Low temperature water-gas shift: Examining the efficiency of Au as a promoter for ceria-based...

www.elsevier.com/locate/apcata

Applied Catalysis A: General 292 (2005) 229–243

Low temperature water-gas shift: Examining the efficiency

of Au as a promoter for ceria-based catalysts prepared

by CVD of a Au precursor

Gary Jacobs a, Sandrine Ricote a, Patricia M. Patterson a, Uschi M. Graham a,Alan Dozier b, Syed Khalid c, Elin Rhodus d, Burtron H. Davis a,*

a Center for Applied Energy Research, 2540 Research Park Drive, Lexington, KY 40511, USAb University of Kentucky Electron Microscopy Center, Chemical and Materials Engineering Department,

A004 ASTeCC Building 0286, Lexington, KY 40506, USAc National Synchrotron Light Source, Brookhaven National Laboratory, Upton, NY 11973, USA

d EBCE Program, Lafayette High School, 400 Reed Lane, Lexington, KY 40503, USA

Received 11 April 2005; received in revised form 3 June 2005; accepted 12 June 2005

Abstract

Increasing the Au loading had a significant positive impact on the catalytic activity. The partial reduction of ceria is necessary for

generating bridging OH groups on the surface of ceria, which serve as the active sites. The surface shell reduction process in H2 was monitored

by TPR, XANES, and in situ DRIFTS spectroscopy. Either the oxygen deficiencies are first formed, followed by dissociative adsorption of

H2O to generate the Type II bridging OH groups or, they may be formed directly by spillover of dissociated H2 from the metal to the ceria

surface. For each pair of bridging OH groups formed, two cerium atoms in the surface shell change from the Ce4+ to Ce3+ oxidation state.

Addition of Au facilitates the surface reduction process and thereby decreases the reduction temperature from 450 8C for the unpromoted

catalyst to 100 8C for the 5% Au/ceria catalyst. A systematic decrease in the required temperature for ceria surface shell reduction was

observed by increasing the Au promoter loading as follows: 0.1, 0.25, 0.5, 1, 2.5, 5%.

Gold and platinum promoted catalysts were compared in a suitable reaction temperature range after first ensuring that metal–oxide

interactions were overcome (ca. 5% metal loading). Approximately 20 times the amount of 5% Au/ceria catalyst was required to achieve a

lightoff curve similar to that of a 5% Pt/ceria in the temperature range 200–300 8C, and 5% Pt/ceria also exhibited higher steady-state activity

(about double that of Au) at 175 8C. This result suggests that, in addition to the role the metal plays in facilitating the formation of the active

site bridging OH groups, it also influences the intrinsic rate of the WGS reaction. Transient formate decomposition experiments carried out at

140 8C indicated that the rate of formate decomposition was approximately 20 times higher for 2.5% Pt/ceria than that of 2.5% Au/ceria,

suggesting that the metal (in addition to the promoting effect of H2O previously reported) plays a role in facilitating the decomposition of

surface formate intermediates, the proposed rate limiting step of the reaction mechanism.

# 2005 Published by Elsevier B.V.

Keywords: Au loading; Au/ceria catalyst; Pt/ceria catalyst

1. Introduction

The implementation of PEM fuel cells will require very

active and robust fuel processing catalysts for such reactions

as hydrocarbon steam reforming, low temperature water-gas

shift, and preferential oxidation [1–3]. The latter two reactions

* Corresponding author. Tel.: +1 606 257 0253; fax: +1 606 257 0302.

E-mail address: [email protected] (B.H. Davis).

0926-860X/$ – see front matter # 2005 Published by Elsevier B.V.

doi:10.1016/j.apcata.2005.06.003

are receiving widespread interest from catalyst developers for

the conversion of CO, which acts as a poison for fuel cell

electrode catalysts [4]. This study centers on comparing Au

relative to Pt, as a promoter of ceria-based low temperature

water-gas shift catalysts, as it has been claimed to be an

excellent water-gas shift catalyst [5–11].

There is much controversy in the literature regarding how

Au and Pt promoted ceria catalysts operate for water-gas

shift, and this probably stems in part from: (1) differences in

G. Jacobs et al. / Applied Catalysis A: General 292 (2005) 229–243230

Scheme 1. Surface formate mechanism.

Scheme 2. Ceria-mediated redox process.

feed conditions, as some researchers continue to utilize a

high temperature shift feed for low temperature shift studies;

(2) the absence of H2O, H2, or CO2 in many mechanistic

evaluations; and (3) lack of verification to determine

whether or not the surface of a particular catalyst is in an

active state.

Regarding the first point, the feed utilized in this study

is considered to approximate the exit stream of a high

temperature shift section of a fuel processor, with the caveat

that we do not include CO2 in this particular study. The reason

is that we are attempting to construct a mechanistic picture

for these catalysts on an incremental basis. The feed that we

have chosen includes: 3.75 cm3/min CO, 125 cm3/min H2O

(steam), 100 cm3/min H2, and a small amount of N2 for the

purpose of internal calibration. The feed matches that of our

previous study of the impact of Pt loading on low temperature

water-gas shift activity [12].

Considering the second point, H2O has been found in

many studies to promote the rate of the water-gas shift

reaction not only for metal promoted ceria systems [13–15],

but also for other common active materials, including ZnO

[16], MgO [17,18], ThO2 [19], and ZrO2 [18] containing

catalysts. Unfortunately, many mechanistic studies based on

IR spectroscopy have neglected the promoting effect of H2O

entirely. This is a critical point, especially considering that

the feed ratio for low temperature water-gas shift has a very

high H2O/CO ratio, and this is considered in Section 4. The

impact of CO2 to the rate also should be considered, since

excellent water-gas shift catalysts when excluding CO2 in

the feed can be inhibited when CO2 is present [20], leading

to potentially unfair comparisons among catalysts. Finally,

as low-temperature shift is essentially a hydrogen purifica-

tion step, adequate amounts of hydrogen were included in

the feed.

The final point also merits a brief discussion. It is

important for researchers comparing catalysts to ensure

that the catalyst is tested in a range of temperature that

ensures a suitably active surface is present. Most

researchers agree that defects (i.e., oxygen deficiencies)

in the surface shell of ceria play an important role in the

water-gas shift mechanism. Yet, it is necessary to use

techniques, like TPR or XANES, to make certain that the

defect-surface is generated. For example, in examining our

previous Pt loading study [12], 1% Pt/ceria is activated by

250 8C, while 5% Pt/ceria is activated at much lower

temperature (i.e., ca. 50 8C lower temperature). Therefore,

to benchmark Au, which typically is activated at a lower

temperature than Pt on an equivalent loading basis [21], to

Pt in a range where the Pt catalyst is not yet in an activated

state, should be avoided.

There are numerous inconsistencies in the literature

regarding the WGS mechanism over metal promoted ceria

catalysts, as well as the nature of the active sites involved.

Our results continue to support the initial surface formate

mechanism of Shido and Iwasawa [13] for Rh/ceria, which is

not unlike a mechanism considered by Tamaru [22] for ZnO-

based catalysts, and stems from the now classic studies of

formic acid decomposition [23]. In this mechanism [13],

depicted in Scheme 1, Type II bridging OH groups, defined

in [24,25], are the active sites and react with CO to generate

surface formate intermediates. The decomposition of

surface formates to unidentate carbonate is proposed to

be promoted by steam (leading to the inclusion of H2O in the

transition state complex by Shido and Iwasawa [13]) and is

considered to be the rate limiting step of the mechanism. On

the other hand, a support-mediated redox process has also

been proposed for precious metal loaded ceria catalysts (e.g.,

for Pd) [26,27], shown in Scheme 2, whereby adsorbed CO

on the metal promoter reacts with ceria to generate the

cerous oxide form and CO2, with H2O replenishing the

oxygen vacancy with liberation of H2. One basis for this

mechanism was the fact that the group [26] did not observe

surface formate species in their FTIR studies with Pd/ceria.

This mechanism, initially found support by another group to

describe the catalysis of Au/ceria [7], but their viewpoint has

recently changed to favor a mechanism relying on non-

metallic Au and Pt species in close association with ceria as

the active sites for the WGS reaction [5,6,8]. A basis for this

modification was that after a large fraction of metal promoter

was leached from the catalyst, the catalyst still retained its

activity. In this new viewpoint, one can infer that the metal is

proposed not to play an important role in the catalytic cycle.

Others support the view that the active sites are solely

associated with Au, in line with Haruta’s initial findings that

Au particles smaller than ca. 5 nm possessed intrinsic

activity for the CO oxidation reaction [28,29]. Recent work

by Haruta and co-workers, does, however, support a WGS

mechanism operating via formate intermediates [30].

G. Jacobs et al. / Applied Catalysis A: General 292 (2005) 229–243 231

In this contribution, the aim is to explore the efficiency of

Au as a promoter for low temperature water-gas shift relative

to its neighbor, Pt, and to make comparisons under

conditions where the defect concentrations in ceria are

similar, in order to shed light as to whether or not the metal is

involved in the actual catalytic mechanism. As such, we rely

on techniques such as TPR and XANES to make

determinations regarding the required temperature for,

and extent of, activation as well as reaction testing and in

situ DRIFTS spectroscopy to shed light onto the effective-

ness of each promoter. XANES is fast becoming a preferred

method for quantifying oxidation state changes for ceria

catalysts [31]. To ensure a small Au crystallite size, we

utilize a CVD technique previously employed for the

preparation of effective Pt dehydrocyclization catalysts [32].

Benefits of the technique are that small, uniform, well

dispersed, nano-crystallites are formed at typical low

promoter loadings, and that the precursor is virtually

completely loaded onto the substrate. It is very important to

note that this procedure is different from the deposition-

precipitation techniques used by other researchers [5–11].

2. Experimental

2.1. Catalyst preparation

High surface area ceria was prepared via homogeneous

precipitation of the nitrate in urea with aqueous ammonia in

a procedure similar to Li et al. [33], whereby urea

decomposition is a slow process resulting in a more

homogeneous precipitation. Appropriate amounts of Ce(N-

O3)3�6H2O (Alfa Aesar, 99.5%) and urea (Alfa Aesar,

99.5%) were dissolved in 900 cm3 of deionized water, and to

this solution about 30 cm3 NH4OH (Alfa Aesar, 28–30%

NH3) was added drop wise (�1 cm3/min). The mixture was

then heated at 100 8C with constant stirring for 8 h. The

precipitate was filtered, washed with 600 cm3 of boiling

deionized water, and dried in an oven at 110 8C overnight.

The dried precipitate was then crushed and calcined in a

muffle furnace at 400 8C for 4 h.

Dimethyl(acetylacetonate gold III) was purchased from

Strem Chemicals (stock no. 79-1500). The precursor will

hereafter be referred to as AuAcAc, and was successfully

utilized previously in a vapor phase grafting procedure to

prepare nano-crystalline Au clusters [34]. The cerium oxide

(BET SA, 120 m2/g) was heated under high vacuum

(1 � 10�7 Torr) at 350 8C to drive water out of the pores

and subsequently cooled to room temperature. The sample

tube was backfilled with nitrogen and transferred to an inert

atmosphere, where the ceria was well mixed with the gold

precursor. A sample of the physical mixture was retrieved

for FTIR analysis. The physical mixture was loaded into a

sample tube and attached to the high vacuum line. The

sample was heated in stages (1 8C/min) to 80 8C, and finally,

100 8C with 15 min ramps between each temperature and

holding for 1 h at each temperature. Finally, the catalyst was

ramped to 130 8C, held for 15 min, and the catalyst was then

quenched to room temperature. The ceria is yellow in color,

but the sample retrieved after the sublimation procedure (for

the 1% Au catalyst) had a slightly orange, highly uniform

color. This sample was utilized for FTIR study of the

precursor decomposition, as well as the catalyst activation

step. Based on these experiments, a calcination temperature

of 250 8C in O2 was selected, while a temperature of 175 8Cin H2 was chosen as the initial temperature for catalyst

activation. The calcined catalysts ranged in color from a very

light, grayish-purple to dark black.

2.2. BET surface area

BET surface area measurements were carried out using a

Micromeritics Tristar 3000 gas adsorption analyzer. In each

trial, a weight of approximately 0.25 g of sample was used.

Nitrogen adsorption was carried out at its boiling temperature.

2.3. Temperature programmed reduction (TPR)

TPR was conducted on unpromoted and gold promoted

ceria catalysts in a Zeton-Altamira AMI-200 unit, which was

equipped with a thermal conductivity detector (TCD).

Argon was used as the reference gas, and 10% H2 (balance

Ar) was flowed at 30 cm3/min as the temperature was

increased from 50 to 1100 8C at a ramp rate of 10 8C/min.

2.4. X-ray absorption near edge spectroscopy (XANES)

XANES spectra at the Au LIII (11,919 eV) and Ce LIII

(5723 eV) edges were recorded at the National Synchrotron

Light Source (NSLS) at Brookhaven National Laboratory

(BNL) in Upton, New York at Beamline X-18b. Experi-

mental details for the beamline and the sample preparation

are provided in our previous investigation of Pt/ceria

catalysts [12]. After purging the cell for a long duration of

time with a high flow rate of helium to ensure removal of air,

the samples were treated in situ at the beamline with a

hydrogen/helium mixture (60 cm3/min H2 and 300 cm3/min

He) while heating at 10 8C/min in the temperature range 50–

300 8C. Scans were obtained in the transmission mode at

50 8C intervals to explore the partial reduction of ceria and

the reduction of the gold promoter. The UHP H2 and He

gases were mixed in a manifold and passed through an

oxygen/water trap prior to feeding to the cell. Raw data were

processed to give the normalized XANES spectra. Linear

combination XANES fits of treated catalysts with reference

standards were carried out using the WinXAS program [35].

2.5. High resolution transmission electron microscopy

(HRTEM)

High resolution transmission electron microscopy

(HRTEM) measurements were carried out using a JEOL


Fig. 1. IR spectra of: (a) physical mixture of dimethylAu(acac) and ceria

and (b) catalyst after sublimation and deposition of dimethylAu(acac) onto

ceria surface. Results indicate that the gold compound remained intact.

2010 TEM field emission electron microscope, equipped

with an energy dispersive X-ray (EDX) detector, and

operated at an accelerating voltage of 200 kV. Furthermore,

Emisec Control is used for digital beam control and the

integration of STEM (scanning transmission) images. The

electron beam has a point-to-point resolution of 0.5 nm.

Prior to HRTEM analysis the Au promoted samples were

reduced ex situ in flowing hydrogen at 300 8C and

subsequently passivated at room temperature. Catalyst

powder was lightly dusted onto 200 mesh Cu grids coated

with lacy carbon. The higher temperature condition was

chosen to ensure that the low loading catalysts (e.g., 0.1

wt.% Au) were activated; that is, that the temperature was

high enough to overcome the Au-ceria interactions that are

described later.

2.6. Diffuse reflectance infrared Fourier transform

spectroscopy (DRIFTS)

A Nicolet Nexus 870 was used, equipped with a DTGS-

TEC detector. A high pressure/high temperature chamber

fitted with ZnSe windows was utilized as the WGS reactor

for in situ reaction measurements. The gas lines leading to

and from the reactor were heat traced, insulated with

ceramic fiber tape, and covered with general purpose

insulating wrap. Scans were taken at a resolution of 4 to give

a data spacing of 1.928 cm�1. Typically, 64–256 scans were

taken to improve the signal to noise ratio. The sample

amount utilized was 33 mg.

A steam generator consisted of a downflow tube packed

with quartz beads and quartz wool heated by a ceramic oven

and equipped with an internal thermocouple. The lines after

the steam addition were heat traced. The steam generator

and lines were run at the same temperature as that of the in

situ sample holder of the DRIFTS cell. Water was pumped

by a precision ISCO Model 500D syringe pump into a steam

generator via a thin needle welded to a 1.6 mm line.

Feed gases (UHP) were controlled by using Brooks 5850

series E mass flow controllers. Iron carbonyl traps,

consisting of lead oxide on alumina (Calsicat), were placed

on the CO gas line. All gas lines were filtered with Supelco

O2/moisture traps. During CO adsorption, the flows were

maintained at 3.75 cm3/min CO and 135 cm3/min N2.

During formate decomposition experiments, 125 cm3/min

of H2O (i.e., steam) was utilized with 100 cm3/min of N2.

2.7. Testing in a fixed bed reactor

Steady-state CO conversion measurements were con-

ducted in a fixed bed reactor consisting of a 0.5 in. stainless

steel tube with an internal thermocouple. Experiments were

conducted at two different space velocities using either

33 mg of catalyst diluted to 0.4 g with silica (high space

velocity tests) or 660 mg of catalyst diluted to 1.5 g with

silica (low space velocity tests). The catalyst bed was

supported on glass wool. The description of the steam

generator, gas delivery system, and ancillary equipment is

provided in Section 2.6. As in our previous study of Pt/ceria

[12], conditions were chosen to mimic those of the low

temperature shift reactor of a fuel processor, with the ex-

ception that CO2 was not included in the tests. The gas flows

were 3.75 cm3/min CO, 125 cm3/min H2O, 100 cm3/min H2,

and 10 cm3/min of N2. Catalysts were activated in H2

(100 cm3/min) prior to reaction testing at the temperature of

interest over the range 175 8C (initial) to 300 8C (final).

3. Results

3.1. Selection of catalyst pretreatment and activation

conditions

First, it was important to assess whether the AuAcAc

compound decomposed during the sublimation step under

high vacuum and heating at a maximum temperature of

130 8C. Therefore, after physically mixing the AuAcAc with

ceria in an inert atmosphere, a sample was taken for infrared

analysis. Likewise, a sample was taken of the catalyst material

after the sublimation procedure. Fig. 1 shows absorbance

spectra for the two catalysts referenced to unpromoted ceria.

The presence of C–H bands 2800–3000 cm�1 range and a

complex series of bands in the 1000–1600 cm�1 provide

evidence of the AuAcAc compound [34]. The color of the

precursor is purplish. The presence of AuAcAc after

sublimation lent a lightly orangish, yet uniform, hue to the

ceria, which was previously yellow in color. By comparison

of the catalyst after sublimation of AuAcAc with the

references of the physical mixture of ceria and AuAcAc, it

is suggested from Fig. 1 that the sublimation procedure did not

cause the AuAcAc to decompose, rather only to disperse.


Secondly, it was appropriate to assess the decomposition

temperature of AuAcAc, the decomposition temperature of

ceria surface carbonates, and the activation temperature of

the Type II bridging OH groups [24,25], the proposed active

sites [13], following pretreatments for the catalyst, the latter

two steps being important during the ceria surface shell

reduction step in hydrogen [12]. These again were

monitored by infrared spectroscopy.

In previous work with Pt vapor phase impregnation (VPI)

catalysts [32], calcination in air, followed by reduction in H2

led to finely dispersed Pt clusters. Therefore, O2 treatment

was followed up to 250 8C. Previous Au catalysts were

found to promote surface ceria reduction in hydrogen flow at

between 100 and 200 8C [21,36]. Therefore, after O2

calcination at 250 8C, the catalyst was cooled to 175 8C in a

N2 purge, and a H2 treatment was carried out. Finally, with a

separate sample, we conducted direct treatment with H2 to

carry out the AuAcAc decomposition and surface ceria shell

activation step, foregoing calcination in O2.

Fig. 2 provides a picture for the role of Au. On the left,

unpromoted ceria displays a series of absorption bands in the

1200–1700 cm�1 range. These are due in part to a variety of

carbonates which cover the surface of the material [24]. A

band at about 1630 cm�1 is likely due to the bending mode

of adsorbed H2O. After O2 treatment at 250 8C, followed by

H2 treatment at 175 8C, the carbonates [7,21,36], however,

remain close to saturation on the surface for the unpromoted

ceria sample. However, on the right, Au promotes the

decomposition of the carbonate (CO2 was observed to be

liberated during treatment) and Type II OH groups are

generated on the surface. A small band at 2120 cm�1 also

developed during catalyst activation; this band is always

observed during H2 treatment and carbonate decomposition

and is an important indicator of ceria surface shell reduction

[25,37]. In one viewpoint the band is assigned to Ce3+–CO,

Fig. 2. Residual carbonate region and hydroxyl group region for: (left)

unpromoted catalyst (solid) at room temperature in N2 flow and (dashed)

after O2 treatment at 250 8C followed by H2 treatment at 175 8C; and (right)

1% Au/ceria catalyst after decomposition of Au precursor and catalyst

activation by (solid) O2 treatment at 250 8C followed by H2 treatment at

175 8C or (dashed) after direct H2 treatment at 175 8C (no calcination).

formed during surface carbonate decomposition [35], while

in another view it is due to subsurface Ce3+ ions partly

surrounded by oxygen vacancies [25]. The presence of the

2120 cm�1 band at low temperature is indicative of the role

Au plays in promoting ceria surface shell reduction. It is

important to note also that the bridging OH groups are also

indicative of partially reduced ceria, as the Ce atoms

associated with them must be in the Ce3+ oxidation state

[12]. The fact that the bridging OH groups are generated at a

low temperature in the presence of Au has led us to postulate

two routes for the Type II bridging OH group formation, as

previously suggested for Pt [12]:

As the carbonates need to be removed from the surface of

the oxide prior to the formation of the Type II bridging OH

groups, there is an important inverse relationship between

surface residual carbonate and bridging OH band intensities.

Since the residual carbonate band intensities are smaller for

the O2 treated catalysts followed by H2 treatment relative to

the bands remaining after direct H2 treatment (Fig. 2), we

decided to include the low temperature O2 calcination step in

the catalyst pretreatment.

3.2. Standard characterization

Table 1 provides the results of nitrogen physisorption.

The BET surface area of the starting unpromoted ceria

material was 121 m2/g. After loading Au and following O2

calcination, the surface areas of all the catalysts were

between 111 and 120 m2/g. There was a slight increase in the

average pore radius from 1.43 to 1.47 nm for the more

heavily loaded Au catalysts, probably attributed to blocking

of the narrowest pores by the metal.

Temperature programmed reduction profiles of the

unpromoted and Au-promoted ceria samples are displayed

Table 1

BET surface area and porosity measurements from physisorption of nitro-

gen at 77 K

Sample name BET SA

(m2/g)

Pore volume

(cm3/g)

Average pore

radius (nm)

Unpromoted ceria 121 0.087 1.43

0.1% Au/ceria 120 0.086 1.43

0.25% Au/ceria 118 0.084 1.43

0.50% Au/ceria 118 0.085 1.43

1.0% Au/ceria 119 0.085 1.42

2.5% Au/ceria 116 0.085 1.47

5.0% Au/ceria 111 0.082 1.47


Fig. 3. TPR profiles of, moving upward: (a) unpromoted ceria; (b) 0.1% Au/

ceria; (c) 0.25% Au/ceria; (d) 0.5% Au/ceria; (e) 1.0% Au/ceria; (f) 2.5%

Au/ceria; and (g) 5.0% Au/ceria.

in Fig. 3. High surface area ceria materials usually have two

distinct features inhydrogenTPR,apeakclose to750 8Cthat is

assigned to the bulk reduction from CeO2 to Ce2O3, and a

broader peak situated close to 450 8C that is typically assigned

to a surface reduction process. As has been noted in previous

work [7,12,13,21,33,36], addition of metals can catalyze the

surface reduction process, shifting the broad peak to lower

temperatures and sharpening its features, while not impacting

bulk ceria reduction. Yao and Yao [38] indicated that surface

reduction probably involves the removal of surface capping

oxygen atoms from ceria. However, it has also been reported

that bridging-type OH groups are formed after a hydrogen

surface reduction process not only for metal/ceria catalysts,

but for other oxides in which partial reduction has been

claimed, including zirconia-based [18,39] and thoria-based

[19,40–42] catalysts. In these cases, noble metal addition has

also been found to facilitate bridging OH group formation

[18,19]. It is noted that the partial reduction property of

zirconia, facilitated by a metal promoter, has been strongly

suggested to be important in the dry reforming of methane

reaction [43,44].

3.3. X-ray absorption near edge spectroscopy (XANES)

To directly monitor the reduction of the catalyst in H2, in

situ XANES was employed, as the white line provides

valuable information about the density of unoccupied states

above the Fermi level. LIII edges of both Au and Ce were

scanned, the transitions being from 2p3/2 core states to

unoccupied 5d levels.

In the case of Au (LIII = 11,919 eV), the reference spectra

are shown in Fig. 4a, and the white line indicates that an

oxidized form of Au predominates for the O2 calcined

catalyst. Fig. 4b and c compare the changes during reduction

in H2 to the Au white line of the Au/ceria catalysts with 1 and

5 wt.% Au loading. The more heavily loaded 5% Au catalyst

offered slightly more facile reduction of the Au. Au

reduction is nearly complete at 100 8C (there is a slight

presence of an oxidized state) and certainly completed by

150 8C for the 1.0% Au loaded sample, while Au reduction

is complete by 100 8C for 5.0% Au/ceria. The results are

consistent with a small but detectable metal–support

interaction between Au and ceria, which hinders Au

reduction at low loadings, where interactions between

metal and ceria will be the greatest.

The effectiveness of Au in promoting reduction of the

ceria surface shell to produce the active bridging OH groups

is also important, as formation of the Type II bridging OH

groups is a surface reduction process, accompanied by a

change in the oxidation state of the Ce atoms involved with

the bridging OH group sites from Ce4+ to Ce3+ (i.e., partial

reduction of ceria). Therefore, the Ce LIII edge

(LIII = 5723 eV) was also examined for each catalyst over

a series of reduction temperatures in flowing hydrogen. As

displayed in Fig. 5A, the XANES lineshapes of Ce3+ and

Ce4+ oxidation states are very different. Peak C for Ce4+ is

usually assigned to absorption into the 5d level where there

is no occupancy in 4f for either the initial or final state and is

used to detect the presence of completely oxidized CeO2; it

is absent for the Ce3+ oxidation state [45,46]. The final state

configuration for the transition is normally written Ce

[2p54f05d1] O [2p6]. The peak identified as B is split into at

least two separate assignments [45,46]. Peak B1, also present

in the CeO2 sample (i.e., Ce4+), has been assigned to

absorption into the 4f level in the final state. That is, in

addition to an electron excited from the Ce 2p shell to the 5d

shell, another electron is excited, coming from the valence

band (i.e., the O 2p shell) to the Ce 4f shell, leaving a hole in

the valence band. The simplest description for the final state

configuration associated with B1 is normally written Ce

[2p54f15d1] O [2p5]. During reduction, peaks B1 and C

decrease and a new peak, B0, develops, occurring just below

that of B1, and associated with absorption into the 5d level,

with 4f occupancy in the initial state. Its final state

configuration may be written Ce [2p54f15d1] O [2p6]. The

intensity of B0 is very high for any Ce3+ reference, with one

example (cerous carbonate) provided in Fig. 5A. The

importance of these identifying features (B0 for Ce3+ and

both B1 and C for Ce4+) is that they can be used as a

fingerprint to obtain insight into the extent of reduction of

ceria during oxidative/reduction treatments.

The degree of reduction of ceria was quantified by

carrying out a linear combination fitting of XANES spectra

for treated catalysts with WinXAS [47] using reference

compounds for Ce3+ and Ce4+ oxidation states, between the

range 5.70 and 5.77 keV. This technique has generally been


Fig. 4. XANES spectra at Au LIII edge for: (a) Au reference compounds; (b) 1% Au/ceria; and (c) 5.0% Au/ceria ranging from 50 to 150 8C in hydrogen.

found to be more consistent than alternate methods, like

XPS [31]. Results of the fitting procedure are reported in

Table 2. As displayed in Fig. 5B, for unpromoted ceria, the

ceria remains relatively unreduced with H2 treatment at

300 8C, and only about 3% of cerium atoms are in the Ce3+

oxidation state. However, with just 0.1% Au, changes are

observed consistent with enhanced partial reduction of ceria,

with 10.8% Ce3+ detected by 300 8C (Fig. 5C). With

increasing Au promoter loading, the reduction of the ceria

surface shell is further facilitated (i.e., shifts to lower

temperature). For 0.25% Au, 21.1% Ce3+ is present at

300 8C (Fig. 5D). An upper limit of about 1/4 of Ce atoms in

the 3+ oxidation state is obtained with 0.5% Au at 300 8C

(Fig. 5E). Above that loading, this limit is obtained at lower

and lower temperatures, with approximately 1/4 of the atoms

in the Ce3+ oxidation state at 250 8C for 1.0% Au, 200 8C for

2.5% Au, and 150 8C for the 5.0% Au loading (Fig. 5F–H,

respectively).

The results are consistent with the hypothesis that

enhanced partial reduction of ceria is attributed to the

promotion by Au. If one examines the XANES spectra of the

Au and Ce edges closely for each of the catalysts, partial

reduction of ceria is not facilitated until Au0 is present, and

we therefore suggest that the partial reduction of ceria may

be due to the result of the spillover of dissociated H2 from

Au0 to the surface of ceria, although one cannot rule out a


Fig. 5. XANES spectra at Ce LIII edge for: (A) references and, the reduction of Au/ceria catalysts, including (B) unpromoted, (C) 0.1% Au/ceria, and (D) 0.25%

Au/ceria. Note that the shifts in energy are not real. The spectra are staggered for ease of viewing the lineshape differences. XANES spectra at Ce LIII edge for

the reduction of Au/ceria catalysts continued, including (E) 0.5% Au/ceria, (F) 1.0% Au/ceria, (G) 2.5% Au/ceria and (H) 5.0% Au/ceria.

chemical effect, whereby O removal is facilitated by the

presence of the metal, allowing for H2O to dissociate at the

vacancy. The former would allow the bridging OH groups to

be formed directly, although both processes must be

accompanied by reduction of the cerium ions in the surface

shell from Ce4+ to Ce3+, as discussed previously, and similar

to what we proposed for Pt/ceria [12].

3.4. High resolution transmission electron microscopy

(HRTEM)

Fig. 6 presents STEM images with a 5 nm scale for most

of the Au/ceria samples after reduction in H2 at 300 8C,

including 0.25% Au, 0.5% Au, 1% Au, 2.5% Au, and 5.0%

Au. The ceria agglomerates are composed of 5–8 nm

diameter ceria particles which exhibit a strong tendency to

align and form elongated ridges and channels, resulting in

what appears to be an almost corrugated surface. The Au

nanoparticles are clearly visible in Fig. 6 (their presence was

confirmed by an EDX analysis). For the 0.25% Au and 0.5%

Au samples, very small, well-dispersed Au crystallites are

observed, mainly in the 1–2 nm diameter and 1–3 nm

diameter range, respectively. For the 1% Au/ceria catalyst,

some larger clusters in the 5–8 nm range were also found (in

addition to the smaller crystallites). For the 2.5% Au/ceria

and 5.0% Au/ceria catalysts, a heterogeneous distribution


Fig. 5. (Continued ).

Table 2

Linear combination fits of XANES spectra with the Ce3+ and Ce4+ reference compounds

T (o.c.) No Au 0.1% Au 0.25% Au 0.5% Au 1.0% Au 2.5% Au 5.0% Au

Ce4+ Ce3+ Ce4+ Ce3+ Ce4+ Ce3+ Ce4+ Ce3+ Ce4+ Ce3+ Ce4+ Ce3+ Ce4+ Ce3+

50 100 0 99.3 0.7 99.5 0.5 99.2 0.8 99.6 0.4 99.5 0.5 99.3 0.7

100 – – 98.4 1.6 99.3 0.7 97.9 2.1 99.2 0.8 98.7 1.3 81.2 18.8

150 – – 97.9 2.1 93.6 6.4 86.8 13.2 87.9 12.1 79.1 20.9 76.1 23.9

200 – – 94.7 5.3 86.9 13.1 78.7 21.3 78.2 21.8 76.5 23.5 74.4 25.6

250 – – 92.1 7.9 81.8 18.2 77.0 23.0 76.5 23.5 75.1 24.9 73.7 26.3

300 97.0 3.0 89.2 10.8 78.9 21.1 74.7 25.3 75.8 24.2 75.2 24.8 73.3 26.6

Relative percentages as a function of reduction temperature. Reduction carried out in hydrogen.


Fig. 6. STEM images of the Au/CeO2 catalysts after O2 calcination at 250 8C and reduction in H2 at 300 8C.

was obtained. While 1–5 nm Au crystallites are found, there

are also much larger particles (10–20 nm for 2.5% Au and up

to 30 nm at 5.0% Au loading).

3.5. Catalytic activity tests

Catalyst activity measurements were conducted at two

space velocities, first at a high space velocity in order to

compare with our most active Pt/ceria catalyst with a loading

Fig. 7. High space velocity condition: comparison of CO conversion

between (filled circles) 5% Au/ceria and (open circles) 5% Pt/ceria

(33 mg catalyst, 3.75 cm3/min CO:125 cm3/min H2O:100 cm3/min

H2:10 cm3/min N2).

of 5 wt.% Pt [12]. As shown in Fig. 7, the activity of the 5%

Pt/ceria catalyst far exceeds that of the 5% Au/ceria catalyst.

In order to obtain a lightoff curve for comparison among the

Au loaded catalysts, we lowered the space velocity by a

factor of 20. Fig. 8 shows that there is a consistent increase in

Fig. 8. Low space velocity condition: CO conversion as a function of

temperature for different Au loadings benchmarked to Pt/ceria (660 mg

catalyst, 3.75 cm3/min CO:125 cm3/min H2O:100 cm3/min H2:10 cm3/min

N2) for moving upward: (X) unpromoted ceria; (squares) 0.1% Au/ceria;

(upright triangles) 0.25% Au/ceria; (inverted triangles) 0.5% Au/ceria;

(diamonds) 1.0% Au/ceria; (crossmarks) 2.5% Au/ceria; (filled circles)

5.0% Au/ceria; and (open circles) 5.0% Pt/ceria. Data obtained from fixed

bed reactor testing.


the catalyst activity as a function of the Au loading. At

175 8C, the 5% Pt catalyst exhibited CO conversion

at slightly more than double that of the 5% Au loaded

catalyst.

3.6. Assessment of the turnover rate of surface formate

In contrast to the recently proposed ceria-mediated redox

mechanism, some favor a mechanistic picture that relies on

the decomposition of surface formate intermediates as the

rate limiting step for water-gas shift over metal promoted

ceria catalysts [13,48,49]. Therefore, we selected two

catalysts with high metal loading, but that still provided a

good signal in infrared spectroscopy—2.5 wt.% Au and

2.5 wt.% Pt-loaded catalysts for these studies. The catalysts

were reduced at 225 8C in H2, cooled to 140 8C, and purged

with N2 prior to CO adsorption. It is important to note that

the Au/ceria catalyst had slightly higher intensity bands for

residual carbonate on the surface, as we have observed in a

previous study [21,36] (Fig. 9). The positions of these bands

indicate that this residual carbonate is likely present on the

partially reduced ceria surface [24]. The bridging OH group

bands, with an identifying band situated at ca. 3650 cm�1,

were slightly lower in intensity for the Au catalyst relative to

Pt, indicating a lower active site density. Formate bands

(Fig. 10) arose from the reaction of CO with the Type II

bridging OH groups [13,48,49], and their intensities provide

Fig. 9. Comparison of residual carbonates on 2.5% loaded Au and Pt/ceria

after H2 reduction at 225 8C.

Fig. 10. DRIFTS spectra of catalysts after reduction in H2 at 225 8C, cooled

to 140 8C, and purged in N2, including (bottom) 2.5% Pt/ceria after (solid

line) CO adsorption (3.75 cm3/min CO:135 cm3/min N2) followed by

(dotted line) formate decomposition in H2O:N2 (125 cm3/min

H2O:100 cm3/min N2) for 6 min, purged in N2; and (top) 2.5% Au/ceria

after (solid line) CO adsorption (3.75 cm3/min CO:135 cm3/min N2) fol-

lowed by (dotted line) formate decomposition in H2O:N2 (125 cm3/min

H2O:100 cm3/min N2) for 40 min, purged in N2.

a second important relative measurement of the active sites.

Formate bands included the C–H stretching bands at ca.

2950 and 2850 cm�1, with OCO asymmetric and symmetric

stretching bands at 1580 cm�1 and ca. 1290–1340 cm�1,

respectively [50]. The C–H band intensity at 2850 cm�1 was

about 1.3 times higher for the Pt/ceria catalyst than that of

Au, suggesting a slightly higher active site density, in

agreement with the earlier statements regarding the relative

bridging OH and residual carbonate intensities, and

consistent with our earlier findings [36]. Nevertheless,

although we have reported lower activity for 1% Au/ceria

relative to Pt before [11], during a rapid screening of

promoters, results which are consistent with the current


investigation, it is apparent that the difference in relative

active site density cannot adequately account for the

significant difference in activity between the two promoters.

Therefore, in addition to the role that the metal promoter

plays in activating the catalyst at lower temperature, it

appears that the metal promoter may also be directly

involved in the mechanism [11,12,36].

Since the forward decomposition of formates rapidly

occurs in the presence of H2O at shift temperatures, the very

low temperature of 140 8C was selected so that the formate

decomposition could be monitored at a measurable rate. The

data in Fig. 10 are consistent with our earlier transient

formate decomposition study [15]. Over 2.5% Pt/ceria, the

formate decomposed to unidentate carbonate (OCO vibra-

tions at ca. 1460 and 1390 cm�1, respectively [50]) rapidly

(in approximately 6 min), and a band for gas phase CO2 was

observed, indicating that some WGS product was liberated

from the catalyst. The intense C–H stretching band of

formate at 2850 cm�1 was used to assess the changes. In

contrast, for the 2.5% Au/ceria catalyst, the formate

decomposition proceeded at a much slower rate, and even

after 40 min, after purging in N2 and using the C–H

stretching band at 2850 cm�1 as a reference, about 63% of

the original formate intensity was still observed.

4. Discussion

In this work, the goal was to fairly compare Au relative to

Pt as a promoter for the low temperature water-gas shift

reaction. As such, conditions are chosen (with the exception

of added CO2) to mimic the feed entering the low

temperature water-gas shift section of a fuel processor for

PEM fuel cell applications. That is, the feed is assumed to be

close to that of the tail gas exiting the high temperature shift

reactor, which is typically operated at a temperature

resulting in close to equilibrium conversion at that higher

reaction temperature [4]. In this case, the feed conditions

contain a very high H2O/CO ratio. An appropriate amount of

hydrogen was also included in the feed. In order to make a

fair comparison of Au with Pt, we therefore include the

5 wt.% Pt-loaded catalyst in the study, because at 5 wt.%

loading, the metal–oxide interactions are overcome by

200 8C. This is supported by a comparison of the TPR and

XANES results with those of our earlier loading study of Pt/

ceria catalysts [12]. That is to say, one should not select a Pt

catalyst with a low Pt loading, and test it in a low

temperature shift range where the catalyst is not in an active

state. For example, a 1 wt.% Pt catalyst is fully activated at

about 250 8C, whereas the 1% Au catalyst is fully activated

at a temperature that is about 50–100 8C lower.

Therefore, on the basis of the 5% metal loading, the first

important finding in this work is that, for the Au/ceria

catalysts prepared by CVD in this work Au was found to be

far less effective than Pt as a promoter for the low

temperature water-gas shift reaction, as only 1/20th of the

amount of Pt/ceria catalyst (33 mg) was needed to achieve a

lightoff curve comparable to that of Au (using 660 mg). This

is in contrast to recent reports that maintain that Au/ceria is

an excellent low temperature water-gas shift catalyst relative

to Pt [5,6,8]. In fairness, however, differences in preparation

procedure may play an important role in determining the

cluster size and/or the nature of the active site, which could

conceivably alter the mechanistic picture (e.g., Au cationic

site [6–8]). Regarding cluster size, for example, Tabakova

et al. reported highly active Au/ceria catalysts with clusters

<1 nm were more active than larger clusters [9]. Rather, we

found that in the range 200–300 8C, Au/ceria is only

marginally active relative to Pt/ceria, in agreement with our

earlier findings [21]. It is only at the 175 8C condition that

Au performs relatively better, yet still maintaining only a

little less than half of the activity of the 5% Pt/ceria catalyst.

It is important to note that low temperature shift catalysts

considered for commercial applications are typically

screened at conditions close to those exhibited by our high

space velocity condition.

A first glance at the TPR and XANES results shows that

Au appears to be somewhat of an anomaly. Au is certainly an

excellent promoter for the reduction of the surface shell of

ceria, as previously observed [7,11,36]. Yet if Au promotes

reduction of ceria, then why are the catalysts so much less

active than Pt/ceria? To obtain answers to these questions, a

characterization of these materials using TPR, in situ

XANES, and in situ FTIR spectroscopy was conducted.

Also, we have good evidence that a role of the metal

promoter is to assist in generating the active sites for WGS at

low temperature. However, a fundamental question that has

remained elusive and requires a reasonable explanation is

whether the metal promoter participates in the WGS

reaction. And, if this is indeed the case, then at which

step in the mechanism does the metal play a role, and exactly

how does it function?

In addressing these questions, however, it is helpful to

cast the answers in terms of a particular mechanism, the two

most often cited being the recently claimed ceria-mediated

redox process [7,26,27] and the earlier, surface formate

mechanism [13,14,48,49]. We select the latter mechanism to

describe our results, as all of our previous findings point to a

mechanism involving Type II bridging OH groups as active

sites [12] and surface formate intermediate decomposition

as the rate limiting step, in agreement with a mechanism like

that proposed by Shido and Iwasawa for Rh/ceria [13].

Arguments to support this mechanism are provided in our

earlier work and those of others, and are therefore not

detailed here. We refer the reader to the TPD and kinetic

studies of Shido and Iwasawa [13,14]; as well as our steady-

state dynamic studies of the formate coverage with respect to

the WGS reaction [48,49]; our steady-state [51] and

transient [15] kinetic isotope effect studies; and most

recently, our steady-state isotopic transient kinetic analyses

utilizing 12C and 13C isotopes in conjunction with in situ

DRIFTS spectroscopy for both the forward shift [52] and


reverse water-gas shift [53] reactions. One interesting note

from the work at hand, however, is the fact that Au/ceria

possesses water-gas shift activity [4–11], and yet there is no

evidence for a Au–CO band by either this contribution or

those of others (e.g., see Fig. 11 in [54]). Recall that it is the

reaction of Au–CO with CeO2 in the Ce4+ oxidation state

that was proposed by proponents of the ceria-mediated

redox process [7] to generate CO2. This observation suggests

that one feature of a redox mechanism is not present.

Returning to the formate perspective, we consider first the

activation of the catalysts in hydrogen. Reduction of the

catalyst does not lead to free oxygen vacancies when H2O or

H2 is present, but rather to the formation of Type II bridging

OH groups, which are strongly related to vacancies, and we

have noted that the Ce atoms involved with either vacancies

or the Type II bridging OH groups must be in the Ce3+

oxidation state (i.e., in the ceria surface shell). They are

produced either via oxygen vacancy formation followed by

dissociation of H2O or via hydrogen dissociation on a metal

promoter and spillover to the oxide surface, as described

previously. Both the TPR profiles and the XANES spectra

provide information on the temperature at which activation

of the surface occurs, as well as the extent of the reduction as

a function of the temperature.

As mentioned earlier, low loadings of Au activate the

reduction process at a temperature that is lower than that of

equivalent Pt-loaded catalysts (e.g., compare with [12]).

Examining results of XANES for 0.5 wt.% metal loadings,

for example [12], the 1/4 ceria reduction threshold is reached

by 300 8C for 0.5% Pt/ceria, while the same extent of

reduction occurs at roughly 200 8C for 0.5% Au/ceria. This

is why a fair comparison of Au and Pt/ceria catalysts should

only be made in a temperature range where the two catalysts

exhibit comparable extents of ceria reduction.

Next consider the metal–oxide interaction. As shown in

both the TPR and XANES studies of Pt/ceria [12] and those

in this investigation of Au/ceria, ceria interactions with the

promoter are important and increasing the loading assists in

decreasing this interaction. This suggests that moving to a

larger cluster size assists in overcoming the interaction. That

is, the larger metal clusters for the more heavily loaded

catalysts are more easily reduced, and once Pt or Au is

reduced, hydrogen can dissociate and spillover to the surface

of ceria, to facilitate reduction from Ce4+ to Ce3+ in the

surface shell by bridging OH group formation. This

explanation is supported by the XANES results at the Au

LIII edge, which show that the gold, appearing close to 1+

oxidation state (by comparison with reference spectra) when

initially loaded into the cell, is virtually completely reduced

to Au0 by 100 8C for the 1 and 5 wt.% Au loaded catalysts,

and this reduction occurs below the temperature at which

major changes in the cerium surface shell reduction occurs

(e.g., in the range of 150–200 8C for 1 wt.% Au/ceria). It is

often noted that Au/ceria catalysts require no activation, but

it is probably more appropriate to say that the WGS feed,

through surface reduction, activates the catalyst, since the

reduction process occurs at such a low temperature. The

results do not support a recent proposal [5,6,8] that active

non-metallic species are responsible for WGS activity,

although differences regarding preparation methods used

among researchers may change the nature of the active site.

For example, in certain cases, Au in a cationic state, strongly

interacting with ceria, has been proposed to be an active

center for WGS. Rather, for our catalysts from the XANES

results, though cationic after calcination, Au and Pt [12] are

in a reduced form when they promote the surface shell

reduction of ceria. We do not rule out the possibility that a

fraction of Au, not detectable by XANES, might be in

cationic form, as suggested by others [5,6,8]. Au species

with greater than zero oxidation state have been reported for

catalysts for the selective oxidation of CO by Calla and

Davis [55], although the oxide support in that case was

Al2O3. However, using XANES, the authors observed only

metallic Au after activation.

The comparison of the 5% loaded Pt and Au catalysts,

with equivalent extents of ceria surface shell reduction,

raises questions. Why is the 5% Pt/ceria catalyst 20 times

more active in the temperature range between 200 and

300 8C than gold? The result suggests that the metal plays a

role not only in activating the Type II bridging OH groups,

but also in turning over the intermediates involved in water-

gas shift. This conclusion is in contrast to recent reports that

suggest metallic species play no role in the catalysis [5,6,8].

For example, after cyanide leaching of metal (i.e., Au and Pt)

from the catalyst, one group [5,6,8] found that the water-gas

shift activity is not changed, leading them to the conclusion

that active non-metallic species in close association with

ceria are responsible for catalytic activity. As discussed

earlier, this is not supported by our XANES results, either for

Au in this work or Pt from our earlier study [12]. Therefore,

to test the possibility of a participation of the metal in the

reaction mechanism, we examined the turnover rate of

formate by transient decomposition with H2O.

To preface the decision to use formate decomposition, we

are aware that recently, two different groups have suggested

that formates are not intermediates in the water-gas shift

reaction. For example, Tibiletti et al. [56] have used steady-

state isotopic transient kinetic analysis of the RWGS

reaction with 12C and 13C isotopes for labeling CO2. They

have observed that Pt carbonyl exchanges faster than the

formate (which exchanged very slowly) in RWGS. They

have claimed that, therefore, formates are merely spectators

on the surface, and that a redox mechanism is operating. By

the principle of microscopic reversibility, they further claim

that formates are not intermediates to the forward shift,

either. We have recently repeated their study of RWGS, and

obtained the same data. However, the principle of

microscopic reversibility must be invoked with caution,

as reactants or products can promote the rate of the reaction.

Shido and Iwasawa [13] have shown, and we have confirmed

[15], that while formate is quite stable in the absence (or

under low concentrations) of H2O, when H2O is present, the


forward decomposition of formate to unidentate carbonate

and H2 occurs rapidly, even at a low temperature. In fact, we

recently performed the RWGS experiments in the presence

of co-adsorbed H2O, and the formate exchanges very

rapidly, as it does for the forward shift reaction as well

[52,53]. Moreover, the Pt carbonyl exchanges rapidly for

both 12CO2 to 13CO2 switching and 12CO to 13CO switching

even when the second reactant (i.e., H2 for RWGS and H2O

for forward shift) is absent, being replaced by inert gas,

suggesting that the Pt-carbonyl exchange occurs rapidly

even in the absence of the RWGS or forward WGS reactions

[52,53]. There has been, more recently [54], a second claim

that formates are responsible for deactivation, as CO + H2

led to their appearance in in situ spectroscopy, and, more-

over, switching from WGS, to a CO + H2 feed, and back to

WGS led to decreased activity. Again, however, when H2O is

present [13,15], formates decompose very rapidly at shift

temperatures to CO2 and H2. We have found (unpublished

results) that the Pt/ceria catalysts have some activity for

Fischer–Tropsch synthesis, so it is more likely that the latter

authors were generating hydrogenation products (and there-

fore, carbonaceous residue), by running a mixture of only

CO + H2 that likely led to the decreased activity.

Returning to the formate decomposition experiments,

therefore, we are in agreement with Shido and Iwasawa that,

while formates are quite stable when CO is reacted with the

Type II bridging OH groups, they are likely intermediates to

the WGS reaction when H2O is present. In our previous

switching experiments, one can observe the rapid formation

of CO2 when the stabilized formates are decomposed with

H2O by FTIR spectroscopy [15,57] at shift temperatures.

And, in fact, for Pt/ceria, we had to lower the temperature to

140 8C in order to slow the formate forward decomposition

rate enough so that we could make measurements by FTIR to

monitor their decomposition rate [15]. In those studies, the

intensity of the Pt-carbonyl band remained little affected,

and it was suggested that intermediate carbonate species

(i.e., precursor to CO2) were produced from the formate

decomposition in steam.

In good agreement with our previous results for Pt/ceria,

the formate decomposition by H2O was complete in about

6 min. Here the result is slightly faster than previously

observed, since we are using a 2.5% Pt/ceria catalyst versus

a 1% Pt/ceria [15], which took about 8 min to complete the

decomposition. However, even after 40 min, a considerable

amount of formate (about 63%) remained on the surface of

the 2.5% Au/ceria catalyst after the H2O was purged from

the reactor with N2. This is the first direct evidence that

suggests that the metal does play a role in the mechanism,

and apparently during the proposed rate limiting step [13],

the forward decomposition of formate, which involves C–H

bond breaking. A relative rate comparison of the two

catalysts, based on these measurements, can be estimated, as

follows:

Xð1�0:63Þ=40 ¼ ð1�0Þ=6; Pt : Au ¼ X ¼ 18

The results suggest that on a relative basis, the formate

decomposes at approximately 18 times faster for the 5 wt.%

Pt/ceria catalyst, than it does for the 5 wt.% Au/ceria

catalyst.

It is still not clear how Pt accelerates the rate of the

formate decomposition in the presence of H2O. We consider

at least three possibilities. (1) Pt may assist in dehydro-

genating the formate intermediate. Certainly, Pt has been

found to exhibit far greater hydrogenation/dehydrogenation

ability than Au, leading to the extensive study of Au for

partial hydrogenation reactions, including acetylene [58]

and butadiene [59,60]. If the formate intermediate is a

mobile species and the metal assists in formate C–H bond

breaking, the hypothesis suggests that the promoter type and

loading should impact the WGS rate. (2) Another possibility

is an electronic effect, which may impact the tilting of the

formate, proposed to be associated with the transition state

of formate decomposition [13]. (3) In re-considering the

work of Tamaru, H+ may promote the decomposition rate of

formate. It is not clear at this time whether or not Pt may be

able to promote the heterolytic dissociation of H2O, but

dissociation of a fraction of adsorbed H2O on Pt (1 1 1) has

been suggested recently [61].

5. Conclusions

Increasing the Au loading had a significant positive

impact on the catalytic activity for WGS. The partial

reduction of ceria is necessary for generating bridging OH

groups on the surface of ceria, which serve as the active sites.

During this surface shell reduction process in hydrogen,

bridging OH groups are formed, and two cerium atoms for

each pair of Type II OH groups change from the Ce4+ to Ce3+

oxidation state. Addition of Au to ceria catalyzes the surface

reduction process and reduces the peak reduction tempera-

ture from 450 8C for the unpromoted catalyst to 100 8C for

the 5 wt.% Au/ceria catalyst. A systematic decrease in the

temperature required for ceria surface shell reduction was

observed by increasing the Au promoter loading as follows:

0.1, 0.25, 0.5, 1.0, 2.5, and 5.0 wt.%.

After ensuring that we accounted for metal–oxide

interactions and obtained similar extents of ceria surface

shell reduction to provide a valid comparison, 5 wt.% Au/

ceria was found to have about 1/20th the activity of 5 wt.%

Pt/ceria in the range 200–300 8C, and 5 wt.% Pt/ceria

exhibited a higher steady-state activity (about double that of

Au) at 175 8C. Transient formate decomposition experi-

ments carried out showed that the rate of formate

decomposition was approximately 20 times faster for Pt/

ceria than that observed with Au/ceria, suggesting that the

metal (in addition to the rate promotion of H2O) is also

involved in promoting the formate decomposition, which is

the proposed rate limiting step of the formate reaction

mechanism. It is suggested that Pt accomplishes this higher

turnover rate by nature of its greater dehydrogenating ability


over Au, or via an electronic influence, or that it may

possibly be involved in dissociating H2O to promote the

decomposition of formate intermediate.

Acknowledgments

The work was supported by the Commonwealth of

Kentucky. We would like to thank Joel Young at the

University of Oklahoma’s Department of Physics for

fabrication of the in situ X-ray spectroscopy cell. Research

carried out (in part) at the NSLS, at BNL, is supported by the

U.S. Department of Energy, Division of Materials Science

and Division of Chemical Sciences, under Contract No. DE-

AC02-98CH10886.

References

[1] National Hydrogen Energy Roadmap, US DOE, 2002.

[2] Fuel Cell Report to Congress, US DOE, 2003.

[3] D.C. Dayton, M. Ratcliff, R. Bain, Fuel Cell Integration—A Study of

the Impact of Gas Quality and Impurities, NREL/MP-510-30298,

2001.

[4] A.F. Ghenciu, Curr. Opin. Solid State Mater. Sci. 6 (2002) 389.

[5] Q. Fu, W. Deng, H. Saltsburg, M. Flytzani-Stephanopoulos, Appl.

Catal. B: Environ. 56 (2005) 57.

[6] Q. Fu, H. Saltsburg, M. Flytzani-Stephanopoulos, Science 301 (2003)

935.

[7] Q. Fu, A. Weber, M. Flytzani-Stephanopoulos, Catal. Lett. 77 (2001) 87.

[8] Q. Fu, S. Kudriavtseva, H. Saltsburg, M. Flytzani-Stephanopoulos,

Chem. Eng. J. 93 (2003) 41.

[9] T. Tabakova, F. Boccuzzi, M. Manzoli, J.W. Sobczak, V. Idakiev, D.

Andreeva, Appl. Catal. B: Environ. 49 (2004) 73.

[10] T. Tabakova, F. Boccuzzi, M. Manzoli, V. Idakiev, D. Andreeva, Appl.

Catal. A: Gen. 252 (2003) 385.

[11] D. Andreeva, V. Idakiev, T. Tabakova, L. Ilieva, P. Falaras, A.

Bourlinos, A. Travlos, Catal. Today 72 (2002) 51.

[12] G. Jacobs, U.M. Graham, E. Chenu, P.M. Patterson, B.H. Davis, J.

Catal. 229 (2005) 499.

[13] T. Shido, Y. Iwasawa, J. Catal. 141 (1993) 71.


[15] G. Jacobs, P.M. Patterson, U.M. Graham, D.E. Sparks, B.H. Davis,

Appl. Catal. A: Gen. 269 (2004) 63.


[17] T. Shido, K. Asakura, Y. Iwasawa, J. Catal. 122 (1990) 55.

[18] E. Chenu, G. Jacobs, A.C. Crawford, R.A. Keogh, P.M. Patterson, D.E.

Sparks, B.H. Davis, Appl. Catal. B: Environ. 59 (2004) 45.

[19] G. Jacobs, A.C. Crawford, L. Williams, P.M. Patterson, B.H. Davis,

Appl. Catal. A: Gen. 267 (2004) 27.

[20] M.W. Balakos, M.R. Madden, T.L. Walsh, J.P. Wagner, Water gas

shift: alternatives for fuel processing, in: Proceedings of the AICHE

Spring National Meeting, New Orleans, LA, 2004.

[21] G. Jacobs, E. Chenu, P.M. Patterson, L. Williams, D.E. Sparks, G.

Thomas, B.H. Davis, Appl. Catal. A: Gen. 258 (2004) 203.

[22] K. Tamaru, Dynamic Heterogeneous Catalysis, Academic Press, 1977.

[23] P. Mars, J.J.F. Scholten, P. Zwietering, Adv. Catal. 14 (1963) 35.

[24] A. Laachir, V. Perrichon, A. Badri, J. Lamotte, E. Catherine, J.C.

Lavalley, J. El Fallah, L. Hilaire, F. Le Normand, E. Quemere, G.N.

Sauvion, O. Touret, J. Chem. Soc., Faraday Trans. 87 (1991)

1601.

[25] C. Binet, M. Daturi, J.C. Lavalley, Catal. Today 50 (1999) 207.

[26] S. Hilaire, X. Wang, T. Luo, R.J. Gorte, J. Wagner, Appl. Catal. 215

(2001) 271.

[27] T. Bunluesin, R. Gorte, G. Graham, Appl. Catal. B 15 (1998) 107.

[28] M. Haruta, S. Tsubota, T. Kobayashi, H. Kageyama, M.J. Genet, B.

Delmon, J. Catal. 144 (1993) 175.

[29] M. Haruta, S. Tsubota, T. Kobayashi, S. Iijima, J. Catal. 115 (1989)

301.

[30] H. Sakurai, T. Akita, S. Tsubota, M. Kiuchi, M. Haruta, Appl. Catal. A:

Gen. 291 (2005) 179.

[31] F. Zhang, P. Wang, J. Koberstein, S. Khalid, Siu-Wai Chan, Surf. Sci.

563 (2004) 74.

[32] S. Jongpatiwut, P. Sackamduang, T. Rirksomboon, S. Osuwan, D.E.

Resasco, J. Catal. 218 (2003) 1.

[33] Y. Li, Q. Fu, M. Flytzani-Stephanopoulos, Appl. Catal. B 27 (2000)

179.

[34] M. Okumura, S. Tsubota, M. Haruta, J. Mol. Catal. A: Chem. 199

(2003) 73.

[35] T. Ressler, WinXAS 97 Version 1.0, 1997.

[36] G. Jacobs, P.M. Patterson, L. Williams, E. Chenu, D.E. Sparks, G.

Thomas, B.H. Davis, Appl. Catal. A: Gen. 262 (2004) 177.

[37] F. Bozon-Verduraz, A. Bensalem, J. Chem. Soc., Faraday Trans. 90

(1994) 653.

[38] H.C. Yao, Y.F.Y. Yao, J. Catal. 86 (1984) 254.

[39] K. Jung, A.T. Bell, J. Catal. 193 (2000) 207.

[40] J. Lamotte, J.C. Lavalley, E. Druet, E. Freund, J. Chem. Soc., Faraday

Trans. 1 79 (1983) 2219.

[41] J. Lamotte, J.C. Lavalley, V. Lorenzelli, E. Freund, J. Chem. Soc.,

Faraday Trans. 1 81 (1985) 215.

[42] J. Lamotte, V. Moravek, M. Bensitel, J.C. Lavalley, React. Kinet.

Catal. Lett. 36 (1988) 113.

[43] S.M. Stagg-Williams, D.E. Resasco, Stud. Surf. Sci. Catal. 130 (2000)

3663.

[44] F.B. Noronha, C. Taylor, E.C. Fendley, S.M. Stagg-Williams, D.E.

Resasco, Catal. Lett. 90 (2003) 13.

[45] S. Overbury, D. Huntley, D. Mullins, G. Glavee, Catal. Lett. 51 (1998)

133.

[46] J. El Fallah, S. Boujani, H. Dexpert, A. Kiennemann, J. Majerns, O.

Touret, F. Villain, F. Le Normand, J. Phys. Chem. 98 (1994)

5522.

[47] T. Ressler, WinXAS 97 Version 1.0, 1997.

[48] G. Jacobs, L. Williams, U. Graham, D.E. Sparks, B.H. Davis, J. Phys.

Chem. B 107 (2003) 10398.

[49] G. Jacobs, L. Williams, U. Graham, D.E. Sparks, G. Thomas, B.H.

Davis, Appl. Catal. A 252 (2003) 107.

[50] A. Holmgren, B. Andersson, D. Duprez, Appl. Catal. B: Environ. 22

(1999) 215.

[51] G. Jacobs, S. Khalid, P.M. Patterson, L. Williams, D.E. Sparks, B.H.

Davis, Appl. Catal. 268 (2004) 255.

[52] G. Jacobs, A.C. Crawford, B.H. Davis, Catal. Lett. 100 (2005)

147.

[53] G. Jacobs, B.H. Davis, Appl. Catal. A: Gen. 284 (2005) 31.

[54] C.H. Kim, L.T. Thompson, J. Catal. 230 (2005) 66.

[55] J.T. Calla, R.J. Davis, Catal. Lett. 99 (2005) 21.

[56] D. Tibiletti, A. Goguet, F.C. Meunier, J.P. Breen, R. Burch, Chem.

Commun. 14 (2004) 1636.

[57] G. Jacobs, P.M. Patterson, L. Williams, D.E. Sparks, B.H. Davis,

Catal. Lett. 96 (2004) 97.

[58] T.V. Choudhary, C. Sivadinarayana, A.K. Datye, D. Kumar, D.W.

Goodman, Catal. Lett. 86 (2003) 1.

[59] P.A. Sermon, G.C. Bond, J. Chem. Soc., Faraday Trans. I 74 (1978)

385.

[60] M. Okumura, T. Akita, M. Haruta, Catal. Today 74 (2002) 265.

[61] T. Jacob, W.A. Goddard III, J. Am. Chem. Soc. 126 (2004)

9360.

Low temperature water-gas shift: Examining the efficiency of Au as a promoter for ceria-based...

Documents

Transcript of Low temperature water-gas shift: Examining the efficiency of Au as a promoter for ceria-based...