Water Gas Shift Catalysis
Transcript of Water Gas Shift Catalysis
This article was downloaded by: [University of Utah]On: 12 April 2013, At: 23:18Publisher: Taylor & FrancisInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Catalysis Reviews: Science andEngineeringPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/lctr20
Water Gas Shift CatalysisChandra Ratnasamy a & Jon P. Wagner aa Sud Chemie, Louisville, KY, USAVersion of record first published: 05 Aug 2009.
To cite this article: Chandra Ratnasamy & Jon P. Wagner (2009): Water Gas Shift Catalysis, CatalysisReviews: Science and Engineering, 51:3, 325-440
To link to this article: http://dx.doi.org/10.1080/01614940903048661
PLEASE SCROLL DOWN FOR ARTICLE
Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions
This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden.
The publisher does not give any warranty express or implied or make any representationthat the contents will be complete or accurate or up to date. The accuracy of anyinstructions, formulae, and drug doses should be independently verified with primarysources. The publisher shall not be liable for any loss, actions, claims, proceedings,demand, or costs or damages whatsoever or howsoever caused arising directly orindirectly in connection with or arising out of the use of this material.
Water Gas Shift Catalysis
Chandra Ratnasamy and Jon P. Wagner
Sud Chemie, Louisville, KY, USA
Developments in water gas shift (WGS) catalysis, especially during the last decade, arereviewed. Recent developments include the development of (1) chromium-free catalyststhat can operate at lower steam to gas ratios and (2) more active catalysts that canoperate at gas hourly space velocities above 40,000 h21. A current challenge is todevelop catalysts for use in fuel cell applications. Precious metal catalysts supported onpartially reducible oxide supports (Pt-ceria, Pt-titania, Au-ceria, etc.) are the currentfront runners. A critical review of the mechanism of the WGS reaction is alsopresented.
Keywords Water gas shift, Hydrogen production, CO conversions, Fuel processor,Fuel cell, Iron oxide catalysts, Copper-Zinc oxide catalysts, Pt catalysts,Redox mechanism, Formate mechanism, High temperature shift, Lowtemperature shift, Sour gas shift, Chromium-free catalysts
1. INTRODUCTION
‘‘Water gas’’ is a mixture of hydrogen and carbon monoxide. It is used
extensively in the industry for the manufacture of ammonia, methanol,
hydrogen (for hydrotreating, hydrocracking of petroleum fractions and other
hydrogenations in the petroleum refining and petrochemical industry),
hydrocarbons (by the Fischer-Tropsch process) and metals (by the reduction
of the oxide ore). It is manufactured by the reaction of a carbonaceous material
(coal, coke, natural gas, naphtha, etc.) with steam [Eqs. (1, 2)], oxygen [Eq. (3)]
or carbon dioxide [Eq. (4)]:
CzH2O<COzH2 H2=CO~1; DH~131:2 kJ=molð Þ, ð1Þ
CH4zH2O<COz3 H2 H2=CO~3; DH~206:3 kJ=molð Þ, ð2Þ
Received 24 August 2008; accepted 17 February 2009.Address correspondence to Chandra Ratnasamy, Sud Chemie, Louisville, KY 40210,USA. E-mail: [email protected]
Catalysis Reviews, 51:325–440, 2009
Copyright # Taylor & Francis Group, LLC
ISSN: 0161-4940 print 1520-5703 online
DOI: 10.1080/01614940903048661
325
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
CH4z0:5 O2<COz2H2 H2=CO~2; DH~{35:6 kJ=molð Þ, and ð3Þ
CH4zCO2<2COz2H2 H2=CO~1; DH~247:4 kJ=molð Þ: ð4Þ
Reactions 1, 2, and 4 are endothermic while reaction 3 is exothermic. It may be
noted that the molar ratio of H2 to CO varies depending on the source of
carbon/oxygen.
Steam reforming [Eq. (2)] is the most popular mode of generating water
gas, especially if the ultimate objective is the generation of pure hydrogen
since it provides the highest molar ratio of H2/CO of 3. The exothermic partial
oxidation [Eq. (3)] providing a H2/CO molar ratio of 2 is used in the
manufacture of water gas when (a) a lower H2/CO ratio (( 2, for example)
is needed (e.g., dimethyl ether or Fischer Tropsch synthesis), or (b) due to
difficulties in external heat supply, internal heat generation (autothermal
reforming) is needed as in the case of fuel processors for fuel cell applications.
‘‘Dry reforming’’ or ‘‘CO2 reforming’’ (Eq. 4) is an additional source of water gas
with a very low H2/CO molar ratio of one. This process is used in the
manufacture of water gas from natural gas for the reduction of iron ore
wherein CO has been found to be as good a reductant (if not better) as H2.
The water gas shift reaction [Eq. (5)] was first reported in 1888 (1), but it
came into popular usage later, as a source of hydrogen for the Haber process
for the manufacture of ammonia:
CO gð ÞzH2O gð Þ<CO2 gð ÞzH2 gð Þ DH~{41:1 kJ=molð Þ: ð5Þ
In the initial stages of the Haber ammonia process, the hydrogen needed for
the process was obtained from the water gas generated by Eq. (1). In coal or
coke gasification [Eq. (1)], when steam is contacted with incandescent coke (at
about 1000uC), CO2 is an additional product [Eq. (6)] especially at lower
temperatures:
Cz2H2O<CO2z2H2 DH~90 kJ=molð Þ: ð6Þ
While CO2 could be easily removed from the products of the reaction (by
absorption in water), CO had to be removed by liquefaction or copper liquor
scrubbing. A catalytic process to remove the CO from (CO + H2) mixtures was
needed. In 1914, Bosch and Wild (2) discovered that the oxides of iron and
chromium could convert a mixture of steam and CO into CO2 at 400–500uC,
according to Eq. 5, and, in the process, generate additional hydrogen for the
Haber process. Thenceforth, the water gas produced from the carbonaceous
source by steam reforming was passed over the iron-chromium catalyst to shift
326 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
the CO to CO2 by the water gas shift reaction. Iron-based catalysts are still
used today industrially. There are four general types of water gas shift
catalysts. One of them is the promoted iron oxide catalyst. Catalysts of this
type promote the shift reaction at moderately high temperatures (350–450uC)
and are therefore called high temperature shift (HTS) catalysts. The second
type is copper-zinc oxide catalyst and is called the low temperature shift (LTS)
catalyst because it is used at relatively low temperatures (190–250uC). The
third type employs cobalt and molybdenum sulfides as the active ingredients.
Catalysts of this type are sulfur-tolerant and can be used in sulfur-containing
‘‘sour gas’’ streams and are therefore called sour gas shift catalysts. There was
interest (in the past) in a fourth type of catalyst, medium temperature shift or
MTS catalyst that operates at temperatures between the HTS and LTS
catalysts. Normally, these are copper-zinc catalysts that are actually LTS
catalysts modified (usually with iron oxide) to operate at slightly higher
temperatures (275–350uC) than a standard LTS catalyst. In addition to the
above four, precious metal- based catalysts (mainly platinum and gold) have
been under intensive investigation during the last decade for use in fuel cell
applications. Promoters, like Cu and Al2O3 are added to the conventional iron
oxide - chromium oxide HTS catalyst compositions in some modern versions.
At lower temperatures, the iron-based catalysts are less active.
Equilibrium concentrations of CO are lower only at low temperatures (section
II), however. Hence, to achieve higher conversions of CO at lower
temperatures (190–250uC), a second, more active catalyst, based on Cu-ZnO,
was developed in the early 1960s and is used in the industry extensively. It
may be noted that Cu- based catalysts had been patented as early as 1931 (3).
Today, the industrial WGS process takes place in a series of adiabatic
converters where the effluent from the reformer system is converted in two
WGS reactors (HTS and LTS converters, respectively), with the second WGS
reactor at a significantly lower temperature in order to shift the equilibrium
towards the favored hydrogen product (Fig. 1). The modern, two stage WGS
converter systems reduce the CO concentrations to about 0.3%(wt) from the
high levels (10–50%) in the outlet from the reformers. During the last couple of
decades, fuel cells, generating electricity from the reaction of hydrogen with
oxygen, for stationary and mobile applications, have become popular. A crucial
prerequisite for the techno-economic success of fuel cells, especially those that
operate either at low temperatures (like the Polymer Electrolyte Membrane
Fuel Cells, PEMFC) or in mobile applications (as in automobiles), is the
discovery of improved reforming and WGS catalysts for the generation of
hydrogen which are much more active than those used in chemical plants. The
requirements of WGS catalysts for fuel cell applications are quite different
from those of the traditional Fe2O3-Cr2O3 or Cu-ZnO based catalysts. The
catalyst bed must have a reduced volume and weight to be economical and
Water Gas Shift Catalysis 327
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
have sufficient durability to withstand rapid start-up and shut-down
conditions. In addition, the catalyst must not require controlled and elaborate,
pre-reduction procedures (as is the case with the Cu- based LTS catalysts),
must be non-pyrophoric and oxidation- tolerant on exposure to air. The two
HTS and LTS catalysts are extremely pyrophoric when activated (reduced)
and, therefore, safety from runaway heat generation and fires cannot be
ensured upon air exposure. In response to these needs, noble metal- based
reforming and WGS catalysts are under intense development worldwide for
fuel cell applications.
Newsome had provided an excellent review of the WGS literature up to
1980 (4). Lloyd et al. have reviewed the industrial developments in this area
up to 1996 (5). A comprehensive review by Kochlofl in 1997 (6, 7) covers both
the fundamental and applied aspects of the field. More recently, in 2003,
Ladebeck and Wagner have given a brief survey of the WGS catalyst
developments, especially for fuel cell applications (8, 9).
This review, after highlighting the significant features of the conventional
HTS and LTS catalysts and processes based on the Fe2O3-Cr2O3 and Cu-ZnO-
Al2O3 catalysts, respectively, critically examines the extensive results (both in
the journal and patent literature) from the study of noble metal based WGS
catalysts in the last decade or so, with particular emphasis on catalyst surface
Figure 1: Syngas generation and water gas shift reactors for NH3 Synthesis (8–9).
328 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
structures, active sites, reaction intermediates and mechanisms. This subject
has developed significantly, mainly, in the last decade due to its relevance to
the fuel cell industry.
2. THERMODYNAMICS
The feed composition to the HTS reactor can vary depending on the end-
application of the outlet of the WGS stage. Table 1 gives the composition at the
inlet to the WGS stage for some typical applications. The high N2 content in
the feed, for eventual application in ammonia synthesis, comes from the
addition of air in the secondary reformer to provide the N2 reactant for NH3.
The water gas shift reaction is moderately exothermic (Eq. 5) and conversions
are equilibrium- controlled. The equilibrium constant decreases with increas-
ing temperature (Fig. 2) and, in the temperature range 315–480uC, is given (8)
by Eq. 7:
Kp~exp 4577:8=Tð Þ{4:33½ �, ð7Þ
where T is in K. Accordingly, high conversions are favored at low
temperatures and are not affected, significantly, by changes in total pressure.
The reaction is reversible and the forward rate is strongly inhibited by the
reaction products, H2 and CO2. When operated under adiabatic conditions
(typical in industry), the exothermic rise in the catalyst bed temperature can
Table 1: Feed compositions and process conditions at the inlet to the WGS stagefor some typical applications.
Application Code 1 A B C
Feed Composition (mole %, dry)CO 12.8 10.3 46CO2 7.8 11.4 6.9H2 56.4 74.5 47N2 22.4 0.1 —CH4 0.3 3.7 0.1Ar 0.3 — —
Inlet steam/gas, molar ratio 0.6 0.9–1.0 1.0–2.2Pressure, bar 25–30 20–30 12–30Inlet Temperature, uC 343–399 343–399 343–399Outlet Temperature, uC 399–466 399–454 371–454Space Velocity, h21 2500 1500–2000 500–1400Number of beds 1 1 3Outlet CO, mole %, dry 2.0–3.5 2.0–3.0 1.5–3.5
(1) Application Code. A: NH3 Plant based on steam-hydrocarbon reforming; B: H2 Plant basedon steam-hydrocarbon reforming; and C: H2 plant based on partial oxidation of oil feed.
Water Gas Shift Catalysis 329
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
inhibit, due to thermodynamic reasons, CO conversions. This limitation can,
however, be mitigated by using two or more beds with heat removal between
them. CO levels at the exit of the HTS reactor are around 3–5 wt% while
values around 0.3 wt% can be achieved at an exit temperature of 200uC in the
LTS reactor. The lower limit of the operating temperature in the LTS reactor
is the dew point of water at the operating pressure (190–200uC at 30 bar).
Condensed steam affects, adversely, the catalytic activity of the Cu-based,
LTS catalysts. Even in the case of HTS, exposure to liquid water originating
from condensation should be avoided, since leaching of the water-soluble Cr 6+
ion is equally unwelcome. Similarly, the lower limit on the pressure is the
operating pressure of the downstream units (10–60 bars). The water content
has a strong influence on CO conversion. The water entering the WGS reactor
can be varied by controlling the amount added upstream at the reforming
stage or by injecting water before or between the stages of the WGS reaction.
In contrast, the CO, CO2, and H2 concentrations at the inlet to the HTS
reactor are more dependent on the reformer operation, which, in turn,
Figure 2: Variation of equilibrium constant (Kp) for the water-gas shift reaction withtemperature (5).
330 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
determines the thermodynamic equilibrium conditions. The effect of water
concentrations at various temperatures on the equilibrium CO concentration
is shown in Figs. 3 and 4 for typical HTS and LTS operations, respectively (8,
9). The gas composition used for these calculations are shown in Table 2 and
represents a syngas generated from autothermal reforming (ATR) and
excludes any residual hydrocarbons which may also be present. By increasing
the molar steam to dry gas (CO+CO2) ratio from 0.25 (20% H2O) to 0.75 (42.9%
H2O), the equilibrium temperature (for 1%CO) increases by 100uC. By
operating at 100uC higher temperature, a significant reduction of the reactor
size can be achieved by utilizing the more favorable kinetics at the higher
temperature. The CO concentration at the inlet to the HTS reactor can vary
widely in the range 12–40% (dry basis) depending on the raw material
(natural gas or coal) and the reforming process (steam or autothermal
reforming) utilized to generate the CO. The HTS exit (and LTS inlet)
concentrations are in the range of 3–5% (dry basis) and depend on the
operating temperature of the HTS catalyst bed. Too low of values of the steam/
dry gas ratio can lead to catalyst deactivation (due to coke laydown) while
values much higher than that stoichiometrically needed by Eq. (5) increase the
energy costs and adversely affect the process economy.
The method of producing the syngas will also affect the WGS equilibrium
compositions (8, 9). Autothermal reforming produces a syngas with lower H2
Figure 3: Equilibrium CO concentrations in HTS gas from an autothermal reformer at varioussteam/gas ratios (8–9).
Water Gas Shift Catalysis 331
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
concentration (due to the dilution with nitrogen) compared to steam
reforming. The lower H2 concentration increases the equilibrium CO
conversion whereas the high H2 concentrations expected with steam
reforming lower the equilibrium CO conversions. Figures 5 and 6 (8, 9) show
the equilibrium CO composition as a function of H2 content at constant CO
and CO2 concentrations for HTS and LTS gases, respectively. To achieve 1%
CO at the reactor outlet, the temperature must be decreased by nearly 40uCwhen the H2 is increased from 35 to 74%. In other words, the outlet CO
concentration would be 1.66% for steam reforming at the same temperature
required to achieve 1% CO for a feedstock from autothermal reforming. The
effect of H2 concentration is not as significant as the steam/dry gas ratio, but,
it is not trivial and must be considered when trying to maximize efficiency and
Figure 4: Equilibrium CO concentrations in LTS gas from an autothermal reformer as afunction of steam/gas ratios (8–9).
Table 2: Representative, methane-free, inlet gas compositions from autothermalreforming of methane (vol%) (7).
HTS (%) LTS (%)
CO 9 3CO2 7 13H2 24 30N2 28 28H2O 32 26
332 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
Figure 5: Equilibrium CO in HTS as a function of H2 concentration (8–9).
Figure 6: Equilibrium CO in LTS as a function of H2 concentration (8–9).
Water Gas Shift Catalysis 333
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
minimize the volume of the WGS reactor, especially in fuel processors for fuel
cell applications.
3. HIGH TEMPERATURE SHIFT CATALYSTS
3.1. Iron Oxide– Chromium Oxide CatalystsThe high temperature shift reaction using Fe2O3-Cr2O3 catalysts has been
in commercial use for more than 60 years. Many excellent reviews are
available (4–11). The important structural and textural roles of Cr2O3 in the
catalyst formulation has also been investigated in detail (11). Two stage CO
conversion systems employing WGS using Fe2O3-Cr2O3 catalysts and
methanation using nickel-based catalysts for CO removal was the common
and economical design in ammonia synthesis up to the late 1950s. Most of
those plants employed the Fe2O3-Cr2O3 HTS catalyst in the first, high
temperature reactor as well as in the second stage converter at temperatures
as low as 320uC. The conventional Fe2O3-Cr2O3 catalysts worked extremely
well for these high temperature applications but their relatively poor
performance in the lower temperature, second bed of these reactors motivated
further investigations. The early development of unsupported metallic copper
catalysts or copper supported on Al2O3, SiO2, MgO, pumice or Cr2O3 were
characterized by relatively short life and low space velocity operations (400–
1000 h21). Important progress was made by the addition of ZnO or ZnO-Al2O3.
These Cu-ZnO-Al2O3 catalysts exhibited not only a considerable increase in
lifetime, but also an increase in the turnover number by an order of
magnitude. Today, in industrial adiabatic converters, the syngas effluent
from the reformer system is converted in two steps, with the second step at a
significantly lower temperature in order to shift the equilibrium towards the
favored hydrogen product. In the first step, catalysts based on the Fe2O3 –
Cr2O3 oxides are applied at a reactor inlet temperature of 300–360uC and a
total pressure between 10 and 60 bars. Under normal operating conditions,
the temperature rises, progressively, through the reactor bed and can increase
up to 500uC. At exit gas temperatures of 400 to 500uC, the CO content can be
reduced in an industrial HTS converter to 5 vol% or lower. In this section, we
review the main features of the HTS catalyst/process and highlight the
developments during the last decade.
/Conventional Fe2O3-Cr2O3 catalysts contain about 80–90%(wt) of Fe2O3,
8–10% Cr2O3, the balance being promoters and stabilizers like copper oxide,
Al2O3, alkali, MgO, ZnO, etc. The BET surface areas of these catalysts vary
between 30–100 m2/g depending on the Cr2O3 and Al2O3 contents and
calcination temperatures. One of the major functions of Cr2O3 and Al2O3 is
to prevent the sintering, and, consequent loss of surface area of the iron oxide
334 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
crystallites during the start – up and further operation (11). Pure Fe2O3, when
used as a HTS catalyst, deactivates fast due to sintering of the iron oxide
crystallites. In addition to being a textural promoter preventing the sintering
of iron oxide crystallites, Cr2O3 also functions as a structural promoter to
enhance the intrinsic catalytic activity of Fe2O3. As supplied, the Fe2O3-Cr2O3
catalyst is a solid solution of a - Fe2O3 and Cr2O3, wherein the Cr3+ ion
substitutes, isomorphously and partially, the Fe3+ ions in the a - Fe2O3 lattice
framework. Even though most of the chromium ions in the fresh catalyst are
present in the Cr3+ state, a small fraction, especially on the surface, is present
in the hexavalent state, as CrO3. During start-up in the industrial reactor,
Fe2O3 is reduced to Fe3O4 in syngas at 300–450uC (5) (Eqs. 8–9):
3Fe2O3zH2<2Fe3O4zH2O DH~{16:3 kJ=molð Þ, and ð8Þ
3Fe2O3zCO<2Fe3O4zCO2 DH~z24:8 kJ=molð Þ: ð9Þ
The reduction has to be done carefully and the reaction heat removed, to avoid
further reduction of Fe3O4 (Eqs. 10–14).
Fe3O4zH2<3FeOzH2O DH~{63:8 kJ=molð Þ, ð10Þ
Fe3O4zCO<3FeOzCO2 DH~{22:6 kJ=molð Þ, ð11Þ
FeOzH2<FezH2O DH~{24:5 kJ=molð Þ, ð12Þ
FeOzCO<FezCO2 DH~{12:6 kJ=molð Þ, and ð13Þ
Fe3O4z4H2<3Fez4H2O DH~{149:4 kJ=molð Þ ð14Þ
Importantly, neither pure hydrogen nor H2-N2 mixtures should be used to
reduce the HTS catalysts to avoid the occurrence of the strongly exothermic
reduction to metallic Fe (Eq. 14). It is Fe3O4 that is the active phase
responsible for the WGS reaction. The CrO3 phase present must also be
reduced to Cr2O3 during start-up (Eqs. 15 and 16):
2 CrO3z3H2<Cr2O3z3 H2O DH~{684:7 kJ=molð Þ, and ð15Þ
2CrO3z3 CO<Cr2O3z3 CO2 DH~{808:2 kJ=molð Þ ð16Þ
Water Gas Shift Catalysis 335
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
The ratios H2O/H2 and CO2 /CO determine the relative stabilities of the Fe2O3
and Fe3O4 as well as those of Cr2O3/CrO3 phases. Under normal operating
conditions of HTS (H2O/H2 . 0.4 and CO2/CO . 1.2), the Fe3O4 and Cr2O3
phases are more stable; neither FeO nor metallic Fe are formed under these
conditions. Formation of metallic Fe (due to low H2O/H2 ratios) can trigger the
highly exothermic methanation and Boudouard reactions (Eqs. 17 and 18,
respectively) and lead to runaway conditions and catalyst deactivation:
COz3 H2<CH4zH2O DH~{206:2 kJ=molð Þ, and ð17Þ
2CO<CzCO2 DH~{172:5 kJ=molð Þ ð18Þ
While maintaining sufficiently high H2O/H2 ratios is important, passing
steam, in the absence of reductants like H2 and CO, over the reduced iron-
oxide – chromium oxide catalyst, can reoxidize the Fe3O4 to Fe2O3 (Eq. 19) and
thereby lower catalytic activity:
2Fe3O4zH2O<3Fe2O3zH2 ð19Þ
The Fe2O3-Cr2O3 catalysts are rugged and have a lifetime of 3–5 years
depending, mainly, on the temperature of operation. Unlike the Cu-ZnO (LTS)
catalyst, the Fe2O3-Cr2O3 catalyst is not extremely sensitive to the presence of
sulfur and can tolerate the presence of substantial amounts of sulfur due to
the facile reversibility of the sulfidation reaction (Eq. 20):
Fe3O4z3H2SzH2<3FeSz4H2O DH~{75:0 kJ=molð Þ ð20Þ
The value of the equilibrium constant (Kp 5 PH2P3H2S/P4
H2O) varies from
3 6 10210 to 45 6 10210 in the range of 300–450uC. The rate of the HTS
reaction is limited by pore diffusion and linearly dependent on the steam
partial pressures under industrial conditions (12). A power law – type rate
Eq.satisfactorily fits the experimental data (13). Apart from an increase in the
pressure-drop across the catalyst bed during use due to inadequate
mechanical crushing strength of the catalyst pellets, catalyst deactivation is
mainly due to loss of iron oxide surface area by thermal sintering. In addition
to the abovementioned risks of thermal sintering during start- up reduction of
the catalyst, the exothermic nature of the WGS reaction (Eq. 5) also generates
a large amount of heat during the operational phase of the process. For
example, it has been estimated (5) that the conversion of 1 vol% of CO results
in a temperature rise of about 7–10uC in the catalyst bed. The syngas
feedstocks to the HTS reactor may contain anywhere from 8 vol% (steam
reformer) to 45% (partial oxidation/ autothermal reforming of methane or coal/
coke) of CO. Hence, temperatures in the catalyst bed may rise by 500uC (to
336 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
about 800–850uC) if heat removal is inadequate. Distributing the catalyst in
two or three beds and providing inter-bed coolers can restrict the exit
temperatures to about 450uC and outlet CO content to 3–5 wt%. The Fe2O3-
Cr2O3 catalysts can tolerate sulfur up to, about, even 1000 ppm. Their major
drawbacks are: (a) the toxicity of the water-soluble Cr6+ ions posing health
hazards during catalyst manufacture and handling, and (b) the low volumetric
catalytic activity (GHSV510,000 – 15,000h21), especially at low temperatures,
when CO conversion is favored thermodynamically, necessitating the use of
large catalyst bed volumes. The latter handicap is of crucial importance in fuel
cell applications.
3.1.1. Influence of Catalyst Composition and Preparation Methods
The preparation method of the Fe2O3-Cr2O3 catalyst has a strong
influence on their properties (14, 15). They are usually prepared by
coprecipitation of the hydroxides followed by drying and calcining them to
the corresponding oxides. The oxides are reduced in situ before use. The
precipitation method involves the conventional coprecipitation of the mixed
iron and chromium nitrates with ammonium hydroxides. In an alternate
impregnation method, iron hydroxide gel is first prepared and then
impregnated with chromium nitrate solution. Chromium retards the sintering
of the iron oxide crystallites both during activation and the WGS reaction (14,
15). X-ray photoelectron spectroscopy revealed that there was surface
enrichment of Cr ions in fresh samples prepared by both the coprecipitation
and impregnation routes. The surface concentration of chromium was higher
in the impregnated samples. However, after activation and running the WGS
reaction, it was observed that the relative surface concentration of chromium
had decreased significantly in both the samples, suggesting that, during the
activation and WGS reaction, some Cr ions had migrated from the surface into
the bulk. This is in agreement with the earlier findings of Edwards et al. (16)
who had shown that Cr 3+ (d3) goes into the magnetite (Fe3O4) spinel lattice
and occupies, preferentially, the octahedral sites because of its high crystal
field stabilization, in contrast to the Fe3+ ion (d5) which does not have any
preferred site. When the Cr ions occupy tetrahedral sites, they cause strain in
the magnetite lattice, thereby decreasing the average particle size of the iron
oxide. The lower particle size, in turn, increases the surface area of the iron
oxide and leads to enhanced catalytic activity. The presence of chromium also
caused (16) an increase in the surface Fe2+/Fe3+ ratio (from XPS data) in the
fresh samples and this effect was more pronounced for the impregnated
sample. After reaction, however, these ratios were smaller for both the
precipitated and impregnated samples than the fresh sample. The Mossbauer
spectra of the pure magnetite sample indicated that the Fe3+ ions are in
Water Gas Shift Catalysis 337
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
tetrahedral sites (A sites) and Fe2+ ions are in octahedral sites (B sites) of Fe3O4.
Chromium addition to magnetite (both by the precipitation and impregnation
methods) produced a decrease in the hyperfine magnetic fields in both sites due
to the partial replacement of iron ions by chromium. This decrease was more
pronounced in B sites, indicating that Cr 3+ ions entered preferentially the
octahedral B sites. The sample prepared by the impregnation method was more
active in the WGS reaction. The role of chromium is twofold: (a) as a textural
spacer/stabilizer for iron oxide crystallites (stabilization of their smaller
crystallite size) and (b) as a structural promoter in increasing the intrinsic
catalytic activity of iron oxide crystallites due to (a) the increase in lattice strain
caused by the substitution of Fe3+ by Cr3+ ions in the magnetite lattice and (b) an
increase in the surface area of the iron oxide crystallites as mentioned above.
During the last two decades, due to the rising cost of hydrocarbon
feedstocks, plants have been forced to keep operating costs low by being as
energy efficient as possible. One method used in improving energy efficiency is
by reducing the overall steam to gas ratio of the plant (usually starting at the
inlet to the reformer). Depending on the level, these lower steam-to-gas ratios
can cause over-reduction of the Fe2O3 in HTS catalyst and result in the
formation of iron carbides. Iron carbides are very effective catalysts for the
formation of hydrocarbons by the Fischer-Tropsch reactions. Products from
the Fischer-Tropsch reactions would negatively impact both LTS catalyst
performance and plant efficiency. The over-reduction can, also, result in a
volume shrinkage within the catalyst pellet that weakens it and may lead to
an increased rate of mechanical breakdown. To minimize the low steam to gas
ratio effects on HTS catalysts, the Sud-Chemie Inc. group developed and
introduced, in the late eighties, a copper promoted iron oxide- chromium oxide
formulation that successfully suppressed the Fischer- Tropsch reactions in
commercial operation. Figures 7 and 8 compare the by-products formation
Figure 7: Comparison of CH4 formation over standard iron oxide - chromium oxide andcopper promoted, iron oxide-chromium oxide catalysts.
338 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
(methane and C2+ hydrocarbons, respectively) across a conventional iron
oxide-chromium oxide catalyst and a copper - promoted, iron oxide- chromium
oxide catalyst. It may be seen that there is a significantly lower by-product
formation in the latter. It is speculated that the presence of copper suppresses
C-O cleavage (in CO), prevents the formation of iron carbides and thereby
avoids the hydrogenation of the adsorbed carbon (to hydrocarbons) and
facilitates its desorption as CO or CO2.
3.2. Influence of Process Variables on Reaction RatesKey process variables affecting the performance of the HTS converter
involve the temperature and inlet steam/dry gas ratio since these influence
both the equilibrium CO content and reaction kinetics. Other factors to be
considered are pressure and catalyst activity.
Temperature: Since the reaction is exothermic, higher CO conversions can
be obtained by reducing the temperature at which the gas leaves the reactor.
However, this principle applies only to a catalyst that is equilibrium- and not
kinetically- limited. A reactor that is operating with the exit gas CO
concentration above equilibrium (kinetically limited) may benefit from higher
gas/ bed temperatures. The exit gas temperature determines both the catalyst
reaction rate in the bottom of the bed and the CO equilibrium value of the
outlet gas. For a reactor loaded with a highly active catalyst, the exit
temperature is determined primarily by the inlet temperature, CO concentra-
tion, and steam/ gas ratio. Exit CO equilibrium is usually achieved in these
cases. The higher the inlet CO concentration and lower the inlet steam/gas
ratio, the larger will be the overall temperature rise through the bed. The
temperature rise is also somewhat dependent on the other gas components
and their composition because of heat capacity effects. Temperature rises of
Figure 8: C2+ hydrocarbon production over iron oxide - chromium oxide and copper -
promoted, iron oxide - chromium oxide catalysts.
Water Gas Shift Catalysis 339
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
30–75uC are common across commercial reactors. For some of the recent
copper-promoted iron oxide-chromium oxide catalysts, the maximum operat-
ing temperature is around 510uC.
Steam to gas ratio: Both laboratory and commercial data indicate that
higher steam/dry gas ratios in commercial ranges also increase the water gas
shift reaction rate. As a result of the steam/ dry gas ratio effect on both the
thermodynamic and kinetic properties of the process, higher values give
higher CO conversions and a lower exit CO content in the gas. In most plant
configurations, the inlet steam/gas ratio cannot be independently controlled in
the HTS reactor. Other considerations, such as downstream gas purity
requirements and the overall site energy balance determine the inlet reformer
steam/ gas ratio and, as a result, fix the value at the inlet of the HTS
converter. Commercial operating conditions are such that the equilibrium CO
concentrations at the exit of the HTS reactor are usually about 2.0 to 5.0%.
Figures 9–12 show the relative impacts of both inlet steam/ gas ratio and exit
temperature on the equilibrium CO concentration at the exit of the HTS
reactor. They also illustrate the differences in achievable CO levels as a
function of the end-product (ammonia or hydrogen; Figs, 9 and 10,
respectively) as well as the type of reformer feedstock (partial oxidation of
natural gas or fuel oil; Figures 11 and 12, respectively) used to generate the
HTS feed gas. In addition to CO conversions, the steam to gas ratio can also
affect the production of hydrocarbons (mainly methane) by the Fischer-
Tropsch reaction. To minimize such undesirable reactions, a minimum steam
to gas ratio of 0.4 and a maximum CO/CO2 ratio of 1.6 is ensured in the HTS
reactor.
Figure 9: Influence of HTS reactor inlet steam/gas ratio and exit temperature on equilibriumCO concentrations.
340 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
Pressure: The equilibrium CO concentration is virtually unaffected by
system pressure. The pressure will, however, have an impact on the system
kinetics due to pore diffusion limitations and partial pressure effects of the
reactants. Higher pressures will improve overall CO conversion in kinetically-
limited applications.
Figure 10: CO equilibrium vs Inlet S/Gas Ratio and Exit Temperature – Hydrogen Plant.
Figure 11: CO equilibrium Vs Inlet Ratio and Exit Temperature – Partial Oxidation of naturalgas feed.
Water Gas Shift Catalysis 341
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
3.3. Catalyst DeactivationThe primary deactivation mechanism of the HTS catalyst is due to
thermal sintering of the iron crystallites. The degree of thermal sintering is a
function of time and operating temperature and is irreversible. Thermal
sintering occurs more rapidly at higher temperatures. Hence, the maximum
bed/exit gas temperatures are usually limited to less than 510uC. Some
thermal sintering is unavoidable in the start-up and normal operation of the
catalyst. Commercial data suggest that there is approximately a 50% loss in
total surface area over the first few months of operation and, then, a further
25% loss throughout the remaining life of the catalyst. As a result of these
changes in surface area, the deactivation rate for the catalyst is faster during
the first few months of operation and then stabilizes with very gradual aging
after the first year. Unlike LTS catalysts which can show distinct zones of
deactivation (completely inactive, partially deactivated and essentially fresh,
non-deactivated), the HTS catalyst undergoes a more gradual deactivation
that is spread throughout the bed. The more typical symptom of activity loss is
a gradual spreading out of the reactor temperature profile and an increase in
the CO leakage. Figure 13 shows a typical change in the temperature profile
with time- on- stream in commercial reactors. For a relatively new catalyst,
the temperature increases more sharply in the top 50–60% of the bed. After
some time-on-stream, the catalyst in the top is less active and the temperature
profile changes. Throughout the catalyst aging period, however, there
continues to be some catalyst activity through all of the beds. As the profile
spreads more throughout the bed, it may become necessary to increase the
inlet gas temperature in order to maintain acceptable exit CO levels.
Apart from thermal sintering, activity loss for HTS catalyst is most
commonly due to the presence of poisons in the feedstock and the deposition of
solids on the catalyst. The latter (mainly entrained carbon, boiler water solids,
Figure 12: CO equilibrium vs Inlet S/G and Exit Temperature—Partial Oxidation of Fuel Oil.
342 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
and/or silica) will coat the outer surface of the catalyst and block available
pores. In most modern plants, the HTS catalyst is downstream of equipment
operating at higher temperatures such as reformers and heat exchangers. As a
result, there may be a gradual depositing of steam- volatile compounds into
the HTS bed. In time, these deposits can plug the pores in the catalyst and/ or
the void space between catalyst tablets. As a consequence, the activity declines
and the pressure drop may increase. Proper design of the pore size distribution
and geometric shape of the catalyst pellets can minimize such effects. The
presence of sulfur in the feed gas will affect the size of the converter, as
allowances must be made for the adverse effect of sulfur on catalytic activity.
The presence of oxygen (from the secondary reformer or the partial oxidation
reactor) may also influence the design since the oxygen will be converted to
water through an exothermic reaction. Thus, when the shift feed contains
appreciable oxygen, an allowance may be necessary for the accompanying
temperature rise due to this reaction. Any saturates, e.g. methane, ethane,
propane or unsaturates (ethylene, propylene) in the process gas will
essentially pass through the shift converter unchanged. There is no conclusive
evidence to indicate that the saturates will crack, or that the unsaturates will
be hydrogenated to any significant degree. Even if the unsaturates do
hydrogenate, this side reaction apparently does not affect the catalytic
activity. Acetylene, on the other hand, can be troublesome, because it does
hydrogenate and impair the catalytic activity. When both diolefins and nitric
oxide are present together in the feed stream, polymeric gums are usually
formed and the shift catalyst could be subjected to a serious fouling problem.
3.4. HTS Catalytic Reactor Design ConsiderationsIt must be noted that, unlike the LTS catalyst, the HTS reactor system is
not designed to achieve equilibrium CO leakages for the major part of the
catalyst life. Although equilibrium CO leakages are often experienced at the
Figure 13: HTS Bed Temperature Profile at Start, Middle, and End of Run.
Water Gas Shift Catalysis 343
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
start of run and for some period of time, the reactor size and catalyst volume to
achieve and maintain equilibrium throughout the total charge life would be
substantially higher and cost prohibitive. As a result of these cost considera-
tions, the reactor is usually designed to be kinetically, rather than
equilibrium, limited. This means that any factor influencing the overall
reaction kinetics will have a much more important impact on the required
catalyst volume for a given application. Typical design lives for a HTS catalyst
are 3–5 years before there is a need for catalyst replacement.
Since HTS catalysts/ reactors are usually designed to be kinetically
limited, the inlet gas temperature will have a significant impact on the
required catalyst volume. Lower inlet gas temperatures will require
increased catalyst volumes to achieve similar levels of performance.
Figure 14 shows the impact of temperature on reaction rates across a typical
Fe2O3-Cr2O3 HTS catalyst. At the higher operating temperatures for HTS
reactors, the WGS reaction is much more pore diffusion limited compared to
the LTS reaction. An increase in reaction rates can be achieved by
incorporating a catalyst with high geometric surface area per unit loaded
volume of the reactor and/or increasing the size of the pores. Although
pressure has no impact on the WGS equilibrium CO levels, there is a
significant influence on the reaction rate because of pore diffusion
considerations. Figure 15 shows the relative influence of system pressure
on reaction rates and the corresponding required catalyst volumes. The rate
increases with reactor pressure up to about 21 bar. Required catalyst
volumes would correspondingly decrease with increasing pressure. Operating
pressure for a HTS plant is usually set more by an examination of overall
economics (related to feedstock supply pressure and equipment costs) rather
than catalyst reaction rate effects.
Figure 14: Effect of Temperature on relative HTS reaction rates over a commercial Fe2O3 -Cr2O3 catalyst.
344 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
3.5. Reaction Mechanisms over Iron Oxide-Chromium OxideCatalystsThe kinetics and reaction mechanism of the HTS WGS reaction has been
studied extensively and various mechanisms proposed (3–5). Temkin et al.
proposed, more than 50 years ago, that the WGS reaction proceeds by an Eley-
Rideal type mechanism, via alternate reduction and oxidation of the surface of
iron oxide (17–19):
COz Oð Þ<CO2zðÞ, and ð21Þ
H2OzðÞ<H2z Oð Þ, ð22Þ
where (O) is an oxygen atom on the oxide surface; and ( ) is a vacant site (an
oxygen anion vacancy) on the surface caused by the removal of an oxygen
atom. The surface is reduced by CO (Eq. 21), and subsequently, oxidized by
H2O [Eq. (22)}. This mechanism is referred to, in subsequent literature, as the
‘‘redox mechanism’’. The term redox mechanism denotes that the catalyst
itself undergoes changes in oxidation state during the course of the
mechanism. It does not refer to the oxidation state changes of the reactants
or products or associated intermediates. It should be pointed out, here, that
the Eley-Rideal mechanisms refer to an adsorbed species reacting with a gas-
phase species. The redox mechanism implied in Eqs. (21) and (22) is
essentially the same as the Mars - Van Krevelen mechanism, with the
difference that the oxygen used to oxidize the catalyst [Eq. (22)] comes from
the water rather than the gas-phase oxygen. Whether this oxygen atom
(extracted from gas-phase H2O) is better referred to as an ‘‘adsorbed’’ species
or, rather as an occupant of the surface lattice position is a moot point.
Figure 15: Effect of pressure on relative LTS reaction rates over a commercial Cu-ZnOcatalyst.
Water Gas Shift Catalysis 345
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
A multistep Langmuir- Hinshelwood type mechanism (Eqs. 23–27) was
proposed by Oki et al.(20–22) in 1973. From simultaneous exchange rate
measurements, they concluded that while the evolution of gaseous H2 from
adsorbed H atoms [Eq. (27)] is the rate determining step at low CO
conversions, adsorption of CO [Eq. (23)} controls the overall reaction rate at
steady state, near-equilibrium, conditions prevalent in industrial reactors:
CO gð Þza<CO að Þ, ð23Þ
H2O gð Þz3a<2H að ÞzO að Þ, ð24Þ
CO að ÞzO að Þ<CO2 að Þza, ð25Þ
CO2 að Þ<CO2 gð Þza:, and ð26Þ
2H að Þ<H2 gð Þz2a ð27Þ
In the above Eqs., ‘‘a’ refers to an adsorption site. It can be located either on
the support or the metal oxide. The HTS reaction on Fe2O3-Cr2O3 catalysts
probably proceeds by an oxidation- reduction mechanism (See Section 9).
3.6. Metal- promoted Iron Oxide –Chromium Oxide HTSCatalystsThe possibility of increasing the activity of Fe2O3-Cr2O3 catalysts by
promotion has been studied by Trimm and coworkers (23–25). Small amounts
of precious metals were found to increase the rate of the forward reaction CO +H2O R CO2 + H2 and to increase the rate (Fig. 16). Platinum was found to
increase the reactivity of all the oxides with the promotional effect being most
pronounced with Cr2O3, U3O8 and CeO2-ZrO2 supports. Comparisons were
also made with Pt-U3O8 which was as efficient (on a weight basis) as Pt-
Fe2O3-Cr2O3. In fact, the specific activity, on an area basis, of Pt-U3O8 (BET
area 5 2.3 m2/g) was more than 25 times that of Pt-Fe2O3-Cr2O3 (BET area 5
63 m2/g) (Table 3). However, catalytic activity over this catalyst dropped
quickly as temperature was reduced. Trimm (19, 21) has also compared the
kinetics of the WGS reaction for the promoted and unpromoted Fe2O3-Cr2O3
catalysts. The general power rate law expression remained unchanged in the
absence and presence of the noble metal promoter indicating that it is the
number of active sites that has increased by promotion (by noble metals).
346 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
Figure 16: Apparent activation energy plots for promoted iron-chromia catalysts (23).
Table 3: Rates and apparent activation energies for water gas shift over 1%Pt/oxide catalysts (21).
Catalyst BET area (m2/g21)Rate at 450uC
(mmol(CO) gcatalyst21s21) Ea(kJ/mol)
Pt/Cr2O3 22 0.174 41 ¡ 2Cr2O3 0.022 78 ¡ 1Pt/Cr2O3-Fe3O4
a63 0.149 50 ¡ 3
Cr2O3-Fe3O4a 0.124 70 ¡ 2
Pt/U3O8 2.3 0.142 59 ¡ 3U3O8 0.01 24 ¡ 2Pt/CeO2-ZrO2
a67 0.079 28 ¡ 1
CeO2-ZrO2a 0.008 55 ¡ 1
Pt/CeO2-Fe3O4
an.m.b 0.07 50 ¡ 1
Pt/CeO2 122 0.055 52 ¡ 1Pt/MgO 77 0.034 41 ¡ 1Pt/V2O5 6 0.032 52 ¡ 3Pt/ZrO2 n.m.b 0.026 24 ¡ 1Pt/Fe3O4 29 0.022 55 ¡ 3Fe3O4 0.023 48 ¡ 2Pt/MoO3 1.6 0.02 49 ¡ 3Pt/Bi2MoO6 2.1 0.018c 62 ¡ 4Pt/MnO2 17 0.016c 53 ¡ 2Pt/Al2O3 272 0.014 47 ¡ 1
aThe composition of the mixed oxides were as follows: 8 wt% Cr2O3-Fe3O4, 8 wt% CeO2 – Fe3O4,50 wt% CeO2-ZrO2. bNot measured. cNo measurement at 450uC; calculated from Arrheniusparameters.
Water Gas Shift Catalysis 347
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
Apparent activation energies were found to be similar (, 50 kJ/mole) for Pt
supported on CeO2, Fe3O4-Cr2O3, CeO2-Fe3O4, Fe3O4, V2O5, MgO, MnO, and
Al2O3, despite up to 15-fold differences in rates of reactions (Table 3;
Figure 16). Since it is unlikely that surface diffusion of oxygen will have
similar activation energies for such a variety of solids, the authors suggested
that diffusion of oxygen on the surface or across the surface of the support to
react with CO adsorbed on the metal cannot be rate controlling in WGS
reactions on Fe3O4-Cr2O3 promoted with noble metals. Rhodium was found to
be the most active promoter for Fe2O3-Cr2O3 oxides. More recently, this group
had probed (24) the origin of rhodium promotion of Fe3O4-Cr2O3 catalysts for
the HTS reaction using various kinetic techniques and concluded that, of the
two steps that may restrict the rate of the WGS reaction over iron- chromium
oxide catalysts (reduction by CO and H2 generation through reoxidation by
water), rhodium acts primarily by accelerating the latter.
Although the promoted catalysts are more efficient than the unpromoted
Fe2O3-Cr2O3 catalysts above 300uC, they are still less active than the copper-
based catalysts at temperatures below 300uC. As a result, the CO concentra-
tion is reduced but, at about 3–4%, it is still too high for many applications
(e.g., fuel cells, ammonia synthesis). Low temperature WGS is required to
reduce CO concentrations still further.
During the last decade, attempts to develop improved HTS catalysts have
been along two main lines:(A) replacing, at least partially, Fe by more active
elements (like noble metals), and (B) replacing Cr, partially or completely, by
non-toxic elements like Cu, Ca, Ce, Zr, La etc. (26–34). Promotion of the
Fe2O3-Cr2O3 catalysts by 2 wt% Ag, Cu, Ba, Pb and Hg was explored by
Rhodes et al. (31). The catalysts were prepared by coprecipitation. Boron was
found to poison the activity slightly whereas the others did increase the
activity between 350–440uC, with the relative order being Hg.Ag, Ba
.Cu.Pb. unpromoted Fe2O3-Cr2O3 . B. From their results, the barium or
silver- promoted Fe2O3-Cr2O3 catalysts appear promising: a 10–15% increase
in CO conversion was observed (when Ag or Ba was incorporated in the
conventional Fe2O3-Cr2O3 catalysts) at their reaction conditions (400uC, 27
bar, GHSV 5 1.2 6 10 6 h21; the volume of steam was 75 volume% of the dry
gas). The promoters decreased the activation energy of the reaction (Table 4).
Based on the compensation effect (Fig. 17) seen when the activation energies
were plotted against the corresponding pre-exponential factors (in the
Arrhenius Eq.), the authors concluded that CO adsorption is an important
factor controlling the relative catalytic activities of the various samples in the
WGS reaction. Andreev et al.(29) studied the effect of the addition of CuO,
CoO, and ZnO (5 wt%) on the activity of Fe2O3-Cr2O3 catalysts. The Cu –
promoted sample was found to be the most active at 380uC. Kappen et al. (30)
investigated the state of their Cu promoter (0.17–1.5 wt%) in Fe2O3-Cr2O3
348 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
catalysts and found that Cu was in the metallic state under the WGS reaction
conditions. However, it was reoxidized easily when exposed to the atmosphere.
Most of the current generation, industrial, HTS catalysts contain oxides of Fe,
Cr and Cu.
3.7. Chromium-free HTS CatalystsWhen chromium oxide is used as a component of a catalyst, especially in
hexavalent form which is soluble in water, expenditures must be incurred to
guarantee worker safety both during production and later handling of the
Table 4: Effect of additives on the performance of Fe3O4/Cr2O3 water gas shiftcatalysts (27).
Additive CO conversiona (%) Activation energyb (kJ/mol)
None 18.8 112B 18.7 108Pb 25 90Cu 27.9 81Ag 32.9 74Ba 33.5 83Hg 37.4 82
a CO conversion at 400uC, 27 bar, GHSV51.2 6 106 h21. b ¡ 4kJ/mol.
Figure 17: ‘‘Compensation effect’’ plot for the modified Fe2O3/Cr2O3 catalysts (31).
Water Gas Shift Catalysis 349
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
catalyst, and health hazards cannot be fully ruled out despite considerable
effort. In addition, the spent catalyst ultimately poses a hazard to man and the
environment and must be disposed of in accordance with the laws for the
disposal of toxic waste. HTS catalysts completely free from Cr and containing
Ca, Ce or Zr were first claimed by Chinchen (34). Their catalytic activities,
however, were low. Similar low- active catalysts, based on Mg and Zn ferrites,
were also reported by Rethwisch and Dumesic (35). More active catalysts
based on alkali- promoted Co-, Cu-and Fe- manganese oxide systems were
reported by Hutchings and coworkers (36, 37). The relative first order rate
constants in the WGS reaction for Fe-Cr, Fe-Mn, Cu-Mn and Co-Mn catalysts
were found to be (36, 37) 1.0, 0.06, 0.75 and 1.75, respectively. Their Co-Mn
catalyst, however exhibited significant methanation activity and the Cu-Mn
catalysts were more sensitive to sulfur than the Fe-Cr formulations. Ladebeck
and Kochloefl (38) had found that chromia-free, iron oxide catalysts containing
about 5 wt% of Al2O3, 2 wt% of Cu and 2.5 wt% of CeO2 were very active for the
HTS reaction. The incorporation of ZrO2, La2O3 or MnO instead of CeO2
resulted in catalysts with a high initial activity but with a poorer stability.
Araujo and Rangel (28) investigated the catalytic performance of Al-doped, Fe-
based catalysts with small amounts of copper (3 wt%), prepared by the
coprecipitation (for Al and Fe) - impregnation (for Cu) method, in the HTS
reaction. The aluminium and copper-doped iron catalyst was studied at 370uCand showed similar activity compared to the commercial Fe-Cr-Cu catalyst.
Costa et al. subsequently examined (33) the use of thorium, instead of
chromium, in Fe- Cu – based catalysts for the HTS reaction. These Fe-Th-Cu
catalysts were more active than the commercial Fe-Cr-Cu catalyst at H2O/CO
5 0.6 and 370uC. Its high activity was attributed to an increase in surface
area due to the presence of thorium. From a detailed study of chromium-free,
iron-based HTS WGS catalysts, Natesakhawat et al. (26) concluded that a
combination of copper and aluminum is a potential replacement for Cr in HTS
catalysts. Further improvements in HTS activity of Fe-Al catalysts could be
achieved by the addition of small amounts of copper or cobalt. The CO
conversions (at 400uC, CO/H2O/N2 5 1/1/18 (vol) and feed GHSV of 6000 h21)
were 43, 46, 27, 12 and 16% for Fe-Cr, Fe-Cu-Al, Fe-Al, Fe-Ga and Fe-Mn,
respectively. As a textural promoter, aluminum oxide (like chromium oxide)
prevented the sintering of the iron oxide crystallites and stabilized the active
phase, magnetite (Fe3O4) by retarding its further reduction to FeO and
metallic Fe. The promotional effect of Cu was found to be strongly dependent
on the preparation method. Fe-Cu-Al catalysts prepared by a one-step method
(simultaneous coprecipitation of all the three component hydroxides) had
higher CO conversions than those prepared by a two-step, coprecipitation –
impregnation method (coprecipitation of the Fe and Al components followed by
impregnation of Cu on the precipitate obtained by coprecipitation) (Fig. 18).
350 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
Although the activities were similar at 250uC, the Fe-Cu-Al catalyst pre-
pared by the one-step method was more active at higher temperatures. The
better stabilization against sintering, at higher temperatures, of the copper
crystallites in the coprecipitated samples is probably the reason for its
superior performance. A significant difference in the temperature-pro-
grammed-reduction profile was also observed between the two samples
(Fig. 19). The low temperature reduction profile has contributions from
three different reduction sites. The reduction of hematite to magnetite
appears to have shifted from 300–290uC. The major peak from reduction of
Cu species appears at 260uC with a very weak shoulder around 220uC. This
shoulder is possibly due to reduction of the Cu species which are on the
external surface of the catalyst and which can be reduced easily. The rest of
the copper species appear to be more difficult to reduce as seen by a shift in
reduction temperature from 220–260uC, possibly due to stronger interaction
with the hematite matrix. While the peak resulting from the reduction of
hematite to magnetite shifts to lower temperatures (from 300–290uC), the
peak from a further reduction of magnetite changes very little compared to
the non-promoted, Fe-Al sample. These results suggest that the preparation
method makes a significant difference in the way Cu promoter is
incorporated into the catalyst structure. During the last decade, more
active and chromium-free, noble metal-based HTS catalysts are under
development for use in fuel cell applications. These will be described later,
in Section 6.
Figure 18: Effect of preparation method (one vs. two steps) of Fe2O3-Al2O3-CuO catalysts onCO conversion (26).
Water Gas Shift Catalysis 351
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
4. LOW TEMPERATURE WATER GAS SHIFT CATALYSTS
4.1. Cu-ZnO-Al2O3 CatalystsAn excellent perspective of the historical background for the evolution of
the low temperature water gas shift catalysts has been provided by Twigg et
al. (5). The development of highly efficient sulfur removal hydrodesulfurisa-
tion technologies using Co(Ni)- MoO3- Al2O3 catalysts in the 1960s provided
ammonia manufacturers with syngas streams containing less than 1.0 ppm
sulfur. This, in turn, enabled the use of the otherwise sulfur-sensitive Cu-ZnO
catalysts at sufficiently low temperatures (190–200uC) when the equilibrium
CO concentrations, at the exit of the LTS converters, can be below 0.3%. It
may be recalled that the Fe-Cr catalysts are not active enough below 350uC, at
which temperature, the equilibrium CO concentrations are around 3–5%. As a
direct consequence of having such low levels of CO (below 0.3%wt) from the
Cu-ZnO catalysts, it was economic to incorporate a methanation stage in the
process in place of the more complicated copper liquor scrubbing system that
Figure 19: Effect of preparation method (one vs. two steps) on the temperatureprogrammed reduction profiles, in H2, of Fe2O3-Al2O3-CuO catalysts (26).
352 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
was formerly used to remove residual CO, thereby enhancing the technoeco-
nomic viability of large ammonia plants. The activity of metallic copper in the
WGS reaction has, of course, been known for a long time (4–7). The problem
was the easy copper sintering and the subsequent loss of copper surface area
during the activation (reduction of the precursor copper oxide) and use of
copper catalysts. Various stabilizers, like SiO2, Cr2O3, Mn - Cr2O3 etc., were
evaluated for their ability to stabilize the copper surface area (39). The
introduction of the Cu-ZnO catalyst in the early 1960s (5) and, later, the Cu-
ZnO-Cr2O3, the Cu-Zn-Mn-Cr2O3 and, especially Cu-ZnO-Al2O3 formulations
enabled the production of catalysts with high and stable copper surface areas
and established the LTS process as a standard operation in any scheme of
hydrogen production from carbonaceous raw material. Currently, Cu-ZnO-
Al2O3 based catalysts are used almost exclusively for industrial LTS
operations.
The feed gas to the LTS reactor is that exiting the HTS unit cooled, either
by direct or indirect heat exchange, to approximately 200uC. LTS converters
are employed more frequently in hydrogen and ammonia - producing plants
than in methanol or hydrocarbon (Fischer-Tropsch) plants. Hydrogen plants
normally begin with primary steam/hydrocarbon reforming of natural gas to
syngas which is then water gas shifted over HTS and LTS catalysts to
maximize the hydrogen mole fraction of the effluent. Following CO2 scrubbing
and methanation to remove unreacted CO, the product hydrogen is then
utilized for hydrocracking, hydrogenation or other service. In ammonia plants,
there is a secondary reformer between the primary reformer and the HTS
units for the introduction of the requisite nitrogen. Process conditions in the
WGS section are more severe than in hydrogen plants. This is because the
downstream ammonia process is considerably more sensitive to the purity of
the hydrogen produced. Not only does a lower level of hydrogen reduce
ammonia production, but also the corresponding higher level of inerts (like
CH4 and CH3OH) increases the purge rate from the synthesis loop. The
principal deactivation mechanism for LTS catalysts is poisoning by sulfur and
chlorides contained in the process gas. If only a single bed of LTS catalyst is
employed, this deactivation process begins as soon as the catalyst is placed on
stream and, normally, within 6–12 months, a rise in CO leakage will be
detected. In an ammonia plant, a rise of 0.1% CO in the LTS converter effluent
is roughly equivalent to a production loss of 30 T/day in a 3000 T/day of
ammonia plant. To minimize this production loss and maintain a low CO
leakage for a long period of time, many plants have installed guard beds of
LTS catalysts immediately ahead of the main LTS unit. These beds are
usually about 1/4 the size of the main LTS bed and serve to sacrificially screen
poisons from the main bed and to promote additional water gas shift. In
general an ammonia process will tolerate up to 0.4% CO in the LTS effluent
Water Gas Shift Catalysis 353
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
before process economics dictate a catalyst change. Hydrogen plants may
tolerate a slightly higher CO leakage.
The preparative chemistry of the Cu-ZnO, with or without Al2O3 or Cr2O3,
has been studied extensively and is still a subject of interest since the nature
of the precursor mixture and its evolution during the preparation steps seem
to influence the catalytic properties. In an early publication, Uchida et al.(40)
tested several catalyst combinations, Cu/Zn, Cu/Al, Cu/Al/Zn, Cu/Fe and Cu/
Cr, prepared by the coprecipitation of their mixed hydroxides/carbonates/
hydroxycarbonates, and compared their catalytic activity and stability in the
WGS reaction. Addition of zinc to copper increased the catalytic activity which
reached a maximum around a Cu/Zn ratio of 0.4. They observed that the
method of preparation of the Cu-ZnO catalyst is extremely important in
determining the catalytic activity. They also established, using x-ray
diffraction, that the major constituents of a Cu-ZnO catalyst after use were
copper metal and zinc oxide (41). It was speculated, even at that early stage,
that copper metal can be the active ingredient (42, 43). Highly active catalysts
are prepared by coprecipitation from the corresponding aqueous solutions of
metal nitrates with sodium carbonate and having Cu/Zn atomic ratios
between 0.4 and 2.0 (42–45). To avoid extensive washing of the filter cake
for a reduction of its Na content, ammonium carbonate or hydroxide as a
precipitating agent was recommended by Sengupta et al. (46). The thermal
decomposition of the resultant, aqueous (Cu,Zn)(NH3)4(HCO3)2 complexes by
steam provides alkali-free Cu-Zn hydroxycarbonates. Petrini et al. (44, 45) had
also noted that highly active catalysts can be prepared if Al(OH)3 is added
during the Cu-Zn precipitation.
Gines et al. (47) reported a detailed study of the influence of preparation
methods on the activity and structure sensitivity of the Cu-ZnO-Al2O3 mixed
oxide catalysts. Samples were prepared by coprecipitation from aqueous
solutions of the nitrates of Cu, Zn and Al with sodium carbonate at 60uC and a
constant pH around 7 in a stirred batch reactor. The precipitates were filtered,
washed with distilled water at 60uC until no sodium ions were detected
and dried at 90–100uC overnight. Finally, the samples were decomposed in
air for 8 h at temperatures between 400–700uC. Depending on the ratio of
Cu, Zn, and Al cations, different hydroxycarbonate phases were formed:
malachite[Cu2(OH)2CO3], which is capable of substituting Cu by zinc and
is called zincian-malachite or rosasite [(Cu,Zn)2 (OH)2CO3],hydrotalcite
[(Cu,Zn)6Al2CO3(OH)16. 4H2O], aurichalcite [Cu,Zn)5(CO3)2(OH)6] and hydro-
zincite [Zn5(CO3)2(OH)6]. The rosasite phase can transform to the aurichalcite
phase for Zn concentrations greater than 40 mol%. No trace of copper
hydroxynitrates like gherardite was observed. An important observation
was that hydrotalcite was selectively obtained as a single phase only in
preparations using a (Cu+Zn)/Al atomic ratio of 3, the stoichiometric, M2+/M 3+
354 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
metal cation ratio in hydrotalcite. The BET surface areas increased with Al
content. On thermal decomposition, the mixed oxide, CuO-ZnO-Al2O3, was
obtained. X-ray diffraction data revealed the presence of crystalline CuO and
ZnO. Additionally, amorphous alumina was also present. Crystalline, spinel-
like, ZnAl2O4 was detected in trace amounts only in samples containing .13%
Al2O3. The concentration of free CuO and ZnO crystallite sizes were related to
the hydrotalcite content in the hydroxycarbonate precursor: higher the
amount of hydrotalcite in the precursor, the lower the CuO and ZnO
crystallite sizes in the resulting mixed, ternary oxide. The influence of the
hydroxycarbonate precursor structure was preserved throughout the calcina-
tion step and manifested itself in a different reducibility of the CuO/ZnO
precursors. After activating and reducing the samples, they were tested in the
WGS reaction. Cu (I) oxide is a probable intermediate in the reduction of CuO
to Cu metal. Completely reduced Cu clusters on ZnO constitute the active bulk
phase for the WGS reaction. A gas mixture consisting of 10%CO/30% N2/
30%H2/ 30% H2O was fed to the fixed bed reactor at a volumetric flow of 750 ml
STP/ min (catalyst weight 5 0.5 g). The reaction was carried out at 230uC and
1 bar. It should be pointed out that CO2, one of the products of the reversible
WGS reaction, was not included in the inlet gas mixture. A remarkable feature
of their catalytic result is that the turnover frequency (number of CO2
molecules produced per surface copper atom per second) was essentially
constant, not only when the copper metal surface area was varied between 3–
35 m2/g Cu, but also when the CuO loading was varied between 30 and 50 wt%,
the Al/Zn atomic ratios between 0 and 2.5, the copper dispersion between 0.5
and 5.0%, and the calcination temperature between 400–700uC, clearly
suggesting that the specific reaction rate is proportional to the copper metal
surface area. Based on these results, the authors concluded that (a) the WGS
reaction is a structure insensitive reaction and linearly proportional to the
surface area of metallic copper; and that (b) both the metallic copper
dispersion and catalytic activity were related to the amount of hydrotalcite
contained in the precursor precipitate; the higher the content of the
hydrotalcite in the precursor, the higher the catalytic activity of the resulting
catalyst.
Contrary results, namely, that the turnover frequency does vary, by an
order of magnitude, when the copper metal surface area was changed from 10
to 40 m2 /g, for the Cu-ZnO-Al2O3 system had been reported earlier, by
Chinchen and Spencer (48). These authors had carried out the WGS reaction
at 30 bar and their reaction mixture had included CO2 under conditions closer
to those in practice in the industry. Even though it is well established (4, 5)
that the catalytic activity of Cu-ZnO-Al2O3 catalysts in WGS reactions
increases with the surface area of metallic copper, there are no reports that,
under industrial conditions, the rate of the reaction correlates linearly with
Water Gas Shift Catalysis 355
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
the metallic Cu area over the entire Cu-Zn composition range. While a high Cu
surface area is a necessary prerequisite for catalytic activity, additional
factors like the ‘‘microstrain’’ in the copper nanocrystallites due to the
presence of Zn ions probably affect catalytic activity. The hydroxycarbonate
precursors mentioned earlier probably influence the residual concentrations of
Zn in the Cu metal crystallites in Cu-ZnO. Similarly, oxygen vacancies in ZnO
formed, for example, during the reduction/activation of the catalyst or during
the WGS reaction will also influence the catalytic activity indirectly by
influencing the wetting behavior at the Cu/ZnO interface and, thereby, the
‘‘microstrain’’ in the Cu crystallites. Hence, bulk structural changes in the
ZnO or Cu metal crystallites resulting from the preparation procedures cannot
be ignored.
4.2. Promoted Cu-based LTS CatalystsAttempts have been made during the past decade to prepare alternate
base metal catalysts which are superior to the conventional Cu-ZnO-Al2O3
catalysts. Tanaka et al. (49–51) have explored the performance of Cu-Mn
spinel oxides in the LTS reaction. They had originally found (50, 51) that Cu-
Mn spinel catalysts which were prepared by coprecipitation with NH3, showed
a WGS activity comparable to that of Cu-ZnO-Al2O3 catalysts in spite of their
low surface area. Since Cu and Mn ions may not have coprecipitated
homogeneously due to formation of the copper amine complex, [Cu(NH3)4]2+
by the NH3 coprecipitation method, they later prepared (49) their catalysts by
citric acid complex, urea homogeneous coprecipitation or the Pechini method.
The last method involves the polymerization accompanied with esterification
of ethylene glycol and citric acid during the precipitation of the hydroxides of
Cu and Mn. Higher CO conversions were obtained for samples prepared by the
citric acid method. CO conversion was enhanced with a rise in the calcination
temperature of the Cu-Mn spinel prepared by the citric acid method. Partial
substitution of Fe or Al for Mn in the spinel lattice enhanced their CO
conversion activity to levels higher than that of conventional Cu-ZnO-Al2O3
catalysts when the temperature was increased to 300uC Fig. 20.
4.3. KineticsThe kinetics of the water gas shift reaction has been studied extensively
(52–59). An accurate description of the measured reaction rates from a data
set can be obtained from an expression where all kinetic parameters are fitted,
for example, to a power law. Such empirical kinetic expressions are essential
in reactor design calculations where it is necessary to have a very accurate
description of the reaction rate. Different mechanisms, however, can lead to
356 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
the same overall kinetic expression. Hence, it is difficult to determine the
mechanism from an empirical kinetic expression alone. Microkinetic models
are useful here as they are based on the knowledge about elementary steps
and their energetics. They enable us to estimate surface coverages, reaction
orders, and activation enthalpy during reaction conditions. Ovesen et al. (58)
had analysed the microkinetics of the WGS reaction under industrial
conditions based on a model developed by them earlier (59). The reaction
was studied over three different Cu- based catalysts, Cu-ZnO-Al2O3, Cu-Al2O3
and Cu-SiO2. The Cu-ZnO-Al2O3 catalysts contained about 40% Cu, 22% Zn
and 5% Al. Ovesen et al.’s model (58, 59) is based on the surface redox
mechanism:
1. H2O (g) + * u H2O*
2. H2O* + * u OH* + H*
3. 2OH* u H2O* + O*
4. OH* + * u O* + H*
5. 2H* u H2 (g) + 2*
6. CO (g) + * u CO*
7. CO* + O* u CO2* + *
8. CO2* u CO2 (g) + *
Figure 20: CO conversion over Cu-Mn catalysts. Reaction conditions: H2 37.5%; CO, 5.0%;H2O, 25.0%; CO2, 12.5%; space velocity, 6400 h21 (49).
Water Gas Shift Catalysis 357
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
where the asterisk signifies a free surface site and X* is an adsorbed species,
X. The expressions for the rate and equilibrium Eqs. that constitute the model
are shown in Table 5. The model describes the coverage of surface species in
addition to the overall rate. When this model was tested against measure-
ments for an industrial (Cu-ZnO-Fe2O3) catalyst at 1 bar by Van Herwijnen
and de Jong (52), a good agreement was found (58). From parallel
physicochemical measurements, it was deduced that the catalyst exposed
nanocrystallites of Cu (111) facets almost exclusively. The rate-determining
step was dependent, critically, on the composition of the feed gas mixture. It
was found that reaction step 2 above is rate limiting in a gas with a low ratio of
water to carbon monoxide whereas reaction step 7 is rate limiting in a gas with
a high ratio of water to CO. Reaction 4 was significant in a CO2 + H2 mixture.
However, when this model was tested against the high pressure data,
deviation between the calculated and experimental rates was found (58). To
describe the kinetics of the water gas shift reaction at industrial conditions it
was necessary to include the synthesis and hydrogenation of formate (reaction
steps 9–11 below):
9. CO2* + H* u HCOO* + *
10. HCOO* + H* u H2COO* + *
11. H2COO* + 4H* u CH3OH (g) + H2O(g) + 5*
The reaction step 9 was in equilibrium under the industrial conditions (high
pressure). The coverage of HCOO* was always low. Ovesen et al.’s reaction
Table 5: Rate and Equilibrium Eqs. for Kinetic Model (51).
K1PH2O
P0
~HH2O
r2~k2HH2OH{k2
K2
HOHHH
K2H2OH~HH2OHO
r4~k4HOHHO{k4
K4HOHH
K5H2H~
PH2
P0H2
K6PCO
P0H~HCO
r7~k7HCOHO{k7
K7HCO2
H
K8HCO2~
PCO2
P0H
Note: ki is Forward Rate Constant, Ki Equilibrium Constant, and hi Surface Coverage of Species i).
358 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
sequence for the industrial LTS reaction (58) consists of the steps 1 through
11 with steps 2, 4, 7, and 10 as possible slow steps. The formate may be
present on the surface but, it is not a species in the catalytic cycle for CO
conversion to CO2. These conclusions from kinetic studies are similar to those
of a combined kinetic and DRIFTS study of Pt- and Au-based catalysts by
Meunier et al. (60–62). It should be pointed out, here, that Ovesen et al.’s rate
equations do not consider that co-adsorbed water molecules may influence the
rate of decomposition of the formate intermediate. It will be interesting to
explore the changes if this issue is taken into consideration (see sections 9 and
10). The satisfactory agreement between the calculated exit mole fraction of
CO from the microkinetic model and the experimental exit mole fraction of
CO for Cu-ZnO-Al2O3 is shown in Fig. (21). This model was refined further in a
later publication by Schumacher et al. (54). It was established that the
adsorption energies for CO and oxygen (the latter arising from H2O) can
describe, to a large extent, changes in the remaining activation and adsorption
energies through linear correlations. The model predicted well the order of
catalytic activities for transition metals although it failed to describe the
experimental data quantitatively. The discrepancy was due to the neglect of
adsorbate-adsorbate interactions which play an important role at high
coverages. The model also predicted that the activity of copper can be
improved by increasing the strength with which CO and oxygen are bound to
the surface, thus suggesting possible directions for improving the LTS
catalyst.
Figure 21: Calculated (from the microkinetic model) and experimental exit mole fraction (inwet gas) of CO for Cu/ZnO/Al2O3 (51).
Water Gas Shift Catalysis 359
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
4.4. Deactivation of LTS Catalysts
4.4.1. Thermal Sintering
When formulated properly and operated under standard LTS condi-
tions, the Cu-ZnO-Al2O3 catalyst is quite rugged and lasts a few years.
The major sources of catalyst deactivation are thermal sintering of the
copper crystallites and poisoning by sulfur and chlorine compounds. Twigg
and Spencer have reviewed the deactivation of copper-based catalysts in
the WGS reaction (64). Due to the low melting point of copper metal
(1083uC), copper has low Tammam and Huttig temperatures. Cu-ZnO-
Al2O3 catalysts sinter and lose copper surface area, and, hence, catalytic
activity, when heated above 300uC. Indeed, one of the major roles of Al2O3
is to retard such growth of copper crystallites and function as a textural
promoter. Details of the mechanism of the thermal sintering of Cu
catalysts under hydrogen at elevated temperatures were studied by Tohji
et al. (65) using EXAFS techniques. As the temperature was increased in
hydrogen, a quasi- two-dimensional layer of copper metal epitaxially
developed over the ZnO support below 127uC. Between 127–230uC, small
copper metal clusters dispersed over ZnO start to appear. Above 250–
300uC, the small clusters fuse to give larger copper metal crystals
agglomerated on the support. Since these catalysts begin to lose copper
surface area and catalytic activity also above 250uC, it is reasonable to
assume that the active sites for the WGS reaction are associated with the
small copper clusters and their concentrations are diminished when the
small crystallites grow into larger ones. Thermal sintering leads to their
growth and consequent catalytic deactivation (66).
4.4.2. Sulfur Poisoning
The second major cause of deactivation of these catalysts is poisoning by
sulfur compounds present in the reaction gas stream. The LTS, Cu-ZnO-Al2O3
catalyst operates at 190–250uC, a temperature sufficiently low wherein
thermodynamics favors strong adsorption of poisons. Sulfur is a powerful
poison for Cu, as indicated by the change in enthalpy of the sulfidation
reaction [Eq. (28)] (64):
2 CuzH2S[Cu2SzH2 DH~{59:4 kJ=mol ð28Þ
Sulfiding of copper, hence, occurs. The corresponding equilibrium constant is
(64) about 1 6 10+5. Sulfur, accumulating on the surface, blocks the pores and
the active sites leading to catalytic deactivation. To retain the long term
360 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
activity of copper catalysts, it is usual to maintain gas phase sulfur
concentrations below 0.1 ppm of S. In addition to keeping gas phase
concentrations of sulfur low, the ZnO component of the catalysts is also
engineered during catalyst manufacture to divert the sulfur away from the
small Cu crystallites and absorb it in ZnO as ZnS. It is essential to keep the
crystallite size of ZnO as small as possible to accomplish this absorption. The
reaction of H2S with ZnO [Eq. (29)] is quite exothermic (66) and proceeds
readily:
ZnO sð ÞzH2S gð Þ[ZnS sð ÞzH2O gð Þ ð29Þ
DHu 5 276.7 kJ /mol; DSu 5 23.0 J.mol21 K21.
There are two forms of zinc sulfide, wurtzite (a-ZnS), and sphalerite (b- ZnS)
and both forms are seen in discharged plant samples of zinc oxide absorbents.
Sphalerite is the more stable form and the above data refer to this form. The
equilibrium constant at 500 K is 7.4 6 10 7. The reaction is strongly favored
thermodynamically.
4.4.3. Chloride Poisoning
Chlorine compounds, like HCl, form low-melting cuprous chloride (m.p. 5
430uC) on reaction with copper in the Cu-ZnO catalyst. The ZnCl2 formed also
has a low melting point (283uC). Their formation is favored thermodynami-
cally (64) under the WGS reaction conditions [Eqs. (30, 31)]:
Cu sð ÞzHCl gð Þ[CuCl sð Þz0:5 H2 gð Þ ð30Þ
DHu 5 2 43.5 kJ /mol, and
ZnO sð Þz2HCl gð Þ[ZnCl2 sð ÞzH2O gð Þ ð31Þ
DHu 5 2 121.8 kJ /mol; DSu 5 117.2 J.mol21 K21.
These mobile chlorides facilitate the movement and sintering of copper as well
as the ZnO crystallites on the catalyst surface. The limits on HCl content to
avoid catalyst poisoning are more severe than for H2S poisoning, on the order
of 1 ppb. Unlike the case of sulfur poisoning, the ZnO cannot offer any
protection in the case of chloride poisoning.
In addition to the major poisons, sulfur and chloride, the Cu-ZnO catalysts
are also deactivated by the presence of As, trivalent phosphorous, silica, and
transition metals like Fe, Co and Ni, in the feed stream. Due to their low
temperature of operation, Cu-ZnO catalysts do not form significant amounts of
coke when operated with purified feedstock (65).
Water Gas Shift Catalysis 361
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
5. SULFUR TOLERANT WGS CATALYSTS
The sulfur levels in natural gas or light petroleum naphtha are in the range of
5–50 ppm and conventional hydrodesulfurisation of the feedstock with Co (Ni)-
Mo- alumina catalysts is used before steam reforming them with nickel- based
catalysts. The latter are deactivated in the presence of sulfur. H2S can be
removed from natural gas as well as hydrodesulfuriser effluents by reaction
with ZnO at 370uC. Other sulfur compounds can be removed from natural gas
also by absorption at ambient temperatures on activated charcoal (loaded with
copper) or molecular sieves. The efficiency of these absorption systems
depends both on the type of sulfur compounds and on the amount of high
molecular weight hydrocarbons in the natural gas. Low boiling sulfur
compounds, like COS, are not strongly absorbed and condensable hydro-
carbons can rapidly saturate the absorbent. Catalytic hydrodesulfurisation
can remove COS. The removal of H2S by absorption in a hot ZnO bed is usually
not complete. Approximately 50 ppb H2S slips through and enters the
reformer upstream of the WGS reactor. After the volume expansion due to
the reforming reaction, the resulting H2S concentration in the gas entering
the WGS reactor is about 10 ppb. Due to the high temperatures in the reformer
and the low capacity of modern Fe2O3-Cr2O3-Cu HTS catalysts for sulfur
absorption, nearly all this residual sulfur exits the HTS stage and is removed
from the syngas by the Cu-ZnO-Al2O3 LTS catalyst located downstream.
Unlike Fe2O3-Cr2O3 catalysts, the Cu-ZnO-Al2O3 catalysts are adversely
affected by the presence of sulfur compounds in concentrations greater than
about 0.1 ppm. The deactivation is irreversible even when the sulfur is
removed from the feed gas stream. Normally the Cu-ZnO-Al2O3, LTS catalyst
reactors are designed for a space velocity of 1000–2500 h21 to take into account
poisoning by this sulfur. The actual catalyst volume in the reactor represents
approximately three times the volume needed by the kinetics. The Cu-ZnO-
Al2O3 catalysts, thus, serve also as a total sulfur absorber protecting
downstream processes (ammonia, methanol and Fischer-Tropsch syntheses,
hydrogenations, fuel cell electrodes, etc.) in industrial applications.
It is necessary to consider sulfur-tolerant WGS catalysts mainly when the
syngas is generated by the gasification and partial oxidation of heavy fuel oil,
tar sands, oil shale, coal, coke or biomass. Syngas from these raw materials
contain much larger concentrations of CO (up to 50%) (Table 1) and sulfur (up
to 3% wt) (5). In such cases, the Fe2O3-Cr2O3 catalyst had to be used as the
only WGS catalyst; conventional, Cu-based LTS catalysts cannot be used at
the high sulfur concentrations at the exit of HTS reactors in such operations.
The Fe2O3-Cr2O3 catalyst is sulfided during use and, in the sulfided state its
activity is much lower than in the oxide state. It is, hence, necessary to either
operate at higher H2O/CO ratios or remove the sulfur compounds from the
process gas over sulfided Co-Mo-alumina catalysts before it enters the HTS
362 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
reactor. In view of the high energy costs of operating at high H2O/CO ratios,
the latter option is usually adopted (5). In addition to removal of sulfur
compounds, these Co-Mo-alumina catalysts also serve to remove CO (by the
WGS reaction) from the process gas and, thus, serve as sulfur-tolerant, sour
gas shift catalysts. In fact, these catalysts are active mainly in the sulfided
form. When such sour gas catalysts based on Co-Mo sulfides are used, the
preferred minimum inlet sulfur in the feed for acceptable perfomance is about
300 ppm. If these catalysts are adequately presulfided before use, then they
can operate satisfactorily even in feed streams that contain H2S at a level as
low as 35 ppm. Non-sulfided Co-Mo catalyst exhibits very little WGS activity.
Commercial sour gas converters with Co-Mo catalysts operate in the
temperature range of 250–350uC and at pressures from atmospheric to 40
bar. Typical process conditions in a Co-Mo- based sour gas shift catalytic
reactor in a H2 plant using Texaco partial oxidation process to generate syngas
from heavy oil are shown in Table 6. The syngas from the partial oxidation
reactor contains 0.25% sulfur. The sulfided Co-Mo catalyst is deployed in 3
beds. The CO content is reduced from 46% (vol) at the inlet to the first bed to
1% at the exit of the third bed. It should be noted that all the cobalt moly-based
sour gas shift catalysts convert H2S in the presence of CO into COS. Therefore,
the COS concentration at the outlet of the last sour gas shift reactor is at
equilibrium. At high operating pressures and relatively high steam/dry gas
ratio, the resulting COS concentration is usually well below 0.1 ppmv.
However, under certain circumstances, the COS concentration can be much
higher and downstream COS hydrolysis has to be considered. One of the
advantages of the sour gas shift reaction using sulfided, cobalt molybdenum
catalysts is that they operate at much lower temperatures (250–350uC) than
conventional HTS, iron oxide- chromium oxide catalysts (350–450uC).
Table 6: Typical process conditions for cobalt-molybdenum catalyst-based sourgas shift reactor.
Bed 1 Bed 2 Bed 3
Inlet Feed Composition (mole %)CO 46 16 3.1CO2 6.9 26 34.2H2 47 57.9 62.6CH4 0.1 0.1 0.1Sulfur 0.25 — —
Inlet steam/gas, molar ratio 0.96 0.7 0.61Pressure, bar 35 34 33Inlet Temperature, uC 266 288 278Outlet Temperature, uC 411 367 292Space Velocity, h21 2940 2220 1785Outlet CO, mole % 16 3.1 1
Water Gas Shift Catalysis 363
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
Therefore, the water gas shift equilibrium is favored resulting in lower outlet
CO concentrations. The sour gas shift catalysts also need much less steam for
the same or even higher CO conversion since the possibility of metal formation
(and accompanying methanation and Fischer-Tropsch reactions) from these
sulfided Co-Mo catalysts is remote. However, these catalysts operate at lower
space velocities and, hence, need about 20% more catalyst than the
corresponding iron oxide- chromium oxide HTS catalysts. Additionally, they
also need sulfur in the syngas to be, and remain, in the active sulfided state.
They are used mainly for production of syngas from coal and heavy oil
gasification.
Addition of alkali to these sulfided catalysts promotes their WGS activity
(67, 68). It has also been reported (69, 70) that Co-Mo-based catalysts
promoted by Ti improve the WGS activity of the former in the presence of
sulfur compounds. The sulfided Co-Mo catalysts are not affected by poisons,
like NH3 or HCN when they are present in low concentrations (below about
0.5%). Phenol is a catalyst poison but the rate of deactivation is relatively low
at low concentrations of phenol. Phenol poisoning is reversible and the
catalyst can be regenerated with steam - air regeneration. A high benzene
concentration (above 10%) tends to decrease the catalyst’s activity. Chloride is
a major poison for these catalysts. Even at a 1–2 ppb level, chlorides have an
adverse effect on catalyst performance. The effect of chlorides is cumulative
and catalyst regeneration will not restore catalyst activity.
Mellor et al. (72) reported novel Co-MnO and CoCr2O4 catalysts tolerant to
sulfur up to levels of 220 ppm under WGS reaction conditions. However, in
coal-derived process gases containing between 0.25 and 0.3 mol% sulfur, and
at a reaction temperature of 400uC, the Co-MnO catalyst deactivated rapidly
and irreversibly with formation of bulk Co9S8 and a surface manganese sulfide
species. The CoCr2O4 catalyst deactivated only partially under similar
conditions. Bulk sulfiding of the CoCr2O4 catalyst to CoCr2S4 occurred at
550uC and this catalyst gave near equilibrium CO conversions in the WGS
reaction. A pre-sulfided cobalt chromium catalyst demonstrated typical sulfur
dependent mechanistic characteristics, with a maximum activity above 400uC(72). It may be noted that these sulfided Co-MnO and CoCr2O4 catalysts are
active in the WGS reaction only at temperatures considerably higher than the
sulfided cobalt moly catalysts. They are, hence, under a thermodynamic
handicap, vis-a-vis the latter regarding CO conversion.
6. Pt GROUP METAL-BASED WGS CATALYSTS
Even though the high WGS activity of the platinum group of metals was
known for many decades, their high price precluded their adoption in
364 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
commercial practice. The need for compact catalyst beds in automobile
applications of fuel cells provided an impetus for intense research in this field
during the last decade. Based on the experience from earlier studies of
automotive exhaust catalysts, gold, platinum and other metal –on- partially
reducible metal oxide supports have been the frontrunners in this area (73–
115). It should, however, be noted that auto exhaust catalysis operates under
an oxidative atmosphere above 400uC while the WGS fuel processing
environment is a reducing atmosphere at temperatures between 180–450uCand wherein the partial pressures of H2 and CO2 are much greater. In
addition, noble metals exhibit lower activity in WGS reactions below about
250uC which limits the CO exit levels to about 0.5–1.0 wt%. Of the many
catalysts that have been studied, precious metals (mainly Pt, Rh, Ru, Au, and
Pd) deposited on partially reducible oxides (ceria, zirconia, titania, iron oxides,
and mixed oxides of ceria, like ceria- zirconia) have been the most
investigated. These catalysts are quite active in the 250–400uC range. Pre-
reduction of these catalysts is not required and they can be safely exposed to
air during cool down or start-up without significant loss of performance, a
crucial requirement of fuel processor catalysts. The reaction rate for these
catalysts is close to zero order for CO and, hence, advantageous in driving the
reaction to equilibrium with minimal volume as compared to conventional Cu-
ZnO, where the order for CO is close to one (101). A large number of different
formulations, combining precious metals with partially reducible oxides, have
been proposed as promising catalysts in the literature for the WGS reaction.
Some typical examples are : Au-Fe2O3(73, 74), Au-CeO2(74, 75), Au-TiO2(76),
Ru-ZrO2(77), Rh-CeO2(68) , Pt- CeO2 (74, 78–81), Pt-ZrO2(82)] ,Pt-TiO2(83)
,Pt-Fe2O3(85), and Pd-CeO2(86, 87). Some non-noble metal-based catalysts,
with partially reducible metal oxide supports, have also been reported [Cu-
CeO2 (88, 89), Ag-TiO2 (76), Cu-TiO2 (76), Cu-ZrO2 (90), Cu-Fe2O3 (91).
Grenoble et al. (53) and Panagiotopoulou and Kondaridis (92) had shown that
the precious metal-based catalysts are bifunctional; both the metal and
support have a significant influence on the overall performance. Ceria and
ceria- zirconia have been explored extensively as supports for LTS catalysts in
the last decade. The incorporation of Zr improves the thermal stability
(against sintering), oxygen storage capacity and WGS activity of the ceria
crystallites. The bulk structure of zirconia-doped ceria is well known. Several
different tetragonal phases of varying degrees of stability can be formed
depending upon the Zr doping level, preparation technique, crystallite size,
shape and thermal history. The metastable ‘‘t’’ phase is commonly formed
when Zr doping ranges from 15–28 wt%. It has a tetragonal oxygen ion
sublattice and a cubic Ce/Zr fluorite sublattice. Another metastable t’ phase,
that can form between 28–63 wt% Zr doping, is tetragonal (P42/nmc space
group) on both sublattices having a c/a lattice parameter ratio greater than
Water Gas Shift Catalysis 365
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
unity. Fluorite structures, such as CeO2, commonly occur as crystallites that
maximize the most stable {111} surface face.
Apart from the noble metals and Au, other transition metals, such as
cobalt and nickel have also been investigated as WGS catalysts. They,
however, cause methanation of CO to CH4 under typical WGS reaction
conditions, especially, below 350uC [Eq. (17)].
A meaningful comparison and rating of all the reported catalysts is
difficult since the various authors had prepared their catalysts by different
methods, with different catalyst precursors, and had evaluated those using
different compositions of the feedstocks and at different reaction conditions.
In an attempt to bring some order in the picture, Thinon et al. (93) have
recently screened about 20 metal-on-oxide catalysts for the WGS reaction
under identical conditions using a model reformate as the reaction mixture.
They used commercial high-throughput equipment consisting of 16 parallel
reactors set-up to compare the activity and selectivity of these bifunctional
WGS catalysts. The catalysts were prepared by impregnation of the supports
with a solution of the corresponding metal precursors. The supports used were
commercial metal oxide powders with surface areas between 30–80 m2/g
except Fe2O3 (7 m2/g). The impregnated material were dried and calcined at
400uC. The feed stream to simulate a typical reformate consisted of 10% CO,
10% CO2, 20% H2O, 30% H2 and 30% Ar. Additional runs were also made with
a feed of 10% CO and 20% H2O diluted in Ar to investigate the forward
reaction in the WGS equilibrium. The catalytic activity was evaluated at 1 bar.
Catalysts based on Pt, Au, Cu, Rh, Pd, and Ru supported on ceria, alumina,
zirconia, Fe2O3 and TiO2 were evaluated. The Rh and Ru- based catalysts
were found to promote methanation reactions. The salient features of
their results are shown in Table 7. In this Table, catalytic activity has been
Table 7: Apparent activation energies and catalytic activities (at 300uC) of Pt, Auand Cu based catalysts ([81).
Catalyst Ea(kJ/mol) Activity (mmol/kg cats)
0.9%Pt/CeO2-Al2O3 70 271.5%Pt/ZrO2 58 202%Pt/CeO2 65 151.9%Pt/TiO2 23 391.5%Pt/Fe2O3 44 61.7% Pd/CeO2 43 85%Au/CeO2 9 271.5%Au/TiO2 29 125%Au/Fe2O3 21 121.5%Au/ZrO2 15 122.1%Cu/CeO2 43 168.9%Cu/CeO2 49 189.1%Au/Fe2O3 23 13
366 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
defined as:
Activity mmol=kgcat:sð Þ~FCO| XCO=Wcatð Þ, ð32Þ
where FCO (mol/s) is the molar flow rate of CO, XCO is the fractional conversionand Wcat (kg) is the weight of the catalyst. Pt/TiO2 and Pt/CeO2-Al2O3 are themost active catalysts at 300uC. It must be pointed out that the inhibitingeffects of the products on the reaction rates are neglected in calculating thevalues in Table 7. Hydrogen and carbon dioxide have, generally, a negativeeffect on the activity and they can also be the reactants for the methanationreactions. The Pt-based catalysts show the highest values for the activationenergies, Cu- based catalysts intermediate values and Au low values. The lowactivation energy observed for the gold catalyst should make it attractive atlow temperatures, especially in combination with Pt/TiO2, provided poisoningby the products, H2O and CO2, is not significant. These conclusions, however,have to be validated by experiments (a) at higher pressures (10–40 bars) and(b) for longer periods of time before application in industry.
One of the drawbacks of TiO2 (vis-a-vis ceria) as a support for Pt in this
reaction is the higher temperatures needed to partially reduce the former. In
an attempt to address this problem, Gonzalez et al. (94) were able to improve
the low temperature activity of Pt –TiO2 catalysts by incorporating ceria in the
support. Pt supported on ceria - modified TiO2 catalyst showed better thermal
stability and lower temperature reducibility compared to TiO2 and a higher
WGS activity than titania or ceria supports (Fig. 22). The catalytic activity of
Figure 22: CO conversion for the WGS reaction on supported Pt catalysts: (m) Pt/TiO2, (&)Pt/Ce-TiO2 ($) Pt/CeO2 (reference). Reaction conditions: total pressure 1 atm, GHSV 5 21200Lh21 kgcat21, feed gas composition (mol%): H2 28%, CH4 0.1%, CO 4.4%, CO2 8.7%, N2 29.2%,H2O 29.6%. Dotted line shows thermodynamic equilibrium limit (94).
Water Gas Shift Catalysis 367
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
Pt-TiO2 (as well as those of Pd- and Ir-TiO2) was also improved by the addition
of Re by Sato et al. (95) (Figure 23). Catalytic activities were evaluated by
them, in a closed gas circulation system, as an initial H2 formation rate in 10
Torr of CO and 10 Torr of H2O at 50–200uC. Among the Pt, Pd and Ir catalysts,
Pt-TiO2 was the most active catalyst lending further support to Thinon et al.’s
conclusions (81). Two important features were observed: (a) possible formation
of bimetallic surface clusters with Re in the case of Pt and Pd; and (b)
anchoring and ‘spacing’ of metal nanoparticles by highly dispersed Re over
TiO2 in the case of Ir. Panagiotopoulou et al. (96) studied the influence of the
source of the TiO2 support in Pt-TiO2 catalysts and found that the WGS
activity depended strongly also on the phase composition and particle size of
the TiO2 support; the activity increased with increasing reducibility of TiO2.
Both TPR and Raman spectroscopy data indicated that the titania could be
reduced by H2 or CO at temperatures as low as 150uC. Based on their results,
the authors suggested that the titania surface undergoes successive reduction
and oxidation by adsorbed CO and water, respectively, thereby cycling
between TiO22x and TiO2. Sato et al. (97), from CO adsorption, X-ray
photoelectron spectroscopy, in situ IR spectroscopy of adsorbed CO molecules
and catalytic studies of the Pt-Re-TiO2 system, observed that a bimetallic Pt-
Re alloy is formed under reaction conditions and that an additional surface
compound is formed between Pt and Re during the WGS reaction. It is not a
Figure 23: Influence of Re content on the H2 formation rates of WGS reaction over 2 wt% Pt-Re/TiO2 (100uC), 1 wt% Pd – Re/TiO2 (200uC) and 1 wt% Ir-Re/TiO2 (100uC) (95).
368 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
mixture of Pt and ReOx. This surface compound, probably, accounts for the
greater activity of Pt-Re-TiO2. The binding energy of the Pt electrons is lower
in Pt-Re-TiO2 than in Pt-TiO2 suggesting that the Pt crystallites are slightly
negatively charged. CO is adsorbed more strongly on Pt in Pt-Re-TiO2 than on
Pt-TiO2. This is understandable since a more negative Pt will transfer
electrons more easily to the antibonding orbitals of CO, thereby stabilizing CO
in the adsorbed state. IR spectra of CO on Pt-TiO2 reveals only linearly
adsorbed CO. On Pt-Re-TiO2, bridged CO as well as formate ions are seen
additionally indicating that CO is more activated on Pt-Re-TiO2 than on Pt-
TiO2. A more activated CO is more likely to undergo further conversion to
CO2. Hence, the Pt-Re-TiO2 is more active.
When long-term catalytic runs were carried out over the promising Pt-
TiO2 catalysts, Azzam et al. (98) found that even though their Pt-TiO2 was a
very active and selective catalyst for the WGS reaction, they deactivated with
time on stream (Fig. 24). Catalyst deactivation during the WGS reaction was
also a problem with Pt- and Pd- ceria catalysts (86–87, 101). Wang et al. (86)
investigated the mechanism responsible for the irreversible deactivation of
ceria- supported precious metals for the WGS reaction through accelerated
aging tests. They showed that deactivation of Pd- ceria occurs more rapidly at
400uC than 250uC when operating with an integral reactor in 25 Torr each of
CO and H2O. By heating a fresh catalyst in H2, H2O, CO or CO2, it was
discovered that deactivation occurs due to the presence of CO. Similar
conclusions were also reached by Ruettinger et al. (86). Measurements of
Figure 24: WGS CO conversion for Pt/TiO2 with time on stream at 300uC. After 22 h tests, thecatalyst was subjected to the following treatment for 1 h each: (a) O2 at 450uC, (b) H2 at300uC, (c) N2 at 300uC, then tested in WGS. Testing conditions: PCO5 60mbar, PH2O5150mbar,P52 bar, and GHSV5410,000 h21, mcat 551 mg (98).
Water Gas Shift Catalysis 369
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
metal dispersion by CO adsorption and by X-ray diffraction show (86) that
deactivation on Pd- ceria and Pt - ceria catalysts was due to loss of noble metal
surface area. Pd dispersion values, for example, decreased from 23% (fresh
catalyst) to 3% after a 10 hr treatment in CO at 400uC causing the CO
conversion to decrease from 25% (fresh catalyst) to 3% after treatment for only
2 hr in CO at 400uC. The corresponding dispersion values after similar
treatments in H2O and CO2 were 25 and 16, respectively. The CO conversions
were also not significantly decreased by similar treatments in H2, H2O or CO2
at 400uC. Finally, water gas shift rates on a series of Pd- Ceria catalysts with
ceria crystallite sizes ranging from 7.2 to 40 nm and Pd loadings of either 1 or
6 wt% demonstrated that the rates were strictly proportional to the CO
adsorption capacity and, hence, Pd surface area (Fig. 25). Later, using in-situ
FTIR spectroscopy in the OCO stretching region, Gorte et al. (106) observed
strongly – held, carbonate-like species on the surface, formed from CO. These
were postulated to be the major cause of catalyst deactivation. The authors,
however, do not show the C-H stretching region to confirm the presence/
absence of the formate. Since (a) high temperature treatments in H2 did not
reduce catalytic activity, and (b) high temperature oxidation also did not
restore the activity of deactivated catalysts, they ruled out the over-reduction
of ceria as a contributory factor to the deactivation of the catalyst. Just
because there is a coverage of a surface by a species does not mean that the
Figure 25: Differential water gas shift rates as a function of CO adsorption capacity for aseries of Pd/ceria catalysts in 25 TOrr each of CO and H2O at 250uC. X, 1 wt%Pd/ceria, withceria calcined at 600uC, &, 6 wt%Pd/ceria, with ceria calcined at 600uC, D, 1 wt% Pd/ceria,with ceria precipitated and calcined at 350uC; $, 1 wt% Pd/ceria, with ceria calcined at800uC; m, 1 wt% Pd/ceria, with ceria calcined at 950uC (86–87).
370 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
species in question is causing deactivation. First, one has to see the increase of
that species as a function of time. Secondly, loss of metal support interaction
(e.g., growth of metal particle size) can cause the steady state coverage of the
intermediate (e.g., formate, carbonate) to increase because the metal may
assist in decomposing that intermediate and any loss in the metal’s interaction
with the support (e.g., ceria) would cause the inventory of the intermediate to
build up (since it would be reacting more slowly with loss of the metal’s
interaction). So, while the first point is necessary, the second point explains
why the first point is not sufficient. Zalc et al. (109) observed a strong
dependence of the deactivation rate on the presence of hydrogen in the feed
and suggested that irreversible over-reduction of ceria by hydrogen may,
under certain circumstances, be yet another cause of deactivation of the Pt-
ceria catalysts. It has also been proposed (110) that, yet another potential
cause of deactivation was the growth of ceria crystallites and occlusion of Pt
crystallites, and the consequent decrease of the BET surface area during the
reaction. While all the above-mentioned factors may potentially lead to
catalyst deactivation, it is difficult to extrapolate the validity and relevance of
these conclusions to the WGS reaction in industrial reactors in view of the
widely different methods of catalyst preparation, activation and reaction
conditions used by the various authors. Noble metal- based catalysts
containing a combination of Pt, CeO2 and TiO2 have, recently, been claimed
to be superior WGS catalysts (111). Baidya et al. (112) have compared the
structure, reducibility and catalytic activity of various solid solution oxides
containing cerium, titanium and platinum. Nanocrystalline Ce12xTixO2 (0, x
, 0.4) and Ce12x2yTixPtyO22d (x 5 0.15, y 5 0.01, 0.02) solid solutions,
crystallizing in the fluorite structure, were prepared by a novel, single step
solution combustion method. Their fluorite structure and solid solution
formation were confirmed by XRD Rietveld calculations. Temperature
programmed reduction and XPS study of Ce12x TixO2 (x 5 0.00–0.04) showed
complete reduction of Ti4+ to Ti3+ and reduction of , 20% of Ce4+ to Ce3+ state,
compared to 8% Ce4+ to Ce3+ reduction in the case of pure CeO2, below 675uC.
The insertion of both Pt and Ti ions in the ceria lattice enhanced the
reducibility of CeO2. Ce0.84Ti0.15 Pt0.01O22d crystallized with a fluorite
structure and Pt was ionically substituted with 2+ and 4+ oxidation states.
The amount of hydrogen adsorbed at 30uC over Ce0.84Ti0.15Pt0.01O22d was two
orders of magnitude larger than that over pure 8 nm Pt metal crystallites. CO
and hydrocarbon oxidation activities were also much higher over the Pt-Ti-
Ceria sample, Ce12x2yTixPtyO2 (x 5 0.15, y 5 0.01, 0.02), compared to the Pt-
Ceria sample, Ce12x PtxO2 (x 5 0.01, 0.02). Synergistic involvement of the
Pt2+/Pt0 and Ti4+/Ti3+ redox couples in addition to Ce4+/Ce3+ were held
responsible for the higher reducibility and catalytic activity in the oxidation of
CO (Figure 26). As may be seen from the figure, the Ce0.84Ti0.15Pt0.01O2-d has a
Water Gas Shift Catalysis 371
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
much lower light-off temperature with T50 5 170uC compared to T50 5 260uCfor the Ce0.99Pt0.01O22d sample indicating that the incorporation of Ti in Pt-
ceria has enhanced the CO oxidation activity probably by increasing the
oxygen ion vacancy concentration and the consequent increase in the oxygen
storage capacity of the material. Parallel TPR and XPS measurements also
confirmed the greater reducibility of the Pt-titania-ceria samples. XPS data
also indicated the presence of Pt in the ionic (Pt2+) state. In view of the known
deactivation of both the Pt-TiO2 and Pt-CeO2 samples during prolonged WGS
reactions, it will be interesting to study the long term stability of the Pt-TiO2-
CeO2 solid solution catalyst in the WGS reaction.
The long-term stability of ceria-based catalysts for WGS operation in fuel
cell applications was investigated by Zalc et al. (109) who prepared a variety of
Pt-ceria WGS catalysts and tested them in the range 250–450uC under feed
and reaction conditions typical of a reformer outlet. They observed first order
deactivation. Virtually identical deactivation rates were found for all the Pt-
ceria catalysts tested. Significantly lower deactivation rates were observed
when hydrogen was not present in the feed. Attempts to rejuvenate the
catalyst by heating under steam and under air were unsuccessful (109).
Catalyst deactivation is still a major obstacle in the commercialization of WGS
catalysts for fuel processing. The goal for most programs is 40,000 hours of
catalyst life. This is an ambitious goal of about 4.5 years of continuous
operation. There are indications, from studies in the industry that addition of
other rare earth elements like lanthanum, praseodymium, etc., to the ceria-
zirconia support can reduce, to some extent, the agglomeration of ceria
crystallites and, the consequent deactivation of Pt-Ceria-Zirconia catalysts.
Another potential cause of catalyst deactivation in the case of Pt group-based
Figure 26: (a)CO oxidation over Ce0.85Ti0.15O2, and Pt, Ti substituted oxides. CO52 vol%,O252 vol%, Flow rate 5 100sccm, GHSV543,000h21, W525mg (112).
372 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
catalysts is the relatively larger formation of hydrocarbons, including
methane, over these catalysts compared to the conventional Cu-ZnO-Al2O3
catalysts. Once formed, these hydrocarbons can undergo further reactions,
like dehydrogenations/ hydrogenolysis/hydrogenation (over Pt), oligomerisa-
tion, carbon formation etc. This problem will be more significant at lower
temperatures? The known higher Fischer-Tropsch activity of the Pt group
metals (compared to copper) for the synthesis of hydrocarbons from CO and
H2, at 200–350uC, is a handicap in the WGS reaction. We may, perhaps, have
to reduce the Fischer-Tropsch activity of the noble metal component to the
level of copper without, however, sacrificing its greater catalytic activity in
WGS reactions. Incorporation of the Pt in the lattice sites of the partially
reducible cerium oxide (as in the work of Baidya et al. (99)) and preserving the
Pt in the ionic state under the WGS reaction conditions, may be one potential
solution since ionic Pt, while active in redox reactions (109), is not known to
possess Fischer –Tropsch activity.
7. Au-BASED WGS CATALYSTS
The low WGS activity of Pt-, Rh- and Pd-based catalysts below 250uC had led
to increased interest in more active catalysts to take advantage of the
favorable thermodynamics at these temperatures. In the past 10 years,
supported gold catalysts with remarkably high activity for the WGS reaction
have been discovered (116, 117). Gold catalysts can offer some advantages in
the range of 180–250uC where the Pt group metals are insufficiently active
(118–143). They are, also, not pyrophoric if exposed to air and require no
exceptional pre-treatment before use. Figure 27 illustrates the high activity of
gold when compared to the Pt and the Cu-Zn-based catalysts. First developed
as a low temperature catalyst for the preferential oxidation of carbon
monoxide (in a mixture of CO and H2) by Haruta et al. (118), it was soon
recognized that the catalytic activity was high only when the particle size of
gold was very small, of the order of 1–5 nanometers (119). Extension of the
studies to low temperature WGS over Au/ a-Fe2O3 (120) and Au-Fe2O3 - MOx
(121) showed that the catalysts were active at temperatures as low as 160uC.
Again, the activity was associated with highly dispersed gold (about 2 nm
particles) (120). The dissociative adsorption of water on the nano gold particles
followed by spillover of hydroxyl groups onto adjacent ferric oxide sites,
involving the redox couple Fe3+/ Fe2+, was postulated. Promotion of Au/Fe2O3
by Ru increased the WGS rate threefold at 120uC (122). Gold nano particles
supported on other supports including TiO2 (123–125), and ceria (126–128)
were also found to lead to catalysts active at temperatures below 200uC. Based
on their studies of Au-TiO2, Andreeva et al.(120) postulated the existence of
Water Gas Shift Catalysis 373
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
gold in an ionic form at the interface between the Au and the TiO2 phases,
probably as Au ions inserted in the surface regions of the TiO2 lattice. In order
to estimate the relative contributions of the metallic and ‘‘ionic’’ gold to
catalytic activity, Fu et al. (137) measured the rates of the WGS reaction after
leaching out the metallic Au from Au-ceria with NaCN. The rates and
apparent activation energies were the same, before and after leaching with
NaCN, highlighting the importance of the fraction of the gold (presumably
ionic) that was not leached out by the NaCN treatment. The amount of such
‘‘ionic’’ gold inserted in ceria was found to increase with decreasing crystallite
size of ceria. Large crystallites of ceria did not retain any gold. Incorporation of
gold also increased the stability of the ceria microcrystallites. Catalytic
activity in the WGS reaction was also reasonably stable. When this idea was
extended and gold ions were stabilized in the framework of an ionic lattice, as
in Au2Sr5O8 or La2Au0.5O4 (128), not only was the sintering reduced and the
thermal stability of the catalyst increased but the catalytic activity was also
enhanced. It should be mentioned here that recent studies indicate that ionic
gold is unlikely to be present, in the steady state reducing conditions during
WGS reaction, especially, at higher pressures (129). In-situ time-resolved X-
ray diffraction and X-ray absorption spectroscopy were used by Rodriguez et
al. (129) to monitor the behavior of nanostructured Au-CeO2 catalysts under
the WGS reaction. Above 250uC, a complete AuOx ) Au transformation was
observed with high catalytic activity. Photoemission results for the oxidation
and reduction of Au nanoparticles supported on rough ceria films or a CeO2
(111) single crystal corroborated that cationic Aud+ species cannot be the key
Figure 27: CO conversion over supported Au and Pt catalysts (116).
374 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
sites responsible for the WGS activity at high temperatures. They suggested
that the active sites in Au - ceria catalysts involve pure gold nanoparticles in
contact with O vacancies on the ceria. The role of cationic Au3+ and nonionic
Au0 species in the LTS reaction over Au-ceria catalysts was also studied by
Karpenko et al. (130–131) by comparing the reaction behavior of a cyanide-
leached catalyst with that of non-leached catalysts. Using rate measurements
as well as in situ spectroscopic and structure-sensitive techniques, they found
that, based on the Au mass balance, cyanide leaching removed all the Au
except for ionic Au3+ species, and that leaching resulted in pronounced decay
of the catalyst mass- normalized activity to 1–25% of that of a non-leached
catalyst. The extent of the activity loss strongly depended on the post- leaching
treatment of the leached catalyst. Both the catalyst pretreatment after the
leaching and, in particular, the WGS reaction resulted in considerable
reformation of Au0 aggregates and metallic Au0 nanoparticles as indicated
by Au(4f) signals at 85.8 ev(Au3+), 84.0 – 84.6 ev (up-shifted signal of small Au0
aggregates), and 84.0 ev (metallic Au0). Hence, they concluded (130) that (a)
Au0 species, including both small aggregates and metallic nanoparticles
contribute predominantly to the WGS activity, and (b) cationic gold has a
negligible contribution to the WGS activity in the steady state. Au ions are,
expectedly reduced to Au0 atoms in the reducing atmosphere during the WGS
reaction. Combining TEM, XRD, XPS, DRIFTS and activity studies, they
concluded (131), further, that for reaction up to 200uC, catalyst deactivation
was dominated by the formation of stable adsorbed monodentate carbonate
species. The influence of other effects, such as catalyst reduction/ oxidation
were less significant.
Au-CeO2 and Au-CeO2-Al2O3 catalysts were also investigated by Andreeva
et al. (132) who compared samples, prepared by a mechanochemical
activation, with those prepared by a conventional coprecipitation; the former
were more active. This was attributed to the smaller size of the Au and ceria
crystallites in the former. The main role of alumina was that of a textural
promoter in stabilizing the Au and ceria crystallites against agglomeration
during the WGS reaction and, thereby, maintaining a high catalytic activity in
the steady state. The addition of alumina to ceria results in smaller ceria
crystallites and, consequently, an increase in the number of oxygen vacancies
and oxygen storage capacity of ceria, as estimated from temperature-
programmed reduction experiments. A correlation was found between WGS
activity and the oxygen storage capacity of the samples (132).
In an attempt to gain insights into the reactivity of supported Au
nanoparticles, Janssens et al. (133) applied density – functional calculations,
adsorption studies of CO and oxygen on single crystal surfaces and WGS
activity measurements on well-characterised, supported gold particles. They
attributed the increasing activity of supported Au catalysts, with decreasing
Water Gas Shift Catalysis 375
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
Au particle size, to the increasing number of low - coordinated Au atoms
present in such small particles. Their DFT calculations indicate that
adsorption of CO and oxygen on the densely–packed surfaces (which expose
Au atoms with high coordination numbers, like 8) is generally difficult or
thermodynamically not possible. On the other hand, adsorption was favored
on Au atoms with a lower coordination number. The effect of the Au
coordination number on the adsorption strength of CO and oxygen was found
to be larger than other electronic effects or strain and was, therefore, a crucial
parameter for the catalytic activity. The smaller particle size and support
effects influence the catalytic activity only indirectly through their influence
on exposing a larger number of low-coordinated Au atoms. Among such atoms,
the Au atoms located at the corners of Au crystallites (and, hence, with the
lowest coordination numbers) were the most reactive. During the synthesis of
the various supported Au catalysts, the properties of the support surface (i.e.,
quality and number of nucleation sites) influence the size, dispersion and
morphology of the Au nanoparticles, and, thereby, the concentration of active,
low coordinated sites. Moreover, during catalytic operation, the metal-support
interface energy, which is influenced largely by the support, has a significant
influence on the stability of the particles. A large interface (metal – support)
energy probably can retard the sintering of the Au nanoparticles. Figure 28
(133) shows that there is a clear relation between the adsorption energy of CO
(and oxygen) and the coordination number of the Au atoms to which these
molecules are attached. The lower the coordination number of Au, the stronger
Figure 28: Correlation between the binding energies for CO, O2, and O atoms on Au andthe coordination number of the Au atoms. The solid blue dots indicate experimentallydetermined values for CO adsorption energy on steps, edges and the Au (110)-(162) surface(133).
376 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
the Au-CO bond. The coordination number effect on the adsorption energies
(Fig. 28) (133) can, in turn, be related to the changes in the surface electronic
structure. The low- coordinated Au atoms have high-lying metal d states,
which are in a better position to interact with the adsorbate valence state than
the low- lying states of the high coordination number Au sites of the close-
packed structure (133). This is one of the main reasons why the low-
coordinated transition metal atoms on surfaces are, generally, more active in
catalytic reactions. The trend that the CO adsorption strength on Au increases
with decreasing Au coordination number is also reflected in temperature-
programmed desorption spectra of CO on Au single crystal surfaces and well-
defined nanoparticles (Fig. 29). Janssens et al. (133) attributed the desorption
around 2103 to 283uC (in Fig. 29) to CO adsorbed on defect or corner sites, the
desorption around 2 123uC to CO adsorption on the (110)-(162) surface, and
Figure 29: Temperature programmed desorption of CO on various Au samples (133).
Water Gas Shift Catalysis 377
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
the desorption around 2163 to 2143uC to CO adsorbed on the edge of the
nanoparticles, or step sites on the single crystal surfaces. For particles with a
given shape, the number of corner atoms per gold particle is independent of
the particle diameter. Hence, for a given amount of gold in the catalyst, the
larger the number of metal particles (i.e., larger the dispersion), the larger will
be the number of corner atoms and, hence, the catalytic activity. This
explanation (133) is fundamentally different from the quantum size effects
conventionally invoked and which ascribes the higher catalytic activity of
small Au particles to changes in electronic structure as the particle size
decreases. Though quantum size effects are important for very small particles
containing only a few atoms (134), they seem not to be necessary to explain the
catalytic effect for supported Au particles larger than about 1.5 nm. It may be
noted that Bond and Thompson had also pointed out earlier that significant
chemisorption of molecules like CO occurs only when an adequate number of
low-coordination surface Au atoms are present (135).
One of the major drawbacks of the gold catalyst is catalytic deactivation
during use. There are two potential causes of deactivation of Au catalysts
(135). The first is Au particle growth, giving larger, but less active particles,
and the second is the formation of unreactive species formed during the WGS
reaction and physical blocking of those sites at which participation of the
support is essential for high activity. Such species include carbonates,
bicarbonates, formates etc. In attempts to prolong the catalytic activity of
supported gold catalysts (for CO oxidation), Moreau and Bond (136) have
recently found that inclusion of Fe(OH)3 or lanthanum oxide during the
preparation of Au catalysts supported on ceria and zirconia, gave better
activity and much improved stability with time-on-stream. This effect was
linked to the ability of the FeOx phase to provide hydroxyl groups, stable at
the reaction temperatures, that are needed for the catalytic action and to form
anion vacancies (by replacement of a tetra- by a trivalent metal cation) at
which O2 or H2O molecules can chemisorb. The effect is similar to those seen
when La3+ or, Fe3+ are dispersed in the ceria lattice (136).
The relatively high activity of gold catalysts has been challenged. Jacobs
et al. (138) reported that a 5%Pt- ceria catalyst was much more active than a
5% Au-ceria catalyst. They attributed their distinctive results, essentially, to
(a) the higher content of Pt in their catalysts, (b) the complete reduction of
platinum oxide, and (c) the ‘‘careful activation’’ of their samples. Differences in
the details of catalyst preparation and determination of the dispersion of noble
metals on partially reducible oxides (see Section 10), adopted by the various
researchers, perhaps, explain such differences. In addition, differences in the
composition of the feedstocks used to evaluate the catalytic activities of the
Au- and Pt-based catalysts and the reaction conditions will also influence the
conclusions. The two metals, Pt and Au, respond differently (116) to changes
378 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
in the concentrations of the reactants. The power law dependencies, on the
concentrations of the reactants and products, are different for Pt and Au (116).
Thus, under certain conditions, the order with respect to carbon monoxide is
negative for Pt but positive for Au. The rate for Au, for example, was given by:
R~k CO½ �0:7 H2O½ �0:6 CO2½ �{0:3H2½ �{0:9: ð33Þ
The positive order with respect to CO reflects the weak adsorption of CO on Au
in contrast to Pt where CO is much more strongly adsorbed, at least, up to
200–250uC, accounting for the negative order with respect to CO on Pt. In
addition to catalytic activity, the stability of the catalyst during prolonged use
under various process conditions is also of major importance. Here, Pt-based
catalysts are quite rugged and have a distinct advantage over Au-based
catalysts. The performance of the latter is more sensitive to conditions of
storage and operation. In view of the importance of resolving this issue (the
relative superiority of Pt- and Au- based catalysts) in the design of fuel
processors for fuel cells, a quantitative comparison of their kinetic behavior
using industrial feedstock and under identical, but, realistic conditions, is
desirable.
Can the performance of gold-based catalysts be improved? Two approaches
have been taken in the last few years: (a) Improving the metal function by
combining Au with another metal (like Pt) to form bimetallic catalysts, and (b)
incorporating promoters in the ceria support. Juan et al. (139) have reported
that when Pt, Pd, W, or Ni is added to Au – ceria, there is a synergistic effect
and the resultant bimetallic catalysts are more active than Au-Ceria or Pt-
Ceria. The catalysts were tested in the temperature range 150–500uC, with a
H2O/ CO ratio of 13.5 and GHSV of 52,000 h21 (Fig. 30). The Au-Pt -Ceria
clearly displayed a much higher activity compared to Au-Ceria at the same
Figure 30: CO conversion over Au-M bimetallic promoters on CeO2 (139).
Water Gas Shift Catalysis 379
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
temperature. The WGS activities over these samples were ranked as: Au-Pt .
Pt . Au-Pd . Au-W . Pd . Au-Ni . Au. For a quantitative estimate of the
synergy between the two metallic functions (Au and Pt), the atomic loadings
as well as the dispersion of the two metals must be kept the same.
It has to be noted that the catalytic activities (in Fig. 30) of the bimetallic
gold and other metal- supported catalysts are expressed in terms of CO
fractional conversion without using their normalized specific activities
(activity per metal site) since the interactions between different metals over
these bimetallic gold catalysts is not, sufficiently, clear to differentiate
between CO adsorption on Au or Pt or both. The total number of metal sites
per gram of the catalyst (in mmoles per g), evaluated via CO chemisorption
with the assumption of 1:1 CO to metal site ratio, were 2.7(Au), 3.8(Pt) and 2.5
(Au-Pt), all of them supported on ceria (139). Among all the catalysts, a
3 wt%(Au-Pt)- CeO2 displayed the best catalytic activity in the WGS reaction.
Very interestingly, the WGS activity was strongly correlated with the surface
reducibility data from temperature-programmed reduction experiments
(Figure 31). The reducibility of the catalysts, in turn, depended on the
modified local electronic band structure of the promoted ceria. CeO2 shows two
distinct reduction peaks, one at 440uC (assigned to reduction of surface
oxygen) and another at 800uC (reduction of bulk ceria oxygen).The incorpora-
tion of gold and the metallic promoters in the ceria catalyst facilitated the
reduction of surface oxygen at lower temperatures while the reduction of bulk
oxygen remained unchanged. Among the different promoters studied by them,
the Au-Pt-Ceria combination was more effective than Pt, Pd, or Au alone on
ceria in giving higher WGS activity at lower temperatures. The temperature of
the surface oxygen reduction peaks, in Au-Pt, was 120uC. The corresponding
temperatures in the case of Pt and Pd supported on ceria were 130uC and
135uC, respectively. It may be noted that Fu et al. (140) and Andreeva et al.
(141) had also reported similar low temperatures (around 150uC) in theTPR
profile for reduction, in hydrogen, of their Au - ceria prepared by a deposition –
precipitation technique. There is also additional support from the literature
(140, 142) that Au facilitates reduction of surface oxygen at temperatures
lower than even noble metals. The ranked order (139) of the lowering in
temperature (Fig. 31) of the first reduction peak (from surface oxygen loss
from ceria) was Au-Pt . Pt . Au-Pd . Au-Ca . Au-W . Pd . Au-Ni . Au.
This order matched, closely, the relative order of the WGS catalytic activity of
these samples.
Ceria is a n-type semiconductor whose electronic band structure can be
modified by promoters. From the UV diffuse reflectance spectra of the
samples, a clear, blue shift of the absorption edge (O2p R Ce 4f) of the ceria
upon doping with Au or Au-Pd was observed. The degree of band gap widening
was found to relate to different bimetallic promoters (139). The order of
380 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
bandgap widening was remarkably similar to both the orders of surface
oxygen reducibility and WGS activity suggesting an electron transfer
mechanism at the interface between ceria and the metallic components
facilitating the redox transformations occurring in ceria (139):
2Mz2 O½ �u2 Mzz2½ �zO2z2e, and ð34Þ
Figure 31: TPR profiles of CeO2 (a) monometallic and gold bimetallic doped CeO2 samples:Au-Pt/CeO2 (b); Au-Pd/CeO2 (c); Pd/CeO2 (d); Pt/CeO2 (e); Au-W/CeO2 (f); Au-Ca /CeO2
(g); Au-Ni/CeO2 (h) and Au/CeO2 (i) (139).
Water Gas Shift Catalysis 381
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
2ez2Ce4zu2 Ce3z, ð35Þ
where M represents a metal, like Pt, Pd, Au etc., [ ] an oxygen anion vacancy
and [O] denotes an oxygen anion on the surface.
One of the problems in comparing, quantitatively, the specific WGS
reaction rate data over precious metal- reducible oxide catalysts from different
laboratories is the absence of a reference method for determining the
dispersion of the metal on the support. In the case of a metal dispersed over
a non-reducible oxide, like Pt-silica or Pt-alumina, the experimental
procedures (temperature of reduction, temperature of H2 or CO adsorption,
etc.) and the stoichiometry of the chemisorption for determination of
reproducible and accurate metal dispersion values are accepted and have
been standardized. An important difficulty originates when the support can
also adsorb (and, even react with) the probe molecules, H2 or CO in quantities
comparable or even more than the metal itself. Such is the case for redox
supports such as ceria, titania or ceria-zirconia on which hydrogen spillover
processes (from the metal to the oxide) occur easily in the presence of a
metallic phase, like Pt or Au. Perrichon et al. (144) determined the Pt
dispersion by chemisorption of H2 and CO in a series of Pt-ceria-zirconia
catalysts covering the full range of composition between ceria and zirconia
using volumetric techniques and FTIR spectroscopy. Using IR spectroscopy of
adsorbed CO to distinguish the CO adsorption on the Pt surface from that on
the ceria-zirconia support allowed them to validate a protocol of hydrogen
chemisorption for measuring the metal dispersion. A first method is based on
the CO adsorption isotherm analysis, using IR spectroscopy as the detection
tool. Apart from the quantitative analysis of the adsorbed/desorbed gas phase,
this method also gives information about the coordination mode of the CO
molecule on the metal particle, linear or bridged. The second hydrogen
chemisorption method is based on the use of a double isotherm of hydrogen
adsorption at 278uC , this low temperature being required to suppress the
hydrogen spillover from the metal to the ceria- zirconia support. The
irreversible adsorption of H2, measured either at saturation or by extrapola-
tion to zero pressure, leads to the most reliable metal dispersion values which
can be independently confirmed by FTIR spectroscopy of adsorbed CO. Pt
dispersion, measured by this method, was always higher on the mixed oxides,
ceria- zirconia, than on the pure ceria or zirconia supports (144).
8. MONOLITH-COATED WGS CATALYSTS FOR FUEL CELLS
Current fuel cells use hydrogen, produced by reforming (steam or auto-
thermal) and partial oxidation of natural gas or liquid fuel, to generate
382 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
electricity. The WGS reaction is a critical step in reducing the CO
concentration in such H2 feed streams, especially in low temperature fuel
cells which are not tolerant to CO in concentrations more than 50– 100 ppm
(Fig. 32). The design of fuel processors for stationary fuel cells is less
constrained by the need for compaction and fast response as it is for
automotive applications. Compared to conventional industrial WGS plants for
the generation of hydrogen, however, a reduction in reformer and water gas
catalytic reactor sizes by over two orders of magnitude is necessary before fuel
cells can compete techno-economically with other modes of electricity
generation in automobile applications.
Among the various fuel cells, the Proton Exchange Membrane Fuel Cells
(PEMFC) offer great promise as an alternative to traditional fuel combustion
for generation of electricity for mobile and stationary applications. This H2
must have a CO concentration lower than 50 ppm. In a typical fuel processor
for a PEMFC, the hydrocarbon undergoes reforming by the steam reforming
(SR), autothermal reforming (ATR) or catalytic partial oxidation (CPO). The
reformate then undergoes a series of reactions with the goal of reducing the
concentration of CO and increasing the concentration of H2. The first is the
water gas shift reaction which reduces the concentration of CO in the
reformate from about 10% to less than 1% while increasing the hydrogen
concentration. Further CO clean-up methods, such as preferential CO
Figure 32: Water gas shift in a Fuel Processor for fuel cells (8–9).
Water Gas Shift Catalysis 383
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
oxidation or selective CO methanation are needed to reduce CO levels to below
50 ppm.
Most of the papers on precious Pt- group or gold catalysts, described in the
earlier sections, report results with powder catalysts or those in the form of
extrudates or tablets. However, in many fuel cell applications, due to
requirements of very high space velocities (to reduce the volume of the
catalyst bed), low pressure drops and mechanical strength, the use of monolith
catalysts is almost mandatory. In response to this requirement, during the
last decade, many publications and patents (especially from industrial
research laboratories) have appeared that describe results with noble metal-
based WGS catalysts washcoated on ceramic or metallic monoliths (113–115).
The requirements of WGS catalysts for vehicular fuel cell applications are
quite different from those needed in NH3 or H2 plants (Table 8). The
development of robust WGS catalysts that can operate in such demanding
conditions is critical in the development of hydrogen generators for fuel cells.
Furthermore, much more active catalysts are sought to make the fuel
processor as compact as possible. The challenge is formidable to achieve such
high catalytic activity at low temperatures (111, 145–148). It is also desirable
to replace the two, HTS and LTS, reactors operating in the 350–450uC and
200–300uC , respectively, by a single, medium temperature, shift reactor in
the 200–350uC range. The activity of the Pt-group catalysts is inadequate
below 250uC. Since CO concentrations as high as1% are tolerated by recent
preferential oxidation (PROX) catalysts used in mobile fuel processors without
sacrificing too much efficiency, this enables one to run monolithic WGS
catalysts at temperatures as high as 300–350uC to reduce the residual CO
content to about 1% in the reformate gas. Even for stationary applications,
this concept of replacing the two HTS and LTS reactors by a single shift
reactor is appealing because of the immense volume/ weight savings and the
Table 8: WGS catalyst requirements for mobile and stationary applications (7).
WGS catalyst attribute Mobile application Stationary application
Volume reduction Critical, ,0.11kW-1 Not as constrainedWeight reduction Critical, ,0.11kgkW-1 Not as constrainedCost Critical, ,$1kW-1 Not as criticalRapid response Critical, , 15 secs Load followingNonpyrophoric Important Eliminate purgingAttrition resistance Critical No constraintSelectivity Critical ImportantNo reduction required Critical ImportantOxidation tolerant Critical ImportantCondensation tolerant Important ImportantPoison tolerant Desired DesiredPressure drop Important Important
384 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
ruggedness of the monolith catalysts. While the efficiency of the monolith
catalysts is slightly lower (about 2–5%), the catalyst bed volume is about 90%
smaller than a comparable system with a particulate catalyst. At the present
stage of development of the Pt-group metal-based, single stage shift reactor, a
two-stage PROX or selective methanation catalyst may be necessary to reduce
the CO content of the reformate to , 10 ppm. The discovery of PEMFC anodes
that can tolerate higher amounts of CO in the H2 stream (Pt-Ru instead of Pt)
may improve the situation further. Additionally, replacement of the low
temperature (100uC) polyvinyl styrene – based electrolyte membrane by the
high temperature (200uC) polybenzimidazole - based membranes in the
PEMFC (enabling the fuel cell operation at 200uC) can also lead to a greater
tolerance of CO by the fuel cell Pt anodes since the poisoning by strongly-held
CO is less at higher temperatures.
8.1. Preparation of Monolith-Coated WGS CatalystsMonolith-coated WGS catalysts are comprised of essentially three
elements: (a) the honeycomb monolith made of cordierite or a metal, (b) the
active metal (the metals Pt, Pd, Rh, Au or their mixtures), and (c) the support
metal oxide (ceria, zirconia, lanthana, titania, alumina or their mixtures)
powder (the ‘washcoat’). The high geometric surface area of honeycomb
monoliths combined with their good mechanical strength and low pressure
drop make them particularly attractive for vehicular applications. The
performance and durability of the finished catalyst depends significantly on
the quality of the washcoat. It is extremely important that the washcoating
process produces a reproducible, uniform, layer of the washcoat. Apart from
the active metals and the support metal oxide, there are several other
materials which can act as additive, binder or adhesive to the washcoat slurry
which is deposited on the monolith prior to impregnation of the noble metal.
Acetic acid is added to most washcoating slurries as a peptizing and dispersing
agent to maintain an adequately low viscosity of the washcoating solution.
Major steps in slurry preparation are particle size reduction of the support
metal oxide powder and addition of appropriate acid/sol to adjust the pH,
viscosity and homogeneity of the support metal oxide slurry in water. Size
reduction of powder particles in water down to a few microns to achieve well
dispersed, homogeneous aqueous slurry can be accomplished by ball milling.
The particle size distribution of the washcoat affects the mechanical strength
of the finished washcoat and its adhesion to the monolith, as well as the
rheological properties of the slurry during the washcoating process. In the
next step, the materials are dispersed in an acidic medium in a tank with a
high-shear mixer. The solids content in the slurry is typically 30–50%wt. After
prolonged mixing, the slurry suspension becomes a stable colloidal system.
Water Gas Shift Catalysis 385
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
The amount of washcoat that can be deposited on the monolith depends on the
properties of the monolith and the slurry. The catalyst samples are then
prepared by dipping the cordierite monolith in this slurry until the desired
loading is reached. The monolith is then air-knifed to remove excess slurry
followed by drying at 110uC and calcining at 550uC. The monolith is then
impregnated, with an aqueous solution containing the precious metal complex,
dried and calcined. The physicochemical properties of the washcoated monolith
catalyst, like chemical composition, BET surface area, pore volume, metal
dispersion and washcoat adhesion are then measured. When ceria-zirconia is
used as the support oxide for the platinum group metals, their content, in g/liter
volume of the final monolith catalyst, is between 200–500. The noble metal
content is between 2–10 g/liter of the monolith. The costs of the noble metal and
the rare earth oxides constitute a significant part of the cost of the fuel processor
in a fuel cell and efforts are in progress to reduce them further.
The ceramic honeycomb monolith, which is generally used, has some
disadvantages. It has a non-uniform flow distribution due to unidirectional
channels and a closed structure between channels and slow diffusion rate of
reactants to the catalyst surface due to low turbulence in the channels.
Further, when the catalyst particles are washcoated into the channels, the
catalyst particles are not uniformly deposited and are mainly deposited at
corners of square-shaped channels in the monolith, thus leading to lower
catalytic activity. In addition, the low thermal conductivity of the ceramic
material is also a disadvantage in dissipating away the heat generated in
exothermic reactions like the WGS reaction. Metallic monoliths have high
thermal conductivity, larger geometric surface area per unit volume, easy
fabrication and have uniformly deposited catalyst particles. The gas flux flows
in the channel direction as well as in the direction perpendicular to the
channels. Thus, a turbulent flow of the reactant gases results, leading to high
mass transfer rates. Consequently, the required reactor volumes are
decreased. The metallic monolith can be made of a refractory metal like
stainless steel or other iron-based corrosion resistant alloys (e.g., iron-
chromium alloy). They are typically fabricated from such materials by placing
a flat and corrugated metal sheet, one over the other, and rolling the stacked
sheets into a tubular configuration about an axis parallel to the configuration,
to provide a cylindrical –shaped body having a plurality of fine, parallel gas
flow passages, which can range from 200 to 1200 per square inch of face area
compared to about 400 for the cordierite monolith.
8.2. Catalytic PropertiesThe catalytic properties of the monolith catalyst are, broadly, similar to
those of the corresponding powder catalysts described in earlier sections. Only
386 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
a few examples will be given here to illustrate the advantages of the monolith
catalyst. Figure 33 compares the catalytic activity of Pt- CeO2, Pt-ZrO2, Pt-Re-
CeO2, Pt-CeO2-ZrO2 and Pt-Re-CeO2-ZrO2. The Pt-Re-CeO2-ZrO2 catalyst
reaches equilibrium conversion levels around 275uC. The high space velocity
of operation (GHSV 5 20,000 h21) should especially be noted. The advantages
of the noble metal-washcoated, monolith catalysts are apparent from these
results. A conventional Cu-ZnO-Al2O3 LTS catalyst does not give equilibrium
conversions at such high space velocities. Particulate (extrudate or sphere)
catalysts will give rise to a very high pressure drop at these space velocities.
As anticipated, the Pt-Re combination is superior to the Pt-alone composition
and ceria-zirconia is a superior support to ceria or zirconia alone. The
influence of monolith geometry and external geometric surface area on CO
conversion is shown in Fig. 34. The 600 and 400 cpsi (cells per square inch)
monoliths have the same CO conversion while the catalyst with 225 cpsi has a
lower activity suggesting that mass transfer from the bulk fluid to the catalyst
surface, and, not the surface reaction, is controlling the rate of the reaction in
monoliths with lower geometric surface areas (like the 225 cpsi monolith). As
expected, catalytic activity decreases at high space velocities (Fig. 35). It may,
however, be noted that even at such high space velocities the catalyst has,
still, a fairly good catalytic activity. The amount of ceria-zirconia washcoated
per unit volume of the monolith (keeping the Pt content constant) varies,
usually, between 200–500 g/L of monolith. Ceria-zirconia amounts beyond
500 g/liter of the monolith do not increase the catalytic activity. The superior
Figure 33: CO conversion over supported Pt, Re and PtRe catalysts.
Water Gas Shift Catalysis 387
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
Figure 35: Influence of gas velocities on CO conversion over monolith catalysts.
Figure 34: Influence of monolith geometry (CPSI) on CO conversion.
388 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
activity of TiO2- based supports, illustrated earlier (Section 6), is also
reproduced in the case of TiO2 supports washcoated on monoliths (Fig. 36).
Pt-Re-TiO2 has a higher activity than Pt-Re-ceria-zirconia. It must be
mentioned that the catalytic activities of the monoliths shown in Figs. 33–36
above are initial activities. Like their powder and particulate analogs (Fig. 37),
Figure 36: CO conversion over Pt-Re-CeO2-ZrO2 and Pt-Re-TiO2 (Initial activity).
Figure 37: Deactivation of 1% Pt/CeO2($) and 1% Pt/MgCeO2(#) at 300uC with time;(6%CO, 16%H2, 1.6%CO2, 60%H2O, and 0.4% CH4)(v/v); (168).
Water Gas Shift Catalysis 389
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
the monolith catalysts also undergo deactivation on prolonged use.
Deactivation of Pt group metal–ceria catalysts with time-on-stream in realistic
reformate gas streams have been reported (149, 150) and attributed to various
reasons, such as metal-induced over- reduction of ceria, precious metal
sintering after high temperature reaction aging of Pd and Pt- CeO2 catalysts
(86, 87) and carbonate formation (106). Deactivation due to carbonate
formation has also been reported for Pt-CeO2 (151) and Au-CeO2 (137).
Apart from their high cost, some of the technical drawbacks of noble metal
catalysts in WGS applications at low temperatures include (a) lower catalytic
activity (compared to Cu and Au) below 250uC, (b) formation of CH4 (up to 1%)
below 300uC (152), and (c) formation of strongly–held formates and carbonate
species on the surface, which, eventually cause catalyst deactivation. The
known Fischer-Tropsch activity of the noble metals in the 200 – 300uC range
(153) is, perhaps, relevant in accounting for their methanation activity under
WGS conditions. In 2007, researchers from the Honda Motor Company
reported (152) results of combinatorial catalysis for over 250,000 materials
and claimed that catalysts containing a combination of (a) one noble metal like
Pt or Rh, (b) one group 11 metal like Cu, Ag, Au, and (c) one partially reducible
oxide like ceria, zirconia, titania, lanthana,vanadia or their mixtures, form
superior WGS catalysts active and stable at low temperatures. While most of
the other elements of the above combination were well-known for their
importance in WGS catalysis, the inclusion of vanadia as a promoting support
for the WGS reaction is interesting and may open future possibilities.
Suppression of methanation activity of monolith catalysts during WGS
reactions at low temperatures by inclusion of basic oxides, like ZnO, MgO,
CaO, SrO, and BaO, in the catalyst support, has been claimed by Farrauto’s
group (154). Inclusion of any of these basic oxides in the catalyst formulation
has been claimed to reduce the methane content in the outlet of the WGS
reactor to less than 5 ppm.
A comparison of the catalytic activity, in the WGS reaction, of Pt-CeO2-
Al2O3 in the pellet form vis-a-vis the metal platelets-washcoated formulation
has been published (155). The authors prepared their Pt-CeO2-Al2O3 catalyst
by a sol-gel method, washcoated it in the micro channels of stainless steel
platelets and evaluated them for catalytic activity in the WGS reaction.
Microstructured metal platelets offer excellent temperature control due to
their good thermal conductivity and small dimensions. Moreover, the use of
thin washcoat layers of catalysts eliminates the intraparticle diffusion
limitations that occur for fast reactions. The superior catalytic activity of
the platelets compared to the pellet samples (Fig. 38) was attributed to the
diffusion limitations inside the pellet samples. The conversion over the pellet
samples above 290uC is lower than that of the platelet samples indicating that
there might be diffusion limitations inside the pellets, as the pellet size
390 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
(250 mm) is substantially larger than the equivalent particle diameter of the
coated, catalyst layer (37 mm) inside the micro channel. To verify this
hypothesis, a simulation was performed that took into account explicitly the
diffusion of matter inside the pellets. The simulation was based on a value of
the average particle diameter of 250 mm, a tortuosity factor of 5 and the mean
pore diameter and pellet porosity as calculated from the N2 adsorption data.
Figure 38 compares the experimental data and the simulation. It can be
observed that at 260uC, the powder and the platelet indeed give similar initial
rates. Above this temperature, the rates (for the pellet) are lower due to
diffusion limitations inside pores, until the thermodynamic equilibrium is
reached, at which point, the rate for the reverse water gas shift is close to the
forward WGS rate and, hence, the overall rate is lower. They concluded that,
catalysts deposited on micro structured platelets lead to a better utilization of
the Pt metal.
9. SURFACE STRUCTURES, ACTIVE SITES AND REACTIONMECHANISMS
The mechanism of the WGS reaction has been studied thoroughly for many
decades. Even though there is some consensus on the redox mechanism
prevailing over the iron-chromia catalysts at high temperatures, there is
considerable uncertainty about the operative mechanism at low temperatures,
over the Cu-ZnO and precious metal- partially reducible metal oxide catalysts.
Figure 38: Comparison of CO conversion over pellets (&) and metallic platelet (m)substrates. symbols: experimental data, solid line: model calculations; (155).
Water Gas Shift Catalysis 391
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
Newsome (4) and Kochloffl (6, 7) have reviewed the literature in this field upto
1975 and 1996, respectively. The mechanism over the Au based catalysts has
been reviewed by Burch (116). The WGS reaction involves the removal of an
oxygen atom from the H2O molecule to liberate H2 and addition of the oxygen
to CO to form CO2. The dissociation of H2O can occur on the metal, the support
or both. Similarly, the CO can react with the oxygen-containing species (H2O,
OH or O) either from the gas phase, the adsorbed state or the surface lattice.
All the mechanisms that have been proposed for the WGS reaction can be,
broadly, divided into two categories: (a) those that involve a rate –
determining step in which a molecule of H2O or CO, from the gas phase,
reacts with a surface species (oxygen vacancy or a surface oxygen atom) as
exemplified in Eqs. 21 and 22 or (b) those that involve a rate – determining
step in which the reaction takes place between two adsorbed species (the
Langmuir-Hinshelwood mechanism), illustrated, for example, in Eqs. 23–27.
The first is exemplified by the redox mechanism proposed long ago, in 1949, by
Temkin (17–19) and developed further during the past decades. The multistep
mechanism (Eqs. 23–27) proposed by Oki et al., in 1973 (20–22), as well as the
formate mechanism fall in the second category (the L-H mechanism). The
redox mechanism for the HTS reaction was supported by the results of
Boreskov et al., (156–157) who established that the rates of reduction and
oxidation of an iron oxide-based WGS catalyst were in good agreement with
reaction rates calculated from Eqs. 21 (surface oxidation of the catalyst) and
22 (surface reduction). Additional support for this mechanism was derived
from the results of Tinkle and Dumesic (158) who, from rates of adsorption/
desorption and interconversion of CO and CO2 (using isotope exchange
techniques) over iron oxide – chromium oxide catalysts, concluded that CO/
CO2 interconversion is fast compared with adsorption/desorption of CO and
CO2. Thus, Eq. 25 (the surface reaction between adsorbed CO and O moeities)
is fast and, not the rate determining step. The picture is more complex for the
mechanism of the LTS reaction. This issue is discussed below in more detail.
9.1. High Temperature Shift CatalystsIron oxide can exist in three forms: hematite (Fe2O3), magnetite (Fe3O4)
and wustite (FeO). FeO is unstable below 570uC , when it decomposes to a- Fe
and Fe3O4. Below 570uC, the reduction of Fe2O3 to Fe metal proceeds in two
steps via an Fe3O4 intermediate. The reduction of Fe2O3 to Fe3O4 is
exothermic, whereas further reduction to the metal is endothermic.
Hematite crystallizes in the Al2O3 (corundum) structure with a closely packed
oxygen lattice, with Fe3+ cations occupying octahedral sites. Its structure can
be visualized (159) as being composed of Fe-O3-Fe units (triplets) of closely
packed oxygen atoms with Fe(III) on either side. The Fe(III) atoms in each of
392 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
these Fe-O3-Fe units have opposite spins, being antiferromagnetically coupled
as a result of superexchange interaction through the triad of oxygen atoms.
Substitution of Fe cation sites by other metal (M) ion substituents in the
structure promotes formation of mixed or inverse A (12d)B d [A d[B(22 d)]O4
structures, where d is the degree of inversion. For a normal spinel, AB2O4, d 5
0, whereas for an inverse spinel, B[AB]O4, d 5 1. In magnetite- type
structures, octahedral sites are occupied by 2+ and 3+ ions, whereas
tetrahedral sites are occupied only by 3+ ions. Boreskov et al. (156–157)
had, earlier, demonstrated that the octahedral Fe2+ and Fe3+ ions located in
the magnetite-based structure function as a redox couple, and that magnetite-
based catalysts can be highly effective for the complete dissociation of H2O
into H2 and adsorbed oxygen under HTS reaction conditions. Water
dissociation causes the oxidation of Fe2+ to Fe3+ and liberates H2. The Fe3+
centers may, subsequently, be reduced to Fe2+ by CO or H2, thereby producing
CO2 (or H2O) to complete the reaction loop. Many of these substitutions
improve the thermal and textural stability of the structure while promoting
the reducibility of Fe2O3 to Fe3O4. Inverse and mixed spinel structures readily
undergo rapid electron exchange between the 2+ and 3+ states, thereby
catalyzing the WGS reaction. A detailed investigation of the structural
properties of the magnetite (Fe3O4) - based HTS catalyst system has been
published, recently, by Khan et al. (159). These authors prepared metal-doped,
iron oxide-based catalysts with nominal composition of Fe1.82 M0.18O3, (where
M5 Cr, Mn, Co, Ni, Cu, Zn and Ce) by the coprecipitation of the nitrates.
Dilute ammonia was used to precipitate the hydroxides at pH 5 8.5. The
resulting cake was dried at 80uC and further calcined at 500uC in an inert
environment. The structure of the materials was studied by various structural
techniques and evaluated for their catalytic activity in the WGS reaction at
350 – 550uC, different steam to CO ratios (1, 3.5, and 7) and a GHSV of
60,000 h 21. On activation, the hematite- like Fe1.82M0.18O3 phase transformed
into either an inverse or mixed Fe2.73M0.27O4 magnetite - like spinel phase.
The activity of Cr- and Ce- substituted Fe3O4 materials approached
equilibrium levels at high temperatures. At lower temperatures, the activity
of these magnetite- based catalysts was limited by the dissociation of steam.
Interestingly, they discovered that Ce- substituted Fe3O4 spinels are quite
promising HTS catalysts under steam-rich and high temperature conditions.
Khan et al. also carried out the temperature- programmed reduction in
hydrogen of their various promoted iron catalysts. Each promoter ion
influenced the reduction profile of iron oxide in a unique manner. In the case
of the Fe2O3-Cr2O3 catalyst, the first reduction peak at 225uC corresponded to
the reduction of Cr6+ to Cr3+. Further partial reduction of Cr3+ to Cr2+, which
would be expected at 490uC, was not observed. Reduction of Fe2O3 to Fe3O4
was observed at Tmax 350uC, whereas further reduction to FeO occurred at
Water Gas Shift Catalysis 393
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
higher temperatures. Adding chromium to Fe2O3 did not improve the
reducibility of hematite to magnetite. Based on XRD, TPR and Mossbauer
studies of the Fe2O3-Cr2O3 spinel and earlier information, Khan et al.
proposed that iron – chromia forms an inverse spinel structure and that Cr3+
replaces equal amounts of Fe2+ and Fe3+ from the octahedral sites, with the
displaced Fe2+, consequently located in tetrahedral sites. In the Fe2O3-CuO
catalyst, the reduction of Cu2+ to Cu+ occurred at 143uC. An interesting
observation was that the addition of Cu to Fe2O3 decreased the reduction
temperature of hematite to magnetite considerably to 190uC, compared to
348uC for the pristine hematite sample. The addition of Cu to iron oxide thus
improved reducibility of Fe3+ to Fe2+ species. On doping with Cu, the mobility
of lattice oxygen and hydroxyl groups increased, due to the greater
electronegativity of Cu (1.9) compared to Fe (1.8), thereby perhaps improving
catalytic activity. Promoting iron oxide with cerium causes a shift in the
reduction temperature peak maxima of both hematite - to - magnetite and
magnetite – to - wustite to lower temperatures. In the Fe2O3-Cr2O3, the ceria
surface shell reduction occurs at 380uC, instead of 485uC as in pure ceria. The
temperature of bulk reduction of ceria, however, was not affected by the
presence of iron. These results are significant in the development of Cr-free
iron oxide-based, HTS catalysts.
The propensity of metallic iron and the lower oxides of iron to be oxidized
by steam and evolve hydrogen at high temperatures are well known.
Additionally, the adsorption and surface concentrations of CO at high
temperatures will also be low. The redox mechanism will thus be favored at
high temperatures and over those catalysts which can dissociate H2O into H2
and O2, a crucial requirement of the redox mechanism. Reviewing this area in
1996, Kochloffl (6, 7) concluded that the WGS reaction at high temperatures
over Fe2O3-Cr2O3 catalysts proceeds, most probably, by the redox mechanism.
9.2. Low Temperature Shift CatalystsFrom a mechanistic viewpoint, the LTS catalysts that have been studied
extensively may be divided into three groups; (a) Cu-ZnO-Al2O3, that is
currently used as the standard, industrial catalyst, (b) the Pt group metals
supported on partially reducible oxides, like ceria, titania, zirconia or their
mixtures, and (c) Au supported on the same, above-mentioned oxides. The
reaction mechanism on oxide-supported Au catalysts may be different from
that on supported platinum metal catalysts, because of (a) the lower
adsorption energy of CO on the Au nanoparticles; and (b) the inactivity of
Au (unlike the Pt- group of metals) for H2O dissociation. Some of the features
of the LTS reaction over Cu-ZnO catalysts that distinguish them from the HTS
reaction over iron oxide – chromia are: (1) the dissociation of H2O to H2 and O
394 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
over copper metal or ZnO, at these low temperatures, is less documented
compared to that on the iron oxide, Fe3O4, at the higher temperatures (120);
(2) the amount of CO adsorbed on metallic copper, even at the low
temperatures, is less than that on the platinum group of metals at similar
temperatures and (3) the WGS rate is proportional to the CO partial pressure
to the first order over the Cu-ZnO compared to the zero order observed over
the Pt-based catalysts. The last feature implies that, reducing the CO
concentration from 1.0% (a typical value in the LTS reactor) to 0.5% requires
twice as much catalyst as reducing it from 2% to 1% thus leading to large
second stage LTS reactors. Hence, major efforts have been made in the last
two decades, in the field of fuel processors, to develop cost-effective catalysts
that are tolerant to oxygen exposure, have robust, high volumetric activities at
low temperatures and whose CO conversion rate is independent of the CO
concentration (zero order) in the range 3–0.3%. As mentioned earlier (Section
6), noble metal- based catalysts, like Pt- ceria, Pt-ceria-zirconia, Pt- titania
and their modified versions, like Pt-Na-ceria, Pt-Na-titania, Pt-Na-ceria-
titania appear promising. Though these catalysts have high initial activities,
they still undergo deactivation at temperatures below 250uC on prolonged use.
Currently, efforts are in progress to find the root cause of deactivation of these
catalysts so that they can be used successfully in fuel cells. It is in this context
that a better understanding of the basic mechanism of the LTS reaction over
these noble metal- based catalysts is of importance. A redox mechanism,
involving the reduction of the catalyst by CO and reoxidation (and, thereby the
regeneration of the catalyst by H2O), similar to the one described above for the
HTS reaction over the iron oxide- chromium oxide catalyst, has also been
proposed for the LTS reaction over the Pt-ceria catalysts. In this mechanism,
the CO abstracts an oxygen (forming CO2) from the ceria lattice at the Pt-ceria
interface. Two Ce4+ ions are reduced to the Ce3+ state in this process. The
resulting reduced ceria lattice is then reoxidised through the dissociation of
the incoming H2O. The O vacancy is, thereby, refilled, formally oxidizing two
Ce3+ to Ce4+ and releasing molecular H2 in the process. If the sites for the
adsorption of CO and H2O are different, there will be no competition between
CO and H2O for adsorption and the zero order rate dependence for CO may be
observed. If the sites for the adsorption of CO or H2O is occupied by the other
reactant (H2O and CO, respectively), or, by strongly adsorbed species (like the
formates, carbonates or carboxylates), the reaction order, may be different.
However, it should be noted that at the lower temperatures characteristic of
the LTS reactions over catalysts like Cu-ZnO, and Pt-ceria (190–250uC), the
ability of steam to reoxidise the partially reducible oxide supports (with or
without the presence of the noble metals) has not been demonstrated so far; In
addition, in the case of fuel cell conditions (for Pt-Ceria), this reoxidation of
Ce3+ to Ce4+ has to occur in the presence of a considerable amount of hydrogen.
Water Gas Shift Catalysis 395
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
Techniques, like XANES, can, perhaps, be used to resolve this issue. In
addition to the redox mechanism, an associative mechanism involving a
surface formate intermediate has also been proposed.
9.2.1. The ‘Formate Surface Intermediate’ Hypothesis
From studies of the WGS reaction over copper chromite catalysts,
Armstrong and Hilditch, had, already in 1920, proposed (160) that the
reaction involved the adsorption of CO and H2O on the catalyst surface to form
a surface intermediate that subsequently decomposed to CO2 and H2. To
isolate and identify the ‘‘surface intermediate’’, they reacted it with NH3 and
obtained ammonium formate. The formation and decomposition of a formate-
or formic acid – type surface species was speculated to lead to the formation of
the products, H2 and CO2, over copper – based catalysts, like Cu-Cr2O3.
Similarly, using dimethyl sulphate as a methylating agent to trap the surface
formate, Deluzarche et al. (161) obtained dimethyl formate further supporting
the presence of formates on the surface during the WGS reaction. During the
last few decades, extensive research using a variety of techniques has,
conclusively, proven the existence of a formate species at low temperatures
over Cu-ZnO as well as precious metal- reducible metal oxide catalysts (162).
Boreskov and Davydov (163–164) had earlier carried out pioneering IR
spectroscopic studies supporting the associative mechanism over a wide
number of copper- based catalysts including the industrially important Cu-
ZnO. Additional early work supporting the formate associative mechanism for
Cu-ZnO include those of Herwijnen et al.(165–166) who observed the nearly
identical conversion rates for the water gas shift reaction and formate
decomposition. Rhodes et al.(167) have, however, raised some doubts as to
whether these surface formates constitute the only, or, even the main
intermediates in the reaction path of CO and H2O to H2 and CO2 or are merely
spectators formed by a parallel route from the reactants and/ or products
(167). Shido and Iwasawa (107, 108) investigated the WGS reaction over ZnO,
CeO2 and MgO using in-situ FTIR spectroscopy of the surface species. Their
results indicated that surface OH groups, formed by reaction of H2O with
oxygen vacancies on partially reduced CeO2, reacted with CO to form bridged
formates. The bridged formates were converted to bidentate formates above
170uC. This transformation occurred at room temperatures in the presence of
water. About 30% of these adsorbed bidentate formates were, in turn,
decomposed to the final products (H2 and CO2) and adsorbed, unidentate
carbonates. The rest decomposed back to the reactants, CO and H2O. These
transformations of the bidentate formates were also influenced by the
presence of H2O. Coadsorbed H2O also promoted the decomposition of
the unidentate carbonates to CO2. In addition to the unidentate carbonates,
396 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
the presence of surface carboxylates and bidentate carbonates were also
identified by Shido and Iwasawa (107, 108). Binet et al. (168) and Fallah et al.
(169) observed IR bands at , 3650 cm21 on partially reduced (with H2) CeO2
samples and assigned it to Type II bridging OH groups. Jacobs et al. have,
indeed confirmed, using in-situ IR spectroscopy, (170–172) that the formates,
formed on Pt (1%) on ceria, decomposed to CO and OH in the absence of steam
in about 6 min at 300uC , while in the presence of steam, they decomposed
completely and much more rapidly in 8 min even at 140uC to produce H2 and
unidentate carbonates. Based on mechanistic studies, including kinetic
isotope effect (170) and isotope tracer studies (172), they, further, suggested
that the rate-limiting step in the LTS reaction is the cleavage of the C-H bond
of the surface formate. Bridging OH groups and surface formates have also
been identified, earlier, by IR spectroscopy over thoria and zirconia-based
catalysts (173–175). Jacobs et al. have also observed that Pt-thoria (176) and
Pt-zirconia (177) possess much higher WGS activity than the corresponding
oxides without the precious metals and attributed it to the more facile
formation of the Type II bridged OH groups, and the surface formates derived
from them, over the Pt-loaded catalysts. In addition, kinetic isotope effects
similar to those observed in the case of Pt-ceria were also observed for Pt
supported on thoria and zirconia suggesting that the rate-limiting step was
likely to be the C-H bond scission of the formate intermediate in these cases
also. To summarize, in the formate mechanism, a bidentate formate produced
from CO and surface OH groups acts as an intermediate. The bidentate
formate, then, decomposes to gaseous hydrogen and a surface unidentate
carbonate, which further decomposes to gaseous CO2. One of the major
contributions of Jacob et.al., is the discovery that co-adsorbed water plays a
crucial role in the selective decomposition of the formate intermediate to CO2
and H2. The main roles of Pt are (1) to catalyze the reduction of ceria, leading
to the formation of surface, terminal OH groups on ceria and (2) to catalyse the
decomposition of the formate to H2 and CO2. The rate determining step is
likely to be the decomposition of the unidentate formate to CO2 and H2.
If the decomposition of the formates to H2 and CO2 is the rate-limiting
step, then, factors that facilitate the cleavage of the formate C-H bond should
enhance the reaction rate. More specifically, addition of bases, like the alkali
ions, which are known to accelerate the decomposition of formates, should
enhance the WGS rates. Pigos et al. (178) have recently observed that the
incorporation of Na in Pt- zirconia catalysts does, indeed, enhance the WGS
reaction rates. Interestingly, their DRIFTS investigations suggest that
incorporation of Na in Pt- zirconia modifies the electronic properties of the
surface formate and weakens its C-H bond. Two significant features of their
DRIFTS results (178) are (a) The C-H band of the formate species was
shifted to lower wavenumbers from 2880 cm21 (Pt-ZrO2) to 2842 and
Water Gas Shift Catalysis 397
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
2804 cm2 1 (Pt-Na-ZrO2), respectively, indicating a weakening of the C-H bond
on Na incorporation; and (b) The ratio of the intensity of the bridged to linear
Pt carbonyls increased from 1:5 to 4:5, thus, favoring bridge-bonded CO in the
Pt-Na-ZrO2 catalyst. The influence of catalyst basicity in increasing the
concentration of the bridged carbonyls had, already, been reported by Mojet et
al., who found (179) that, in Pt-SiO2 and Pt-K-L (LTL) zeolites, increasing the
K+ ion content also increased the concentration of the bridged carbonyls. The
faster decomposition of the surface formate over Pt-Na-ZrO2 is illustrated in
Fig. 39. 20% of the initial intensity of the IR band of formate is decreased in
5.2 min for the Pt-ZrO2 and 2.4 min over Pt-Na-ZrO2. CO2 was also seen in the
presence of steam. In addition to the formate, carbonates and carboxylates
were also seen after steaming. Pigos et al. (178) also compared the stability of
the formate species, on Pt-ZrO2 and PtNa-ZrO2, in the absence of steam, by
monitoring their thermal decomposition. It may be recalled that Shido and
Iwasawa (107, 108) had, earlier, observed that in the absence of steam,
formates on metal-ceria catalysts decompose, primarily, in the reverse
direction, back to CO and OH. To follow the thermal decomposition of the
formate in the absence of steam, C-H bond breaking was probed by Pigos et al.
(178) by flowing D2 and monitoring the exchange of the C-H and C-D formate
bands. The areas of the formate C-H bands (at 2880 and 2966 cm21) and their
corresponding C-D bands were quantified and plotted as a function of time
(Fig. 40). The formate H-D exchange rates were very close to the overall
formate decay rate. Moreover, the half-life for H-D exchange for Pt-Na-ZrO2
was approximately half that of the Pt-ZrO2 (Fig. 40) indicating that C-H bond
breaking in the formate is more facile for the Na-promoted catalyst. It may be
observed that in none of the aforementioned studies were the formate
Figure 39: Formate area response to steaming at 130uC for Pt/ZrO2 and PtNa/ZrO2 (178).
398 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
decomposition rate constant and surface coverage determined simultaneously
at the steady state under WGS reaction conditions. Based on DRIFTS analyses
combined with the utilization of isotopic tracers, it had been shown (180) that
formates were less reactive than carbonyl and carbonate species under steady
state conditions whereas the reverse trend was observed during the non-
steady state, desorption experiments.
The reactivity of the species formed at the surface of a Au-Ce(La)O2
catalyst during the WGS reaction, in the steady state, was investigated by
Meunier et al.(60–62) using simultaneous DRIFTS and kinetic analysis. The
exchange of the product CO2 and formate and carbonate surface species were
followed during an isotope exchange of the reactant, CO, using a DRIFTS cell
as the reactor. In independent experiments, the DRIFTS cell yielded identical
reaction rates to that measured in a quartz, plug-flow reactor. The DRIFTS
signal was used to quantify the relative concentrations of the surface species
as well as that of CO2. The analysis of the formate exchange curves between
Figure 40: Formate area to D2:N2 at 225uC for (top) Pt/ZrO2 and (bottom) PtNa/ZrO2. The K-life of formate is indicated for formate decay and formate exchange from H to D. Fasterdecay and exchange rates are observed for PtNa/ZrO2, indicating a higher reactivity forformate C-H bond breaking. Reverse decomposition (178).
Water Gas Shift Catalysis 399
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
155–220uC suggested the presence of two types of surface formates: ‘‘slow
formates’’ that display an exchange rate constant 10- to 20 fold slower than
that of CO2 and ‘‘fast formates’’ that exchanged on a time scale similar to that
of CO2. Figure 41 compares the molar rate of formate decomposition to that of
CO2 formation per unit mass of catalyst. The specific rate of CO2 formation
was determined from the CO2 concentration in the DRIFTS cell exhaust gas
(by gas chromatography), the sample weight in the DRIFTS crucible and the
flow rate of the reactants. The rate of formate decomposition was calculated
from the DRIFTS data as the sum of the decomposition of the ‘‘fast’’ and ‘‘slow’’
surface formates. The semilog plot shows that the rate of CO2 formation was
more than an order of magnitude (about 60-fold) higher than the rate of
decomposition of the (slow + fast) formates, indicating that the formates,
detected by DRIFTS, cannot be the only reaction intermediates in the
production of CO2.
Thus, while there is sufficient experimental evidence to conclude that (a)
formate-like species are present, under WGS conditions, on the surface of Cu-
ZnO and precious metal-partially reducible, metal oxides; and (b) the
decomposition of these formate species under WGS conditions leads to the
products, CO2 and H2, it is not established that CO2 and H2 are derived only
from the surface formates and not, also, additionally, from other intermedi-
ates, like the carbonates/caboxylates or by a completely different mechanism,
like the redox mechanism, which does not involve any long-lived and
experimentally observable, surface intermediate. This viewpoint is further
supported by the investigations of Gokhale et al. (196–199) and Mhadeshwar
and Vlachos (209–210) (described later, Section 9.2.4).
Figure 41: Rate of CO2 production and rate of formate decomposition over the 0.6 AuCl atthree different temperatures under 2% 12CO + 7%H2O (60–62).
400 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
9.2.2. The Redox Mechanism on Pt-based Catalysts
A redox mechanism is also being advocated (106, 148) for the LTS reaction
on Pt-ceria and other platinum group metals on ceria. According to this
picture, CO adsorbs on transition metal sites and reacts with oxygen from the
ceria, which, in turn, is reoxidised by H2O. In other words, it involves the
reaction of the reactants, CO and H2O, with the surface: CO with the oxide ion
of the ceria (to yield CO2) and H2O with the anion vacancies on ceria
(generating OH groups and, eventually, H). An important role of the metal is
to adsorb/activate CO and create of oxygen vacancies at the metal ceria
interface. In contrast to the formate theory, there is no postulate of a stable,
experimentally observable and kinetically relevant, surface intermediate.
Evidence for this mechanism came, initially, from kinetic studies (181). TPD
studies (182) have demonstrated that oxygen from ceria can react with CO
adsorbed on metals. It has also been established (183) that reduced ceria can
be oxidized by CO2. While the redox mechanism is well-established at high
temperatures in the case of the iron oxide-chromium oxide catalysts, its
applicability to LTS over Cu-ZnO and Pt-ceria catalysts is uncertain and
depends on confirmation of the ability of H2O to reoxidize the partially
reduced support oxide at temperatures below 250uC, especially in the presence
of significant amounts of hydrogen, as is the case for fuel cell applications (see
also sections 9.2 and 10). Such an unambiguous experimental confirmation is
desirable. Another feature of the ceria- based catalysts, namely, that high
temperature calcination lowers, not only the concentration of oxygen
vacancies and loosely – bound surface oxygen atoms, but also their WGS
activity lends additional support to the redox mechanism. Reaction orders on
Pd-ceria were, approximately, zeroth-order in CO, half-order in H2O, inverse-
half order in CO2 and inverse first order in H2 (106). The rate limiting step
was believed to be the dissociation of H2O on the ceria support. Diffuse
reflectance and FTIR spectroscopic measurements on Pd-ceria indicated that
the ceria existed (not surprisingly) in a reduced state under WGS conditions
and is covered by carbonate species that are removed only by reoxidation of
ceria (106). Such surface carbonates were also a cause of catalyst deactivation.
It may be noted, however, that, under their WGS reaction conditions, the Cu-
ZnO catalyst was much more active than all their metal-supported ceria
catalysts.
Azzam et al. (184–185) studied the WGS reaction on catalysts based on
ReO2-TiO2. Results pointed to contributions of an associative formate route
with redox regeneration and two classical redox routes involving TiO2 and
ReOx, respectively. Under their WGS reaction condition, rhenium was
present, at least partly, as ReOx providing an additional redox route for
WGS reaction in which ReOx is reduced by CO generating CO2 and re-oxidized
by H2O forming H2. The reaction between CO adsorbed on Pt and OH groups
Water Gas Shift Catalysis 401
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
on titania was the rate-determining step. Gold nanoparticles supported on
reducible and non-reducible oxides with comparable gold particle size were
studied in the WGS reaction by Sandoval et al. (186). Not surprisingly, the
activity of Au on reducible oxides was much higher than the one observed on
non-reducible oxides. The optimum calcination temperature was 300uC. For
samples calcined at 300uC and reaction temperatures below 225uC the activity
varied (Fig. 42) as follows: TiO2 . CeO2 . Al2O3 . SiO2. A novel catalyst
consisting of platinum deposited over a cerium-modified titania substrate has
been, recently, reported by Gonzalez et al. (187). They showed better thermal
stability with respect to the bare TiO2 support and higher WGS activity than
those corresponding to individual titania or ceria supports. XPS and TPR
results revealed the intimate contact between Pt and cerium entities in the Pt/
CeO2–TiO2 catalyst that facilitates the reducibility of the support at lower
temperatures. The importance of CO adsorption on ceria in influencing the
rates of the WGS reaction was investigated by Li et al. (188). Au nanoparticles
on monoclinic ZrO2 showed much higher catalytic activity for the low-
temperature water gas shift reaction than those on tetragonal zirconia, mainly
due to the high CO adsorption capacity of monoclinic ZrO2. Formate species
formed by the reaction of adsorbed CO on gold nanoparticle with hydroxyl
groups on ZrO2 were postulated to be the reaction intermediates.
Figure 42: CO conversion over Au nanoparticles supported on TiO2, CeO2, Al2O3, and SiO2
(186).
402 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
9.2.3. Mechanism over Cu- and Au- based Catalysts
One of the important issues in the mechanism of the LTS reaction is the
role of the copper and gold metal nanoparticles supported on ZnO or CeO2. Cu-
CeO2 and Au- CeO2 are two of the promising LTS catalysts. Cu and Au are
present, mostly as nanosized metallic particles, during the WGS reaction.
What is the intrinsic reactivity of these nanoparticles? Can nanoparticles of
Cu and Au catalyze the WGS reaction on their own without the aid of an oxide
support (such as ceria or ZnO)? Results from catalytic studies over bulk Cu
and Au may not be directly applicable to the nanoparticles. On the pure, bulk
metals, the WGS reaction, probably, proceeds by a redox mechanism (63, 173).
The mechanism may, however, be modified by the presence of the support
oxide (especially by a partially reducible one, like ceria) in intimate contact
with the metal nanoparticles and wherein metal-support interactions will be
more important. Rodriguez et al. (189–195) have addressed this issue. They
investigated the WGS reaction on Cu and Au nanoparticles supported on CeO2
(111) and ZnO (0001) surfaces (189–190). Pristine CeO2 (111) and ZnO (0001)
surfaces did not display any catalytic activity under their reaction conditions
(300–375uC, PCO 5 20 Torr, PH2O 5 10 Torr). Significant catalytic activity was
measured when Au or Cu particles (2–4 nm) were deposited (Fig. 43). The
deposition of Cu nanoparticles on ZnO (0001) produced a catalyst that was
clearly more active than the pure extended Cu surfaces. An even better
catalyst was obtained when the nanoparticles were supported on CeO2(111).
They found negligible WGS activity on Au (111) (Fig. 43) or polycrystalline Au.
Figure 43: Amounts of H2 produced during the WGS reaction on 0.5ML of gold or copperdeposited on CeO2 (111) and ZnO (0001). For comparison, the activities of Au (111) and Cu(100) are also included. The catalysts were exposed to a mixture of 20 Torr CO and 10 Torr H2Oat 625 K for 5 minutes in a batch reactor. A reaction time of 2–3 minutes was enough to reacha steady - state regime in the reactor (189–190).
Water Gas Shift Catalysis 403
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
The Au-ZnO(0001) system displayed catalytic activity that was worse than
that of Cu-ZnO(0001). In contrast, Au-CeO2 had an activity similar to that of
Cu- CeO2 (111). Their XPS data pointed to a lack of oxidation of the metals,
and, a reduction of the ceria support. They also evaluated the importance of
surface intermediates for the pure metal surface Cu (110), and the metal- ceria
catalyst. Importantly, in the case of pure, metallic Cu (110), analysis of the
surface after the WGS reaction showed it to be essentially free of formate and
carbonate species, suggesting that, on the pure metal surfaces, the WGS
reaction proceeds by the redox mechanism. In agreement with others, they
also identified adsorbed formate and carbonate-like species on the metal-ceria
surfaces after the WGS reaction. Using density functional calculations, they
had also investigated (193, 194) the WGS reaction on Cu29 and Au29 clusters
(representative of the metal nanoparticles formed on deposition on ceria or
ZnO supports) and on Cu (100) and Au (100) surfaces (representative of the
surface of the bulk metals). Figure 44 shows the calculated energy changes for
the WGS reaction on a Cu29 cluster. The reaction pathway with the minimum
energy barriers involves the following steps (Eqs. 36–41):
CO gð Þ<CO adsð Þ, ð36Þ
H2O gð Þ<H2O adsð Þ, ð37Þ
H2O adsð Þ?OH adsð ÞzH adsð Þ, ð38Þ
Figure 44: Reaction profile and structure for the WGS reaction on a Cu29 nanoparticle. Thezero energy is taken as the sum of the energies of the bare nanoparticles, gas-phase water,and carbon monoxide. The red bars represent the transition states, and the black barsrepresent reactants, intermediates, or products. Cluster side view: yellow - Cu, red -O, gray-C, white- H (189–190).
404 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
CO adsð ÞzOH adsð Þ?OCOH adsð Þ, ð39Þ
OCOH adsð Þ?CO2 gð ÞzH adsð Þ, and ð40Þ
2H adsð Þ?H2 gð Þ: ð41Þ
The adsorption of CO or H2O on the Cu particles is exothermic. The first and
the most important energy barrier is for the dissociation of water into
adsorbed OH and H (reaction 38). Then, the reaction of OH and CO produces
an OCOH, carboxyl species. The final important energy barrier is for the
decomposition of this OCOH carboxyl intermediate into CO2 gas and adsorbed
H, which eventually yields the H2 gas. The DFT results indicated that a free,
metallic, nanoparticle of copper can catalyze the WGS reaction easily. A
comparison with the corresponding results on the Cu(100) surface of bulk
copper shows that the dissociation of H2O on the surface of bulk copper has a
larger activation energy barrier (1.13 ev vs. 0.94 ev on the nanoparticle) and
that no stable OCOH, carboxyl intermediate, is formed, as a redox mechanism
operates. The presence of corner or edge atoms in Cu29 favors the dissociation
of H2O. The Au29 nanoparticles and the bulk Au (110) surface could not
catalyze (169) the WGS action. Neither surface was able to adsorb and
dissociate water molecules. Figure 45 (189, 190) shows a correlation between
the calculated barrier (y axis) and the calculated energy (x axis) for water
dissociation on Au(100), Cu(100), as well as the ionic and neutral Au29 and
Cu29 particles. All the gold systems are characterized by a large activation
Figure 45: Correlation between the calculated barrier (DEa) and the calculated reactionenergy (DE) for water dissolution on several copper and gold systems (189–190).
Water Gas Shift Catalysis 405
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
barrier and an endothermic DE value. There is significant improvement in
chemical reactivity in going from Au(100) to Au29, but not enough for
dissociation of the water molecule. These results, of course, cannot explain the
large catalytic activity of Au-ceria (Fig. 43) and highlight the important role of
ceria in the activation of the system. A perfect CeO2 (111) surface does not
dissociate water at low or, even, high temperatures. When O vacancies are
present, however, the H2O molecules dissociate on the partially reduced ceria
surface. Au and Cu particles facilitate the reduction of the ceria surface by the
CO/H2O mixture and, thereby, facilitate the most difficult step in the WGS
reaction, namely, the dissociation of H2O (189, 190).
9.2.4. The Carboxylate Mechanism
Surface carboxylic species had been observed earlier by spectroscopic
techniques on LTS catalysts (107, 108). Mhadeshwar and Vlachos (210) and
Gokhale et al. (196–199) proposed from theoretical calculations, that they are
important, reactive intermediates, which play a central role in the WGS
reaction. Gokhale et al. used periodic, self-consistent, density functional
theory (DFT-GGA) to investigate the WGS reaction mechanism on Cu(111),
the dominant facet of copper crystallites in the Cu-ZnO industrial WGS
catalysts. Their proposal for an alternate WGS reaction mechanism, involving
the oxidation of adsorbed CO by adsorbed OH, to form carboxyl (COOH)
species is compared with the conventional redox mechanism in Table 9. The
crucial difference is that, while in the conventional redox mechanism the
adsorbed CO is oxidized to CO2 by adsorbed O atoms, CO2 is formed by the
decomposition of an adsorbed carboxyl group (formed by the reaction of an
adsorbed CO with an adsorbed OH group) in the new proposal. CO2 may also
be generated by the reaction of the carboxyl with a second adsorbed OH group
(Table 9). They also suggested that although it is possible to form the carboxyl
group, COOH, in a single, elementary reaction step (reaction of CO with OH
Table 9: Redox and carboxyl mechanisms on Cu (111)a (175).
Redox mechanism Carboxyl mechanism
CO + * ) CO* CO + * ) CO*H2O+ * ) H2O* H2O+ * ) H2O*H2O + * ) H* + OH* H2O + * ) H* + OH*OH* + * ) O* + H* CO* + OH* ) COOH* + *OH* + OH* ) H2O* + O* COOH* + * ) CO2* + H*CO* + O* )CO2* + * COOH* + OH* ) CO2* + H2O*CO2* ) CO2 + * CO2* ) CO2 + *H* + H* ) H2 + 2* H* + H* ) H2 + 2*
aSteps in italics highlight differences between the two mechanisms.
406 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
as postulated in the formate mechanism), that is less likely. That is because
OH binds to the surface through its O atom, and CO through its C atom,
whereas the formate binds through its two O atoms, not its C atom. Hence,
two O atoms of the formate will have to bind to the surface forming a bidentate
species, either sequentially (via unidentate formates) or, less likely,
simultaneously. Their calculations suggested that the easiest way to form
the formate, HCOO, is by reacting CO2 with atomic H. The 16 elementary
steps involved in their mechanistic model of the WGS reaction on Cu (111) are
shown in Table 10 and the corresponding reaction network in Figure 46. Using
the DFT-derived parameters as initial estimates for the microkinetic model
parameters, they fitted the 16- step model to the experimental WGS reaction
rate data published, earlier, by Koryabkin et al. (200). As may be seen from
Fig. 47, the agreement is satisfactory. Their model also tested well against the
kinetic data of Herwijnen and Jong on a Cu-ZnO-Al2O3 catalyst (52). Based on
the good ‘‘fit’’ between the calculated and observed data, they suggested that
Cu(111) may be a dominant active site for the WGS reaction on realistic
industrial catalysts. An alternate explanation may be that the WGS reaction
on these catalysts is not structure sensitive, and therefore, the reaction rate is
comparable on different Cu facets (196). From their model calculations they
also predicted that steps 5 (H2O* + * u H* + OH*) and 9 ( CO* + OH* u cis-
COOH * + *), in Table 10, are rate controlling under industrial conditions. In
the absence of CO2 and H2 co-feed, step 5 has a considerably stronger
influence on the overall reaction. To summarize their results on Cu(111): (a) H
abstraction from H2O appears to be the rate-controlling step for the entire
Table 10: Elementary Steps involved n the water gas shift reaction on Cu (111)(175).
Step No Elementary Step
1 CO + * R CO*2 H2 + 2* R 2H*3 H2O + * R H2O*4. CO2 + * R CO2*5 H2O* + * ROH* + H*6 OH* + * R O* + H*7 2OH* R O* + H2O*8 CO* + * R CO2* +*9 CO* + OH * R cis-COOH* + *10 cis-COOH* R COOH*11 COOH* +* R CO2* + H*12 COOH* + OH* R CO2* +H2O*13 CO2* + H* R HCOO* + *14 HCOO* + * R HCOO**15 CO2* + H2O*+ *R HCOO** + OH*16 CO2* + OH* + * R HCOO** + O*
Water Gas Shift Catalysis 407
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
WGS reaction network, (b) carboxyl(COOH) is a very reactive, but short-lived ,
intermediate, and (c) formate (HCOO) formed, probably, from CO2 and H, is a
spectator species which tends to block active sites, and can reach substantial
surface coverages, particularly at high pressures. This site- blocking by
formate can, also, explain the observed negative WGS reaction order with
respect to CO2 (196–199).
This approach has, recently, been extended by the same group (197) to the
water gas reaction on Pt(111) surface of bulk Pt metal. It should be borne in
Figure 47: Experimental WGSR rates versus rates predicted by the microkinetic model(196–199).
Figure 46: Reaction rate for the water gas reaction. A reaction scheme including both thesurface redox mechanism and the carboxyl mechanism is outlined. The thermochemistry andthe kinetic barriers for all the elementary steps are given in electron volts. For reactionsinvolving bond making, the activation barriers are reported with respect to the adsorbedreactants at infinite separation from each other. The minimum energy pathway for the WGSRis highlighted with green (196–199).
408 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
mind that the influence of the support on the electronic and textural
characteristics of the metal surface OH groups or reaction intermediates
was not explicitly taken into account in these calculations. The results are,
broadly, similar to those described above for copper (196). The contribution of
the surface redox mechanism to the water gas reaction on Pt, involving CO
oxidation by atomic O, is negligible under the range of conditions studied
(250–300uC, 1 bar, feed 5 (CO+H2O+CO2+ H2)). The lowest energy path
involves the formation of the carboxyl (COOH) intermediate, which is
subsequently, decomposed by reaction with OH (COOH + OH R CO2 +H2O). The OH species is, then, regenerated by dissociation of the formed H2O.
When the concentration of the OH groups is limited, the direct decomposition
route (COOH R CO2 + H) dominates. Additional H2O in the feed increases the
OH coverage and makes the OH 2 mediated COOH + OH, low energy
decomposition path more kinetically accessible, thus, enhancing the water gas
reaction rates.
9.2.5. Alkali-doped, Pt-based, LTS Catalysts
One of the key, rate- determining, steps in the formate mechanism is the
C-H bond scission in the surface formate intermediate. Evin et al., have
recently found in accord with the results of Pigos et al. (178), that alkali
doping weakens the C-H bond, as demonstrated by a shift to lower
wavenumbers of the n(CH) vibrational mode, and enhances the LTS reaction
over Pt-Ceria catalysts significantly (201). However, with high alkalinity (,2.5% Na or equimolar amounts of K, Rb, or Cs), a trade- off was observed such
that while the formate became more reactive, the stability of the adsorbed
carbonate species, which arises from the decomposition of the initially- formed
formate intermediate, was found to increase. This was observed by TPD-MS
measurements of the adsorbed CO2 probe molecule. Increasing the amount of
alkali to levels that were too high also led to (a) lower catalyst BET surface
area, (b) the blocking of the Pt surface sites as observed in infrared
measurements, and (c) a shift to higher temperature of the surface shell
reduction step of ceria during TPR. When the alkalinity was too high, the CO
conversion rate during the water-gas shift reaction also decreased in
comparison with the undoped Pt-ceria catalyst. However, at lower levels of
the alkali, the above-mentioned inhibiting factors on the water-gas shift rate
were suppressed such that the weakening of the formate C-H bond could be
utilized to improve the overall turnover efficiency during the water-gas shift
cycle. This was demonstrated at 0.5% Na (or equimolar levels of K) doping
levels. Not only was the formate turnover rate found to increase significantly
during both transient and steady state DRIFT tests, but this effect was
accompanied by a notable increase in the CO conversion rate during low
Water Gas Shift Catalysis 409
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
temperature water-gas shift. Evin et al., (202) had also observed, using in-situ
DRIFTS spectroscopy, that, on steaming pre-adsorbed formate species, these
species were more reactive in the decomposition step in the catalytic cycle (i.e.,
decomposition of the formate to H2 + CO2) for the Li- and Na- doped catalysts
relative to undoped Pt- ceria (202). For example, the relative formate
decomposition rates were 1.0, 1.2, and 1.41 for the undoped, 0.15% Li- doped
and 0.5% Na- doped Pt-Ceria, catalysts respectively. The CO conversion at
225uC, correspondingly, increased from 12% for the undoped 2% Pt- ceria to
14% for 2%Pt- 0.15%Li-ceria and 24.3% for the 2% Pt-ceria sample doped with
0.5% Na. However, with increasing atomic number over the series of alkali –
doped catalysts, the stability of the carbonate species (another surface species
formed during the WGS reaction) was also found to increase. This was
observed during TPD-MS measurements of the adsorbed CO2 probe molecule
by a systematic increase of a high temperature peak for a fraction of the CO2
desorbed. This result indicates that alkali-doping is an optimization problem-
that is, while improving the decomposition rates of formate species, the
carbonate intermediate stability also increases, making it difficult to liberate
the CO2. An optimal amount of basicity, sufficient to decompose the formate,
but, not enough to stabilize too much the carbonate, is needed. Infrared
spectroscopy results of CO adsorbed on Pt and ceria suggested that the alkali
dopant is located on, and electronically modifies, both the Pt and ceria
components. Alkali doping may, thus, provide a path forward for improving
the WGS rate by means other than resorting to higher noble metal loadings.
It has, of course, been known for a long time that alkali metals promote
the WGS reaction rate. In 1981, Sato and White (203) doped Pt- TiO2 with
NaOH and found an improvement in the photocatalyzed water gas shift rate.
Klier (204) also highlighted the promoting influence of alkali dopants, their
relative efficiency being, Cs. Rb. K.Na, Li. Klier also suggested that the
alkali should be present at concentrations less than a monolayer. Campbell et
al., (205), observed a promotion of the WGS activity of Cu (110) by Cs ions. Cs
1.5–2.0 CO3 was found after the reaction by surface analysis (TDS, XPS, AES) of
the catalyst. In kinetic studies, using a low H2O/CO ratio, they found that on
the optimally Cs- promoted surface, the reactant orders were zero order for
H2O and 0.5 order for CO, suggesting that H2O dissociation was not rate
controlling on the alkali-promoted catalyst. They proposed a redox mechanism
to describe the catalysis of both the clean and Cs- doped surfaces with Cs
playing the role of O mediator among CO2, H2O, and CO, where Cs is,
primarily, in the form of a carbonate. Honda Research Inc., has also claimed
(152) a remarkable improvement in the WGS activity of Pt- ZrO2 catalysts for
fuel processors for use in fuel cell applications by doping the catalysts with
alkali. Among the promising compositions discovered was an important
improvement when Pt- ZrO2 was doped with Na alone or in combination with
410 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
vanadium. In their DRIFTS spectroscopic study of metal- ceria catalysts,
Pigos et al. (178) found that formate C-H stretching bands were strongly
shifted to lower wavenumbers upon CO adsorption indicative of a weakening
of the C-H bond. Further, in transient formate decomposition experiments,
both in the presence and absence of steam, they reported that formates, over
Pt-Na- ZrO2, decomposed at twice the rate of those observed on Pt-ZrO2
without Na. Among the alkali dopants, Na was found to provide the most
benefit.
9.2.6. LTS over Pt Supported on Non-reducible Oxides
What is the mechanism prevailing over catalysts comprising of noble
metals supported on non-reducible oxide supports, like alumina or silica? The
fact that gamma alumina is an ‘‘irreducible’’ oxide at the WGS reaction
conditions will seem to exclude the redox mechanism involving oxygen ion
vacancies on the support, extensively discussed in the literature for partially
reducible metal oxides and supported metal catalysts on such carriers.
Olympiou et al. (206) studied the mechanism of the WGS reaction over
alumina-supported Pt, Pd, and Rh catalysts, using steady state isotopic
transient kinetic analysis (SSITKA) techniques coupled with mass spectro-
metry. In particular, the concentrations (mmol g21) of active intermediate
species found in the carbon-path from CO to the CO2 product (using 13CO),
and in the hydrogen- path from H2O to H2 (using D2O) were determined
(Table 11). It was found that by increasing the reaction temperature from 350
to 500uC, the concentrations of the active species in both the carbon and
hydrogen paths increased significantly. Based on (a) the large concentration of
the active species present in the hydrogen- path (OH/H located on the alumina
support), which was larger than six equivalent monolayers (based on the
exposed platinum metal surface area), (b) the small concentration of OH
groups along the periphery of the metal-support interface, and (c) the
Table 11: Concentration of active ‘‘H-containing’’ (H-pool) and ‘‘C-containing’’(C-pool) surface species at water gas shift conditions (181).
Catalyst T (uC)H-pool
(mmol gcat21) or (h)a
C-pool(mmol gcat
21) or (h)a
0.5 wt% Pt/c-Al2O3 350 350 (28.5) 1.3 (0.1)500 1664 (135.6) 31.7(2.6)
0.5 wt% Pd/c-Al2O3 350 235 (9.8) 0.5 (0.02)500 3194 (132.7) 28.6(1.2)
0.5 wt%Rh/c-Al2O3 350 138 (6.2) 2.4(0.1)500 1093 (49.0) 27.3(1.2)
aCoverage in monolayers of exposed surface metal area.
Water Gas Shift Catalysis 411
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
significantly smaller concentration (mmol g21) of active species present in the
carbon-path (adsorbed CO on the noble metal and formate species on the
alumina support and/ or at the metal- support interface), the authors
suggested that the diffusion of OH/H species on the alumina support towards
catalytic sites present in the hydrogen pathway may be the slow step in the
reaction mechanism. The OH/H species were considered to be formed by the
dissociation of H2O on the alumina support. The role of the noble metal was (a)
to activate the CO molecule by chemisorption, and (b) to promote formate
decomposition into CO2 and H2 products. There was also a correlation between
catalytic activity and the surface concentration and binding energy of CO on
the noble metals. Among the alumina-supported noble metals, the order of
activity was found to be Pt . Rh . Pd. It may be remarked that these results
lend strong support to the WGS mechanism proposed by Grenoble et al. (53)
for Pt-alumina in 1981. For the formation of the formate entity, the CO
adsorbed on the Pt metal must react with the OH group adsorbed,
presumably, on the alumina. Whether the formation of the formate is the
result of diffusion of CO from the Pt surface to the Al-OH sites or the diffusion
of the – OH groups from alumina towards CO adsorbed along the
circumference of the metal-support interface is not clear. Duprez (207) has
discussed the mechanism of migration of the OH/H species on metal oxide
surfaces with basic and weak Bronsted acidic character (like gamma alumina).
The problem is still unresolved.
A mechanism based on the interaction of CO with Pt and H2O with ceria
and derived from a kinetic study using a microstructured reactor has been,
recently, proposed for the WGS reaction over a Pt-CeO2-Al2O3 catalyst by
Germani and Schuurman (208). The use of a microstructured reactor, rather
than a packed bed reactor, enabled the measurement of the intrinsic kinetics
of the reaction. The reaction rate was almost zero order in CO and was
strongly inhibited by the partial pressure of hydrogen, and, to a lesser extent
by that of CO2. The rate equation that fitted the data best was based on a dual-
site mechanism with a rate-determining step that involved a species adsorbed
on Pt, a species adsorbed on ceria and a free Pt site. Based on this observation,
a reaction mechanism was proposed where CO, adsorbed on Pt, reacts with
water, dissociatively chemisorbed on ceria, to yield a carboxyl species as an
intermediate. This carboxyl species reacts with a second hydroxyl group and
decomposes over a free Pt site into carbon dioxide and hydrogen as shown
below:
COz�uCO�, ð42Þ
H2OzCe-OuHO-Ce-OH, ð43Þ
412 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
CO�zHO-Ce-OHuCOOH�zCe-OH, ð44Þ
COOH�zCe-OHz�u2H�zCe-O-CO2, ð45Þ
2H�uH2z2�, and ð46Þ
Ce-O-CO2uCO2zCe-O: ð47Þ
In the above equations * represents a Pt site. The rate determining step is the
reaction between the carboxyl species, and the second hydroxyl group on the
ceria (Eq. 45). Once an adjacent Pt site becomes free, this carboxyl complex
decomposes into the reaction products. Hydrogen competes with CO for Pt
adsorption sites and, therefore, retards the rate. Similarly, CO adsorbs
strongly on ceria and has a negative influence on the rate. This mechanism
differs from the mechanism proposed by Shido and Iwasawa (107–108) in that
the surface intermediate is postulated to be a carboxylate and not a formate. It
may be mentioned here that the reaction of CO with Type II bridging OH
groups on ceria to form carboxy species has not been confirmed unambigu-
ously by experiment. We may recall that a carboxyl surface intermediate has
also been postulated, more recently, by Gokhale et al. (196–199). A key feature
of the carboxyl mechanism is the conversion of adsorbed CO to adsorbed
COOH. From microkinetic studies of the WGS reaction system, Mhadeshwar
and Vlachos (210) had, earlier, made an important observation that while the
formation of the carboxyl intermediate from H2O, namely, CO* + H2O* )COOH* + H*, can be the rate-determining step (as per their model), competing
paths for CO* oxidation on Pt by OH* (i.e., CO* + HO*) COOH* + *) instead
of H2O*, cannot completely be ruled out as being important (with the H2O*
decomposition being the rate determining step), due to the relatively small
differences in activation energies of these parallel oxidation paths. They have
also provided a very useful and comprehensive compilation of all the
significant rate expressions and reaction orders for the WGS reaction on
different catalysts postulated in the literature up to 2005 [Table 1 of ref. 185].
One of the drawbacks of the Pt group of metals is that they are less active
in the WGS reaction below about 250uC. On the other hand, Cu-ZnO is an
outstanding WGS catalyst in the range of 200–250uC. Some of the latter’s
drawbacks are (a)the necessity to operate at low GHSV values, (b) its complex
and time-consuming activation protocol before use, and (c) its instability on
contact with air. The Pt group metals do not have these disadvantages. In an
attempt to combine the advantages of both the copper and Pt- group metals in
Water Gas Shift Catalysis 413
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
a single catalyst formulation, Fox et al. (211) have investigated the catalytic
properties of Pd- promoted Cu-Ceria catalysts for the oxygen-assisted, WGS
reaction. It may be noted that the conventional ZnO support has been replaced
by ceria in this formulation. Cu-CeO2, Pd-CeO2, and Cu-Pd-CeO2 catalysts
were prepared and their reduction followed by in-situ XPS to explore the metal
– metal and metal - support interactions in the bimetallic Cu-Pd-CeO2.
Addition of only 1 wt% Pd to 30 wt% Cu-CeO2 greatly enhances the
reducibility of both dispersed CuO as well as the ceria support, presumably
by hydrogen spillover from Pd. In-vacuo reduction (inside the XPS chamber)
up to 400uC results in a continuous growth of metallic copper and Ce3+ surface
species. Support copper, in turn, destabilizes palladium metal (Pdu) with
respect to PdO, this mutual perturbation indicating a strong, intimate
interaction between the Cu-Pd components. The presence of Pd, apparently,
increased the fraction of copper that remains in the metallic state thereby
enhancing its catalytic activity. It may be recalled that the increase of
catalytic activity with metallic copper surface area is well known. Palladium
addition at only 1 wt% significantly improved CO conversion at 180uC,
compared to a monometallic 30 wt% Cu- CeO2 catalyst (Fig. 48). As
anticipated, the Pd-Ceria was the least active compared to Cu-CeO2 and Cu-
Pd-CeO2 at low temperatures. It should be noted that the feedstock used by
Fox et al. (211) (Fig. 48) contained oxygen (2% air) and is not representative of
the conventional feedstock from a steam reformer to the WGS reactor. Reactor
design considerations will be crucial in such a situation to avoid the oxidation
of the hydrogen or CO over the precious metal and the consequent exothermic
temperature rise. In a similar vein, but combining the relative advantages of
two support components, ceria and titania, Gonzalez et al. (212) have recently
Figure 48: CO conversion over 1 wt% Pd, and 30 wt% Cu catalysts and 1 wt% Pd – 30 wt%Cubimetallic catalyst at 180uC. Feed gas: 4% CO, 10% CO2, 26% Ar, 2% air, balance H2. H2O/CO510 (211).
414 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
demonstrated the performance enhancement in the WGS reaction when Pt is
dispersed over a mixed oxide support material containing both ceria and
titania. TiO2 and CeO2 have complementary physical properties which can be
synergized to improve the performance of a catalyst support in the WGS
reaction. For example, the redox properties and thermal stability of titania
can be improved by replacing, partially, Ti4+ ions by the Ce4+ ions in the
titania lattice (213–214). Gonzalez et al.(212) have found that Pt supported on
Ce modified TiO2 catalyst shows better thermal stability (with respect to bare
TiO2 support) and higher WGS activity than those corresponding to individual
titania and ceria supports, indicating a synergistic effect between Pt and the
Ce- modified TiO2 support (Fig. 49). XPS and TPR results revealed the
intimate contact between Pt and cerium entities in the Pt-CeO2-TiO2 catalyst
that facilitates the reducibility of the support at lower temperatures (Fig. 50)
while the Ce-O-Ti interactions decrease the overreduction of TiO2 at high
temperatures. It may be noted that the addition of cerium to TiO2 had also
increased the hydroxyl concentration on the support. This is probably one of
the contributing factors to the greater catalytic activity of the Pt-CeO2-TiO2
compared to Pt-TiO2. This data also underlines the important role of the OH
groups in the mechanism of the LTS reaction.
While discussing the active sites and mechanism of the WGS reaction, it is
instructive to recall two important features of the landmark postulate of Hugh
Taylor in 1926 (215) on active sites over solid catalysts: (a) particular atoms or
groups of atoms on the surface of solids are the active sites responsible for the
Figure 49: CO conversion for the WGS reaction on supported Pt catalysts: (m) Pt/TiO2, (&)Pt/Ce-TiO2, ($) Pt/CeO2 (reference). Reaction conditions: total pressure 1 atm, GHSV521200liter.h21kgcat
21, feed gas composition (mol% ): H2 28%, CH4 0.1%, CO 4.4%, CO2 8.7%, N2
29.2%, H2O 29.6%. Dotted line shows thermodynamic equilibrium limit (212).
Water Gas Shift Catalysis 415
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
catalytic activity and selectivity and, importantly in the present context, (b)
the identity and concentration of the active sites on a catalyst are dependent,
not only on the procedures adopted during its preparation, but also on the
particular reaction conditions, i.e., the relative concentration of the reactants,
temperature, pressure etc. If these conditions are changed, then, the identity
and concentration of the active sites are also likely to change and,
consequently, the dominant reaction mechanistic path from the reactants
(CO and H2O) to the products (CO2 and H2) will be different and depend on
particular reaction conditions. The CO concentration at the inlet to the WGS
section can vary widely in the range 10–40% (dry basis) depending on the raw
material (natural gas or coal) and the reforming process utilized to generate
the syngas. The H2O concentrations will also vary depending on the type of
reformer (steam, partial oxidation or autothermal reformer) that is utilized
upstream of the WGS reactor. Steam reformers utilize higher H2O/ carbon
molar ratios (3–5) than partial oxidation or autothermal reformers (0.5–2.0).
Consequently, the concentration of H2O at the WGS reactor inlet will be
higher when steam reformers are used. Similarly, the concentration of CO2
will be higher when the syngas is generated in a partial oxidation or
Figure 50: Temperature-programmed reduction-MS profiles of (a) Pt/TiO2 and (b) Pt/Ce-TiO2
calcined catalysts (212).
416 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
autothermal reformer than in a steam reformer. It is normal to expect that the
different concentrations of CO, CO2 and H2O in the feedstocks, from these
various H2 generation system configurations, will influence, differently, the
nature and, especially, the concentration of chemical species present on the
catalyst surface (OH groups, H atoms, anion vacancies etc.). Hence, it should
not be surprising that different WGS mechanisms can prevail on the same
catalyst under different reactants concentration and temperature/pressure
conditions, especially in the WGS reaction that is equilibrium-limited at high
temperatures and kinetically limited at low temperatures.
Burch (116) has published a critical discussion of the relative merits of the
various mechanisms that have been proposed for the LTS reaction over metal
– partially reducible oxide supports. He has also presented a ‘‘Universal
mechanism’’ for the WGS reaction that seeks to integrate the salient features
of the formate and redox mechanisms into a single model that is consistent
with all the experimental observations (Fig. 51). Figure 51a shows the
Figure 51: (a) ‘‘carbonate/carboxylate’’mechanism for the reverse WGS reaction (b)‘‘carbonate/carboxylate’’ mechanism for the WGS reaction. (c) ‘‘universal’’ mechanism forthe WGS reaction (116).
Water Gas Shift Catalysis 417
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
mechanism for the reverse water gas reaction, CO2 + H2) CO + H2O (RWGS);
Fig. 51b shows the corresponding mechanism for the forward WGS reaction,
CO + H2O) CO2 + H2. In both cases, the importance of the carbonates and/or
carboxylates is emphasized. Figure 51c shows Burch’s ‘‘universal’’ mechanism.
One crucial feature of Burch’s universal mechanism is that, while in the
formate mechanism, postulated originally by Shido and Iwasawa (107–108)
and elaborated upon by Jacobs et al. (202), the formate intermediate is formed
from insertion of CO into an OH bond, both being adsorbed on the support, it is
formed (in Burch’s postulated mechanism) from the addition of an H to a CO,
both being adsorbed on the metal particle (116). However, there is no
unambiguous experimental evidence for this assumption. In fact, in studies on
transient formate decomposition either with the unpromoted catalyst or the
catalyst promoted with different metals (e.g., Pt, Au) and loadings of metal,
once the surface shell of ceria is reduced, the formate concentration from
reaction of CO with the Type II bridging OH groups is high and, for the most
part, have the same intensity over all the catalysts before the H2O is added to
decompose them. If the formate had been anchored on the metal, a variation of
the intensity with the type and concentration of the metal would have been
observed (223). Both modes of formation of the surface formate intermediate
are shown in Figure 51c. It is important to note that a carbonate–like species
is also involved in the reaction path in all the three postulated mechanisms,
including the formate mechanism, wherein the formate formed initially reacts
with H2O to form a carbonate which finally decomposes to yield CO2. The
dominant mechanism will depend (116) on the reaction conditions, specifically
the temperature and the H2O/CO2 ratio. It can change from a redox-type
process to one dominated by surface intermediate species, including formates,
carbonates and carboxylates. We envision three situations (116):
N At high temperatures, where desorption and/ or decomposition ofintermediates, like formate and carbonate species will be very fast, theredox processes would be expected to be important in determining the rateof the reaction. This is particularly valid in the presence of a highconcentration of H2O when the surface is covered, to a significant extent,by OH groups;
N At low temperatures, and, especially in the presence of a substantialamount of CO2, the final carbonate decomposition step in the mechanismwill be the slow, rate-determining step; and,
N At intermediate temperatures, especially in the presence of a largeconcentration of water, and a low concentration of CO2, the formatedecomposition step in the mechanism would be slow and rate- determining.
As depicted in Fig. 51b, the active sites in the WGS reaction are the oxygen
vacancies which dissociate water into OH adsorbed on the support. Metal sites
418 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
also adsorb and activate CO. In the next step, the OH groups, located on the
support oxide, react with the CO adsorbed on the metal particles to form the
surface intermediates, the formates and carbonates. The latter decompose to
CO2. It must, however, be emphasized that the above mechanism involving
long-lived surface intermediates, like formates or carbonates/ carboxylates, is
valid only below, say, 350uC. At higher temperatures, over HTS catalysts, like
Fe2O3- Cr2O3, the direct oxidation and reduction of the catalyst by H2O and
CO, respectively, by Eley-Rideal-type of reactions are well known, and, hence,
the redox mechanism will, probably be the dominant mechanism. These Eley-
Rideal processes are less favored at low temperatures and, hence, the rates of
formation or decomposition of surface intermediates, like formates and
carbonates/ carboxylates, assume critical importance. One further point in
relation to this mechanism is its relative importance in the case of catalysts,
like Pt-Al2O3, wherein a significant loss of the support hydroxyl groups occurs
only above 400uC (216). Surface oxygen vacancies, that play such an important
role in the above-mentioned mechanisms that invoke the formation of stable
surface intermediates located at these oxygen vacancies, are unlikely to be
present in significant concentrations on the alumina surface in the 190–300uCrange, typical of the LTS reaction, especially in the presence of significant
amounts of H2O. It may be noted that an oxygen anion vacancy is the starting
point in all the mechanisms depicted in Figures 51. In such cases, both the
activation of CO and the dissociation of H2O occur, perhaps, on the Pt metal. It
may, however, be relevant to mention here, that, Chenu et al. (177) had
reported the observation of surface defects, oxygen vacancies and type II
bridging OH groups on non-easily-reducible oxides, like MgO and ZrO2 under
reducing conditions similar to those that prevail during the WGS reaction,
when they were promoted with Pt. After activating the catalyst, DRIFTS of
CO adsorption was used to probe the active OH groups via the generation of
formate species. It is important to note that in the absence of H2O, formates
are quite stable, such that their intensity upon CO adsorption gives a good
qualitative indication of the number of active OH group defect sites. While
formates were observed over both the Pt-ZrO2 and Pt-MgO, the band
intensities were lower as compared with Pt- ceria, suggesting a lower
concentration of defect-associated active OH groups on ZrO2 and MgO. The
WGS rates and formate band intensities from CO adsorption (used to probe
the active OH groups) followed the same trend: Pt-Ceria . Pt-monoclinic ZrO2
. Pt-tetragonal ZrO2 . Pt-magnesia. The Pt content was 1%(wt) in all the
catalysts. On lowering the temperatures, Pt-magnesia was inactive below
400uC, while Pt-Ceria was active up to 250uC, under their reaction conditions
(Feed: 3.75 ml/min CO, 125 ml/min H2O, 100 ml/min H2, 10 ml/min N2; 33 ml of
catalyst). Similar results of generating active OH groups on non-easily
reducible oxides, like ZrO2, on promotion with Pt was also reported by Pigos et
Water Gas Shift Catalysis 419
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
al. (178). They found that the catalytic activity of Pt-ZrO2 was improved,
significantly, when Na was added, either alone or in combination with
vanadium oxide. The results of parallel, DRIFTS spectroscopic experiments
indicated that formate species were more reactive on the Na - promoted
catalysts; The formate C - H stretching bands were shifted to lower
wavenumbers upon CO adsorption. In transient formate decomposition
experiments, both in the presence and absence of steam, the formates over
Pt-Na-ZrO2 decomposed at twice the rate of those observed on Pt-ZrO2 without
Na. This was further supported by steady state WGS rates which confirmed
that the formate species were more reaction rate- limited in DRIFTS for the
Na - promoted catalysts relative to those without Na. These results of Chenu
et al. (177) and Pigos et al. (178) suggest that the active sites for the WGS
reaction, namely, the oxygen anion vacancies and associated OH groups, can
be generated also on otherwise non-reducible oxide supports under reaction
conditions when the catalyst formulation contains elements like the noble
metals.
9.2.7. Catalyst Deactivation
Deactivation during long-term tests has been one of the major drawbacks
of the noble metal- based catalysts with ceria or titania supports. The
facilitating role of bases, like the alkali ions, in the decomposition of formates
had been noted earlier (178). The deactivation of WGS activity along with
strongly held, carbonate surface intermediates had been observed by Gorte et
al. (86–87). At high temperatures (above 350uC) and over the iron oxide-
chromium oxide catalysts, these carbonates decompose more easily and no
deactivation is observed. Do the strongly-held carbonates impede the
reoxidation of the oxide surface or the release of the H2 molecule? The use
of acidic oxides, like those of Nb, Mo, Ta, and W, to enhance the WGS activity
of Pt-Ceria-Zirconia has been reported, recently, by Opalka et al. (217). Aided
by density functional simulations, these authors observed that doping Pt-
ceria-zirconia with acidic, transition metal dopants such as Nb, Mo, Ta and W
oxides increased the oxide surface affinity for water and the turnover rate of
the WGS reaction. The Pt/ Mo-doped –ceria-zirconia, Pt/ Mo0.1Ce 0.7Zr 0.2O2,
for example, was significantly more active (by 15–20% in CO conversion) than
the undoped sample in the temperature range, 200–300uC. The composition of
their feedstock was: 4.9% CO, 10.5% CO2, 33% H2O and 30.3% H2, and the
GHSV was 300,000 h21, simulating the environment in a fuel processor of a
fuel cell. Only initial catalytic activities were reported. Kinetic rate analysis
for the CO conversion yielded reaction orders approaching 0 for CO and 1 for
H2O. They characterised the nature of adsorbed CO and the formate and
carbonate intermediates, formed during the WGS reaction at 200uC over
420 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
catalysts without and with Pt loading, by in-situ infrared spectroscopy. After
oxidizing the nanocrystallites in dry air for 2 h and, then, exposing them to low
CO pressures, only weakly adsorbed CO was detected. When the CO pressures
were increased to 4 bar or higher, small amounts of adsorbed CO2, formates
and carbonates were also observed, indicating that the catalyst was partially
reduced by CO. In the presence of H2O and, under WGS conditions, linearly
adsorbed CO and significant amounts of formates and carbonates were also
observed. The larger, more basic Ce3+ ions formed under reducing conditions
enhance the further reactions of adsorbed CO to form formates and
carbonates. The formates were weakly bonded and could be removed by
outgassing the catalysts in dry nitrogen. The carbonates, on the other hand,
were removed only on oxidation of the catalyst above 270uC. Under the wet,
reducing WGS conditions, on the same oxides with Pt loading, CO linear
adsorption was observed only on the Pt metal (not ceria). Formate and
carbonate formation was observed on the ceria- zirconia oxides. If ligand (CO)
complexation of the oxide surface leading to strongly-held formates and
carbonates is locally specific to the reduced sites, then, CO associative
reactions will compete with or impede the reduction (by CO) or oxidation (by
H2O) of the catalyst and, hence, influence the redox mechanism. Based on
their density functional simulations and IR spectroscopic/kinetic experimental
results, the authors suggested (217) that the associative formation of formates
and carbonates was, indeed, coupled with the bifunctional redox mechanisms
that lead to the reduction of the oxide surface. They observed, further, that the
rate at which O could be removed by CO from the lattice, outstripped the rate
at which H2O could chemisorb and react to replenish the lost O. Hence, the
rate- limiting step, over Pt-ceria and Pt-ceria-zirconia, was not lattice
reduction but rather reoxidation. While the reduction of ceria by CO/H2 is
not in question, its reoxidation by H2O at low temperatures needs to be
verified experimentally by a direct method. To shift the WGS oxidation –
reduction balance towards reoxidation, they added more acidic, less reducible
dopants like Nb, Mo, Ta and W oxides, to make the reoxidation more
favorable. These are strong electron acceptors and are fully oxidized in their
Lewis acid- like oxide phases with generally empty d orbitals (d0 oxides). On
ceria and ceria-zirconia, formate formation was very close in energy to
reoxidation of the reduced surface by H2O. In the presence of acidic transition
metal dopants, however, surface reoxidation was significantly more favorable
than the reaction of adsorbed H2O with CO to form stable formate and
carbonate complexes. For example, while the enthalpy for reoxidation (by
H2O) of the oxide surface of ceria- zirconia was 2134 (eV), it changed to 2340
(eV) on doping with Mo indicating that Mo facilitates the reoxidation of the
surface (217). The transition metal oxide dopants, apparently, shifted the
relative balance of the reaction steps, enhancing the refilling of oxide
Water Gas Shift Catalysis 421
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
vacancies and H2 generation, thereby minimizing the blocking of active sites
by formate and carbonate formation. As a consequence, both catalytic activity
and life were improved (217).
How do dopant ions, like Mo, facilitate surface reoxidation vis-a-vis
formate/carbonate formation? An understanding of the interaction of H2O
with the catalyst surface holds the key to the answer. Atomic simulation
calculations indicated (217) that the H2O molecule dissociated, preferentially,
over the dopant ion to hydroxylate the dopant and to protonate the surface
oxygen ions in the adjacent fluorite lattice. The average H2O adsorption
enthalpy of 257.7 for ceria- zirconia increased to 2 82.3 KJ/mole on doping
with Mo and to 2110.4 KJ/mole on introduction of W in the fluorite lattice
(217).
9.2.8. The LTS Reaction over Non-oxide Supports
While the role of the oxygen anion vacancies and surface OH groups on
metal- metal oxide-based catalysts in the WGS mechanism has been studied
extensively, there is another group of WGS catalysts based on molybdenum
carbide wherein the mechanistic picture is less clear (218–221). Patt et al.
(218) reported high activity for LTS over Mo2C catalysts and obtained higher
activity than over a commercial Cu-ZnO-Al2O3 catalyst. The precursor for
molybdenum was ammonium paramolybdate. The salt was dissolved in warm
water. Then, the liquid was, slowly, evaporated and the solid was calcined in
dry air at 500uC. The oxide was carburized using a temperature-programmed
treatment with CH4 and H2. The LTS activities of the resulting solid (BET
surface area 5 61 m2/g) as well as that of a commercial Cu-ZnO-Al2O3 catalyst
of similar surface area were compared at various temperatures in the range,
220–295uC, using a feed containing 62.5% H2, 31.8% H2O, and 5.7% CO. The
divergence of this feedstock composition from those in commercial practice,
especially the absence of CO2, may be noted. Under these conditions, the CO
conversion over the Mo2C catalyst was at least 50% higher than that over the
Cu-ZnO-Al2O3 sample. Moon and Ryu (219) found that the optimum
carburization temperature was 640–650uC. After repeated thermal cycling
in reductive and oxidative atmospheres, the authors found that even though
there was a decay in catalytic activity of both the Mo2C and a Cu-ZnO-Al2O3
catalysts, the Mo2C was, relatively, more stable. XPS results indicated that
the deactivation of the Mo2C catalyst was linked to the formation of the MoO3
oxide on reaction with H2O. Based on a Density Functional Theory study of
the WGS reaction over Mo2C, combined with infrared spectroscopy experi-
mental results, a redox mechanism was proposed by Tominaga and Nagai
(220). Upon CO adsorption, the authors observed two bands at 1626 cm21 and
1450 cm21. Instead of assigning these to formate species, the authors
422 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
concluded that, since no accompanying symmetric band was observed in the
1350–1365 cm21 range, the 1626 cm21 band was not due to formate but
another vibration. The 1450 cm21 band was assigned to unidentate carbonate.
Based on DFT calculations, the authors assigned the 1626 cm21 band to CO
adsorption on a 3-fold Mo site. A summary of their proposed redox mechanism
is given below where * denotes a free adsorption site:
COz�[CO�, ð48Þ
H2Oz�[H2O�, ð49Þ
H2O�z�[HO�zH�, ð50Þ
HO�z�[O�zH�, ð51Þ
CO�zO�[CO2z2�, and ð52Þ
2H�[H2z2�: ð53Þ
The rate-limiting barrier was found to be the reaction of O with CO to form
CO2 (Eq. 52). This group has, also, extended their study of molybdenum
carbides to include cobalt-containing samples (221). Catalysts were prepared
by combining aqueous solutions of Co(NO3)2 and ammonium heptamolybdate
(NH4)6Mo7O24 and stirring at 80uC to produce a viscous mixture. Solids were
dried in an oven and calcined at 500uC. Carburization was carried out using
20% CH4/H2 mixtures and a temperature-programmed procedure. The
optimum Co content (for catalytic activity using a feed containing 10.5%
CO, 21% H2O the balance being He) was 50% (i.e., Co0.5Mo0.5C). Both the
initial activity and long-term stability of the Co0.5Mo0.5C catalyst was superior
to that of Mo2C. Even though its initial activity was superior to that of a Cu-
ZnO-Al2O3 catalyst, the latter’s long-term stability was better. In view of its
potential as a sulfur-tolerant LTS catalyst, similar to the Co-Mo sour gas shift
catalysts (Section 5) that operate at higher temperatures, further investiga-
tions on this system seem warranted.
10. CONCLUSIONS AND CHALLENGES
The WGS reaction is one of the primary industrial reactions that produce
hydrogen. The utilization of coal and biomass for the production of electrical
Water Gas Shift Catalysis 423
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
power, chemicals, petrochemicals, and hydrogen and transportation fuels is
gaining importance. Declining resources and increasing prices of crude oil are
some of the major driving forces. In future, due to considerations of global
warming, coal may have to be used only in CO2- free power plants, which can
only work in combination with CO2 sequestration. This is only possible when
all carbon compounds in the feedstock are converted to CO2. CO conversion
processes (like the water gas shift reaction) play a key role here. Some sources
predict that by the year 2030, 10% of the yearly consumption of energy will
originate from the WGS reaction (222). The WGS reaction is a well-established
process in conventional chemical plants for the manufacture of ammonia,
methanol, refinery hydrogen, hydrocarbons (by the Fischer - Tropsch process),
etc. In current commercial practice, the WGS conversions are kinetically
limited at low temperatures and thermodynamically limited at high
temperatures. Due to intensive efforts during the last two decades, significant
progress has been made in the study of the mechanism of the WGS reaction
(223). At high temperatures (above 350uC) and over the iron oxide-based
catalysts, the redox mechanism, involving the reduction of the catalyst by CO
and H2 and its reoxidation by H2O probably prevails. At lower temperatures,
even though the detailed mechanism is not established definitively, the
following picture is beginning to emerge: The mechanism and the rate
determining step depends on the nature of the catalyst and process conditions.
On precious metal (Pt, Au) – partially reducible metal oxides (ceria, ceria-
zirconia, titanium oxide), the CO adsorbed, mostly on the metal, reacts with
the surface OH groups on the support to form surface species, like the
formates, carbonates and carboxylates. The concentration and stability of
these species depend on the support oxide, temperature and the partial
pressures of the reactants, especially H2O. Some of these surface species (like
the formates and carbonates) are also intermediates in the reaction path and
decompose to CO2 at higher temperatures and/or partial pressures of H2O.
The decomposition of these various surface species to CO2 is a crucial and slow
step in the reaction path. It is faster at higher temperatures and partial
pressures of H2O. The nature of the catalyst (metal type and loading) and
process conditions have a profound influence on the decomposition of these
intermediates. The accumulation of these species on the surface and, the
consequent, blocking of the catalytically active sites by them lead to loss of
catalytic activity. Factors that hasten their removal, by conversion to CO2, will
improve the catalytic performance of these catalysts. The primary reason for
the accumulation of these intermediates on the surface (and indirectly to loss
of catalytic activity) must be sought in the physicochemical changes under-
gone by the catalyst (e.g., loss of metal-support interface area). Basic additives
like the alkali ions facilitate the decomposition of formates and improve the
low temperature performance of catalysts, like Pt-Ceria, Pt-Ceria-Zirconia
424 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
and Pt-Titania. Similarly, acidic additives, like the oxides of Nb, Mo, Ta and W
facilitate the decomposition of the carbonates to CO2 and accelerate the rate.
The strong adsorption and binding of CO on metals like Pt, at low
temperatures below 250uC, is, also, a rate – inhibiting factor and contribute
to the low activity and slow deactivation of these catalysts at low
temperatures. Development of WGS catalysts kinetically more active in the
190–250uC range is a general challenge. While the Cu-ZnO catalysts are active
in this range, their activity is low necessitating the use of low GHSVs (3000–
5000 h21). A more precise understanding of the mechanism will lead to the
development of better WGS catalysts for hydrogen generation in fuel cells.
There are other challenges to develop improved WGS catalyst and process
versions. Some of them are:
Challenges in High Temperature Shift: (1) Replacing chromium in the
iron oxide- based catalyst by a non-toxic promoter; even though many
chromium-free formulations(containing copper, for example) are in the
pipeline, they are not, yet, proven in commercial usage. (2) Discovery of a
novel, non- noble metal catalyst with higher catalytic activity that will enable
operation at GHSV 5 40,000 h21 and above; the iron oxide- based catalysts
operate at , 15000 h21. This requirement is especially relevant to fuel cell
applications; (3) Discovery of catalysts that can function successfully at low
steam to gas ratios will reduce, significantly, the energy costs associated with
hydrogen generation.
Challenges in Low Temperature Shift: There are, at least, three major
drawbacks in the use of Cu - ZnO – Al2O3 catalysts even in conventional,
stationary applications: (1) their relatively, low catalytic activity (GHSVs of
around only 3000 – 5000 h 21 leading to large – volume catalyst beds), (2) their
elaborate start-up, activation procedures and, (3) their high sensitivity to
sulfur and chlorine compounds as well as to steam below the dew point of H2O
at the operating temperature and pressure. Current pressures for many
downstream applications, for example, restrict the minimum LTS tempera-
tures to 190 – 200uC. Operation at lower pressures in fuel cells, for example,
can benefit from favorable thermodynamics at lower temperatures if a
suitable catalyst can be discovered. Even though well- formulated, modern,
Cu-ZnO-Al2O3 catalysts produce only minor amounts of methanol, it is
desirable to reduce this quantity still further or, even better, eliminate
methanol formation altogether. The above constraints are much more
important for LTS catalysts in mobile fuel cell applications. It is mainly
because of these and other reasons (like the pyrophoricity of the copper –
based catalysts) that the noble metal – reducible oxide catalysts are being
investigated as potential alternatives.
Challenges in Fuel Cell Applications: Even though noble metal –
based catalysts for fuel processors are already in the market, a completely
Water Gas Shift Catalysis 425
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
satisfactory catalyst for WGS applications in fuel cells is not yet in commercial
operation. Ceria and titania- based platinum catalysts are the front runners as
potential water gas catalysts in fuel cell applications. Apart from their high
costs, some of their major drawbacks include their low activity below 250uCand deactivation in long – term operations, especially at lower temperatures
and high pressures. Formation of hydrocarbons at low temperatures and high
pressures (Fischer – Tropsch activity) is yet another drawback of these
catalysts. The present copper – based catalysts do not form significant
amounts of hydrocarbons (like methane) under LTS conditions. Attempts to
improve the long-term life of noble metal catalysts by incorporating acid or
basic additives in the oxide support have been described above (see Section 9).
Modifying the electronic properties of the noble metal, by alloying surface Pt
atoms with those, like Re, Au, Ag and Cu, which do not adsorb CO so strongly,
may, perhaps, be necessary to prevent poisoning by strongly held CO and,
thereby, increase their catalytic activity at low temperatures. Recent
theoretical calculations, by Knudsen et al. (224) indicate that this may indeed
be a promising approach. These authors find that a Cu-Pt surface-alloy binds
CO more weakly than pure Pt (Figure 52) and, hence, CO poisoning at low
temperatures is less likely with the alloy than with the pure metal; in a
temperature programmed desorption experiment, adsorbed CO desorbs at
lower temperatures from Cu-Pt surface alloys than from a pure Pt surface
(Figure 52). Interestingly, the Cu/Pt is also able to activate and dissociate H2O
more easily, the latter being the usual rate-determining step for the WGS on
Figure 52: CO TPD spectra for CuPt (111) surface alloys with varying amounts of copper(ML5 monolayer of Cu) after exposure of 10 Langmuir of CO at 2107uC (224).
426 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
several metal surfaces. The higher WGS activity of Pt-Cu compared to Pt-
alone catalysts had already been claimed by researchers from the Honda
Motor Company (152). Similarly, the higher activity of Pt-Re and Pt-Ag
compared to Pt catalysts has also been claimed by Sud-Chemie Inc. (225).In
addition to the above challenges, development of a single catalyst formulation
for both the hydrocarbon reforming and water gas shift reactions will go a long
way in simplifying the inventory and design of fuel processors. Noble metals,
like Pt, Rh, and Re are the prime candidates for the composite catalyst.
However, materials, that can function as efficient catalyst supports for the
high temperature reforming/ partial oxidation as well as the low temperature
water gas shift reaction, will have to be discovered and developed to meet this
challenge. Major efforts in this direction are in progress worldwide.
Challenges in Fundamental studies: There are also more basic
fundamental problems and key issues that remain to be addressed and
clarified on the mechanistic aspects of the water gas shift, especially the LTS
reaction; (a) The role of oxygen mobility in the oxide component of the noble
metal- partially reducible oxide catalyst needs to be investigated further; in
the redox mechanism, the role of oxygen mobility is very clear and obvious.
However, the role of surface oxygen mobility is also crucial in the associative
mechanism because intermediates (e.g., formates and carbonates) are bound
to the oxide by their surface oxygen atoms and presumably move across the
oxide surface to the metal; information about the nature of the mobile species
(O22/OH 2) and the kinetics of their mobility will benefit both the redox and
associative mechanisms; (b) The need to confirm, unambiguously, that H2O
can indeed reoxidize partially reduced cerium oxide at low temperatures (150–
250uC). Such a reoxidation is a fundamental assumption in the redox
mechanism and its occurrence at high temperatures is well established in
the case of the Fe2O3-Cr2O3 catalysts. There has been no confirmatory
evidence of such a reoxidation process at low temperatures in the case of
partially reduced cerium oxide. Does the presence of a metal (like platinum)
facilitate the reoxidation of partially reduced ceria by water molecules at low
temperatures? While it is known that Pt enhances the reduction of ceria by
hydrogen and creation of surface oxygen vacancies and OH groups, it is not
confirmed that Pt also facilitates the reoxidation of the reduced ceria by H2O
at low temperatures. These experiments are crucial to confirm the redox
mechanism at low temperatures; (c) The need to establish a standard protocol
to estimate the noble metal dispersion when they are supported on partially
reducible oxides like ceria, titania, ceria-zirconia, chromia etc.; while many
labs(especially in the industry) have evolved in-house empirical methods for
catalyst screening purposes, a more scientific foundation is desirable;(d) The
need to model, more accurately, the transition state of formate decomposition
in the presence of co-adsorbed water. In the absence of water, formate is quite
Water Gas Shift Catalysis 427
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
stable and decomposes (reverse thermal decomposition) to CO and OH only at
. 300uC. But, and this is important to note, in the presence of steam (as
during the WGS reaction), the formate decomposes very rapidly even below
150uC in the forward direction to a carbonate, the precursor of CO2 and H2,
picking up the second H, probably from a bridging OH group; most of the
theoretical models developed so far have not explicitly considered this major
influence of co-adsorbed water molecules in enhancing the decomposition of
formate ions to CO2 and H2 at temperatures typical of the WGS reaction and
have treated the decomposition only as a thermal decomposition; hence,
theoretical calculations taking into account the original transition state
picture of Shido and Iwasawa (108) and further elaborated by Jacobs et.al.,
which involves a ‘‘reactant-promoted decomposition of the formate’’ by co-
adsorbed water molecules, are desirable.
ACKNOWLEDGMENTS
We thank the reviewers for useful comments.
REFERENCES
[1] Mond, L., Langer, C. (1888) British Patent, 12608.
[2] Bosch, C., Wild, W. (1914) Canadian Patent, 153379.
[3] Larson, A.T. (1931) US Patent, 1,797,426. Manufacture of hydrogen.
[4] Newsome, D.S. (1980) Water gas shift, reaction. Catalysis Reviews – Science andEng., 21: 275.
[5] Lloyd, L., Ridler, D.E., Twigg, M.V. (1996) The water gas shift reaction, CatalystHandbook, 2nd ed.); Mansion Publishing House: London, 283–338.
[6] Kochloefl, K. (1997) Water gas shift reaction, In Handbook of HeterogeneousCatalysis, Ertl, G., Knozinger, H., and Weitkamp, J. (eds.), Wiley VHS,Publisher: 4: 1831–1840.
[7] Trovarelli, A. (1996) Properties of ceria and ceria-containing materials, CatalysisReviews—SCI ENG, 38: 439–520.
[8] Ladebeck, J.R., Wagner, J.P. (2003) Catalyst development for water gas shift. InHandbook of Fuel Cells—Fundamentals, Technology and Applications, Vol. 3,Vielstich. W., Gasteiger, H., Lamm, A. (eds.) John Wiley and Sons Ltd., 217–229.
[9] Song, C. (2002) Fuel Processing for low temperature and high temperature fuelcells: Challenges and Opportunities for sustainable development in the 21st
century. Catalysis Today, 77(1–2): 17–49.
[10] Moe. J. (1962) M. Design of water-gas shift reactors. Chemical EngineeringProgress, 58: 33–36.
[11] Gonzales, J.C. Gonzales, M.C., Laborde, M. A., Moreno, N. (1986) Effect oftemperature and reduction on the activity of high temperature water gas shiftcatalysts. Appl. Catal., 20: 3–13.
428 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
[12] Chinchan, G.C. Logan, R.H., Spencer, M.S. (1984) Water-gas shift reaction overan iron oxide/chromium oxide catalyst. II: Stability of activity. Appl. Catal., 12:69–88.
[13] Bohlbro, H. (1964) The kinetics of the water gas conversion IV. Influence of thealkali on the rate Eq.. J. Catal., 3: 207–215.
[14] Matijevic, E., Scheider, P. (1978) Ferric hydrous oxide sols: III. Preparation ofuniform particles by hydrolysis of Fe(III)-chloride, nitrate, and perchloratesolutions. J. Colloid Interface Sci., 63: 509–524.
[15] Rangel, M.C., Santos, M.S., Albornoz, A. (2006) The influence of the preparationmethod on the catalytic properties of lanthanum-doped hematite in theethylbenzene dehydrogenation,. Stud. Surf. Sci. Catal., 162: 753–760.
[16] Edwards, M.A., Whittle, D.M., Rhodes, C., Ward, A.M., Rohan, D., Shannon,M.D., Hutchings, G.J., Kiely, C.J. (2002) Microstructural studies of the copperpromoted, iron oxide-chromia water gas shift catalysts. J. Phys. Chem. Phys.,4(15): 3902–3908.
[17] Kulkova, N.V., Temkin, M.I.J. (1949) Physicheck chimii, 23: 695–698.
[18] Temkin, M.I. (1979) The kinetics of some industrial heterogeneous catalyticreactions. In Advances in Catalysis, Vol. 28., Eley, D.D., Pines, H., Weisz, P.B.(eds.), Academic Press: New York, 173–281.
[19] Shchibrya, G.G., Morozov, N.M., Temkin, M.I. (1965) Kinetica i Kataliz, 6, 1057–1059.
[20] Mezaki R., Oki, S. (1973) Locus of the change in the rate determining step.Journal of Catalysis, 30: 488–489.
[21] Amadeo, N.E., Laborde, M.A. (1995) Hydrogen production from the lowtemperature water gas shift reaction: Kinetics and simulation of the industrialreactor. Int. Journal of Hydrogen Energy, 20(12): 949–956.
[22] Oki, S., Happel, J., Hinatow, M., Kancko, Y. (1973) In Proc. 5th Intern. Congresson catalysis, Hightower, J.W. (ed.), North Holland Publ. Co.: Amsterdam,Elsevier, New York, 1: 173–183.
[23] Trimm, D.L. (2005) Minimization of carbon monoxide in a hydrogen stream forfuel cell applications. Applied Catalysis A., 296: 1–11.
[24] Lei, Y., Cant, N.W., Trimm, D.L. (2006) The origin of rhodium promotion ofFe2O3-Cr2O3 catalysts for high temperature water gas shift reaction. J. Catal.,239: 227–236.
[25] Lei, Y., Cant, N.W., Trimm, D.L. (2005) Activity patterns for the water gasshift reaction over supported precious metal catalysts. Catal. Lett., 103(1–2):133–136.
[26] Natesakhawat, S., Wang, X., Zhang, L., Ozkan, U.S. (2006) Development ofchromium-free, iron-based catalysts for high temperature water gas shiftreaction. J. Mol. Catalysis A: Chemical, 260: 82–94.
[27] Araujo, G.C., Rangel, M.C. (2000) An environmental friendly catalyst for thehigh temperature shift reaction. Stud. Surf. Sci. Catal., 130: 1601–1606.
[28] Araujo, G.C., Rangel, M.C. (2000) An environment friendly dopant for hightemperature shift catalysts. Catal. Today, 62: 201–207.
[29] Andreev, A., Idakiev, V., Mihajlova, D., Shopov, D. (1986) Iron-based catalystsfor the water-gas shift reaction promoted by first-row transition metal oxides.Appl. Catal., 22: 385–387.
[30] Kappen, P., Grunwaldt, J.D., Hammershoi, B.S., Trager, L., Clausen, B. S.(2001) The state of copper promoter atoms in high temperature shift catalysts—An in-situ fluorescence XAFS study. J. Catal., 198: 56–65.
Water Gas Shift Catalysis 429
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
[31] Rhodes, C., Williams, B.P., King, F., Hutchings, G.J. (2002) Promotion of Fe3O4-Cr2O3 high temperature water gas shift catalysts. Catalysis Communications,3(8): 381–384.
[32] Rhodes, C., Hutchings, G.J. (2002) Studies of the role of the copper promoter inthe iron oxide-chromia, low temperature water gas shift catalyst. Phy. Chem.Chem. Phys., 5: 2719–2723.
[33] Costa, J.L. R., Marchetti, G.S., Rangel, M.C. (2002) A thorium-doped catalyst forthe high temperature shift reaction. Catal. Today, 77: 205–213.
[34] Chinchen, G.C. Eur. Patent. Appl., A, 1982, 0062410. Catalytic preparation ofhydrogen from carbon monoxide and water.
[35] Rethwisch, D. G., Dumesic, J. A. Effect of metal-oxygen bond strength on theproperties of oxides. II. Water gas shift over bulk oxides, Appl. Catal., 1986,21(1): 97–109.
[36] Hutchings, G. J., Copperwaithet, R.G., Gottschalk, F.M., Hunter, R., Mellor, R.,Orchard, S.W., Sangiorgio, T. (1992) A comparative evaluation of cobaltchromium oxide, cobalt manganese oxide and coppermanganese oxide ascatalysts for the water gas shift reaction. J. Catal., 137: 408–422.
[37] Gottschalk, F.M., Hutchings, G.J. (1989) Manganese oxide water gas shiftcatalysts: Initial Optimization studies. Appl. Catal., 51(1): 127–139.
[38] Ladebeck, J., Kochloefl, K. (1995) Cr-free iron catalysts for water gas shiftreaction, Preparation of Catalysts VI. In, Stud. Surf. Sci. Catal., Vol 91, Issue 5,G. Poncelet, G., Martens, J., Delmon, B., Jacobs, P.A., Grange, P. (eds.), ElsevierSci. Publ.: B. V. Amsterdam, 1079–1083.
[39] Kuijpers, E.G.M., Tjepkema, R.B., Van der Waal, W.J.J., C.M.A.M., Masters,S.F.G.M., Spronck, Geus, J.W. (1986) Structure sensitivity of the water gas shiftreaction over highly active Cu-SiO2 catalysts. Appl. Catal., 25: 139–147.
[40] Uchida, H., Isogai, N., Oba, M., Hasegawa, T. (1967) The zinc oxide-coppercatalyst for carbon monoxide shift conversion.I. The dependency of the catalystactivity on the chemical composition of the catalyst. Bull. Chem. Soc. Japan, 40:1981–1986.
[41] Uchida, H., Isogai, N., Oba, M., Hasegawa, T. (1968) The zinc oxide – coppercatalyst for carbon monoxide shift conversion.II. The catalyst activity and thecatalyst structures. Bull. Chem. Soc. Japan, 41: 479–485.
[42] Bohlbro, H. (1969) An investigation on the kinetics of the conversion of carbonmonoxide with water vapour over iron oxide-based catalysts, 2nd ed., Gellerup:Copenhagen, p. 135.
[43] Bohlbro, H., Jorgensen, M. H. (eds.), Water gas shift reaction. Chem. Eng. World,46: 5–8.
[44] Atake, I., Nishida, K., Li, D., Shishido, T., Oumi, Y., Sano, T., Takehira, K.(2007) Catalytic behavior of ternary Cu-ZnO-Al2O3 system prepared byhomogeneous preparation water gas shift reaction. J. Mol. Catalysis A:Chemical, 275: 130–138.
[45] Petrini, G., Montino, F., Bossi, A., Garbassi, F. (1983) Stud. Surf. Sci. Catal., 16: 735.
[46] Sengupta, G., Das, D.P., Kundu, M.L., Dutta, S., Roy, S.K., Sahay, R.N., Mishra,K.K., Ketchik, S.V. (1989) Study of copper-zinc oxide catalysts, characterizationof the coprecipitate and mixed oxide. Appl. Catal., 55(2): 165–180.
[47] Gines, M.J.L., Amadeo, N., Laborde, M., Apesteguia, C.R. (1995) Activity andstructure sensitivity of the water gas shift reaction over Cu-Zn-Al mixed oxidecatalysts, Appl. Catal., A: General, 131: 283–296.
[48] Chinchen, G.C., Spencer, M.S. (1991) Sensitive and insensitive reactions oncopper catalysts: The water gas shift reaction and methanol synthesis fromcarbon dioxide. Catal. Today, 10: 293–301.
430 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
[49] Tanaka, Y., Takeguchi, T., Kikuchi, R., Eguchi, K. (2005) Influence ofpreparation method and additive for Cu-Mn spinel oxide catalysts on watergas shift reaction of reformed fuels. Appl. Catal., 279: 59–66.
[50] Tanaka, Y., Utaka, T., Kikuchi, R., Takeguchi, T., Sasaki, K., Eguchi, K. (2003)Water gas shift reaction for the reformed fuels over Cu-MnO catalysts preparedvia spinel type oxide. J. Catal., 215: 271–278.
[51] Tanaka, Y., Utaka, T., Kikuchi, R., Sasaki, K., Eguchi, K. (2003) Water gas shiftreaction over Cu-based mixed oxides for CO removal from the reformed fuels.Appl. Catal., 242: 287–295.
[52] Herwijnen, V.T., De Jong, W.A. (1980) Kinetics and mechanism of the CO shifton Cu-ZnO: 1. Kinetics of the forward and reverse CO shift reaction. J. Catalysis,63: 83–93.
[53] Grenoble, D.C., Estadt, M.M. (1981) Ollis, The chemistry and catalysis of thewater gas shift reaction: 1. The kinetics over supported metal catalysts. D. F. J.of Catalysis, 67: 90–102.
[54] Fiolitakis, E., Hofman, H. (1983) Dependence of the kinetics of the lowtemperature water gas shift reaction on the catalyst oxygen activity asinvestigated by wavefront analysis. J. of Catalysis, 80: 328–339.
[55] Hadden, R.A., Vandervell, H.D., Waugh, K.C., Webb, G. (1988) Kinetics andMechanism of the reverse shift reaction on unsupported copper. In Proc. 9th
International Congress on Catalysis, Calgary, Vol. 4, Philips, M.J., Ternan, M.(eds.), Chem. Inst of Canada: Ottawa, 1835.
[56] Leppelt, R., Schumacher, B., Plzak, V., Kinne, M., Behm, R.J. (2006) Kineticsand mechanism of the low temperature water gas shift reaction on Au-CeO2
catalysts in an idealized reaction atmosphere. J. Catalysis, 244: 137–152.
[57] Salmi, T., Hakkarainen, R. (1989) Kinetic study of the low temperature watergas shift reaction over a Cu-ZnO catalyst. Appl. Catal., 49, 285–306.
[58] Ovesen, C.V., Clausen, B.S., Hammershoi, B.S., Steffensen, G., Askgaard, T.,Chorkendorff, I., Norskov, J.K., Rasmussen, P.B., Stolze, P., Taylor, P.A. (1996)Microkinetic analysis of the water gas reaction under industrial conditions. J. ofCatalysis, 158, 170–180.
[59] Ovesen, C.V., Stolze, P., Norskovand, J.K., Campbell, C.T.A. (1992) Kineticmodel of the water gas shift reaction. J. of Catalysis, 134, 445–468.
[60] Meunier, F.C., Reid, D., Goguey, A., Shekhtman, S., Hardacre, C., Burch, R.,Deng, W., Stephanopoulos, M. F. (2007) Quantitative analysis of the reactivity offormate species seen by DRIFTS over a Au/Ce(La)O2 water gas shift catalyst:First unambiguous evidence of the minority role of formates as reactionintermediates. J. of Catalysis, 247, 277–287.
[61] Meunier, F.C., Goguet, A., Hardacre, C., Burch, R., Thompsett, D. (2007)Quantitative DRIFTS investigation of possible reaction mechanisms for thewater gas—shift reaction on high-activity Pt- and Au-based catalysts. J. ofCatalysis, 252, 18–22.
[62] Meunier, F.C., Tibiletti, D., Goguet, A., Shekhtman, S., Hardacre, C., Burch, R.(2007) On the complexity of the water gas shift reaction mechanism over a Pt/CeO2 catalyst: effect of the temperature on the reactivity of formate surfacespecies studied by operando DRIFT during isotopic transient at chemical steadystate. Catalysis Today, 126(1–2), 143–147.
[63] Schumacher, N., Boisen, A., Dahl, D., Gokhale, A. A., Kandoi, S., Grabow, L. C.,Dumesic, J.A., Mavrikakis, M., Chorkendorff, I. (2005) Trends in lowtemperature water gas shift reactivity on transition metals. J. of Catalysis,229, 265–275.
Water Gas Shift Catalysis 431
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
[64] Twigg, M.V., Spencer, M.S. (2001) Deactivation of supported copper metalcatalysts for hydrogenation reactions. Appl. Cat. A: Gen., 212, 161–174.
[65] Tohji, K., Udagawa, Y., Hizushima, T., Vino, A. (1985) The structure of the Cu-ZnO catalysts by an in-situ EXAFS study. J. Phys. Chem., 89(26), 5671–5676.
[66] Spencer, M.S. (1999) The role of Zinc oxide in Cu-ZnO catalysts for methanolsynthesis and water gas shift reaction. Topics in Catalysis, 8(2–4), 259–266.
[67] Aldridge, C.L. (1968) U.S. Patent, 3,615, 216. Water gas shift process forproducing hydrogen using a cesium compound catalyst.
[68] Aldridge, C.L., Kalina, T. (1969) U.K. Patent, 1,325, 172.
[69] Zong, Q., Tang, E., Yang, Y., Zhang, X., Huagong, S. (1995) Chemical Abstracts,137416X.
[70] Zhang, T., Jacobs, P.D., Haynes, Jr, H.W. (1994) Laboratory evaluation of fourcoal liquefaction catalysts prepared from modified alumina supports. CatalysisToday, 19(3), 353–366.
[71] Tang, E., Mao, P. (1994) Hydrogen Energy Prog. X., In Proceedings of the WorldHydrogen Energy Conference, 1, 539.
[72] Mellor, J.R., Copperthwaite, R.G., Coville, N.J. (1997) The selective influence ofsulfur on the performance of novel cobalt-based water gas shift catalysts. Appl.Catal. A: General, 164, 69–79.
[73] Andreeva, D., Idakiev, V., Tabakova, T., Andreeva, A. (1998) Low temperaturewater gas shift reaction over Au-a-Fe2O3. J. of Catalysis, 158, 354–355.
[74] Luengnaruemitchai, A., Osuwan, S., Gulati, E. (2003) Comparative studies oflow temperature water gas shift reaction over Pt-CeO2, Au-CeO2 and Au-Fe2O3
catalysts. Catal.Comm., 4(5), 215–221.
[75] Andreeva, D., Idakiev, V., Tabakova, T., Ilieva, L. Falaras, P., Bourlinosand, A.,Travlos, A. (2002) Low temperature water gas shift reaction over Au-CeO2
catalysts. Catal. Today, 72, 51–57.
[76] Boccuzzi, D.F., Chiorino, A., Manzoli, M., Andreeva, A., Tabakova, T., Ilievab, L.,Ladakiev, L. (2002) Gold, silver and copper catalysts supported on TiO2 for purehydrocarbon production. Catal. Today, 75, 169–175.
[77] Goerke, O., Pfeifer, P., Schubert, K. (2004) Selective oxidation of CO inmicroreactors. Appl. Catal., A, 263, 11–18.
[78] Lywood, W.J., Twigg, M.V. (1991) U.S. Patent 5,030,440. Hydrogen production.
[79] Panagiotopoulou, P., Papavasiliou, J., Avgouropoulos, G., Ionnides, T.,Kondarides, D. I. (2007) Water gas shift activities of doped Pt-GeO2 catalysts.Chem. Eng. Journal, 134 (1–2), 16–22.
[80] Goguet, A., Meunier, F.C., Breen, J.P., Burch, R., Petch, M.I., Ghenciu, A.F.(2004) Study of the origin of the deactivation of a Pt/CeO2 catalyst during inversewater gas shift (RWGS) reaction. J. of Catalysis, 226, 382–392.
[81] Germani, G., Schuurman, Y. (2006) Water gas shift reaction kinetics overstructured Pt-CeO2/ Al2O3 catalysts. AICh.E. Journal, 52(5), 1806–1813.
[82] Xue, E., O’Keefe, M., Ross, J.R.H. (1996) Water gas shift conversion using a feedwith a low steam to carbon monoxide ratio and containing sulfur. CatalysisToday, 30, 107–118.
[83] Panagiotopoulou, P., Kondarides, D. I. (2004) Effect of morphological character-istics of TiO2-supported noble metal catalysts on their activity for water gas shiftreaction. J. of Catalysis, 225, 327–336.
[84] Panagiotopoulou, P., Kondarides, D.I. (2006) Effect of the nature of the supporton the catalytic performance of noble metal catalysts for the water gas reaction.Catalysis Today, 112, 49–52.
432 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
[85] Basinskaa, A., Manieckib, T.P., Jozwiakb, W.K. (2006) Catalytic activity inwater gas shift reaction of platinum group metals supported on iron oxide. React.Kinet. Catal. Lett., 89, 319.
[86] Wang, K., Gorte, R.J., Wagner, J.P. (2002) Deactivation mechanism for Pd-CeO2
for the water gas shift reaction. J. of Catalysis, 212, 225–230.
[87] Ruettinger, W., Liu,X., Farrauto, R.J. (2006) Mechanism of aging for a Pt-ceria-zirconia water gas shift catalyst. Appl. Catal., B, 65(1–2), 135–141.
[88] Kusar, H., Hocevar, S., Levec, J. (2006) Kinetics of the water gas shift reactionover nanostructured copper ceria catalysts. J. Appl. Catal. B., 65(1–2), 135–141.
[89] Quiney, A.S., Schuurman, Y. (2007) Kinetic modeling of CO conversion over aCu/CeO2 catalyst. Chem. Eng. Sci., 62(18–20), 5026–5032.
[90] Ko, J.B., Bae, C.M., Jung, Y.S., Kim, D.H. (2005) Cu-ZrO2 catalysts for water gasshift reaction at low temperatures. Catal. Lett., 105, 157–161.
[91] Schneider, M., Kochloefl, K., Maletz, G.J., Ladebeck, J., Heinisch, C. (1998) U.S.Patent 5,830,425. Chromium-free catalyst based on iron oxide for conversion ofcarbon monoxide.
[92] Panagiotopoulou, P., Kondarides, D.I. (2006) Effect of the nature of the supporton the catalytic performance of noble metal catalysts for the water gas shiftreaction. Catalysis Today, 112, 49–52.
[93] Thinon, O., Diehl, F., Avenier, P., Schuurman, Y. (2008) Screening ofbifunctional water gas shift catalysts. Catalysis Today, 137, 29–35.
[94] Gononzalez, I.D., Navarro, R.M., Galvan, M.C.A., Rosa, F., Fierro, J.L.G. (2008)Performance enhancement in the water gas shift reaction of platinum depositedover a cerium modified TiO2 support. Catal. Comm., 9(8), 1759–1765.
[95] Sato, Y., Terado, K., Soma, Y., Miyao, T., Naito, S. (2006) Marked addition effectof Re upon the water gas shift reaction over TiO2 supported Pt, Pd and Ircatalysts. Catal. Comm., 7(2), 91–95.
[96] Panagiotopoulou, P., Christodoulakis, A., Kondarides, D.I., Boghasian, S. (2006)Particle size effects on the reducibility of titanium oxide and its relation to thewater gas shift activity of Pt-TiO2 catalysts. J. of Catalysis, 240, 114–125.
[97] Sato, Y., Terada, K., Hasegawa, S., Miyao, T., Naito, S. (2005) Mechanisticstudies of water gas shift reactions over TiO2-supported Pt-Re and Pd-Recatalysts. Appl. Catal. A. Gen., 296, 80–89.
[98] Azzam, K.G., Babich, I.V., Seshan, K., Lefferts, L. (2008) Single stage water gasshift conversion over Pt/TiO2—Problem of catalyst deactivation. Appl. Catal., A:General, 338, 66–71.
[99] Lox, E.S.J., Engler, B.H. (1997) Environmental Catalysis, Handbook ofHeterogeneous Catalysis-mobile sources, Ertl, G., Knozinger, H., Weitkamp, J.(eds) Wiley-VCH: 4, 1559–1633.
[100] Wu, X., Fan, J., Ran, R., Weng, D. (2005) Effect of preparation methods on thestructure and redox behavior of Pt-ceria-zirconia catalysts. Chem. Eng. Journal,109(1), 133–139.
[101] Bridger, G.W., Chinchen, G.C. (1970) Water gas shift, Catalyst Handbook, ICI,Wolf. Scientific Books: London, 1970, 97–125.
[102] Sato, Y., Terada, K., Soma, Y., Miyao, T., Naito, S. (2006) Marked addition effectof Re upon the water gas shift reaction over TiO2 supported Pt, Pd and Ircatalysts. Catal. Comm., 7(2), 91–95.
[103] Hagemeyer, A., Brooks, C.J., Calhart, R.E., Yaccato, K., Herrmann, M. (2007)U.S. Patent 7,270,798 B2. Noble metal-free niekel catalyst for hydrogengeneration.
Water Gas Shift Catalysis 433
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
[104] Zerva, C., Philippopoulos, C.J. (2006) Ceria catalysts for water gas shift reaction:Influence of preparation methods on their activity. Appl. Catal. B: Env., 67, 105–112.
[105] Iida, H., Kondo, K., Igarashi, A. (2006) Effect of Pt precursors on catalyticactivity of Pt-TiO2 (rutile) for water gas shift reaction at low temperatures.Catal. Comm., 7(4), 240–244.
[106] Hilaire, S., Wang, X., Luo, T., Gorte, R.T., Wagner, J.P. (2001) A comparativestudy of water gas shift reaction over ceria-supported metallic catalysts. Appl.Catal.., A: Gen, 215, 271–278.
[107] Shido, T., Iwasawa, Y. (1992) Regulation of reaction intermediate by reactant inthe water gas shift reaction on CeO2 in relation to reactant promotedmechanism. J. of Catalysis, 136, 493–503.
[108] Shido, T, Iwasawa, Y. (1993) Reactant promoted reaction mechanism for watergas shift reaction on Rh-doped CeO2, J. Catal., 1993, 141, 71–81.
[109] Zalc, J.M., Sokolovski, V., Loffler, D.G. (2002) Are noble metal based, water gasshift catalysts practical for automotive fuel processing. J. of Catalysis, 206, 169–171.
[110] Iida, H., Igarashi, A. (2006) Difference in the reaction behavior between Pt-Re-TiO2 and Pt-Re-ZrO2 catalysts for water gas shift reaction at low temperatures.Appl.Catal., 303(1) 48–55, , Iida, H., Igarashi, A., Abstracts, Proc. 14th
International Congress on Calalysis, Seoul, Korea, 2008.
[111] Fisher, J.M., Thompsett, D., Walton, R.I., Wright, C.S. (2006) Compound havinga pyrochlore-structure and its use as a catalyst carrier in water gas shiftreaction. WO 2006/030179.
[112] Baidya, T., Gayen, A., Hegde, M. S., Ravishanker, N., Dupont, L. (2006)Enhanced reducibility of Ce 12x TixO2 compared to that of CeO2 and higher redoxcatalytic activity of Ce 12x2y TixPtyO 22d compared to that of Ce 12x PtxO 22 d. J.Phys. Chem. B, 110(11), 5262–5272.
[113] Ruettinger, W., Ilinich, O., Farrauto, R.J. (2003) A new generation of water gasshift catalysts for fuel cell applications. J. of Power Sources, 118(1–2), 61–65.
[114] Sugie, Y., Kimura, K. (2003) U.S. Patent, 6630119. Hydrogen gas generatingmethod.
[115] Wagner, J.P., Cai, Y., Wagner, A.L. (2003) U.S. Patent 2003/0186804 A1.Catalyst for production of hydrogen.
[116] Burch, R. (2006) Gold catalysts for pure hydrogen production in the water-gasshift reaction: activity, structure and reaction mechanism. Phys. Chem. Chem.Phys., 8, 5483–5500.
[117] Tibiletti, D., Amieiro-Fonseca, A., Burch, R., Chen, Y., Fisher, J. M., Goguet, A.,Hardacre, C., Hu, P., Thompsett, D. (2005) DFT and in-situ EXAFS investigationof gold/ceria zirconia low temperature water gas shift catalysts: Identification ofthe nature of active form of gold. J. of Phy. Chem., B, 109, 22553–22559.
[118] Haruta, M., Yamada, N., Kobayashi, T., Iijima, S. (1989) Gold catalysts preparedby coprecipitation for low temperature oxidation of hydrogen and of carbonmonoxide. J. of Catal., 115, 301–309.
[119] Haruta, M., Tsubora, S., Kobayashi, T., Kageyama, H., Genet, M. J., Delmon, B.(1993) Low temperature oxidation of CO over gold supported on TiO2,-a-Fe2O3
and Co3O4. J. of Catal., 144, 175–192.
[120] Andreeva, D., Ivanov, I., Ilieva, L., Sobczak, J.W., Avdeev, G., Tabakova, T.(2007) Nanosized gold catalysts supported on ceria and ceria-alumina for watergas shift reaction. Appl.Catal., A: Gen:., 333(2), 153–160.
[121] Hua, J., Zheng, Q., Zheng, Y., Wei, K., Liu, X. (2005) Influence of modifyingadditives on the catalytic activity and stability of A-Fe2O3-MOx catalysts forwater gas shift reaction. Catal. Lett., 102, 99–108.
434 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
[122] Venugopal, A., Aluha, J., Mogano, D., Scurrell, M.S. (2003) The gold-ruthenium-iron oxide catalytic system for the low temperature water gas shift reaction: Theexamination of gold-ruthenium interactions. App. Cat., A: Gen., 245, 149–158.
[123] Tabakova, T., Idakiev, V., Andreeva, D., Mitov, I. (2000) Influence of themicroscopic properties of the support on the catalytic activity of Au-ZnO, Au-ZrO2, Au-Fe2O3, Au-Fe2O3-ZnO, Au-Fe2O3-ZrO2 catalysts for the water gas shiftreaction. Appl. Catal., A Gen., 202, 91–97.
[124] Sakurai, H., Ueda, A., Kobayashi, T., Haruta, M. (1997) Low temperature watergas shift reaction over gold deposited on TiO2. Chem. Comm., 3, 271–272.
[125] Idakiev, V., Tabakova, T., Yuan, Z.Y., Su, B.L. (2004) Gold catalysts supportedon mesoporous titania for low temperature water gas shift reaction. Appl. Catal.A: Gen. 270, 135–141.
[126] Mohamed, M.M., Salama, T.M., Toman, A.I., El-Shobaky, G.A. (2005) Lowtemperature water gas shift reaction on cerium-containing mordenites preparedby different methods. Appl. Catal. A. Gen., 279, 23–33.
[127] Fu, Q., Kudriavtseva, S., Saltsburg, H., Flytzani-Stephanopoulos, M. (2003) Goldceria catalysts for low temperature water gas shift catalysts. Chem. Eng.Journal, 93, 41–53.
[128] Miyake, Y., Tsuji, S. (2000) European Patent EP 1043059 A1. Catalyst forpurifying an exhaust gas.
[129] Rodriguez, J.A., Wang. X. Liu, P., Wen, W. Hansen, J.C. Hrbk, J., Perez, M.,Evans, J. (2007) Gold nanoparticles on ceria: Importance of O vacancies in theactivation of gold. Topics in Catal., 44(1–2), 73–81.
[130] Karpenko, A., Leppelt, R., Plzak, V., Behm, R.J. (2007) Deactivation of a Au-CeO2 catalyst during the low temperature water gas shift reaction and itsreactivation: A combined TEM, XRD, XPS, DRIFTS and activity study. J. ofCatal., 2007, 252, 231–242.
[131] Karpenko, A., Leppelt, R., Cai, J., Plzak, V., Chuvilin, A., Kaiser, U., Behm, R.J.(2007) Deactivation of a Au/CeO2 catalyst during the low-temperature water-gasshift reaction and its reactivation: A combined TEM, XRD, XPS, DRIFTS andactivity study. J. Catalysis, 250, 139–150.
[132] Andreeva, D., Ivanov, I., Ilieva, L., Sobezak, J.W., Avdeev, G., Petrov, K. (2007)Gold based catalysts on ceria and ceria-alumina for water gas shift reaction.Topics in Catalysis, 44(1–2), 173–182.
[133] Janssens, T.V.W., Clausen, B.S., Hvrolbek, B., Falsig, H., Christensen, C.H.,Bligaard, T., Norskov, J.K. (2007) Insights into the reactivity of supported goldnanoparticles: combining theory and experiments. Topics in Catalysis, 44(1–2),15–26.
[134] Yoon, B., Hakkinen, H., Landman, U. (2003) Interaction of O2 with gold clusters:Molecular and dissociative adsorption. J. Phy. Chem., A107, 4066–4071.
[135] Bond, G.C., Thompson, D.T. (1999) Catalysis Reviews, Catalysis by gold. 41(3–4), 319–388.
[136] Moreau, F., Bond, G.C. (2006) CO oxidation activity of gold catalysts supportedon various oxides and their improvement by inclusion of an iron component.Catalysis Today, 114(4), 362–368.
[137] Fu, Q., Deng, W., Saltsburg, H., Flytzani-Stephanopoulos, M. (2005) Activity andstability of low content gold-cerium oxide catalysts for water gas shift reaction.Appl. Catal:B., Environmental, 56, 57–68.
[138] Jacobs, G., Ricote, S., Patterson, P.M., Graham, W.M., Dozier, A., Khalid, S.,Rhodus, E., Davis, B.H. (2005) Low temperature water gas shift: Examiningefficiency of Au as a promoter for ceria-based catalysts prepared by CVD of a Auprecursor. Appl. Catal., A, Gen, 292, 229–243.
Water Gas Shift Catalysis 435
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
[139] Juan, M.A.H., Yeung, C.M.M., Tsang, S.C. (2008) A study of co-precipitatedbimetallic gold catalysts for water-gas shift reaction. Catal. Comm., 9(7), 1551–1557.
[140] Fu. Q., Weber, A., Flytzani-Stephanopoulos, M. (2001) Nanostructured Au-Ceriacatalysts for the low temperature water gas shift. Catal. Lett., 77, 87–95.
[141] Andreeva, D. (2002) Low temperature water gas shift over gold catalysts. GoldBulletin, 35, 82–88.
[142] Sakurai, H., Akita, T., Tsuboto, S., Kiuchi, M., Haruta, M. (2005) Lowtemperature activity of Au-CeO2 for water gas shift reaction and characteriza-tion by ADF-STEM, temperature – programmed reaction and pulse reaction.Appl. Catal., A, Gen, 291, 179–187.
[143] Rajaram, R.R., Hayes, J.W., Ansell, G.P., Hatcher, H.A. (1999) U.S. Patent,5,993,762. Method of using catalyst containing noble metal and cerium oxide.
[144] Perrichon, V., Retailleau, L., Bazin, P., Daturi, M., Lavalley, J.C. (2004) Metaldispersion of CeO2-ZrO2 supported platinum catalysts measured by H2 and COchemisorption. Appl. Catal., A, Gen: 260, 1–8.
[145] Freund, A., Lang, J., Lehmann, T., Starz, K.A. (1996) Improved Pt alloy catalystsfor fuel cell applications. 27(1–2), 279–283.
[146] Kopasz, J.P., Krause, T.R., Ahmed, S., Krumpelt, M. (2002) Fuel requirementsfor fuel cell systems, American Chemical Society Division of fuel Chemistry.Preprints, 47(2), 489–491.
[147] Qi, X., Stephanopoulos, M. (2004) Activity and stability of Cu-CeO2 catalysts inhigh temperature water gas shift for fuel cell applications. Ind.Eng.Chem.Research, 43(12), 3055–3062.
[148] Deng, W., Stephanapoulos, M. (2006) On the issue of deactivation of Au-ceria, Pt-ceria water gas shift catalysts in practical fuel cell applications. Angew. Chem.Intl. Ed., 45(14), 2285–2289.
[149] Duarte de Farias, A.M., Barandas, A.P.M.G., Perez, R.F., Fraga, M.A. (2007)Water-gas shift reaction over magnesis-modified Pt/CeO2 catalysts. J. of PowerSources, 165(2), 854–860.
[150] Pierre, D., Deng, W., Stephanapoulos, M. (2007) The importance of stronglybound Pt-CeOx species for the water gas reaction: catalytic activity and stabilityevaluation. Topics in Catalysis, 46, 363–373.
[151] Hillaire, S., Ruettinger, W., Xu, X., Farrauto, R. (2005) Deactivation of Pt-CeO2
water gas shift catalysts due to shutdown/startup modes for fuel cellapplications. Applied Catalysis B, Environmental, 56, 69–75.
[152] Hagemeyer, A., Carhart, R.E., Yaccato, K., Strasser, P., Herrmann, M.,Grasselli, R.K., Brooks, C.J., Phillips, C.B. (2007) U.S. Patent 7,179,442 B2.Catalyst formulations containing Groups 11 metals for hydrogen generation.
[153] Ponec, V. (1997) Carbon monoxide and carbon dioxide hydrogenation. InHandbook of Heterogeneous catalysis, Ertl, G., Knozinger, H., Weitkamp, J.(eds.), Wiley VCH Publishers: 4, 1876–1894.
[154] Olga, Ruettinger, W.F., Farrauto, R.J. (2003) U.S. Patent 20030064887.Suppression of methanation activity by a water gas shift reaction catalyst.
[155] Germani, G., Alphonse, P.; Courty, M., Schuurman, Y. Mirodatos, C. (2005)Platinum-ceria-alumina catalysts on nanostructures for carbon monoxideconversions. Catalysis Today, 110, 114–120.
[156] Boreskov, G.K. (1966) Forms of oxygen bonds on the surface of oxidationcatalysts, Disc. Faraday Soc., 1966, 41, 263–276.
[157] Boreskov, G.K.; Yureva, T.M., Morozov, N.M., Sergeeva A.S., Kinet. Katal, 11,1476 (1970).
436 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
[158] Tinkle, M, Dumesic, J.A. (1987) Isotopic exchange measurements of therate of adsorption/desorption and interconversion of CO and CO2 overchromia-promoted magnetite: Implications for water gas shift. J. Catal., 103,65–78.
[159] Khan, A., Chen, P., Boolchand, P., Smirnoitis, P.G. (2008) Modified nanocrystal-line ferrites for high temperature water gas shift membrane reactor applica-tions. J. Catalysis, 253, 91–104.
[160] Armstrong, E.F., Hilditch, T.P. (1920) A study of Catalytic Actions at SolidSurfaces-IV. The interaction of carbon monoxide and steam as conditioned byiron oxide and by copper. Proc. Royal Soc., A97, 265–272.
[161] Deluzarche, A., Hindemann, J.P., Keinemann, A., Keiffer, R. (1985) Applicationof chemical trapping to the determination of surface species and to the study oftheir evolution under reaction conditions in heterogeneous catalysis. J. Mol.Catal., 31(2), 225–250.
[162] Diagne, C., Vos, P.J., Keinneman, A., Perez, M.J. and Portella, F.M. (1990)Water gas shift reaction over chromia-promoted magnetite, use of temperature-programmed-desorption and chemical trapping in the study of the reactionmechanism. React, Kinet, Katal. Lett., 42(1), 25–31.
[163] Davydov, A. (2003) Formate Anions, Molecular Spectroscopy of Oxide CatalystSurfaces. Sheppard, N.T. (ed.), John Wiley Sons, Inc.: 56–78 (hydroxyl groups),447–453 (formate anions).
[164] Davydov, A.V., Boreskov, G.K., Yurieva, T.M., Rubene, N.A. (1977) Associativemechanisms of the water gas shift reactions. Doklady Akademii Nauk USSR,236, 1402.
[165] Van Herwijnen, T., de Jong, W.A. (1980) Kinetics and mechanism of the CO shifton Cu/ZnO:1. Kinetics of the forward and reverse CO shift reactions. J. Catal.,63(I), 83–93.
[166] Van Herwijnen, T., Guczalski, R.T., de Jong, W.A. (1980) Kinetics andmechanism of the CO shift on Cu/ZnO:II. Kinetics of the decomposition offormic acid. J. Catal., 63(1), 94–101.
[167] Rhodes, C., Hutchings, G. J, Ward, A. M. Water gas shift reaction: Finding themechanistic boundary, Catal. Today, 1995, 23, 43–58.
[168] C. . Binet, M. . Daturi, Lavalley, J.C. (1999) Infrared study of polycrystallineceria prepared in oxidized and reduced states. Catal.Today, 50, 207–225.
[169] Fallah, J.El., Boujana, S., Dexpert, H., Kiennemann, A., Marjerus, J., Touret, O.,Villain, F, Le Normand, F. (1994) Redox processes on pure ceria and on Rh-Ceriacatalyst monitored by X-ray absorption (fast acquisition mode). J. Phys. Chem.,98(21), 5522–5533.
[170] Jacobs, G., Patterson, P.M., Graham, U.M., Sparks, D.E., Davis, B.H. (2004) Lowtemperature water gas shift: Kinetic isotope effect observed for decomposition ofsurface formates for Pt-ceria catalysts. Appl. Catal. Gen, 269, 63–73.
[171] Jacobs, G., Graham, U. M., Chenu, E., Patterson, M., Dozier A., Davis, B.H.(2005) Low temperature water gas shift: Impact of Pt promoter loading on thepartial reduction of ceria and consequences for catalyst design. J. Catal., 229,499–512.
[172] Jacobs, G., Crawford, A.C., Davis, B.H. (2005) Water gas shift: Steady stateisotope switching study of the water gas shift reaction over Pt-Ceria using in-situDRIFTS. Catal. Lett., 100, 147–152.
[173] Lamotte, J., Lavalley, J.C., Druet, D, Freund, E. (1983) Infrared studies of acid-base properties of thorium oxide. J. Chem. Soc. Faraday Trans., 79(9), 2219–2227.
Water Gas Shift Catalysis 437
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
[174] Jung, K., Bell, A.T. (2000) Role of hydrogen spillover in methanol synthesis overCu/ZrO2. J. Catal., 193, 207–223.
[175] Kondo, J., Sakata, Y., Domen, K., Maruya, K., Onishi, T. (1990) Infrared study ofhydrogen adsorbed on ZrO2., J.Chem.Soc., Faraday Trans., 86(2), 397–401.
[176] Jacobs, G., Patterson, G. M., Graham, U.M., Crawford, A.C., Dozier, A., Davis,B.H. (2005) Catalytic links among the water gas shift, water-assisted formic aciddecomposition, and methanol steam reforming reactions over Pt-promotedthoria. J. Catal., 235, 79–91.
[177] Chenu, E., Jacobs, G., Crawford, A.C., Keogh, R.A., Patterson, P.M., Sparks,D.E., Davis, B.H. (2005) Water gas shift: An examination of Pt-promoted MgOand tetragonal and monoclinic ZrO2 by in-situ DRIFTS. Appl. Catal., B:Environmental, 59, 45–56.
[178] Pigos, J.M., Brooks, C.J., Jacobs, G., Davis, B.H. (2007) Low temperature watergas shift: Characterization of Pt-based ZrO2 catalyst promoted with Nadiscovered by combinatorial methods. Appl. Catal. A: Gen., 319, 47–57.
[179] Mojet, B.L., Miller, J.T., and Koningsberger, D.C. (1999) The effect of COadsorption at room temperature on the structure of supported Pt particles. J.Phys. Chem., B:103, 2724–2734.
[180] Tibiletti, T., Goguet, A., Meunier, F.C., Breen, J.P., Burch, R. (2004) On theimportance of steady state isotopic techniques for the investigation of themechanism of the reverse water gas shift reaction. Chem. Commun., 1636–1637.
[181] Bunluesin, T., Gorte, R.J., Graham, G.W. (1998) Studies of the water gas shiftreaction on ceria-supported Pt, Pd and Rh: Implications for the oxygen storageproperties. Appl. Catal. B: 15, 107–114.
[182] Zafiris, G.S., Gorte, R J. (1993) Evidence for low temperature oxygen migrationfrom ceria to Rh. J. Catal., 139, 561–567.
[183] Sharma, S., Hilaire, Vohs, J.M., Gorte, R.J., Jen, H.W. (2000) Evidence foroxidation of ceria by CO2. J. Catal., 190, 199–204.
[184] Azzam, K.G., Babich, I.V., Seshan, K., Lafferts, L. (2008) Role of Re in Pt-Re/TiO2 catalysts for water gas shift reaction: A mechanistic and kinetic study.Appl. Catal., B, 80(1–2), 129–140.
[185] Azzam, K.G., Babich, I.V., Seshan, K., Lefferts, L. (2007) Bifunctional catalystsfor single-stage water gas shift reaction in fuel cell applications. Part 1. Effect ofthe support on the reaction sequence. J. Catal., 251 (1), 153–162.
[186] Sandoval, A., Gomez-Cortes, A., Zanella, R., Diaz, G., Saniger, J.M. (2007) Goldnanoparticles: Support effects for the water gas shift reaction. J. Mol. Catal., A:Chem., 278(1–2), 200–208.
[187] Gonzalez, I.D., Navarro, R.M., Alvarez-Galvan, M.C., Rosa, F., Fierro, J.L.G.(2008) Performance enhancement in the water gas shift reaction of platinumdeposited over a cerium-modified TiO2 support. Catal. Commun., 9(8), 1759–1765.
[188] Li, J., Chen, J., Song, W., Liu, J., Shen, W. (2008) Influence of zirconia crystalphase on the catalytic performance of Au-ZrO2 catalysts for low temperaturewater gas shift reaction. Appl. Catal. A: Gen., 334(1–2), 321–329 .
[189] Rodriguez, J. A., Liu, P., Hrbek, J., Evans, J., Perez, M. (2007) Water gas shiftreaction on Cu and Au nanoparticles supported on CeO2(111) and Zn(0001):Intrinsic activity and importance of support interactions, Angew. Chem. Int.Ed.,46(8), 1329–1332.
[190] Rodriguez, J.A., Liu, P., Hrbek ,J., Perez, M., Evans, J. (2008) Water-gas shiftactivity of Au and Cu nanoparticles supported on molybdenum oxides. J. Mol.Catal., A: Chemical, 281(1–2), 59–65.
438 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
[191] Wang, X., Rodriguez, J.A., Hanson, J.C., Gamarra, D., Martinez-Arias, A.,Garcia, M.F. (2006) In-situ studies of the active sites for the water gas shiftreaction over Cu-CeO2 catalysts: Complex interaction between metallic copperand oxygen vacancies of ceria. J. Phys. Chem. B: 110, 428–434.
[192] Wang, X., Rodriguez, J.A., Hanson, J.C., Perez, M., and Evans, J. (2005) In-situtime-resolved characterization of Au-CeO2 and AuOx-CeO2 catalysts during thewater gas reaction: Presence of Au and O vacancies in the active phase. J. Chem.Phys., 123, 22110–22114.
[193] Liu, P., Rodriguez, J.A. (2005) Catalysts for the hydrogen evolution from [NiFe]hydrogenase to the Ni2P(001) surface; The importance of the ensemble effect. J.Amer. Chem.Soc., 127, 14871–14878.
[194] Liu, P., and Rodriguez, J. A. (2006) Water gas shift reaction on molybdenumcarbide surfaces: Essential role of the oxycarbide. J. Phys. Chem. B: 110, 19418–19425.
[195] Barrio, L., Liu, P., Rodriguez, J.A., Campos-Martin, J. M., Fierro, J. L. G. (2006)A density functional theory study of the dissociation of H2 on gold clusters:Importance of fluxionality and ensemble effects. J. Chem. Phys., 125, 164715–164719.
[196] Gokhale, A.A., Dumesic, J. A., Mavrikakis, M.M. (2008) On the mechanism oflow temperature water gas shift reaction on copper. J. Amer. Chem. Soc., 130,1402–1414.
[197] Grabow, L.C., Gokhale, A.A., Evans, S.T., Dumesic, A., Mavrikakis, M. (2008)Mechanism of the water gas shift reaction on Pt: First Principles, Experiments,and microkinetic modeling. J. Phys. Chem. C., 112, 4608–4617.
[198] Korhonen, S.T., Calatayud, M., Krause, O.I. (2008) Structure and stability offormates and carbonates on monoclinic zirconia: A combined study by densityfunctional theory and infra red spectroscopy. J. Phys. Chem. C., 112, 16096–16102.
[199] Van Nutter, R. M., Coleman, J.S., Lund, C.R.F. (2008) DFT models for activesites on high temperature water gas shift catalysts. J. Mol. Catalysis—ACHEMICAL, 292, 76–82.
[200] Koryabkina, N.A., Phatak, A.A., Ruettinger, W. F, Farrauto, R.J., Rebeiro, F.H.(2003) Determination of kinetic parameters for the water gas shift reaction oncopper catalysts under realistic conditions for fuel cell applications. J. Catal.,217, 233–234.
[201] Evin, H.N., Jacobs, G., Martinez, J. R., Thomas, G. A., Davis, B. H. (2008) Lowtemperature water gas shift: Alkali doping to facilitate C-H bond cleaving overPt-Ceria catalysts—An optimization problem. Catal. Lett., 120, 166–178.
[202] Evin, H.N., Jacobs, G., Martinez, J.R., Graham, U.M., Dozier, A., Thomas, G.,Davis, B.H. (2008) Low temperature water-gas shift/methanol steam reforming:Alkali doping to facilitate the scission of formate and methoxy C – H bonds overPt/ceria catalyst. Catal. Lett., 122, 9–19.
[203] Sato, S., White, J.M. (1981) Photocatalytic water decomposition and water gasshift reaction over NaOH-coated, platinized TiO2. J. Catal., 69, 128–139.
[204] Klier, K. (1992) Preparation of bifunctional catalysts. Catal. Today, 15, 361–382.
[205] Campbell, J.M., Nakamura, J., Campbell, C.T. (1992) Model studies of cesiumpromoters in water gas shift catalysts: Cs/Cu(110). J. Catal., 136, 24–42.
[206] Olympiou, G.G., Kalamaras, C.M., Yazdi, C.D.Z., Efstathiou, A.M. (2007)Mechanistic aspects of the water-gas shift reaction on alumina-supported noblemetal catalysts: In-situ DRIFTS and SSITKA–mass spectrometry studies. Catal.Today, 127, 304–318.
Water Gas Shift Catalysis 439
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13
[207] Duprez, D. (2006) Study of surface reaction mechanisms by 16O /18O and H/Dexchange. Catal. Today, 112, 17–22.
[208] Germani, G., Schuurman, Y. (2006) Water gas shift reaction kinetics overstructured Pt-CeO2-Al2O3 catalysts. AIChE Journal, 52(5), 1806–1813.
[209] Mhadeshwar, A.B., Vlachos, D.G. (2005) Is the water gas shift reaction on Ptsimple? Computer aided microkinetic model reaction lumped rate expression andrate determining step. Catal. Today, 105, 162–172.
[210] Mhadeshwar, A.B., Vlachos, D.G. (2005) Hierarchical, multiscale surfacereaction mechanism development: CO and H2 oxidation, water gas shift andpreferential oxidation of CO on Rh. J. Catalysis, 234(1), 48–63.
[211] Fox, E.B., Lee, A.F., Wilson, K., Song, C. (2008) In-situ XPS study on thereducibility of Pd-promoted Cu-CeO2 catalysts for the oxygen-assisted water gasshift reaction. Topics in Catalysis, in Press.
[212] Gonzales, I.D., Navarro, R.M., Galvan, M.C.A., Rosa, F., Fierro, J.L.G. (2008)Performance enhancement in the water-gas shift reaction of Pt deposited over acerium modified TiO2 support. Catal. Communications, 9(8), 1759–1765.
[213] Rynkowski, J., Farbotko, J., Touroude, R., Hillaire, L. (2003) Redox behavior ofceria- titania mixed oxides. Appl. Catal., A Gen., 203, 335–348.
[214] Dutta, G., Waghmare, U.V., Baidya, T., Hegde, M.S., Priolkar, K.R., Sarode, P.R.(2006) Origin of enhanced reducibility/oxygen storage capacity of Ce 1-x TixO2
compared to CeO2 or TiO2. Chem. Materials, 18, 3249–3256.
[215] Taylor, H.S. (1926) Fourth Report of the Committee on Contact catalysis. J.Phys. Chem., 30, 145–171.
[216] Knozinger, H., Ratnasamy, P. (1978) Catalytic aluminas. Catalysis Reviews, 17,31–70.
[217] Opalka, S. M., Vanderspurt, T. H., Radhakrishnan, R., She, Y., Willigan, R.R.(2008) Design of water gas shift catalysts for hydrogen production in fuelprocesses. J. Physics Condensed Matter, 20(6), article no. 064237 (in press).
[218] Patt, J., Moon, D.J., Phillips, C., Thompson, L. (2000) Molybdenum carbidecatalysts for water gas shift. Catal. Lett., 65, 193–195.
[219] Moon, D.J., Ryu, J.W. (2004) Molybdenum carbide catalysts for water gas shiftfor fuel cell powered vehicles application. Catal. Lett., 92,17–24.
[220] Tominaga, H., Nagai, M. (2005) DFT of water gas shift reaction on molybdenumcarbide. J. Phys. Chem., B, 109, 20415–20423.
[221] Nagai, M., Matsuda, K. (2006) Low temperature water gas shift reaction overcobalt-molybdenum carbide catalysts. J. Catal., 238, 489–496.
[222] Chapman, T. (2002) Physics World, July, p. 152.
[223] Jacobs, G., Davis, B.H. (2007) Low temperature water gas shift catalysts.Catalysis, 20, 122–185.
[224] Knudsen, J., Nilekar, A.U., Vang, R.T., Schnadt, J., Kinkes, E.L., Dumesic, J.A.,Mavrikakis, M., and Basenbacher, F. A Cu/Pt near-surface alloy for water gasshift catalysis. J. Amer. Chem. Soc., 129, 6485–6490.
[225] Wagner, J.P., Cai, Y., Wagner, A.L. (2003) U.S. Patent 2003/0186804 A1.Catalyst for production of hydrogen.
440 Ratnasamy and Wagner
Dow
nloa
ded
by [
Uni
vers
ity o
f U
tah]
at 2
3:18
12
Apr
il 20
13