binary permeance data as a function of
55
composition for a third membrane, CMS membrane #3. As can be seen
from Table 3.4, the carbon dioxide composition in the gas mixture has
significant effect on the hydrogen permeance. Normalised hydrogen
permeance (mixed-gas hydrogen permeance/pure gas hydrogen
permeance) was fitted as a function of H2/CO2 composition ratio on the
feed side. Figure 3.3 shows the normalized hydrogen permeance data
for pressures 45 and 75 Psi also shown is the equation obtained by
fitting the permeance data. This hydrogen permeance equation was
used during the modeling studies to incorporate the effect of CO2 on the
hydrogen permeance. As mentioned earlier, the membrane performance
may be different depending on whether the gases are fed, either on the
membrane layer or the support side. Since, during our HAMR
experiments we fed the gas on the membrane support side, the effect on
the membrane performance of feeding the gas on a different side was
also investigated. Table 3.5 shows a comparison between the permeance
data amongst experiments where the gases are either fed on the
support or the membrane layer sides, with data measured for another
membrane, CMS membrane # 4. As shown in this Table, there is no
significant difference in the experiments where the gases are fed on
different sides.
56
Pure Gas
250 °C /446.1 kPa (64.7 psi)
Mixed Gas
250 °C /446.1 kPa (64.7 psi)
Permeance
S. F.
based on H2
Permeance
S. F.
based on H2
H2 1.3446 1 1.2405 1.0
CO 0.0295 45.6 0.0232 53.46
CO2 0.083 16.2 0.1155 10.74
CH4
0.0064
209.7 - -
N2 0.0198 67.9 - -
Ar 0.0248 54.3 - -
H2O - - 0.3883 3.19
Permeance unit: m
3
/(m
2
*hr*bar)
Table 3.2 Single and mixed-gas permeance data for CMS membrane # 1
Pure Gas
250 °C /446.1 kPa (64.7 kPa)
Mixed Gas
250 °C /446.1 kPa (64.7 kPa)
Permeance
S.F.
based on H2
Permeance
S.F.
based on H2
H2 1.6847 1 1.6916 1.0
CO 0.0483 34.87 0.0579 29.21
CO2 0.1069 15.75 0.1182 14.31
N2 0.0352 47.86 - -
H2O - - 1.9005 0.89
Permeance unit: m
3
/(m
2
*hr*bar)
Table 3.3 Single and mixed-gas permeance data for CMS membrane # 2
57
Temperature = 250°C
Binary (H2+ CO2), Total Flow = 30 cc/sec
Gas Delta P
Mol
Fraction Permeance
psi m
3
/m
2
/hr/bar
H2 0 0
CO2
45.5
1 0.0286
H2 0.05 0.17
CO2
45.6
0.95 0.028
H2 0.1 0.1708
CO2
44.8
0.9 0.0289
H2 0.15 0.4308
CO2
45.3
0.85 0.0258
H2 0.2 0.5566
CO2
45.8
0.8 0.0262
H2 0.3 0.6439
CO2
46.4
0.7 0.0278
H2 0.4 0.6667
CO2
45.6
0.6 0.0308
H2 0.5 0.6838
CO2
45.5
0.5 0.0274
H2 0.7 0.6909
CO2
45.8
0.3 0.0289
H2 0.8 0.7133
CO2
44.7
0.2 0.0287
H2 0.85 0.7287
CO2
47.2
0.15 0.0281
H2 0.9 0.751
CO2
45.1
0.1 0.0251
H2 1 0.781
CO2
44
0 0
Table 3.4 H2-CO2 binary gas permeance data for CMS membrane # 3
58
Figure 3.3 Hydrogen permeance as a function of H2/CO2 ratio
Shell Side Feeding Tube Side Feeding
Pure Gas Pure Gas
250 C / 30 psig 250 C / 30 psig
Permeance S.F. Permeance S.F.
[ m
3
/(m
2
*hr*bar) ] based on H2 [ m
3
/(m
2
*hr*bar) ] based on H2
H2 1.1698 1.0 1.1672 1.0
CO 0.0282 41.5 0.0273 42.7
CO2 0.0496 23.6 0.0521 22.4
CH4
0.0223
52.5 0.0212 55.1
H2O 1.0653 1.1 0.1847 6.3
N2 0.0235 49.8 0.0251 46.6
Ar 0.0211 55.4 0.0217 53.8
Table 3.5 Comparison of permeance data for the feed side vs. the tube
side
59
Figures 3.4-3.10 summarize all the CMS membranes,
characterized through the single and mixed-gas permeation
experiments in this study. The single and mixed-gas separation factors
for various gases for all the membranes tested are plotted as a function
of hydrogen permeance. Appendix (B) shows the corresponding
operating conditions and all the data plotted in these figures.
Figure 3.4 H2/CO2 (single gas) vs. H2 permeance
0.
0
0.
5
1.
0
1.
5
2.
0
2.
5
3.
0
3.
5
4.
0
0
1
0
2
0
3
0
4
0
5
0
6
0
H
2
/C
O
2
H
2
Permeance
(m
3
/
m
2
/hr/bar
)
60
Figure 3.5 H2/CO2 (mixed gas) vs. H2 permeance
61
Figure 3.6 H2/CO (single gas) vs. H2 permeance
62
Figure 3.7 H2/CO (mixed gas) vs. H2 permeance
63
Figure 3.8 H2/ CH4 (single gas) vs. H2 permeance
64
Figure 3.9 H2/ CH4 (mixed gas) vs. H2 permeance
65
Figure 3.10 H2/ H2O (mixed gas) vs. H2 permeance
66
3.3.2.2 Palladium membrane data
The flux of the various species through a Pd membrane is
empirically described using the following equation:
F
j
= U
j
(P
F , j
n
! P
P, j
n
) (5)
The permeance U and the exponent n depend on the transport
mechanism that prevails. The exponent n is typically between 0.5 and 1.
For porous membranes a value of 1 is mostly used in calculations, since
the main transport mechanisms are Knudsen diffusion for mesoporous
membranes and molecular sieving for microporous membranes. For
hydrogen diffusion through metals, the Sievert's law with n = 0.5 is
often applicable. Table 3.6 presents the single and mixed-gas
permeances and ideal separation factors obtained for palladium
membrane #1. Table 3.7 indicates the corresponding mixed-gas data
with the same membrane for a different mixed-gas composition.
67
Pure Gas
450 °C /411.61 kPa (59.7 psi)
Mixed Gas
450 °C /411.61 kPa (59.7 psi)
H2 : CH4 : CO : CO2 : H2O
3.0 : 1.88 : 2.03 : 1.97 : 2.0
Permeance
S. F.
based on H2
Permeance
S. F.
based on H2
H2 16.1856 1 17.5213 1
CO 0.2486 65.106 0.1993 87.9
CO2 0.1578 102.6 0.1980 88.5
CH4
0.2242
72.19 0.2288 76.6
Ar 0.1578 102.6 - -
H2O - - 0.6905 25.4
Permeance unit: m
3
/(m
2
*hr*bar)
Table 3.6 Single and mixed-gas permeance data for palladium
membrane # 1
Mixed-gas composition: H2 : CH4 : CO : CO2 : H2O
8.0 : 8.05 : 8.24 : 7.96
T Pressure Permeance
(°C)
Gas
(psi) m
3
/(m
2
*hr*bar)
S.F.
H2 46.2 18.5214 1.0
CO 46.2 0.2512 73.7
CO2 46.2 0.2351 78.8
450
CH4 46.2 0.2339 79.2
Table 3.7 mixed-gas permeance data for palladium membrane # 1
Hydrogen permeation through Pd-based membranes is a complex
process, including hydrogen chemisorption, dissolution into the Pd
lattice, and diffusion in the bulk driven by the concentration gradient
68
and, finally, desorption from the opposite surface. The square-root
dependence of the permeation flux on the pressure originates from
Sieverts' law of hydrogen solubility, which states that the hydrogen
concentration in the metal is proportional to the square root of the
partial pressure of hydrogen on the metal surface. This relationship,
corresponding to an atomic hydrogen diffusion mechanism, has been
proven to hold at low hydrogen partial pressures and low temperatures,
and has sometimes been extended to elevated temperatures.
However, this extension has often been challenged in the
literature. In recent years, deviations from Sieverts' law have been
further confirmed with composite thin Pd membranes. Table 3.8
summarizes the literature data found for the pressure-dependence of
hydrogen flow through palladium and composite palladium membranes.
Figure 3.11 shows the hydrogen flux as a function of pressure across
the membrane for two different palladium membranes that were tested
(#1 and #2). The data in Figure 3.11 are fitted well using a value of the
pressure exponent, n=1, in Equation (5) and is shown in Figure 3.12.
69
Membrane
Pressure
kPa
Temperature
K
n-value reference
Pd 100-700 623-723 0.68
(Hulbert and
Konecny 1961)
Pd/glass 48-290 673 0.76
(Uemiya et al.
1991)
Pd/Al2O3 690 823 0.602
(Alefeld and Volkl
1978)
Pd/-Al2O3 100 573-773 1 (Yan et al. 1994)
Pd/Al2O3 100 673 0.6-0.7
(Xomeritakis and
Lin 1998)
Pd-Ag 0.13-3.33 623 1
(Amandusson et
al. 2001)
Pd77Ag23/Al2O3 250 623-773 0.5 (Guo et al. 2003)
Pd-Ag/-Al2O3 80-250 473-616 0.968 (Chen et al. 2004)
Pd63Cu37/silica/Ni-
PSS
13-79 723 1
(Nam and Lee
2001)
Pd90Cu10/ceramic 34-340 623-723 0.8-1
(Roa and Way
2003a)
Pd60Cu40/ceramic 34-218 623 0.515
(Roa and Way
2003b)
Pd46Cu54/ZrO2-
PSS
100-300 593-753 1 (Gao et al. 2005)
Pd84Cu16/ZrO2-
PSS
50-250 593-753 0.6 (Gao et al. 2005)
Pd/Al2O3 101-410 723 1 This study
Table 3.8 Summary of the literature data for the palladium membranes
70
Figure 3.11 Hydrogen flux as a function of trans-membrane pressure
71
Figure 3.12 Pressure-dependence of hydrogen permeance
Table 3.9 presents the effect of steam on the permeation behavior
of palladium membrane #1. It exhibits the experimental history over a
period of 24 h. Almost immediately after the start of the steam addition
(Time= 0-1 h.), a decrease in the hydrogen permeance was observed. It
is assumed that steam adsorbs at the membrane surface and causes a
reduction in the number of active sites for hydrogen. The reduction in
available sites leads to a reduction in hydrogen flux. After stopping the
steam addition (Time=1.83 h.), the hydrogen permeance increases and
72
levels off to a stable permeance value. The time required for stabilization
is equivalent to the time required to remove the water from the feed and
reactor section. Further exposure to steam (Time = 20.83 -22.83 h.)
indicates further decrease in hydrogen permeance. Membrane exposure
to hydrogen environment revealed positive impact on the hydrogen
permeance (Time=3.83 – 19.83 h.).
Time Action Taken H2 permeance
%
deviation
hr m
3
/m
2
/hr/bar
0
Steam added to
feed
16.66 0.00
1 Steam stopped 16.19 2.93
1.83
Under argon
environment
16.67 -0.03
3.83 Hydrogen started 16.85 -1.13
19.83
Under Hydrogen
environment
18.33 -9.09
20.83 Steam started 16.96 -1.77
21.83 Under steam 16.12 3.37
22.83 Under steam 15.85 5.11
24 Steam stopped 16.94 -1.65
Table 3.9 Effect of steam on the palladium membrane performance
73
Chapter 4
Adsorption Studies
4.1 Introduction
Recent growing interest in CO2 separation from flue-gas and other
gaseous effluents has provided the impetus to investigate CO2
adsorption under a wide range of operating conditions. Literature
studies (Carvill et al. 1996; Hufton et al. 1999; Sircar et al. 1999; Zou
and Rodrigues 2001a; Xiu et al. 2002; Yong and Rodrigues 2002; Xiu et
al. 2003a; Xiu et al. 2003b; Soares et al. 2004; Wang and Rodrigues
2005; Moreira et al. 2006; Lee et al. 2007; Soares et al. 2007), have
identified a large number of adsorbents for CO2 adsorption; these
include various metal oxides (e.g. CaO, MgO), alumina and metal-
promoted alumina, activated carbon, and numerous zeolites (e.g. 4A, 5A,
CrA CrX, CrY, RhA, 13X, and mordenites). Under relatively mild
operating conditions, the zeolite-based adsorbents have relatively high
adsorption capacities, even higher than those of activated carbon. At
higher temperatures of operation, these capacities decline rapidly,
however. In this study a hydrotalcite material was used, instead, as the
CO2 adsorbent. Hydrotalcite has been shown to be the most effective
CO2 adsorbent for the range of temperatures of relevance to the WGS
and reforming reactions. In this chapter, three commercial hydrotalcite
74
materials are investigated. The sorption characteristics and cyclic
reversibility of the materials under both dry and wet atmospheres are
studied as well.
4.2 Studies of CO
2
Adsorbents
Nowadays, due to stricter environmental regulations, the removal
and recovery of carbon dioxide from hot gas streams is becoming
increasingly significant in the field of energy production. A large amount
of carbon dioxide is released in the environment during the combustion
of fossil fuels, such as coal or natural gas, which has become one of the
most serious global environmental problems (Liu et al. 2002; Darwish et
al. 2004; Semelsberger et al. 2004). Recently, the separation of carbon
dioxide from gaseous mixtures has also received attention in other
areas, such as during natural gas treatment, the production of
hydrogen gas, and in the aerospace industry (Ding and Alpay 2000b;
Ding and Alpay 2001; Zou and Rodrigues 2001b; Yong et al. 2002; Yong
and Rodrigues 2002; Abanades et al. 2004). Several options for
reducing carbon dioxide emissions, including substituting nuclear
power for fossil fuels, increasing the efficiency of fossil fuel plants, and
capturing the carbon dioxide prior to emission into the environment,
have been discussed. All such techniques have the attractive feature of
limiting the amount of carbon dioxide emitted into the atmosphere, but
each has certain economical, technical, or societal limitations. The
75
removal and recovery of carbon dioxide from power plant fuel gases can
be an effective approach for reducing total carbon dioxide emissions in
the energy field (Ding and Alpay 2001; Yong et al. 2002; Abanades et al.
2004).
A number of techniques can be used for the separation of carbon
dioxide from hot gas streams. The large-scale separation of carbon
dioxide by absorption is a commercial operation used throughout the
world. Other techniques exist that could be considered for energy-
related applications, such as cryogenic and, membrane separation, and
adsorption processes, such as PSA, vacuum-swing adsorption (VSA),
and temperature-swing adsorption (TSA). PSA is well suited for the
removal and recovery of carbon dioxide from any hot flue-gas. The PSA
process can be operated at elevated temperatures, to remove most of
the carbon dioxide, and it overcomes the need for cooling the flue-gas to
ambient temperature prior to the removal of carbon dioxide. Therefore,
in the last two decades, active research efforts have been directed
towards the separation of carbon dioxide by the PSA process, which has
been industrialized. However, to use the PSA process for the removal
and recovery of carbon dioxide from hot flue gas streams at elevated
temperatures, the first and most important issue is to find the
appropriate adsorbent. The adsorbent must have, (1) high selectively
and adsorption capacity for carbon dioxide at high temperature; (2)
76
adequate adsorption/desorption kinetics under the operating
conditions; (3) stable adsorption capacity for carbon dioxide after
repeated adsorption/desorption cycles, and (4) adequate mechanical
strength after cyclic exposure to high-pressure streams (Carvill et al.
1996; Hufton et al. 1999; Sircar et al. 1999; Ding and Alpay 2000b; Zou
and Rodrigues 2001a; Yong et al. 2002; Yong and Rodrigues 2002).
Recently, it was reported that hydrotalcite-like compounds can well
meet the above requirements and, as such, they are one of the most
promising adsorbents for the sorption-enhanced reaction process
(SERP) for hydrogen production. Cost reduction of 25-30%, compared
with the conventional methane steam reforming, was reported (Hufton
et al. 1999; Sircar et al. 1999).
Hydrotalcites belong to a larger class of anionic and basic clays,
also known as layered double hydroxides (LDH). They are composed of
positively charged brucite-like [Mg(OH)2] layers, with trivalent cations
substituting for the divalent cations at the centers of octahedral sites of
hydroxide sheets, whose vertex contains hydroxide ions. Each OH group
is shared by three octahedral cations and points to the interlayer
regions. The excess positive charge of the material is compensated by
anions and water molecules present in the interstitial position. These
materials are receiving considerable attention in recent years, because
they are used in a wide range of applications, such as catalysts,
77
precursors and supports of catalysts, ion exchangers, filters,
decolorizing agents, industrial adsorbents, etc. The general formula for
the hydrotalcite can be described as [M(II)1-xM(III)x (OH)2]
x+
[A
n-
x/n mH2O].
The structure of these compounds is as shown in Figures 4.1 and 4.2.
Figure 4.1 Structural models for hydrotalcite-like compounds (Yong et
al. 2002)
78
Figure 4.2 The structure of hydrotalcite materials (Reijers et al. 2006)
4.3 Effect of Aluminum content of the hydrotalcite on the
adsorption capacity
In this study three commercial hydrotalcite materials were
investigated as a CO2 adsorbent. Table 4.1 shows the physical and
chemical properties of the materials, as provided by the manufacturer.
Figure 4.3 shows the effect of aluminum content of the hydrotalcite on
CO2 adsorption capacity at 250 ° C and 1 bar. The increase in the
adsorption capacity between MG70 and MG50 is attributed to the
increased charge density in the structure, due to increased aluminum
79
content. In subsequent investigations MG50 was chosen as the CO2
adsorbent.
Product Name PURAL MG30 PURAL MG50 PURAL MG70
Product No. 595030 595050 595070
Chemical data
MgO:Al2O3 % 30:70 50:50 70:30
L.O.I. % 40 max 45 max 45 max
Carbon % 0.5-3 0.5-3 0.5-3
SiO2 ppm 350 max 350 max 350 max
Fe2O3 ppm 200 max 200 max 200 max
Na, Ca, Ti each ppm 50 max 50 max 50 max
Physical Properties
Surface area* m
2
/g 250 min 200 min 180 min
Pore volume ml/g 0.5 min 0.2 min 0.2 min
Loose bulk density g/l 350-550 450-650 350-550
Particle size distribution
< 25 micro m % 20 min 20 min 30 min
< 45 micro m % 40 min 40 min 50 min
< 90 micro m % 85 min 85 min 90 min
* : surface area measured after 3 hr activation at 550C
Table 4.1 The physical and chemical properties of the hydrotalcites
investigated (Sasol 2003)
80
Figure 4.3 Adsorption capacity as a function of Al content of the HTC at
250° C and 1 bar
4.4 Characterization of the hydrotalcites
N2 adsorption at 77K was carried out with a Micrometrics ASAP
2010 apparatus. The adsorption-desorption isotherms, as shown in
Figure 4.4, are of Type II, according to the International Union of Pure
and Applied Chemists (IUPAC) classification. The hysteresis indicates
the presences of slit pores. The analysis of the N2 adsorbed at P/P0
below 0.01 was made according to the Horwath-Kazawo (HK) method,
with the results shown in Figure 4.5. The micropore sizes are in the
range 0.4-2.0 nm with a predominance of 0.75 nm pore width. A
detailed BET report for the adsorption studies is included in Appendix
81
(A).
Figure 4.4 Nitrogen adsorption/desorption isotherm in the hydrotalcite
at 77 K
82
Figure 4.5 HK plot for the hydrotalcite sample
4.5 Adsorption Equilibria
Equilibrium adsorption data were collected in-situ by thermogravimetric
experiments using a Cahn TGA 121 instrument. Before initiating the
sorption experiments, a fresh LDH material was loaded in the
instrument and degassed by heating it from room temperature to 450
°C, at a rate of 5°C/min under vacuum. The sample was kept at 450 °C
for about 4 h, and then cooled down to the intended temperature under
83
vacuum. When the temperature reached the preset point, the sorption
system was kept at the same condition for an additional 90 min in order
to stabilize the TGA microbalance. Only after the microbalance had
stabilized, was the inert flowing Ar gas allowed to fill the system until
atmospheric pressure was reached. Ar was utilized as the inert gas in
order to minimize the disturbances due to the change in the buoyancy
force, when the purge gas was switched to Ar-CO2 mixtures with
predetermined concentrations. The isotherm data for three
temperatures (250°, 350° and 450˚C) are shown in Figure 4.6. Also
shown is the fit of the data to the Sips adsorption isotherm model
described by Equation (5) in Chapter 2. Values of , and n are
summarized in Table 4.2. The temperature-dependence of the Sips
isotherm model for the affinity constant parameter and the
exponent n are given by Equation (6) in Chapter 2.
84
Figure 4.6 CO2 isotherms and the Sips fit
Temperature,
o
C
mCO2, mmol/g
sample bCO2, kPa
-1
n
250 0.9801(± 0.0021) 0.0078 (± 0.0012) 1.8628(± 0.0011)
350 0.7796(± 0.0013) 0.0073(± 0.0026) 2.2867(± 0.0016)
450 0.7443(± 0.0015) 0.0050(± 0.0018) 2.4779(± 0.0014)
Table 4.2 Sips isotherm parameters
85
Previously, Ding and Alpay (2000), who studied CO2 adsorption
on a K-promoted commercial Mg-Al hydrotalcite at 400
o
C, noted a
small (~ 10%) beneficial effect of the presence of water on CO2
adsorption. They also noted, however, that the actual partial water
pressure did not really matter, with even traces of water vapor being
capable of providing the same beneficial effect. Ding and Alpay (2000a)
attribute the beneficial effect to the ability of water vapor to either
maintain the hydroxyl concentration on the surface, or to prevent the
sites from poisoning through carbonate or coke deposition. We
investigated, therefore, the adsorption capacity of the LDH materials as
a function of temperature and atmospheric pressure under wet
conditions. For each series of experiments, 100 - 120 mg of a fresh LDH
sample was utilized. During the adsorption part of the cycle, 70 ml/min
of CO2 was bubbled through a beaker containing distilled water, and the
sample was exposed to the humidified CO2 stream for varying periods of
time. Figure 4.7 shows the comparison of adsorption capacity under dry
and wet conditions for temperatures 250, 350 and 450 °C and
atmospheric pressure. Wet conditions shows higher adsorption capacity
at all three temperatures but the percentage increase in the adsorption
capacity is within experimental error (~5%).
86
Figure 4.7 Adsorption capacities as a function of dry and wet
conditions for MG50
4.6 Adsorption Kinetic Studies
To estimate the diffusion constant of CO2 in hydrotalcite, Crank’s
model (Crank 1975) for diffusion into a spherical particle was used.
According to this model, the governing equation for the concentration C
of a species diffusing into a homogeneous spherical particle of radius r
is given as:
!C
!t
= D(
!
2
C
!r
2
+
2
r
!C
!r
) (3)
87
where D is the diffusivity constant. From the solution of Equation (3),
the uptake Mt at time t is given by:
M
t
M
!
= 1"
6
#
2
1
n
2
exp("n
2
$#
2
$
D
r
2
$t)
n=1
!
%
(4)
where M
is the uptake at large times (equilibrium). For small times (Mt
/M
< 0.25) Equation (4) is approximated as (Crank 1975):
M
t
M
!
=
6D
1/ 2
t
1/ 2
"
1/ 2
r
(5)
Therefore, if Equation (5) is applicable, a plot of Mt /M
versus t
1/2
should yield a straight line, with a slope of 6D
1/2
/
1/2
r. However, as
shown in Figure 4.8.a, the short time data does not indicate linear
behavior. Such a behavior implies the existence of a crystalline
structure in the hydrotalcite particles.
Since nonlinear behavior of Mt /M
with t
1/2
at short times was
observed, the long-time region was selected for the estimation of the
diffusivity. For long times (Mt /M
> 0.5) the higher-order terms in
Equation (4) become negligible so that the expression simplifies to:
M
t
M
!
= 1"
6
#
2
exp("#
2
D$t / r
2
) (6)
Therefore, a plot of ln(1- Mt /M
) versus t is linear with a slope of -!
2
D/r
2
and intercept of ln(6/!
2
). In this region, fairly accurate linearity is
observed for the experimental data at the three temperatures. Figure
88
4.8.b shows a sample of the long-time behavior observed at 250 °C. The
diffusion coefficients for CO2 in LDH were estimated from the slopes of
the plots at long times, are summarized in Table 4.3. The errors in the
diffusivities shown in Table 4.3 are calculated by repeating the
experiment three times and calculating the variation in the measured
diffusivity .
Figure 4.8.a Short-time behavior of CO2 uptake Mt/M∞ at 523 K
89
Figure 4.8.b Long-time behavior of CO2 uptake Mt/M∞ at 523 K
Temperature,
o
C D/ r
2
, s
-1
(x 10
4
)
250
4.52 (± 1.02)
350
4.88(± 0.85)
450
3.92 (± 0.67)
Table 4.3 Diffusivity data for the hydrotalcites
The adsorption rates were also measured by fitting the uptake-
type of experiments using the TGA. For the uptake experiments the
procedure followed was similar to that for the isotherm experiments
However, instead of switching from an inert Ar atmosphere into a
mixture of Ar and CO2, one switches to pure CO2 and monitors the
sample weight change until equilibrium is reached. The kinetic data
90
obtained in the thermogravimetric studies were fitted to Equation (7),
where the mass transfer coefficient was treated as a fitting parameter.
The mass transfer coefficients obtained for various temperatures are
shown in Table 4.4. The difference in the measured mass transfer
coefficient and the diffusivities is explained by the unsteady-state
method described above and used for measuring the diffusivities is
based on the assumption that all the diffusing molecules contribute to
the concentration gradient that gives rise to Fick's first and second law
of diffusion and Equation (3). However, as Weisz and Hicks (1967)
pointed out, adsorption in the material gives rise to a large number of
immobilized molecules that do not contribute to the concentration
gradient.
Temperature,
o
C ka, s
-1
250 0.0051(± 0.0003)
350 0.0064(± 0.0007)
450 0.0098(± 0.0006)
Table 4.4 Mass transfer coefficients for various temperatures using
uptake data from TGA
In another approach, measuring the adsorption kinetic data
involved experiments in a packed-bed column. CO2 breakthrough
measurements were carried out using the experimental set-up shown in
91
Figure 6.1. 120 g of the LDH sample was packed in a tubular reactor,
and a mixture of CO2 and an inert gas (Ar) with a total flow rate of 6
cc/s were used. The CO2 composition was measured at the exit of the
packed-bed reactor using the online mass spectrometer, while the flow
rates were measured using the bubble flow meter. The packed-bed
column data were fitted with the Sips isotherm, and the linear driving
force model to describe the adsorption rate (Amandusson et al. 2000;
Sircar and Hufton 2000; Soares et al. 2004; Soares et al. 2005). The
adsorption rate of CO2 in the LDH was described by following model
!q
i
!t
= k
i
(q
i
*
" q
i
) (7)
where is the effective mass-transfer coefficient, and and are the
equilibrium adsorbed phase concentration and the adsorbed phase
concentration, respectively. In our studies we considered only CO2 to be
adsorbed. A typical breakthrough curve under dry conditions is shown
in Figure 4.9. Also shown in the same figure are model predictions
based on the LDF rate model. The model is consistent with the
experimental data. The value of the calculated effective mass transfer
coefficient is shown in the figure, with its value being within 20% of that
obtained in the TGA experiments. For fitting the reaction experiments
described in chapter 6, we utilized the effective mass transfer coefficient,
92
estimated from the flow experiments. The corresponding mass transfer
coefficients at 350 & 450°C are summarized in Table 4.5.
Figure 4.9 The measured fitted concentration profiles in packed bed.
93
Temperature,
o
C ka, s
-1
250 0.0061(± 0.0008)
350 0.0069(± 0.0006)
450 0.0088(± 0.0003)
Table 4.5 Mass transfer coefficients for various temperatures using the
packed bed.
4.7 Effect of cycle number on adsorption capacity
The effect of cycle number on adsorption capacity of the LDH for
carbon dioxide, under dry conditions, was studied using the TGA
microbalance. A number of experiments were also carried out using the
TGA system in which adsorption (as discussed earlier) was followed by
desorption, where the gas was switched to pure argon (99.9999 %)
under a vacuum environment (100 milliTorr) for one hour. The
adsorption-desorption procedure was repeated for a number of cycles to
study the cyclic adsorption behavior of the LDH material. The CO
2
capacity evaluated over a number of cycles is illustrated in Figure 4.10.
A significant amount of CO
2
is reversibly adsorbed and desorbed during
the experiments; the capacity is around 0.44 mol CO
2
/kg LDH. The
data indicate that the adsorbent can be effectively regenerated by
simply lowering the pressure in the presence of flowing Argon. The CO
2
capacities appear to stabilize after the second cycle. Our group’s prior
94
cyclic adsorption studies (Kim et al. 2004) reported similar trends for
these materials.
Figure 4.10 Effect of cycle number on adsorption capacity of
hydrotalcite at 250°C, Pressure = 1 atm
4.8 Studies of adsorbent regeneration/desorption
Dynamic column experiments for CO2 sorption were conducted to
investigate the regeneration characteristic of the spent adsorbent. A
sample of 100 g of pre-calcined hydrotalcite was packed into a 25.4 cm-
long stainless steel reactor. The reactor was maintained at the
measurement temperatures (250-450 °C) and 1 atm during all the
95
experiments. The breakthrough experiments were conducted by flowing
a feed mixture containing 50% CO2 in Argon at 1 atm and the
experimental temperatures (250-450 °C). The effluent CO2 compositions
were monitored as functions of time by a mass spectrometer, and the
flow rates were measured using Brooks flow sensor. The adsorption run
continued until complete breakthrough of CO2 was observed. The
spent/saturated adsorbent was regenerated by flowing CO2-free Argon
at the experimental temperatures (250-450 °C). The concentration of
CO2 and the flow rate of the effluent were continuously monitored
during regeneration using the mass spectrometer and Brooks mass flow
sensor.
Desorption experiments were carried out by purging the sorption
columns (pre-saturated with CO2 at 1 atm and 250-450 °C) with pure
Argon at different flow rates. Figure 4.11 shows the CO2 desorption
profiles from the hydrotalcite adsorbent (saturated with CO2 at 1 atm
and 250-450 °C) by purging with Argon at 250-450 °C. The Argon inlet
mass flow rate was 100 SCCM. The figure shows the fraction of CO2
desorbed,
F =
NCO
2
(t)
NCO
2
(0)
, as a function of total effluent gas
quantity,
N(t)= Q(t)dt
0
t
!
, at any time t. The quantity,
NCO
2
(t)= Q(t).y
CO
2
(t)
0
t
!
, is the amount of CO2 desorbed from the column
96
at time t. NCO2(o) is the total amount of CO2 in the column at the start
of the desorption test. These values were calculated from the
experimental profiles for desorption of the effluent gas. Almost 100%
desorption was observed at all the three temperature investigated.
Figure 4.11 CO2 desorption profiles using Argon as a purge gas.
97
Chapter 5
Reaction Kinetics Studies
5.1 Introduction
In this chapter, the reaction kinetics for the two reactions, namely,
the methane steam reforming and WGS are investigated. A brief
literature review on the various catalysts used for the reactions and the
proposed reaction mechanisms is also presented. This chapter is
divided into two parts; in part (A) the WGS reaction and kinetic studies
on the commercial Cu/Zn catalyst are discussed. The effect of
temperature, pressure, and composition on the reaction performance
are studied and discussed here. In part (B) the methane reforming
reaction is investigated. A commercial Ni catalyst was used in the
reforming kinetics studies. We employed the kinetic expression by Xu-
Froment (1989) to define the rate of reaction. The reaction rate
constants were evaluated using the kinetic reaction data generated in
the lab. The parameters obtained were compared with the literature
data. Part of the work done in this chapter was in collaboration with my
colleague Hyun Hwang.
98
PART A: THE WGS REACTION KINETICS
5.2 Introduction
Since the early 1940’s the WGS reaction has been an important
step in the industrial hydrogen production. Industrially, the WGS
reaction is used to increase the production of hydrogen for refinery use,
bulk storage and redistribution. Other important uses of hydrogen
include the production of organic bulk chemicals, such as ammonia,
methanol, and alternative hydrocarbon fuels through the Fischer-
Tropsch synthesis. The WGS reaction is also a critical step in reducing
the CO concentration in hydrogen for use in low-temperature fuel cells
that are not tolerant to CO. In most fuel processor designs the WGS
reaction must deliver a CO concentration of <1% to the selective CO
oxidation reactor. Any further reduction in CO concentration eases the
load on the CO oxidizer. High-temperature fuel cells, on the other hand,
are not as dependent on the WGS reaction to reduce CO, since they are
able to withstand the higher concentrations expected from the
reforming reaction. The WGS reaction is very desirable for fuel
processing as, in addition to effectively reducing the CO concentration,
it produces a mole of H2 for every mole of CO that is converts. Recently,
there has been a renewed interest in the WGS reaction, due to its
potential use in supplying hydrogen for fuel cell power generation (Choi
99
and Stenger 2003; Semelsberger et al. 2004; Batista et al. 2005), as
previously discussed. Fuel cells are currently undergoing rapid
development for both stationary and transportation applications.
5.3 Thermodynamic Aspects
One must have a detailed understanding of the thermodynamics
and kinetics in order to fully utilize the benefits of the WGS reaction.
CO+ H
2
O! CO
2
+ H
2
"H =#41.1 kJ / mol (1)
Due to exothermic nature of the WGS reaction, higher CO conversions
are favored at lower temperatures. The WGS equilibrium constant is
nearly 80 times greater when the temperature is decreased from 600°C
to 200°C. The WGS equilibrium constants at various temperatures are
summarized in Table 5.1. The water content also has a strong influence
on converting CO.
Temp, K ! H, kcal Kp
700 -9.05 9.0
600 -9.29 26.65
523 -9.45 87.4
473 -9.56 230
Table 5.1 The WGS equilibrium constant as a function of temperature
(Twigg 1989)
Single-stage WGS reaction is desired but difficult to accomplish,
100
due to the exothermicity of the reaction and the resulting temperature
rise. To avoid the high temperature, the inlet temperature to the
catalyst must be relatively low, where the existing catalysts may
encounter kinetic limitations. Two-stage WGS processes are, therefore,
traditionally used to accommodate both the kinetic and thermodynamic
considerations (Twigg 1989). By operating a catalyst at higher
temperatures (high-temperature shift [HTS]) the favorable kinetics can
be exploited and the volume of the catalyst can be minimized. By
cooling the syngas in between the HTS and low- temperature shift (LTS)
stage an active LTS catalyst can take advantage of the higher reaction
equilibrium conversion at lower temperatures. A two-stage WGS
configuration can produce an exit CO concentration much smaller than
1%. At this time both the single- and two-bed concepts are being
considered in the design of fuel processors.
Since the WGS reaction is equimolar, the effect of pressure is
negligible on the thermodynamics of CO conversion. From a
thermodynamic perspective, the efficiency of the WGS reaction is
maximized at low temperatures, high water, and a low hydrogen
concentration.
5.4 Mechanism and Kinetics
The kinetics and mechanism of the WGS reaction with various
catalyst systems were studied in the past by many authors and
101
excellent summaries are given by Knozinger and Weitkamp (1997).
Because of the industrial significance of the WGS reaction, it has been
investigated by many researchers. The results of several of these
investigations suggest that the WGS reaction largely occurs via four
specific mechanism: (1) the redox mechanism (Eley 1979; Nakamura et
al. 1990; Ovesen et al. 1992; Ovesen et al. 1996; Waugh 1999), 2) the
formate mechanism (Campbell and Daube 1987; Shido and Iwasawa
1993; Askgaard et al. 1995; Ovesen et al. 1996), 3) the associative
mechanism (Rhodes et al. 1995; Callaghan et al. 2003), and, more
recently, 4) the carbonate mechanism (Millar et al. 1992; Tserpe and
Waugh 1997; Waugh 1999; Ma and Lund 2003). The first mechanism
implies successive oxidation by adsorbed oxygen from H2O and
reduction of the reactive catalyst surface by CO occurs as CO is
oxidized to CO2. In the second mechanism, adsorbed water dissociates
into an adsorbed hydroxyl group and adsorbed atomic hydrogen. The
hydroxyl group then combines with adsorbed carbon monoxide to form
adsorbed formate, which eventually decomposes into carbon dioxide
and hydrogen, yielding the WGS products. In the associative
mechanism adsorbed water dissociates into an adsorbed OH and atomic
hydrogen. The adsorbed hydorxyl then oxidizes adsorbed CO, resulting
in adsorbed CO2 and atomic hydrogen. In addition to the redox and
associative mechanism, researchers have also proposed that the WGS
102
reaction may proceed via a carbonate mechanism, which says, CO
oxidation on copper may proceed via a polydentate carbonate structure
identified as a “surface malachite type” species corresponding to the
production of CO2(g) and adsorbed CO2 (Callaghan 2006).
Based on adsorption/desorption measurements and inter-
conversion of CO and CO2, Tinkle and Dumesic (1987) concluded that
the WGS on Fe/Cr catalyst proceeds via the regenerative mechanism
which includes the participation of surface oxygen in the catalytic
reaction. The question of the reaction mechanism over Cu-based
catalysts is still debated today. The catalyst composition, catalyst
precursors, its resulting surface properties, as well as the reaction
conditions, obviously play decisive roles. Most researchers propose the
regenerative mechanism, which is in good agreement with the results
obtained from the WGS reaction carried out on single crystal Cu
surfaces and water adsorption experiments on polycrystalline copper
(Campbell and Daube 1987; Colbourn et al. 1991; Ovesen et al. 1992;
Ladebeck and Wagner 2003).
Ovesen et al. (1996) studied the influence of pressure on the
activation energy and reaction orders, based on macro- and
microkinetic models for Cu/Zn/Al catalyst. Grenoble et al. 1981 studied
the WGS reaction over a number of supported metals at atmospheric
pressures and in a temperature range from 270-380 °C. Alumina
103
supported Groups VIIB, VII and IB metals were examined, and for such
metals the range of activity differed by more than three orders of
magnitude, while the reaction orders varied from 0 to +0.8 for H2O and
from -0.4 to +0.6 for CO. Alumina-supported metals exhibit much
higher activity than catalysts prepared on silica or active carbon. The
turnover number measured on Pt supported on Al2O3 is an order of
magnitude higher than that of Pt-supported on active carbon. Other
researchers reported carbon monoxide reaction of order of 0, and water
reaction orders ranging from 0 to 0.5, -0.5 for carbon dioxide, and -1 for
hydrogen over Pd/CeO2 catalysts. Amadeo et al. (1992) studied the
WGS kinetics on a commercial CuO/ZnO/Al2O3 catalyst and fitted the
experimental data obtained at 0.2 MPa and 453-503 K to both an
associative and a redox mechanism. They found the associative
mechanism to be more probable (Amadeo and Laborde 1995). Ovsen
and co-workers (Ovesen et al. 1992; Ovesen et al. 1996) and Nakamura
et al. (1990), studied the WGS reaction over clean Cu (110) and Cu
(111) surfaces according to a microkinetic analysis. Both groups
proposed that the rate-controlling step of the WGS reaction was the
dissociative chemisorption of water to yield adsorbed oxygen atoms.
Gines and co-workers (Gines et al. 1995; Gines and Apesteguia 1997;
Gines et al. 1997) studied the reverse WGS reaction on a
CuO/ZnO/Al2O3 catalyst and found the dissociative chemisorption of
104
CO2 to be the rate-controlling step when the H2 concentration was
higher than that of CO2. However, when the ratio was reversed, the
main reaction rate was controlled by water formation
A Haldor-Topsoe-manufactured, low temperature shift, LK-821-2,
copper-based catalyst was used in our study. The LK-821-2 catalyst is a
mixture of CuO/ZnO/Al2O3. Table 5.2 presents the catalyst physical
properties provided by the catalyst supplier. According to the
manufacturer, the presence of free zinc oxide combined with an
unsurpassed chemisorption capacity provides the catalyst with an
outstanding tolerance towards sulfur poisoning.
Catalyst type LK-821-2
Catalyst form Tablets
Catalyst size 4.5 X 3.4 mm
Chemical Composition
Cu ~41 wt%
Zn ~21 wt%
Al ~5 wt%
Table 5.2 Physical and chemical properties of the LK-821-2 catalyst
The commercial catalyst is typically supplied as tablets, and it
was mechanically crushed to prepare small particles ~500 micron in
105
size. The particles were irregular in shape, but are considered
“spherical” for estimation of the bed properties. To prepare a “fresh”
catalyst, it must be “activated” by a process, which involves careful
reduction of the copper oxide to copper metal, using a suitable gas
mixture. This is done by heating the catalyst to 220 °C in a flow of 50 %
nitrogen, containing approximately 30 % hydrogen and 20 % water. The
reduction starts above 150 °C. The reaction experiments were carried
out in an isothermal plug-flow reactor; the details of the experimental
set-up will be discussed in Chapter 6. The reaction mixture was
composed of carbon monoxide (>99.99% purity), and hydrogen
(>99.99% purity), and ultra pure deionized water. Water was pumped
into a pre-heater, where it was vaporized. The reaction section consists
of a stainless steel tubular reactor placed inside a furnace. The reactor
was loaded with 0.25-0.5 mm in diameter catalyst particles, which were
diluted with quartz particles of the same size in order to avoid
temperature gradients along the bed. The bed temperature was
measured with J-type thermocouples. Two on line mass spectrometers
were used to perform the analysis of the feed and effluent gas streams
after water condensation.
In this study the most plausible mechanism and most accurate
kinetics for the WGS reaction, over the commercial CuO/ZnO/Al2O3
catalyst that fit the measured experimental data with a high statistical
106
significance were chosen. For accurate measurements of the rate
constants, runs that gave results at or near equilibrium were excluded
from further consideration. These fractional conversions, in
combination with the experimental parameters for each run and the
Hougen-Watson type rate equation, were used to evaluate reaction rate
constants. The effect of feed composition, temperature, and pressure on
the kinetic parameters was investigated. Table 5.3 displays the
experimental conditions investigated in this study.
Temperature (°C) 200-280
Pressure (kPa) 101-506
Weight of catalyst (g) 30
W/FCO (g-cat.h/mol-CO) 12-311
Feed Composition:
H2O/CO 1.1, 2.5, 3.0
H2/CO 4, 5
CO2/CO 0.5
Table 5.3 Experimental conditions investigated
The CO conversion data obtained through experiments were fitted
using nonlinear regression analysis with the error being the difference
between the experimental CO-conversion values and the model
107
prediction values using the fitted reaction rate values; the procedure
minimizes the sum of the squares (< 10
-4
) of the CO conversion values.
The kinetics data for the isothermal reaction were analyzed using the
following Langmuir-Hinshelwood type rate equation (Amadeo and
Laborde 1995),
r =
k ! P
CO
! P
H
2
O
!(1" #)
(1+ K
H
2
! P
H
2
+ K
CO
2
! P
CO
2
+ K
CO
! P
CO
+ K
H
2
O
! P
H
2
O
)
2
(2)
where the approach-to-equilibrium coefficient is defined as
! =
1
K
eq
(P
H
2
" P
CO
2
)
(P
CO
" P
H
2
O
)
(3)
In the rate expression (2), k is the reaction rate constant, Pi the partial
pressure of component i, Ki the surface adsorption equilibrium constant
for component i, while Keq in the Equation (3) is the overall WGS
reaction equilibrium constant. The activation energy and the pre-
exponential factor are calculated using the Arrhenius plot using the rate
Equation (2); the results are shown in Figure 5.1. The constants in the
rate equations were estimated using nonlinear least-square fitting (error
being the sum of squares < 10
-4
) of the experimental data (the fitted
data are also shown in Figure 5.2). Table 5.4 shows the Langmuir-
Hinshelwood reaction rate constants obtained by fitting all the available
experimental data.
108
Figure 5.1 Arrhenius plot for the WGS reaction data
109
Figure 5.2 Experimental vs. fitted CO conversions for the Hougen-
Watson rate expression
110
Hougen-Watson rate model
Rate
r =
k ! P
CO
! P
H
2
O
!(1" #)
(1+ K
H
2
! P
H
2
+ K
CO
2
! P
CO
2
+ K
CO
! P
CO
+ K
H
2
O
! P
H
2
O
)
2
E 30.387 (± 0.09) (kJ/mol)
k
0
48.16 (± 0.03) (mol/g/s/atm
2
)
! H (kJ/mol) K
0
(atm
-1
)
-12 (± 0.25) 0.0178 (± 0.001)
-28 (± 0.13) 0.0410 (± 0.002)
-0.86 (± 0.03) 0.0303 (± 0.011)
-2.42 (± 0.051) 0.005 (± 0.020)
Table 5.4 Hougen-Watson rate model and its kinetic parameters
111
PART B: REFORMING KINETICS
5.5 Introduction
The steam reforming reaction,
CH
4
+ H
2
O ! CO + 3H
2
"H = +206.16 kJ /molCH
4
(4)
is endothermic and requires external heat input. Economics favor
reactor operation at pressures of 3-25 atm and temperatures of 700-
850 °C. The external heat needed to drive the reaction is often provided
by the combustion of a fraction of the incoming natural gas feedstock
(up to 25%), or from burning waste gases, such as the purge gas from
the hydrogen purification system. Heat transfer to the reactants is
accomplished indirectly through a heat exchanger. Methane and steam
react in the catalyst-filled tubes. Typically, the mass ratio of steam-to-
carbon is about 3 or higher in order to avoid "coking" or carbon build-
up on the catalysts surface.
After reforming, as previously noted, the resulting syngas is sent
to one or more shift reactors, where the hydrogen output is increased
via the WGS reaction
CO + H
2
O ! CO
2
+ H
2
"H =#41.15 kJ /molCO (5)
which "converts" CO and H2O into H2. S mentioned above, this reaction
is favored at temperatures of less than 600 °C, and can take place at
temperatures as low as 200 °C, with sufficiently active catalysts. The
112
gas exiting the shift reactor contains mostly H2 (70-80%) plus CO2, CH4,
H2O and smaller quantities of CO. For hydrogen production, the shift
reaction is often accomplished in two stages. A high-temperature shift
reactor operating at about 350-475 °C accomplishes much of the
conversion, followed by a lower temperature (200-250 °C) shift reactor
which brings the CO concentration down to a few percent or less by
volume. Hydrogen is then purified. The degree of purification depends
on the application. For industrial hydrogen production, the PSA
systems or palladium membranes are used to produce hydrogen with
up to 99.999% purity.
5.6 Thermodynamic Aspects
The steam reforming of methane consists of three reversible
reactions: the strongly endothermic reforming reactions (6) and (8), and
the moderately exothermic water-gas-shift (7):
CH
4
+ H
2
O! CO + 3H
2
"H = +206 kJ /mol (6)
CO + H
2
O ! CO
2
+ H
2
"H =#41.1 kJ /mol (7)
CH
4
+ 2H
2
O! CO
2
+ 4H
2
"H = +165 kJ /mol (8)
The SRM equilibrium constants at various temperatures are
summarized in Table 5.4. It should be emphasized that CO2 is not only
produced via the shift reaction (7), but also directly via the steam-
reforming reaction (8). This implies that reaction (8) is not just the
overall reaction, despite the fact that in the literature steam-methane
113
reforming is often considered to be only a combination of reactions (6)
and (7) (Beurden 2004).
Temp, K H, kcal Kp
800 +53.21 0.031
1000 +53.87 26.12
1123 +54.33 530.26
1273 +54.36 9301.00
Table 5.5 Equilibrium constant as a function of temperature (Twigg
1989)
Due to its endothermic character, reforming is favored at high
temperature in the range of 1000 K or more. Moreover, because
reforming is accompanied by a volume expansion, it is favored by low
pressures. In contrast, as noted previously, the exothermic shift
reaction is favored by low temperatures, while it is unaffected by a
change in pressure.
Increasing the amount of steam will enhance the CH4 conversion,
but requires an additional amount of energy to produce the steam. In
practice, steam-to-carbon ratios (S/C) around 3 are applied. Such a
value of S/C will also suppress coke formation during the reaction.
Catalyst manufacturers offer a wide variety of catalysts and are
114
experienced in selecting the best catalyst type based on feedstock
characteristics and furnace design. The reforming reaction is very rapid
and occurs at high temperatures in multiple tubes packed with catalyst
and installed in a direct-fired furnace. In regions of maximum
temperatures, the reaction becomes diffusion-controlled, and therefore
attention must be given to catalyst shape, pore size, and overall
dimensions. Very small pores, although contributing significantly to the
surface area, suffer from diffusion limitations for the fast reforming
reactions and are essentially not utilized.
Chemical analyses for several catalysts typically utilized are
summarized as follows (Howard 2000)
Carrier Nickel as NiO, wt% SiO2, wt%
Refractory alumina 12-20 <0.05
Magnesium aluminate 16-18 <0.2
Calcium aluminate 16-25 <0.2
Calcium aluminate titanate 19-25 <0.2
Table 5.6 Reforming catalysts and supports (Howard 2000)
Table 5.7 summarizes various types of catalysts reported in various
published studies for the methane staem reforming and the ranges of
their operating conditions (Twigg 1989)
115
Catalyst Temperature,
o
C Pressure, bar
Ni 500-900 1-15
Industrial-Ni 500-900 21-41
Ni foil 470-800 1-41
Ni/-Al2O3 350-450 1-2
Commercial-Ni 638 1-18
Ni/Al2O3 or SiO2 670-770 16-26
Rh/SiO2 350-600 1
Table 5.7 Various reforming catalysts and their operating range (Twigg
1989)
5.7 Reaction Kinetics and Mechanism
Because of the thermodynamic constraints of the reforming
reaction, the process is carried out at high temperatures where the
activity of the catalyst is also very high. Table 5.8 shows various
approaches that have been utilized to describe the intrinsic kinetics of
the steam reforming reaction of methane and various hydrocarbons.
116
Investigator Mechanism proposed
Bordov et al (1964) Langmuir-Hinshelwood
Rostrup-Nielsen (1999) Two-step kinetics, power law
Tottrup (1982) pellet kinetics, power law
Xu and Froment (1989a) Langmuir-Hinshelwood
Aparicio (1997) Microkinetic analysis
Table 5.8 Summary of the studies of the kinetics of the steam
reforming reaction and the mechanisms proposed (Beurden 2004)
Xu and Froment (1989) established a complex Langmuir-
Hinshelwood expression, using a classic approach, on the basis of 280
measurements with a Ni/MgAl2O4 catalyst. Its applicability is, however,
restricted to a narrow range of parameters: temperatures of 500-575°C,
pressures of 3-15 bar, and molar H2O/CH4 ratios of 3-5. A number of
other rate equations were established on the basis of a mechanism
involving 13 steps, assuming one of the steps to be rate determining. It
was shown that CO2 is produced, not only by the shift reaction, but also
by steam reforming. Hence, rates for the three reactions were found to
give the best agreement with the measurements
117
Because the three reactions (Equations 6-8) are not independent,
it is necessary to combine the three rate equations into two: one for the
conversion of methane and one for production of CO2:
r
CH
4
= r
1
+ r
3
(9)
r
CO
2
= r
2
+ r
3
(10)
The two expressions include five temperature-dependent constants.
They indicate a small negative reaction order in the overall pressure, in
agreement with the data.
Bordov et al. (1964) performed an alternative analysis of the data
of Xu and Froment (1989), resulting in better agreement with the
calculated and measured values of partial pressures of CO. Many
studies have been performed to investigate the kinetics of steam
reforming, and while there is general agreement on first-order kinetics
with respect to methane, the reported activation energies span a wide
range of values (Bodrov 1964).
In this study a commercial C11-9LDP (Sud-Chemie) catalyst was
utilized. Table 5.9 presents the catalyst’s physical and chemical
properties, as provided by the catalyst supplier.
118
Catalyst type C11-9LDP
Catalyst form Ten hole
Catalyst size 19 X 16 mm
Chemical Composition
Ni 14 1.5 wt%
Al2O3 80-86 wt%
SiO2 < 0.05 wt%
Sulfur <0.05 wt%
Physical Properties
Bulk density 85 5 (lbs/ft
3
)
Crush strength 150
Table 5.9 Physical and chemical properties of the reforming catalyst
A “fresh” catalyst was “activated”, by heating the catalyst to
600 °C in a flow of 20 % nitrogen containing approximately 30%
hydrogen and 50% water for approximately 3 h. The experiments were
carried out in an isothermal plug-flow reactor; the details of the
experimental set-up are discussed later in Chapter 6. The reaction
119
mixture was composed of methane (>99.99% purity), hydrogen
(>99.99% purity), and ultra pure deionized water. Water was pumped
into a preheater, where it was vaporized. The reaction section consists
of a stainless steel tubular reactor placed inside a furnace. The reactor
was loaded with 0.25-0.40 mm in diameter catalyst particles, diluted
with quartz of the same size to avoid temperature gradients along the
bed. The bed temperature was measured with J-type thermocouples.
Analysis of the feed and effluent gas was performed after water
condensation by two on line mass spectrometers.
The experimental data obtained were used to compute the
increase in fractional conversion. These fractional conversions, in
combination with the experimental parameters for each run and the Xu
and Froment (1989) model equation were used to evaluate reaction rate
constants. This mechanism results in the following three rate
equations:
120
i rate expressions equilibrium constant, Keqi
1
r
1
=
k
1
P
H
2
2.5
P
CH
4
P
H
2
O
!
P
H
2
3
P
CO
K
eq1
"
#
$
$
%
&
'
'
DEN
( )
2
,
K
eq1
= exp 30.114 !
26,830
T
"
#
$
%
&
'
2
r
2
=
k
2
P
H
2
P
CO
P
H
2
O
!
P
H
2
P
CO
2
K
eq2
"
#
$
$
%
&
'
'
DEN
( )
2
,
K
eq2
= exp !4.036+
4,400
T
"
#
$
%
&
'
3
r
3
=
k
3
P
H
2
3.5
P
CH
4
P
H
2
O
2
!
P
H
2
4
P
CO
2
K
eq3
"
#
$
$
%
&
'
'
DEN
( )
2
,
K
eq3
= K
eq1
! K
eq2
DEN = 1+ K
CO
P
CO
+ K
H
2
P
H
2
+ K
CH
4
P
CH
4
+ K
H
2
O
! P
H
2
O
/ P
H
2
Table 5.10 Rate expressions and thermodynamic properties for the
methane-steam reforming reaction
ki are the kinetic constants (rate coefficients) and the subscript i
corresponds to the three reaction steps 1, 2, 3, respectively.
k
i
0
and E
i
121
are the pre-exponential factor and the activation energy for the rate step
i , respectively.
The above rate expressions for methane steam reforming have
been extensively studied for both laboratory and industrial scale
reactors. In our studies, since the reactor was operated in the integral
mode, parameters estimation was based on the minimization of the sum
of the weighted residual squares of the conversions. The calculated
conversions were obtained by integrating the set of ordinary differential
equations for the reference components. Material balances for CH4 and
CO2 in the packed-bed plug- flow reactor are as follows
dX
CH
4
d
W
F
CH
4
0
!
"
#
$
%
&
= R
CH
4
= r
1
+ r
3
dX
CO
2
d
W
F
CH
4
0
!
"
#
$
%
&
= R
CO
2
= r
2
+ r
3
(11)
With the corresponding boundary conditions:
At
W
F
CH
4
0
= 0! X
CH
4
= 0, X
CO
2
= 0
where is the overall rate of methane disappearance, and is
the overall rate of CO2 formation. Figure 5.3 presents the experimental
data generated over the temperature range 500-650 °C. Figure 5.4
displays a comparison of the experimental CH4 conversion and the
122
fitted conversion using the Xu-Froment kinetic model. The linearity
indicates a good fit of the data, hence indicating the parameters
obtained are accurate. Table 5.11 presents a comparison between the
fitted kinetic parameters and the literature values. For the regression
analysis, the adsorption equilibrium constants were not considered as
parameters, and values for these constants were those reported in the
Xu and Froment (1989) study.
Figure 5.3 CH4 conversion measured as a function of W/FCH4
123
Figure 5.4 Experimental vs. fitted CH4 conversion for various
temperatures
124
Xu and Froment
Kinetics
This study
k1 (kmol-bar
0.5
/k-cat/h) 4.225x10
15
7.6084x10
15
k2 (kmol/k-cat/h/bar) 1.955x10
6
2.0834x10
6
k3(kmol-bar
0.5
/k-cat/h) 1.020x10
15
1.7244x10
15
KCO (bar
-1
) 8.23x10
-5
KH2 (bar
-1
) 6.12x10
-9
KCH4(bar
-1
) 6.65x10
-4
KH2O 1.77x10
5
Table 5.11 Comparison of the kinetic parameters
125
CHAPTER 6
Results and Conclusions
6.1 Introduction
In this chapter, the experimental and modeling results are
described and discussed. WGS and reforming reactions were
experimentally investigated using commercial catalysts, commercial
CO2 adsorbents, and hydrogen-permselective nanoporous CMS and
composite palladium membranes. The HAMR model presented in
Chapter 3 is used to analyze the experimental data. With the aid of this
model the effect of the various membrane and adsorbent parameters is
discussed. The performance of the HAMR is compared with that of the
more conventional reactor systems, such as the AR, MR, and PBR.
6.2 Experimental set-up
The apparatus used in the experiments is shown in Figure 6.1.
The set-up allows one to carry out the WGS and steam-reforming
reactions with the catalyst and adsorbent being both present, as well as
to regenerate the adsorbent in-situ in a flowing stream of gas, such as
nitrogen and/or steam. The same apparatus is also used for evaluating
the membrane performance. The experimental system is divided into
three sections, namely, (i) the feed section which consists of the gas
cylinders, the mass flow controllers, the high pressure water syringe
126
pumps, and the steam generating units; (ii) the reactor section which
consists of the reactor, a furnace for heating the reactor, back-pressure
regulators for controlling the pressure, two condensers and two
moisture traps to remove water from the exit gas on the feed-side and
the permeate-side streams, and finally (iii) the analysis section that
consists of two on-line mass spectrometers, together with a gas
chromatograph to analyze the exit feed and the permeate-side gas
streams, as well as bubble flow-meters for measuring the total flow-
rates.
Figure 6.1 Experimental set-up
A syringe pump was used to supply a precisely-controlled flow of
water into the heated steam-generating unit (a stainless steel vessel
packed with quartz beads, in order to accelerate water evaporation and
127
to dampen out the fluctuations in the flow of the steam generated).
Mass flow controllers were used to control the flow of other gases into
the steam units where they are mixed with the evaporated water. There
are two steam generating units in the experimental system, one
installed on the feed-side line, and the other on the permeate side in
order to generate steam as the sweep gas. All the lines connecting the
steam-generating unit to the main reactor body are heat-traced using
heating tapes. The feed and sweep gas flows are preheated to the
reaction temperature before entering the reactor. The reactor is made of
stainless steel and has an internal diameter of 3.175 cm and a length of
25.4 cm. During the HAMR experiments, the membrane is installed in
the reactor using graphite o’rings and compression fittings. The catalyst
and/or adsorbent are loaded in the annular space between the
membrane and the reactor body. The reactor is heated by a three-zone
furnace using three individual temperature controllers and three
thermocouples installed in three different locations in the bed. An
additional thermo-well is installed in the bed, and is used to monitor
the temperature along the length of the bed with the aid of a sliding
thermocouple. Back-pressure regulators are used to control the
pressure. In the experiments reported here the feed pressure was varied,
but the permeate pressure was always kept at atmospheric conditions.
The gases exit the reactor and then flow through a condenser and a
128
moisture-trap in order to separate out the moisture. Their flow rates
were then measured by a bubble flow meter. A small slip-stream was
used to measure their composition with the aid of two mass
spectrometers. A gas chromatograph was also installed on line, and was
used to validate the mass spectrometric measurements, and to measure
the CO concentrations. The mass spectrometric measurements were
instantaneous. A number of gas chromatographic measurements were
carried out to measure the ppm level CO concentration in the permeate
side and to verify the mass spectrometer accuracy.
6.3 Experimental results
The WGS and methane-reforming experiments were carried out
according to the HAMR concept described earlier. The feed gas for the
WGS reaction consisted of a mixture of CO, H2 and H2O (1:4:1.1) which
was chosen such that the CO: H2 ratio was close to the equilibrium
steam methane reforming composition at 850 °C and 304.2 kPa (44.1
psi) with H2O/ CH4 ratio of 2. The reaction pressure was 308.2 kPa
(44.7 psi) (feed-side), the membrane-side pressure (permeate-side) was
kept under atmospheric conditions, and a sweep gas (steam) ratio
n
j 0
P
!
n
j 0
F
! ( )
of 0.1 was used, with the reaction temperature being
250 °C. The feed gas for the reforming reaction consisted of a mixture of
CH4, H2 and H2O (1:0.2:3.) The reaction pressure was 308.2 kPa (44.7
psi) (feed-side), the membrane-side pressure (permeate-side) was kept
129
under atmospheric conditions, and a sweep gas (steam) ratio of 0.1 was
used, with the reaction temperature being 450 °C. The reactor was
maintained under isothermal conditions. The hydrogen-selective
membranes characterized in Chapter 3 were used to selectively remove
hydrogen from the reaction products, while the hydrotalcite CO2
adsorbent was packed inside the membrane reactor along with a
commercial catalyst. In the experiments we used 30 g of catalyst and 70
g of adsorbent (pre-calcined at 450 °C).
Figure 6.2 shows the CO conversion profile with respect to time
for the HAMR reactor at 250 °C and for = 282. As expected, a
nearly complete conversion of CO was achieved at the beginning of the
reactor run, i.e., up to about 12 min. Then, the conversion declined
and finally settled at 88.5 % when the adsorbent was saturated.
Compared with the simulation results presented in Figure 6.2, the
overall conversion profiles obtained experimentally are consistent, in
general, with the predicted profile shown in Figure 6.2. The solid lines
in the figures are the results using the model that was previously
discussed in Chapter 2. The dotted lines in Figures 6.2 and 6.3 indicate
the corresponding equilibrium CO conversions (based on the model
predictions
130
Figure 6.2 Comparison of the experimental data with the model
predictions for = 282 at 250 °C
131
Base on the feed composition), which are significantly lower than the
conversions attained by the HAMR. With minor discrepancies, the
HAMR performance prediction is generally consistent with the
experimental results. Also shown in the figure are the exit mole fraction
of CO2 on the feed side (dry basis), and the CO concentration (dry basis)
in the hydrogen stream withdrawn from the permeate side. Figures 6.3-
6.7 show the corresponding profiles for =186, 350, and 300. The CO
conversion ( ), hydrogen yield ( ), and hydrogen recovery ( )
are defined as
X
CO
=
n
CO
0
F
! n
CO,ex
F
+ n
CO,ex
P
( )
n
CO
0
F
Y
H
2
=
(n
H
2
,ex
F
! n
H
2 0
F
)+ (n
H
2
,ex
P
! n
H
2 0
P
)
n
CO
0
F
Re
H
2
=
n
H
2
,ex
P
n
H
2
,ex
F
+ n
H
2
,ex
P
(1)
No adjustable parameters were utilized in these simulations, with
the required adsorbent, catalyst and membrane properties measured in
independent experiments as previously described. The experimental
data and the simulated conversion, exit CO2 mole fraction (dry basis),
and permeate stream CO concentration (dry basis) seem to be matching
well
132
Figure 6.3 Comparison of the experimental data with the model
predictions for = 186 at 250 °C
133
Figure 6.4 Comparison of the experimental data with the model
predictions for = 350 at 250 °C
134
Figure 6.5 Comparison of the experimental with the simulated CO2
breakthrough for = 350 at 250 °C
135
Figure 6.6 Comparison of the experimental data with the model
predictions for = 300 at 250 °C
136
Figure 6.7 Comparison of the experimental data with the simulated
CO2 breakthrough values for = 300 at 250 °C
137
The cyclic adsorbent studies previously discussed have shown,
that the fresh adsorbent capacity decreases, but then stabilizes after
the 2
nd
cycle of adsorption and regeneration. Figure 6.8 shows the CO
conversion profiles for three cycles of reaction-regeneration in the
HAMR system for =466. As expected, the adsorbent stays active for
a longer time during the 1
st
cycle of reaction, when the capacity of the
fresh adsorbent is higher.
Figure 6.8 Cyclic behavior of HAMR for = 466 at 250°C
138
Figure 6.9 Cyclic behavior of HAMR for = 287 at 250°C
A 2
nd
set of cyclic WGS-HAMR experiments were carried out with
an “aged” adsorbent after it had undergone a set of adsorption and
regeneration experiments. The CO conversion (for =287) as a
function of the cycle time is shown in Figure 6.9. During these cyclic
experiments the saturated adsorbent after the reaction step was
regenerated by lowering the pressure (1 atm) and purging the reactor
using Argon for 1 h (similar regeneration procedure was followed during
the 1
st
set of cyclic experiments). In this experiment 4 cycles of reaction
and regeneration were carried out. Fairly reproducible behavior is
139
observed during the different cycles, indicative of stable adsorbent and
catalyst activity.
Figure 6.10 shows the effect of sweep ratio on the CO conversion,
as well as the model predictions. As shown in the figure, upon
increasing the sweep ratio, the conversion increases.
Figure 6.10 Comparison of the effect of sweep ratio on the CO
conversion for = 466 at 250°C
HAMR experiments for the methane steam-reforming reaction
were performed using a composite palladium membrane. The
140
commercial Ni catalyst discussed in Chapter 5 was utilized. For the
analysis of the data in the HAMR model we utilized the Xu-Froment
kinetics with the parameters estimated in Chapter 5. Calcined
hydrotalcite (450°C calcinations temperature) material was used for CO2
adsorption. A mixture of 30g of catalyst and 70g of adsorbent was
packed on the outer side of the membrane. Figure 6.11 shows the CH4
conversion as a function of time for =187 at 450 °C. While the
adsorbent is active the CH4 conversion stays ~ 90% for about 7 min and
then it starts to drop and finally levels to the corresponding MR steady
state conversion value (~56%). The equilibrium conversion under the
operating conditions was ~ 18%, the synergy of reaction-separation
indicates almost 72% increase in conversion for the case of HAMR and
38% increase in conversion for the case of MR.
141
Figure 6.11 Experimental CH4 conversion data for = 187 at 450°C
6.4 Sensitivity analysis
Figures 6.12-6.17 show the effect of membrane and adsorbent
properties on the HAMR performance in terms of CO conversion, H2
recovery, and CO concentration in the hydrogen on the permeate-side.
The hydrogen permeance appears to have the strongest impact on all
the three performance parameters. mCO2 (the Sips isotherm parameter)
and ka (the mass transfer coefficient) have an effect during the transient
period when the adsorbent capacity starts to decline.
142
Figure 6.12 Sensitivity analysis for the H2 permeance.
143
Figure 6.13 Sensitivity analysis for the CO permeance.
144
Figure 6.14 Sensitivity analysis for the CO2 permeance.
145
Figure 6.15 Sensitivity analysis for the H2O permeance.
146
Figure 6.16 Sensitivity analysis for the mCO2.
147
Figure 6.17 Sensitivity analysis for mass transfer coefficient.
148
In the following section the behavior of the HAMR and the
adsorptive reactors are compared. The reactor temperature was 250 °C,
and a CO: H2O: H2 feed ratio of 1:1.1:4 was utilized. Steam was used as
the sweep gas. The empirical power-law kinetics obtained by fitting all
the kinetic data generated was used along with the rate constants
measured under experimental conditions. A hydrogen selective CMS
membrane was used, the properties of which were evaluated in the
laboratory experimental set-up discussed earlier. The behavior of
adsorptive reactor was calculated by assuming all the gas permeabilities
to be zero. Values of the various parameters used in the simulations are
given in Table 6.1.
149
Parameter
Value
Dimension
0.001 m
Da
32.179
(Base case)
Ha
1.082
(Base case)
Pe
1.499
(Base case)
3 bar
1 bar
s
0.1
(Base case)
T
250
°C (Base case)
0.33 m/s
8.019 m/s
!
m
0.1435 m
2
/m
3
!
c
0.1131 -
0.28
-
0.4
-
30.26
-
48.26
-
0.33
-
Table 6.1 Parameters used in the simulations
The effect of , which represents the ratio of catalyst weight
and the inlet CO flow rate, on the hydrogen yield, hydrogen recovery,
and CO impurity in the permeate side is shown in Figure 6.18. ,
150
was varied by keeping the amount of catalyst constant and increasing
or decreasing the inlet flow rate. Hydrogen yield increases with an
increase in . The advantage of the HAMR system over the adsorptive
reactor in the hydrogen yield under similar conditions can be seen in
the figure. Hydrogen recovery which is, of course, a strong function of
the membrane permeation characteristics, membrane surface area, and
the other operating conditions in the reactor is also plotted. The
recovery is found to be increasing with an increase in . The increase
is attributed to the increased conversion that provides higher driving
force for hydrogen permeation. Figure 6.18 also shows the CO
concentration (dry basis) profiles in the permeate side exit of the HAMR,
together with the corresponding exit concentrations for the AR. It
indicates clearly the advantage that the HAMR system provides in terms
of the reduced CO concentrations in the hydrogen product over the AR
system, in addition to improved hydrogen yields. does not seem to
have any effect on the H2 purity for both the HAMR and AR.
151
Figure 6.18 Performance of the HAMR and AR systems for different
152
Figure 6.19 shows the effect of membrane properties, expressed
in terms of the hydrogen permeance on the hydrogen yield, hydrogen
recovery, and CO impurity in the product hydrogen. For the CMS
membrane used in the simulation
! = 48.26 . The other two membranes
have values that are 0.5 and 3 times the base value (since is
inversely proportional to the permeance, the values correspond to
permeances that are 2 and 0.333 times that of the base case
corresponding to
! = 48.26 ). Hydrogen yield increases with an increase
in the hydrogen permeance, but the effect seems to decrease slowly. As
expected, increasing the hydrogen permeance has a very beneficial
effect on its recovery, as seen in the figure
153
Figure 6.19 Effect of membrane properties on the HAMR and AR
systems
154
The effect of using an adsorbent with improved characteristics is
shown in Figure 6.20. The hydrogen yields for the HAMR and AR
systems are compared for three values of , one corresponding to the
adsorbent under investigation (for the reactor temperature and pressure
utilized, this corresponds to 0.03364), and two other cases with the
corresponding values that are three and five times larger. A more
effective adsorbent significantly expands the “time window” of operation
for both the AR and HAMR systems, before regeneration must
commence. It also significantly increases the hydrogen recovery. The
impact of improved adsorbent on CO impurity can also be seen in the
figure.
155
Figure 6.20 Effect of adsorbent properties on the HAMR and AR
systems
156
Figure 6.21 Effect of initial condition for step 1 for AR systems
Before the reaction step starts, the reactor is pressurized with
products to the desired pressure. Effect of pressurizing-gas, used
during the pressurization step, on the adsorptive reactor performance
during the reaction step was investigated. Figure 6.21 shows the
hydrogen molar fraction along the length of the AR over the reaction
step. The molar fraction of hydrogen at the exit of the reactor was
initially very low when the pressurization gases from the reactor voids
were being displaced. To avoid such low purity hydrogen product
157
conditions, it is very important for the AR to use hydrogen as a
pressurization gas during the pressurization and purge steps of the
cyclic hydrogen production process.
In the following, the behavior of the four-step HAMR cyclic
hydrogen production process is further analyzed and discussed. The
various parameters and operating conditions used in these simulations
are indicated in Table 6.2. The base case parameter values match those
of the laboratory HAMR system
158
Base case Range of values
Weight of catalyst (g) 30 -
Weight of adsorbent (g) 70 -
W/F
CO
(g-cat.hr/mol-CO) 200 100-300
(m/s)
0.4747 0.316-0.949
Feed pressure, (atm) 3 -
Permeate pressure (atm) 1 -
Sweep ratio 0.1 -
31.68 19.008-316.8
1.92 0.96-3.84
H
2
/H
2
O - Separation factor 1.125 -
H
2
/CO - Separation factor 80 40-200
H
2
/CO
2
- Separation factor 16.36 -
Length of membrane (m) 0.254 -
Inner diameter of membrane (m) 0.0035 -
Outer diameter of membrane (m) 0.0057 -
Reactor diameter (m) 0.03175 -
Bed porosity 0.4 -
Damköhler number 22.753 11.37-34.13
Hatta number 0.8222 0.411-1.23
Peclet number 1.392 0.83-13.92
STEP1 STEP2 STEP3 STEP4
Feed composition Reaction Blow-down Purge Pressurization
H
2
4 - 0.65 0.65
CO 1 - - -
CO
2
0 - - -
H
2
O 1.1 - 0.35 0.35
Flow rate (mol/h) 0.392 - 0.195 0.195
Pressure (kPa) 303.975 101.325 101.325 303.975
Time (min) 6 2 14 2
Temperature (°C) 250 250 250 250
Table 6.2. Parameters used in the cyclic HAMR simulations
Figure 6.22 shows graphically how we envision the four-step
HAMR process to operate. The selected times of operation in Figure 6.22
are arbitrary, but in practice they need to be selected depending on the
desired product purity requirements (expressed here in terms of the
159
time-averaged CO concentration) and productivity; the latter for the
case of the AR is directly related to the time-averaged conversion, but
for the case of HAMR it also relates to membrane throughput and other
operating conditions, since the product is collected on the permeate
side.
Figure 6.22. Schematic illustration of the four-bed HAMR process
Figure 6.23-a compares the conversion for the HAMR for four
different (Wc/FCO) with that of the AR (to vary the Wc/FCO the amount of
catalyst and adsorbent are kept constant to the base values indicated in
Table 6.2 and the total feed flow rate is changed). Figure 6.23-b shows
the time-averaged CO concentration (Harale et al. 2007). The feed
composition for the HAMR is as indicated in Table 6.2, and the sweep
ratio is equal to 0.1.
160
Figure 6.23-a
Figure 6.23-b
Figure 6.23. Conversion and permeate-side CO concentration (ppm)
profile during the reaction step
161
For the AR the assumption is made that the feed is a (CO:H2:H2O)
mixture with a composition of (1:4:1.7) to match the total H2O used in
the HAMR (both as feed and sweep gas). It is not entirely clear that this
is a fair and equitable basis of comparison to make between the two
reactor systems, as the AR operates at high pressures, whereas the
HAMR permeate side is at a lower pressure, and high-pressure steam is
generally more valuable than its low-pressure counterpart. Doing so, of
course, favors the AR operation as H2O is a reactant for the WGS
reaction (in general, this advantage for the AR operation resulting from
using steam will not be present for other reactions for which steam is
not a reactant).
As Figure 6.23-a indicates, the CO conversion and its time-
averaged concentration (ppm) in the product hydrogen (collected in the
permeate side for the HAMR) during the reaction step (step 1) are strong
functions of the Wc/FCO. Both the AR and the HAMR start at high
conversions, while the adsorbent is still fresh, but their performance
progressively deteriorates as the adsorbent gets saturated. The
advantage of the HAMR is clear from Fig. 6.23-b in terms of product
purity, especially for the lower Wc/FCO ratios. The performance of the
HAMR is significantly impacted by the membrane surface area utilized
and its transport properties, as Figure 6.24 indicates, which shows the
162
conversion (Figure 6.24-a) and the time-averaged CO concentration
(Figure 6.24-b) for
Figure 6.24-a. Effect of varying the DaPe number on CO conversion for
Wc/Fco= 200
163
Figure 6.24-b. Effect of varying the DaPe number on the time-averaged
CO concentration (ppm) for Wc/Fco= 200
three different cases corresponding to three different DaPe values (see
nomenclature and the paper by Harale et. el. (2007) for the definition of
the Da (Damköhler) and Pe (Peclet) numbers; for a given catalyst weight
and under isothermal conditions, DaPe is inversely proportional to the
product of the membrane surface area and the single-gas hydrogen
permeance). Increasing the membrane surface area (per catalyst weight)
or improving the membrane throughput (permeance) significantly
improves conversion and product purity. It has, in addition, a beneficial
effect on hydrogen recovery (see Figure 6.25), defined (Harale et al.
2007) as the fraction of total hydrogen that exits the reactor that leaves
164
from the permeate side (the leftover hydrogen on the feed-side finds use
in the regeneration steps (3 and 4) to purge and pressurize other beds).
Figure 6.25. Effect of varying the DaPe number on H2 recovery for
Wc/Fco= 200
Figure 6.26 shows the effect of the H2/CO membrane separation
factor. For this figure, the single-gas hydrogen, as well as the H2O and
CO2 permeances were kept constant while the CO permeance was
varied. This is equivalent to preparing a CMS membrane with fewer
pinholes and imperfections as it is such defects that allow CO to diffuse
through.
165
Figure 6.26-a. Effect of varying the H2/CO separation factor on CO
conversion for Wc/Fco= 200
Figure 6.26-b. Effect of varying the H2/CO separation factor on the
time-averaged CO concentration in the permeate-side for Wc/Fco= 200
166
Lowering the CO permeance does have a beneficial effect on the
conversion, but it has a significantly more important impact on product
purity (ideally for this HAMR system one should use Pd membranes
which are capable of completing exclude all other species other than
hydrogen from permeating through).
Figure 6.27-a. CO conversion profile as a function of the reaction step
time.
167
Figure 6.27-b. CO concentration profile in the permeate side (ppm) as a
function of the reaction step time.
Figure 6.27 shows the CO conversion (Figure 6.27-a) and its
concentration (ppm) (Figure 6.27-b) in the permeate side along the
length of the reactor during the reaction step (step 1) as a function of
time of operation. Figure 6.28 shows the molar H2 and CO2
compositions along the reactor length during the blow-down step, also
as a function of the operation time. Notice that the rapid pressure swing
during the blow-down step releases adsorbed CO2 from the adsorbent,
as indicated by the increase in the CO2 molar fraction in the exit stream
in the inlet of the reactor. During the blow- down step a maximum in
the CO2 profile develops, which is indicative of the fact that at the end
of the reaction step the bed is not completely saturated with CO2.
168
Figure 6.28-a. Feed-side H2 mole fraction profiles during the blow-
down step.
169
Figure 6.28-b. Feed-side CO2 mole fraction profiles during the blow-
down step
Figure 6.29 shows the CO2 molar composition along the length of
the reactor in the feed-side during the purge step. At the end of this
step the adsorbent is completely regenerated. As with the conventional
AR, the purge step is necessary to assure optimal and uninterrupted
adsorbent operation.
170
Figure 6.29. Feed-side CO2 mole fraction profiles during the purge step
Figure 6.30 shows the molar composition of H2, H2O and CO2
along the length of the reactor on the feed-side, during the
pressurization step. The small drop in the H2 mole fractions along the
length of the reactor during the pressurization step is attributed to the
permeation onto the permeate side.
171
Figure 6.30. Feed-side H2, H2O and CO2 mole fraction profiles during
the pressurization step
172
Figure 6.31 shows the (feed side) pressure history during the
cyclic HAMR operation. During the step I-reaction of the cyclic
operation the pressure stays high at the reaction operating pressure
(303 kPa for this case). Then, during the blow down step the pressure is
dropped to the atmospheric pressure and stays there during the purge
step. The purge step is followed by the pressurization step, where the
pressure in the reactor bed is increased to reaction operating pressure
(303 kPa in this case).
Figure 6.31 Pressure history during the cyclic HAMR operation
173
6.5 Summary and conclusions
In this thesis, the hybrid adsorbent-membrane reactor (HAMR)
was experimentally evaluated for the WGS and reforming reactions. The
HAMR combines the reaction and membrane separation steps with
adsorption on the membrane feed, or permeate sides. The kinetics of
the WGS reaction over the Cu/Zn catalyst was investigated and data-
validated rate expression and kinetic parameters were obtained from
the experimental data. The kinetics of reforming reaction over
commercial Ni catalyst was investigated. Experimental data for
reforming kinetics generated were fitted to the Xu and Froment (1989)
model. A commercial LDH (MG50 Sasol) was utilized for the high-
temperature CO2 adsorption. The adsorption equilibrium and kinetic
studies were performed by thermogravimetric studies and also in flow
experiments. The experimental isotherm data were fitted to the Sips
isotherm. The adsorption data indicate that under the operating
conditions the capacity of the LDH for CO2 adsorption is fairly high. A
linear driving force model was found to give an accurate description of
the adsorption data. The cyclic adsorption behavior of the adsorbent
was also investigated under the operating conditions. Regeneration
characteristics of the adsorbent were investigated. The spent adsorbent
was completely regenerated by simply reducing the pressure.
Nanoporous CMS and composite palladium membranes were used for
174
the in-situ hydrogen separation. The membranes’ performance was
investigated under the operating conditions by single, binary, and
mixed-gas experiments. The mixed-gas permeances were found to be
close to those of the single-gases. In our simulations, therefore, we used
the mixed-gas permeance data for the model predictions.
The experimental studies indicated good agreement with the
model predictions without any adjustable parameters. The reactor
characteristics were investigated for a range of temperatures, pressure,
and other experimental conditions relevant to the aforementioned
applications. They were compared with the behavior of the traditional
packed-bed reactor, the conventional membrane reactor, and an
adsorptive reactor. The HAMR system is of potential interest to pure
hydrogen production for polymer electrolyte membrane fuel-cells for
various mobile and stationary applications.
The effect of membrane and adsorbent properties on the HAMR
performance was also investigated with the help of the model developed.
The results indicate that use of more effective adsorbents results in
increased conversion, yield and broader operational windows. Highly
permeable membranes also increase the reactor yield but, more
importantly, also increase the hydrogen recovery ratio. One of the key
advantages of the HAMR system over the corresponding AR system (in
addition to improvements in yield) is its ability to deliver a product with
175
a significantly lower CO content through the use of membranes, which
preferentially allow the permeation of the hydrogen while excluding CO
and other reactants and products. This may be the primary reason for
adopting such reactors for fuel-cell application, where a CO-free
product is at a premium.
176
Bibliography
Abanades, J. C., et al. (2004). "Sorbent cost and performance in CO2
capture systems." Industrial & Engineering Chemistry Research 43(13):
3462-3466.
Adhikari, S. and S. Fernando (2006). "Hydrogen membrane separation
techniques." Industrial & Engineering Chemistry Research 45(3): 875-
881.
Alefeld, G. and J. Volkl, eds. (1978). Hydrogen in Metals-II. Berlin,
Springer-Verlag.
Amadeo, N. E. and M. A. Laborde (1995). "Hydrogen-Production from
the Low-Temperature Water-Gas Shift Reaction - Kinetics and
Simulation of the Industrial Reactor." International Journal of Hydrogen
Energy 20(12): 949-956.
Amandusson, H., et al. (2000). "The effect of CO and O-2 on hydrogen
permeation through a palladium membrane." Applied Surface Science
153(4): 259-267.
Amandusson, H., et al. (2001). "Hydrogen permeation through surface
modified Pd and PdAg membranes." Journal of Membrane Science
193(1): 35-47.
Aoki, K., et al. (1996). "Preparation of supported palladium membrane
and separation of hydrogen." Korean Journal of Chemical Engineering
13(5): 530-537.
Askgaard, T. S., et al. (1995). "A Kinetic-Model of Methanol Synthesis."
Journal of Catalysis 156(2): 229-242.
Babu, S. P. (2004). Biomass Gasification for Hydrogen Production -
Process Description and Research Needs. Des Plaines, Gas Technology
Institute.
177
Baker, R. (2004). Membrane Technology and Applications. Menlo Park,
Wiley
Balasubramanian, B., et al. (1999). "Hydrogen from methane in a
single-step process." Chemical Engineering Science 54(15-16): 3543-
3552.
Basile, A., et al. (1996). "Membrane reactor for water gas shift reaction."
Gas Separation & Purification 10(4): 243-254.
Basile, A., et al. (2005a). "A dense Pd/Ag membrane reactor for
methanol steam reforming: Experimental study." Catalysis Today 104(2-
4): 244-250.
Basile, A., et al. (2005b). "Hydrogen production from methanol by
oxidative steam reforming carried out in a membrane reactor." Catalysis
Today 104(2-4): 251-259.
Basile, A., et al. (2003a). "Co-current and counter-current modes for
water gas shift membrane reactor." Catalysis Today 82(1-4): 275-281.
Batista, M. S., et al. (2005). "Evaluation of the water-gas shift and CO
methanation processes for purification of reformate gases and the
coupling to a PEM fuel cell system." Journal of Power Sources 145(1):
50-54.
Berseneva, F. N., et al. (1993). "Alloys of Palladium with Metals of the
Platinum Group as Hydrogen-Permeable Membrane-Components at
High-Temperatures of Gas Separation." International Journal of
Hydrogen Energy 18(1): 15-18.
Beurden, P. v. (2004). On the catalytic aspects of steam-methane
reforming a literature survey.
Bodrov, N. M., L.O. Apel'baum and M.I. Temkin (1964). "Kinetics of the
reaction of methane with water vapor on a nickel surface." Kinetics and
Catalysis 5(4): 696–705.
178
Brunetti, A., et al. (2007). "WGS reaction in a membrane reactor using a
porous stainless steel supported silica membrane." Chemical
Engineering and Processing 46(2): 119-126.
Callaghan, C., et al. (2003). "An improved microkinetic model for the
water gas shift reaction on copper." Surface Science 541(1-3): 21-30.
Callaghan, C. A. (2006). Kinetics and Catalysis of the Water-Gas-Shift
Reaction: A Microkinetic and Graph Theoretic Approach. Department of
Chemical Engineering, Worcester Polytechnic Institute. Ph.D. .
Campbell, C. T. and K. A. Daube (1987). "A Surface Science
Investigation of the Water-Gas Shift Reaction on Cu(111)." Journal of
Catalysis 104(1): 109-119.
Carvill, B. T., et al. (1996). "Sorption-enhanced reaction process." Aiche
Journal 42(10): 2765-2772.
Chen, C. S., et al. (2004). "Study of iron-promoted Cu/SiO2 catalyst on
high temperature reverse water gas shift reaction." Applied Catalysis a-
General 257(1): 97-106.
Chen, S. H., et al. (2002). "A palladium alloy composite membrane for
the purification of hydrogen isotopes." Separation Science and
Technology 37(11): 2701-2716.
Chen, Y. Z., et al. (2006). "Hydrogen production from the steam
reforming of liquid hydrocarbons in membrane reactor." Catalysis Today
118(1-2): 136-143.
Chen, Z. X. and S. S. E. H. Elnashaie (2004a). "Steady-state modeling
and bifurcation behavior of Circulating Fluidized Bed Membrane
Reformer-Regenerator for the production of hydrogen for fuel cells from
heptane." Chemical Engineering Science 59(18): 3965-3979.
Chen, Z. X. and S. S. E. H. Elnashaie (2004b). "Bifurcation behavior
and efficient pure hydrogen production for fuel cells using a novel
autothermic membrane circulating fluidized-bed (CFB) reformer:
179
Sequential debottlenecking and the contribution of John Grace."
Industrial & Engineering Chemistry Research 43(18): 5449-5459.
Chen, Z. X. and S. S. E. H. Elnashaie (2005). "Economics of the clean
fuel hydrogen in a novel autothermal reforming process." Industrial &
Engineering Chemistry Research 44(13): 4834-4840.
Chiappetta, G., et al. (2006). "Improvement of membrane reactor
performance in hydrogen production." Desalination 200(1-3): 695-696.
Choi, Y. and H. G. Stenger (2003). "Water gas shift reaction kinetics and
reactor modeling for fuel cell grade hydrogen." Journal of Power Sources
124(2): 432-439.
Colbourn, E., et al. (1991). "Adsorption of Water on Polycrystalline
Copper - Relevance to the Water Gas Shift Reaction." Journal of
Catalysis 130(2): 514-527.
Crank, J. (1975). Mathematics of Diffusion, 2nd edition, Clarendon Press,
Oxford.
Darwish, N. A., et al. (2004). "Feasibility of the direct generation of
hydrogen for fuel-cell-powered vehicles by on-board steam reforming of
naphtha." Fuel 83(4-5): 409-417.
Ding, Y. and E. Alpay (2000a). "Equilibria and kinetics of CO2
adsorption on hydrotalcite adsorbent." Chemical Engineering Science
55(17): 3461-3474.
Ding, Y. and E. Alpay (2000b). "Adsorption-enhanced steam-methane
reforming." Chemical Engineering Science 55(18): 3929-3940.
Ding, Y. and E. Alpay (2001). "High temperature recovery of CO2 from
flue cases using hydrotalcite adsorbent." Process Safety and
Environmental Protection 79(B1): 45-51.
180
Dresselhaus, M. (2003). Basic Research Needs for The Hydrogen
Economy. Chicago, Argonne National Laboratory.
Edwards, M. F. and J. F. Richardson (1968). "Gas Dispersion in Packed
Beds." Chemical Engineering Science 23: 109.
Eley, D. D., Pines, H., Weisz, P. B., ed. (1979). The Kinetics of Some
Industrial Heterogeneous Catalytic Reactions. New York, Academic Press.
Ermilova, M. M., et al. (1988). "Changes in the Catalytic Activity and
Hydrogen Permeability of a Palladium Ruthenium Alloy Membrane
Catalyst under the Influence of Reagents." Bulletin of the Academy of
Sciences of the Ussr Division of Chemical Science 37(4): 637-640.
Estrin, B. M. and Roizenma.R. (1972). "Hydrogen Purification by
Diffusion through a Membrane from a Multicomponent Alloy Based on
Palladium." Journal of Applied Chemistry(MAR): 661-663.
Ferreira-Aparicio, P. and M. J. Benito (2005). "High performance
membrane reactor system for hydrogen production from methane."
Industrial & Engineering Chemistry Research 44(4): 742-748.
Fort, D. and I. R. Harris (1975). "Physical-Properties of Some Palladium
Alloy Hydrogen Diffusion Membrane Materials." Journal of the Less-
Common Metals 41(2): 313-327.
Gallucci, F., et al. (2004). "A simulation study of the steam reforming of
methane in a dense tubular membrane reactor." International Journal
of Hydrogen Energy 29(6): 611-617.
Gallucci, F., et al. (2004). "Experimental study of the methane steam
reforming reaction in a dense Pd/Ag membrane reactor." Industrial &
Engineering Chemistry Research 43(4): 928-933.
Gao, H. Y., et al. (2005). "Electroless plating synthesis, characterization
and permeation properties of Pd-Cu membranes supported on ZrO2
modified porous stainless steel." Journal of Membrane Science 265(1-2):
142-152.
181
Gines, M. J. L., et al. (1995). "Activity and Structure-Sensitivity of the
Water-Gas Shift Reaction over Cu-Zn-Al Mixed-Oxide Catalysts."
Applied Catalysis a-General 131(2): 283-296.
Gines, M. J. L. and C. R. Apesteguia (1997). "Thermal decomposition of
Cu-based hydroxycarbonate catalytic precursors for the low-
temperature CO-shift reaction." Journal of Thermal Analysis and
Calorimetry 50(5-6): 745-756.
Gines, M. J. L., et al. (1997). "Kinetic study of the reverse water-gas
shift reaction over CuO/ZnO/Al2O3 catalysts." Applied Catalysis a-
General 154(1-2): 155-171.
Gluud, W., et al. (1931). Production of Hydrogen. U.S. Patent 1,816,523.
Gregory, D. P. and J. B. Pangborn (1976). Hydrogen energy, Annual
Review Energy: 279-310.
Guazzone, F., et al. (2006). "Effects of surface activity, defects and mass
transfer on hydrogen permeance and n-value in composite palladium-
porous stainless steel membranes." Catalysis Today 118(1-2): 24-31.
Guo, Y. L., et al. (2003). "Preparation and characterization of Pd-
Ag/ceramic composite membrane and application to enhancement of
catalytic dehydrogenation of isobutane." Separation and Purification
Technology 32(1-3): 271-279.
Hemmes K. , G. A., Uil H. (2003). Bio-H2. Petten, Energy Research
Foundation of the Netherlands
Hollein, V., et al. (2001). "Preparation and characterization of palladium
composite membranes for hydrogen removal in hydrocarbon
dehydrogenation membrane reactors." Catalysis Today 67(1-3): 33-42.
Howard, R. (2000). Handbook of Commercial Catalysts : Heterogenous
Catalysis. Boca Arton, CRC.
182
Hsieh, H. P. (1996). Inorganic Membranes for Separation and Reaction.
Amsterdam, Elsevier Science B.V.
Hufton, J. R., et al. (1999). "Sorption-enhanced reaction process for
hydrogen production." Aiche Journal 45(2): 248-256.
Hulbert, R. C. and J. O. Konecny (1961). "Diffusion of hydrogen through
palladium." Journal of Chemical Physics 34: 655-658.
Hwang, S. T. (2001). "Inorganic membranes and membrane reactors."
Korean Journal of Chemical Engineering 18(6): 775-787.
Itoh, N., et al. (2001). "Hydrogen production via methanol steam
reforming in a palladium membrane reactor." Abstracts of Papers of the
American Chemical Society 222: U337-U337.
Iyoha, O., et al. (2007). "Wall-catalyzed water-gas shift reaction in
multi-tubular Pd and 80 wt%Pd-20 wt%Cu membrane reactors at 1173
K." Journal of Membrane Science 298(1-2): 14-23.
Jordal, K., et al. (2004). "Integration of H-2-separating membrane
technology in gas turbine processes for CO2 capture." Energy 29(9-10):
1269-1278.
Jung, S. H., et al. (2000). "Effects of co-existing hydrocarbons on
hydrogen permeation through a palladium membrane." Journal of
Membrane Science 170(1): 53-60.
Karagiannakis, G., et al. (2003). "Hydrogenation of carbon dioxide on
copper in a H+ conducting membrane-reactor." Solid State Ionics 162:
313-318.
Karger, J. R., D.M (1992). Diffusion in Zeolites and Other Microporous
Solids. New York, Wiley
Kikuchi, E. (2000). "Membrane reactor application to hydrogen
production." Catalysis Today 56(1-3): 97-101.
183
Kikuchi, E., et al. (1989). "Membrane Reactor Using Microporous Glass-
Supported Thin-Film of Palladium .1. Application to the Water Gas Shift
Reaction." Chemistry Letters(3): 489-492.
Kim, N., et al. (2005). "Atomistic simulation of nanoporous layered
double hydroxide materials and their properties. I. Structural
modeling." Journal of Chemical Physics 122(21): -.
Kim, Y., et al. (2004). "Thermal evolution of the structure of a Mg-Al-
CO3 layered double hydroxide: Sorption reversibility aspects."
Industrial & Engineering Chemistry Research 43(16): 4559-4570.
Kluiters, S. C. A. (2004). Status review on membrane systems for
hydrogen separation, ECN.
Kulprathipanja, A., et al. (2004). "Effects of water gas shift gases on Pd-
Cu alloy membrane surface morphology and separation properties."
Industrial & Engineering Chemistry Research 43(15): 4188-4198.
Kulprathipanja, A., et al. (2005). "Pd and Pd-Cu membranes: inhibition
of H-2 permeation by H2S." Journal of Membrane Science 254(1-2): 49-
62.
Kusakabe, K., et al. (2006). "Methane steam reforming in a zirconia
membrane reactor." Journal of Chemical Engineering of Japan 39(4):
444-447.
Ladebeck, J. R. and J. P. Wagner (2003). Catalyst Development for
Water-Gas Shift. Handbook of Fuel Cells – Fundamentals, Technology
and Applications. Chichester, Wiley
Lee, K. B., et al. (2007). "Chemisorption of carbon dioxide on
potassium-carbonate-promoted hydrotalcite." Journal of Colloids and
Interface Science 308(1): 30-39.
Levenspiel, O. (1998). Chemical Reaction Engineering. New York, Wiley.
184
Li, A. W., et al. (1999a). "Fabrication of defect-free Pd/alpha-Al2O3
composite membranes for hydrogen separation." Thin Solid Films
350(1-2): 106-112.
Li, A. W., et al. (1999b). "Repair of a Pd/alpha-Al2O3 composite
membrane containing defects." Separation and Purification Technology
15(2): 113-119.
Li, A. W., et al. (2000). "Fabrication of dense palladium composite
membranes for hydrogen separation." Catalysis Today 56(1-3): 45-51.
Li, A. W., et al. (1999). "Repair of Pd/alpha-Al2O3 composite membrane
with defects." Science in China Series B-Chemistry 42(6): 612-616.
Lim, S. Y., et al. (2002). "Design issues of pervaporation membrane
reactors for esterification." Chemical Engineering Science 57(22-23):
4933-4946.
Liu, Z. W., et al. (2002). "Hydrogen production for fuel cells through
methane reforming at low temperatures." Journal of Power Sources
111(2): 283-287.
Ma, D. H. and C. R. F. Lund (2003). "Assessing high-temperature water-
gas shift membrane reactors." Industrial & Engineering Chemistry
Research 42(4): 711-717.
Ma, Y. H. and F. Guazzone (2007). "Metallic membranes for the
separation of hydrogen at high temperatures." Annales De Chimie-
Science Des Materiaux 32(2): 179-195.
Ma, Y. H., et al. (2003). "Thin composite palladium and palladium/alloy
membranes for hydrogen separation." Advanced Membrane Technology
984: 346-360.
Malek, A. and S. Farooq (1997). "Study of a six-bed pressure swing
adsorption process." Aiche Journal 43(10): 2509-2523.
185
Mardilovich, I. P., et al. (2002). "Dependence of hydrogen flux on the
pore size and plating surface topology of asymmetric Pd-porous
stainless steel membranes." Desalination 144(1-3): 85-89.
McHugh, K. (2005). Hydrogen production methods. ALEXANDRIA, MPR
Associates, Inc.: 1-41.
Millar, G. J., et al. (1992). "A Combined Infrared, Temperature
Programmed Desorption and Temperature Programmed Reaction
Spectroscopy Study of Co2 and H2 Interactions on Reduced and
Oxidized Silica-Supported Copper-Catalysts." Molecular Physics 76(4):
833-849.
Moreira, R. F. P. M., et al. (2006). "Adsorption of CO2 on hydrotalcite-
like compounds in a fixed bed." Separation Science and Technology
41(2): 341-357.
Nagamoto, H. and H. Inoue (1985). "A Reactor with Catalytic Membrane
Permeated by Hydrogen." Chemical Engineering Communications 34(1-
6): 315-323.
Nakamura, J., et al. (1990). "Kinetics and Mechanism of the Water-Gas
Shift Reaction Catalyzed by the Clean and Cs-Promoted Cu(110)
Surface - a Comparison with Cu(111)." Journal of the Chemical Society-
Faraday Transactions 86(15): 2725-2734.
Nam, S. E. and K. H. Lee (2001). "Hydrogen separation by Pd alloy
composite membranes: introduction of diffusion barrier." Journal of
Membrane Science 192(1-2): 177-185.
Nam, S. W., et al. (2000). "Methane steam reforming in a Pd-Ru
membrane reactor." Korean Journal of Chemical Engineering 17(3):
288-291.
Noble, R. D. and J. L. Falconer (1995). "Silicalite-1 Zeolite Composite
Membranes." Catalysis Today 25(3-4): 209-212.
186
O'Brien, J., et al. (2001). "Pd/Ag membranes on porous alumina
substrates by unbalanced magnetron sputtering." Surface & Coatings
Technology 142: 253-259.
Ogden, J. (2001). Review of Small Statonary Reformers for Hydrogen
Production. Golden, The International Energy Agency.
Ogden, J. M., et al. (1999). "A comparison of hydrogen, methanal and
gasoline as fuels for fuel cell vehicles: implications for vehicle design
and infrastructure development." Journal of Power Sources 79(2): 143-
168.
Ortiz, A. L. and D. P. Harrison (2001). "Hydrogen production using
sorption-enhanced reaction." Industrial & Engineering Chemistry
Research 40(23): 5102-5109.
Ovesen, C. V., et al. (1996). "A microkinetic analysis of the water-gas
shift reaction under industrial conditions." Journal of Catalysis 158(1):
170-180
Ovesen, C. V., et al. (1992). "A Kinetic-Model of the Water Gas Shift
Reaction." Journal of Catalysis 134(2): 445-468.
Padró, C. E. G. and V. Putsche (1999). Survey of the Economics of
Hydrogen Technologies. Golden, National Renewable Energy Laboratory.
Paglieri, S. N., et al. (1999). "A new preparation technique for Pd
alumina membranes with enhanced high-temperature stability."
Industrial & Engineering Chemistry Research 38(5): 1925-1936.
Paglieri, S. N. and J. D. Way (2002). "Innovations in palladium
membrane research." Separation and Purification Methods 31(1): 1-169.
Park, B., et al. (1998). "Models and simulation of liquid-phase
membrane reactors." Industrial & Engineering Chemistry Research
37(4): 1276-1289.
187
Pavlova, L. F., et al. (1972). "Study of Permeability of Hydrogen through
Palladium - Nickel-Alloy When a Hydrogen Pressure-Gradient Is Present
Along Length of Membrane." Doklady Akademii Nauk Sssr 207(4): 919-
922.
Philcox, J. E. and G. W. Fenner (1996). Gasification-An Attraction for
Chemical Reactions. Danbury, Praxair.
Prasad, P. and S. S. E. H. Elnashaie (2002). "Novel circulating fluidized-
bed membrane reformer for the efficient production of ultraclean fuels
from hydrocarbons." Industrial & Engineering Chemistry Research
41(25): 6518-6527.
Prasad, P. and S. S. E. H. Elnashaie (2003). "Coupled steam and
oxidative reforming for hydrogen production in a novel membrane
circulating fluidized-bed reformer." Industrial & Engineering Chemistry
Research 42(20): 4715-4722.
Prasad, P. and S. S. E. H. Elnashaie (2004). "Novel circulating fluidized-
bed membrane reformer using carbon dioxide sequestration." Industrial
& Engineering Chemistry Research 43(2): 494-501.
Reijers, H. T. J., et al. (2006). "Hydrotalcite as CO2 sorbent for sorption-
enhanced steam reforming of methane." Industrial & Engineering
Chemistry Research 45(8): 2522-2530.
Rhodes, C., et al. (1995). "Water-Gas Shift Reaction - Finding the
Mechanistic Boundary." Catalysis Today 23(1): 43-58.
Roa, F., et al. (2002). "The influence of alloy composition on the H(2)flux
of composite Pd-Cu membranes." Desalination 147(1-3): 411-416.
Roa, F. and J. D. Way (2003a). "Influence of alloy composition and
membrane fabrication on the pressure dependence of the hydrogen flux
of palladium-copper membranes." Industrial & Engineering Chemistry
Research 42(23): 5827-5835.
188
Roa, F. and J. D. Way (2003b). "Alloy composition effect on the N-value
for hydrogen-selective palladium-copper membranes." Abstracts of
Papers of the American Chemical Society 225: U867-U867.
Rostrup-Nielsen, J. R. and I. Alstrup (1999). "Innovation and science in
the process industry - Steam reforming and hydrogenolysis." Catalysis
Today 53(3): 311-316.
Rothenberger, K. S., et al. (2004). "High pressure hydrogen permeance
of porous stainless steel coated with a thin palladium film via
electroless plating." Journal of Membrane Science 244(1-2): 55-68.
Sanchez, J. G. and T. Tsotsis, T. (2002). Catalytic Membranes and
Catalytic Membrane Reactors. Weinheim, Wiley-VCH.
Sasol (2003). Hydrotalcite-PURAL MG. Johannesburg Sasol.
Schiesser, W. E. (1991). The Numerical Method of Lines: Integration of
Partial Differential Equations. San Diego, Academic Press.
Semelsberger, T. A., et al. (2004). "Equilibrium products from
autothermal processes for generating hydrogen-rich fuel-cell feeds."
International Journal of Hydrogen Energy 29(10): 1047-1064.
Sereno, C. and A. E. Rodrigues (1993). "Can steady state momentum
equations be used in modelling pressurization of adsorption beds?" Gas
Separation and Purification 7: 167.
Shido, T. and Y. Iwasawa (1993). "The Effect of Coadsorbates in Reverse
Water Gas Shift Reaction on Zno, in Relation to Reactant-Promoted
Reaction-Mechanism." Journal of Catalysis 140(2): 575-584.
Shoko, E., et al. (2006). "Hydrogen from coal: Production and utilisation
technologies." International Journal of Coal Geology 65(3-4): 213-222.
Sholl, D. S. and Y. H. Ma (2006). "Dense metal membranes for the
production of high-purity hydrogen." Mrs Bulletin 31(10): 770-773.
189
Shu, J., et al. (1993). "Simultaneous Deposition of Pd and Ag on Porous
Stainless-Steel by Electroless Plating." Journal of Membrane Science
77(2-3): 181-195.
Shu, J., et al. (1994b). "Methane Steam Reforming in Asymmetric Pd-Ag
and Pd-Ag/Porous Ss Membrane Reactors." Applied Catalysis a-General
119(2): 305-325.
Shu, J., et al. (1995). "Asymmetric Pd-Ag Stainless-Steel Catalytic
Membranes for Methane Steam Reforming." Catalysis Today 25(3-4):
327-332.
Shu, J., et al. (1991). "Catalytic Palladium-Based Membrane Reactors -
a Review." Canadian Journal of Chemical Engineering 69(5): 1036-1060.
Sircar, S. (2002). "Pressure swing adsorption." Industrial & Engineering
Chemistry Research 41(6): 1389-1392.
Sircar, S. and J. R. Hufton (2000). "Why does the Linear Driving Force
model for adsorption kinetics work?" Adsorption-Journal of the
International Adsorption Society 6(2): 137-147.
Sircar, S., et al. (1999). "Hydrogen production by hybrid SMR-PSA-SSF
membrane system." Separation and Purification Technology 17(1): 11-
20.
Soares, J. L., et al. (2005). "Experimental and theoretical analysis for
the CO2 adsorption on hydrotalcite." Adsorption-Journal of the
International Adsorption Society 11: 237-241.
Soares, J. L., et al. (2004). "Hydrotalcite materials for carbon dioxide
adsorption at high temperatures: Characterization and diffusivity
measurements." Separation Science and Technology 39(9): 1989-2010.
Soares, J. L., et al. (2007). "Carbon dioxide adsorption in Brazilian
coals." Energy & Fuels 21(1): 209-215.
190
Stöcker, J., et al. (2005). 30 Years of PSA Technology for Hydrogen
Purification. Des Plaines, UOP LLC,.
Sun, G. B., et al. (2006). "Ultra thin Pd membrane on alpha-Al2O3
hollow fiber by electroless plating: High permeance and selectivity."
Journal of Membrane Science 284(1-2): 110-119.
Suzuki, Y. and S. Kimura (1984). "Separation and Concentration of
Hydrogen Isotopes by Palladium Alloy Membrane .1. Measurement of
Solubility and Permeation Rate of Hydrogen and Deuterium, and
Analysis of Isotopic and Non-Isotopic Separation by Numerical-Solution
Method of Velocity and Concentration Boundary-Layer Equations."
Journal of the Atomic Energy Society of Japan 26(9): 802-810.
Suzuki, Y. and S. Kimura (1993). "Separation and Concentration of
Hydrogen Isotopes by a Palladium Alloy Membrane." Nuclear
Technology 103(1): 93-100.
Thoen, P. M., et al. (2006). "High flux palladium-copper composite
membranes for hydrogen separations." Desalination 193(1-3): 224-229.
Tien, C. (1994). Adsorption Calculations and Modeling., Butterworth-
Heinemann, Boston.
Timofeev, N. I., et al. (1994). "New Palladium-Based Membrane Alloys
for Separation of Gas-Mixtures to Generate Ultrapure Hydrogen."
International Journal of Hydrogen Energy 19(11): 895-898.
Tong, H. and Y. Matsumura (2006). "Pure hydrogen production by
methane steam reforming with hydrogen-permeable membrane reactor."
Catalysis Today 111(3-4): 147-152.
Tosti , S., et al. (2003). "Pd-Ag membrane reactors for water gas shift
reaction." Chemical Engineering Journal 93 23-30.
Tosti, S., et al. (2002). "Sputtered, electroless, and rolled palladium-
ceramic membranes." Journal of Membrane Science 196(2): 241-249.
191
Tosti, S., et al. (2000). "Rolled thin Pd and Pd-Ag membranes for
hydrogen separation and production." International Journal of
Hydrogen Energy 25(4): 319-325.
Trimm, D. L. and Z. I. Onsan (2001). "Onboard fuel conversion for
hydrogen-fuel-cell-driven vehicles." Catalysis Reviews-Science and
Engineering 43(1-2): 31-84.
Tserpe, E. and K. C. Waugh (1997). "A microkinetic analysis of the
reverse water gas shift reaction." Dynamics of Surfaces and Reaction
Kinetics in Heterogeneous Catalysis 109: 401-416.
Tsuru, T., et al. (2006). "A bimodal catalytic membrane having a
hydrogen-permselective silica layer on a bimodal catalytic support:
Preparation and application to the steam reforming of methane."
Applied Catalysis a-General 302(1): 78-85.
Twigg, M. V., ed. (1989). Catalyst Handbook. London, Wolfe Publishing
Ltd.
Twigg, M. V. and M. S. Spencer (2001). "Deactivation of supported
copper metal catalysts for hydrogenation reactions." Applied Catalysis
a-General 212(1-2): 161-174.
Uemiya, S., et al. (1991). "Hydrogen Permeable Palladium Silver Alloy
Membrane Supported on Porous Ceramics." Journal of Membrane
Science 56(3): 315-325.
Uemiya, S., et al. (1991). "The Water Gas Shift Reaction Assisted by a
Palladium Membrane Reactor." Industrial & Engineering Chemistry
Research 30(3): 585-589.
Waldron, W. E. and S. Sircar (2000). "Parametric study of a pressure
swing adsorption process." Adsorption-Journal of the International
Adsorption Society 6(2): 179-188.
192
Wang, Y. N. and A. E. Rodrigues (2005). "Hydrogen production from
steam methane reforming coupled with in situ CO2 capture: Conceptual
parametric study." Fuel 84(14-15): 1778-1789.
Ward, T. L. and T. Dao (1999). "Model of hydrogen permeation behavior
in palladium membranes." Journal of Membrane Science 153(2): 211-
231.
Waugh, K. C. (1999). "Prediction of global reaction kinetics by solution
of the Arrhenius parameterised component elementary reactions:
microkinetic analysis." Catalysis Today 53(2): 161-176.
Wouwer, A. V., et al. (2004). "Simulation of distributed parameter
systems using a Matlab-based method of lines toolbox: Chemical
engineering applications." Industrial & Engineering Chemistry Research
43(14): 3469-3477.
Xiu, G. H., et al. (2002). "Sorption-enhanced reaction process with
reactive regeneration." Chemical Engineering Science 57(18): 3893-3908.
Xiu, G. H., et al. (2003a). "Adsorption-enhanced steam-methane
reforming with intraparticle-diffusion limitations." Chemical
Engineering Journal 95(1-3): 83-93.
Xiu, G. H., et al. (2003b). "New generalized strategy for improving
sorption-enhanced reaction process." Chemical Engineering Science
58(15): 3425-3437.
Xiu, G. H., et al. (2004). "Subsection-controlling strategy for improving
sorption-enhanced reaction process." Chemical Engineering Research &
Design 82(A2): 192-202.
Xiu, G. H., et al. (2002). "Simulation of five-step one-bed sorption-
enhanced reaction process." Aiche Journal 48(12): 2817-2832.
Xomeritakis, G. and Y. S. Lin (1998). "CVD synthesis and gas
permeation properties of thin palladium/alumina membranes." Aiche
Journal 44(1): 174-183.
193
Yacobucci, B. D. (2004). A Hydrogen Economy and Fuel Cells:An
Overview. CRS report for Congress.
Yan, S. C., et al. (1994). "Thin Palladium Membrane Formed in Support
Pores by Metal-Organic Chemical-Vapor-Deposition Method and
Application to Hydrogen Separation." Industrial & Engineering
Chemistry Research 33(3): 616-622.
Yong, Z., et al. (2002). "Adsorption of carbon dioxide at high
temperature - a review." Separation and Purification Technology 26(2-
3): 195-205.
Yong, Z. and A. E. Rodrigues (2002). "Hydrotalcite-like compounds as
adsorbents for carbon dioxide." Energy Conversion and Management
43(14): 1865-1876.
Zou, Y. and A. E. Rodrigues (2001a). "The separation enhanced reaction
process (SERP) in the production of hydrogen from methane steam
reforming." Adsorption Science & Technology 19(8): 655-671.
Zou, Y. and A. E. Rodrigues (2001b). "Adsorbent materials for carbon
dioxide." Adsorption Science & Technology 19(3): 255-266.
194
APPENDIX A
MG50 BET characterisation report
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
APPENDIX B
CMS Membrane Summary
215
216
217
218
219
220
221
222
223
224
Abstract (if available)
Abstract
As a result of stricter environmental regulations worldwide, hydrogen is progressively becoming an important clean energy source. For H2 to replace fossil fuels in mobile applications, it will require the creation of a production and delivery infrastructure equivalent to that currently existing for fossil fuels, which is an immense task. As an alternative, and as an interim step towards the new hydrogen economy, various groups are currently studying steam reforming of methane (SRM) for the on-board generation of hydrogen, or for on site production, in order to alleviate the need for compressed or liquid hydrogen gas storage. Conventional technologies are, however, neither convenient nor economical to apply for small-scale (on site or on-board) hydrogen generation. Reactive separation processes have, as a result, been attracting renewed interest for application in H2 production through SRM. One such technology is the hybrid adsorbent-membrane reactor (HAMR) system, which couples reaction and membrane separation steps with adsorption on the reactor and/or membrane permeate side. The HAMR concept was originally proposed by USC group for esterification reactions, and it was adapted recently for on-board or onsite hydrogen production applications. Our early studies involved the development of a mathematical model for the HAMR system (applied to hydrogen production through SRM)
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
An integrated 'one-box' process for hydrogen production
PDF
Methanol synthesis in a membrane reactor
PDF
The use of carbon molecule sieve and Pd membranes for conventional and reactive applications
PDF
On the use of membrane reactors in biomass utilization
PDF
Fabrication of nanoporous silicon carbide membranes for gas separation applications
PDF
Studies of transport phenomena in hydrotalcite membranes, and their use in direct methanol fuel cells
PDF
A study of the application of membrane-based reactive separation to the carbon dioxide methanation
PDF
A flow-through membrane reactor for destruction of a chemical warfare simulant
PDF
Process intensification in hydrogen production via membrane-based reactive separations
PDF
A high efficiency, ultra-compact process for pre-combustion CO₂ capture
PDF
Hydrogen storage in carbon and silicon carbide nanotubes
PDF
Methanol synthesis in the membrane reactor
PDF
Fabrication of silicon carbide sintered supports and silicon carbide membranes
PDF
A process-based molecular model of nano-porous silicon carbide membranes
PDF
Molecular modeling of silicon carbide nanoporous membranes and transport and adsorption of gaseous mixtures therein
PDF
Mesoscale SOFC-based power generator system: modeling and experiments
PDF
Preparation of polyetherimide nanoparticles by electrospray drying, and their use in the preparation of mixed-matrix carbon molecular-sieve (CMS) membranes
PDF
Chemical and mechanical deformation of porous media and materials during adsorption and fluid flow
PDF
Novel methods for landfill gas and biogas clean-up
PDF
Performance prediction, state estimation and production optimization of a landfill
Asset Metadata
Creator
Harale, Aadesh
(author)
Core Title
A hybrid adsorbent-membrane reactor (HAMR) system for hydrogen production
School
Viterbi School of Engineering
Degree
Doctor of Philosophy
Degree Program
Chemical Engineering
Publication Date
09/23/2008
Defense Date
10/08/2007
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
adsorption,CO₂ capture,HAMr,hydrogen,membrane reactor,OAI-PMH Harvest
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Sahimi, Muhammad (
committee chair
), Tsotsis, Theodore T. (
committee chair
), Wang, Hai (
committee member
)
Creator Email
aharale@gmail.com,harale@usc.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-m1614
Unique identifier
UC1118402
Identifier
etd-harale-1690 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-96604 (legacy record id),usctheses-m1614 (legacy record id)
Legacy Identifier
etd-harale-1690.pdf
Dmrecord
96604
Document Type
Dissertation
Rights
Harale, Aadesh
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Repository Name
Libraries, University of Southern California
Repository Location
Los Angeles, California
Repository Email
cisadmin@lib.usc.edu
Tags
adsorption
CO₂ capture
HAMr
hydrogen
membrane reactor