You are on page 1of 11

Journal of Catalysis 299 (2013) 90–100

Contents lists available at SciVerse ScienceDirect

Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat

Origin of the high activity of Au/FeOx for low-temperature CO oxidation: Direct


evidence for a redox mechanism
Lin Li, Aiqin Wang, Botao Qiao, Jian Lin, Yanqiang Huang, Xiaodong Wang, Tao Zhang ⇑
State Key Laboratory of Catalysis, Dalian Institute of Chemical Physics, Chinese Academy of Sciences, Dalian 116023, China

a r t i c l e i n f o a b s t r a c t

Article history: FeOx-supported gold nanocatalyst is one of the most active catalysts for low-temperature CO oxidation.
Received 23 August 2012 However, the origin of the high activity is still in debate. In this work, using a combination of surface-
Revised 2 November 2012 sensitive in situ FT-IR, Raman spectroscopy, and microcalorimetry, we provide unambiguous evidence
Accepted 7 November 2012
that the surface lattice oxygen of the FeOx support participates directly in the low-temperature CO
Available online 11 January 2013
oxidation, and the reaction proceeds mainly through a redox mechanism. Both the presence of gold
and the ferrihydrite nature of the FeOx support promote the redox activity greatly. Calcination treatment
Keywords:
has a detrimental effect on the redox activity of the Au/FeOx, which in turn decreases greatly the activity
Gold
Iron oxide
for low-temperature CO oxidation. The gold-assisted redox mechanism was also extended to other metal-
Au/FeOx supported FeOx catalysts, demonstrating the key role of the FeOx support in catalyzing the CO oxidation
CO oxidation reaction.
FT-IR Ó 2012 Elsevier Inc. All rights reserved.
Raman
Microcalorimetry
Redox

1. Introduction materials can also be classified as ‘‘active’’ (reducible oxides such


as TiO2, Fe2O3, or CeO2) and ‘‘inert’’ (nonreducible oxides such as
Low-temperature CO oxidation catalyzed by supported gold SiO2, Al2O3, or MgO) supports [12], and the ‘‘active’’ supports
nanoparticles (Au NPs) has been intensively studied in the funda- usually lead to much more active gold catalysts than the ‘‘inert’’
mental science of catalysis [1–3]. In spite of a great number of con- supports in low-temperature CO oxidation. However, how the
tributions to unraveling the origin of the unexpectedly high oxygen is activated by the reducible oxide support and whether
activity of gold catalysts, the reaction mechanism is still debated. the reducible oxide support participates directly in the reaction
One topic under debate is the active gold species. Since the activity is not clear yet.
of gold catalysts is strongly dependent on the size of Au NPs and Au/FeOx is one of the most active catalysts for low-temperature
also sensitive to the preparation and heat treatment procedures, CO oxidation. Many experimental studies have shown that the
various gold species such as 2–3 nm Au NPs [4,5], gold bilayer as-synthesized Au/FeOx catalyst by a co-precipitation method
structures [6], subnanometer gold clusters [7], low-coordinated was much more active than the heat-treated one [7,13–18]. Since
gold atoms at the perimeter sites of gold particles [8], gold cations the as-synthesized Au/FeOx catalyst was mainly composed of cat-
[9], and pairs of gold atoms and gold cations [10] have all been pro- ionic gold (AuOOHxH2O) and ferrihydrite (Fe5HO84H2O), both cat-
posed as the active sites. Another topic under debate is the activa- ionic gold and OH groups of the ferrihydrite support were
tion of oxygen, which is a central subject for gold-catalyzed suggested to play important roles in CO oxidation [16,17]. Based
oxidation reactions. Since bulk gold is unable to adsorb oxygen, on microcalorimetry results, Gupta and Tripathi proposed that
the adsorption and activation of oxygen are usually believed to CO oxidation over Au/FeOx followed the same redox mechanism
occur at the support [11], the low-coordinated gold atoms [8], or (also called the Mars–van Krevelen mechanism) as that over FeOx,
the interface between Au NPs and the support [1]. According to which involved participation of lattice oxygen of the FeOx support.
their reducibility or their ability to activate oxygen, the substrate However, this redox mechanism has not been accepted widely,
possibly because the removal of lattice oxygen from the FeOx sup-
port usually occurs at a much higher temperature than CO oxida-
⇑ Corresponding author. Address: Dalian Institute of Chemical Physics, Chinese
tion over the Au/FeOx catalyst [19]. Daniells et al. studied the
Academy of Sciences, 457 Zhongshan Road, Dalian 116023, China. Fax: +86 411 mechanism of CO oxidation over as-synthesized Au/FeOx and pro-
84691570. posed a Langmuir–Hinshelwood mechanism where CO adsorbed
E-mail address: taozhang@dicp.ac.cn (T. Zhang). onto Au NPs reacts with oxygen adsorbed molecularly onto the

0021-9517/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.jcat.2012.11.019
L. Li et al. / Journal of Catalysis 299 (2013) 90–100 91

oxygen vacancy of the FeOx support. In this mechanism, the lattice ground spectrum was recorded from the sample using 120 scans
oxygen does not participate in the CO oxidation but is involved in and was then subtracted automatically from the subsequent spec-
an initial activation step to produce oxygen vacancies [16]. Never- tra. The CO adsorption experiments were conducted in a pulse
theless, there is no spectroscopic evidence to support either the re- mode with a manifold of valves, including two four-way valves.
dox mechanism or the Langmuir–Hinshelwood mechanism. In fact, The pretreated sample was pulsed with a gas mixture of 2% CO/
due to the nonuniformity of Au NPs produced by the co-precipita- He at the desired temperature for 1 min and then pulsed with a
tion method and the different forms of the FeOx support, clarifica- 2% O2/He flow for 1 min after being purged with He for 1 min.
tion of the reaction mechanism has become a significant challenge. The gas flow rate was fixed at 20 ml/min. After each pulse, it took
On the other hand, more and more experimental evidence, in ca. 33 s for the pulsed gas to reach the sample in the IR cell. All the
particular that from temporal analysis of products (TAP), has results described here were obtained using 5–6 s intervals and
shown that some special oxygen of the reducible supports (TiO2, summing eight scans in each spectrum with a spectral resolution
CeO2, etc.) could be, with the aid of Au NPs, removed and replen- of 8 cm1. The kinetic studies were performed with the diffuse
ished by sequential exposure to CO and O2 at a reaction tempera- reflectance cell connected to a gas handling system and a high vol-
ture of 353 K [20,21]. This special oxygen was proposed as the umetric system employing Setra Baratron capacitance manometers
surface lattice oxygen that is located at the perimeter sites of Au for precise pressure measurement.
NPs, and the corresponding reaction mechanism was called an Raman spectra were collected using an in situ Raman reactor
Au-assisted Mars–van Krevelen reaction mechanism [20]. How- (Linkam CCR1000) on a LabRam HR800 confocal microprobe Ra-
ever, one cannot distinguish between lattice oxygen and adsorbed man instrument (HORIBA Jobin Yvon, France) with laser excitation
oxygen using a TAP technique, and furthermore, the redox behav- at 633 nm (He–Ne laser) at a laser power of ca. 0.1 mW. In situ Ra-
ior of the support associated with the removal and replenishment man spectra were obtained on Au/FeOx and Pd/FeOx samples by
of the surface lattice oxygen has not been observed so far by any exposure to 2% CO/He or 2% O2/He flow at room temperature and
spectroscopic techniques. Therefore, direct evidence to support on Pt/FeOx at 393 K.
this redox mechanism is still lacking. Actually, to provide spectro- The adsorption heats of CO and O2, as well as the surface reac-
scopic evidence for the Au-assisted Mars–van Krevelen reaction tion with Au/FeOx, were measured with an HT-1000 calorimeter
mechanism is a great challenge, because the removable oxygen ac- (Setaram, France) that is connected to a pulse reactor system
counts for only a very small percentage of the total surface oxygen (Fig. S1). The details of this apparatus have been described in the
of the support [20]. previous work [23]. Before measurement, 20 mg of a sample was
Herein, by using a combination of surface-sensitive in situ spec- purged in flowing He gas at 333 K for 0.5 h, cooled to room temper-
troscopy techniques (Raman and FT-IR) and microcalorimetry, we ature in flowing He, and subsequently isolated from the outside
have for the first time provided compelling evidence for the redox contaminants under He through the switch of the four-way valve.
mechanism of CO oxidation on Au/FeOx catalysts. We show unam- After the sample pretreatment, the calorimetric cell was immersed
biguously that with the promotion of Au NPs, the surface lattice in the isothermal calorimetric block and subsequently connected,
oxygen of the FeOx support participates directly in the CO oxida- by means of the four-way valve, to a gas chromatograph (GC,
tion via an Fe3+ M Fe2+ redox mechanism even well below the Agilent 6890N). After the system reached equilibrium, pulses of
ambient temperature, and this redox mechanism predominates CO and O2 were consecutively admitted to the sample every
in low-temperature CO oxidation on Au/FeOx. More importantly, 30 min. The effluents were analyzed by GC. The resulting heat
the redox mechanism is not limited to gold, but also works quite response for each pulse was recorded as a function of time and
well in FeOx-supported palladium, which is also highly active for integrated to determine the amount of heat generated (mJ). The
low-temperature CO oxidation [22], demonstrating the key role evolved heat (kJ/mol) at a particular sample temperature corre-
of the iron oxide support in catalyzing the CO oxidation reaction. sponds to the overall effect of the catalytic process. For the
CO + O2 (1:1) co-pulsing experiment, the gas amount of every pulse
is twice that in the CO–O2 sequential experiment.
2. Experimental
Temperature-programmed reaction of CO (CO-TPR) was per-
formed in a quartz microreactor. The sample was pretreated under
2.1. Preparation of catalysts
He flow at 333 K for 0.5 h. After the sample cooled to 213 K under
He flow, the gas flow was switched to a flow of 5% CO/He
Au/FeOx catalyst was prepared by a co-precipitation method
(50 ml min1), while the sample was heated to 1073 K at a rate
with HAuCl4 and Fe(NO3)3 as the precursors and Na2CO3 as the pre-
3 K min1.
cipitant, as described in the literature [18]. The resultant precipi-
X-ray photoelectron spectroscopy (XPS) measurements were
tate was washed with distilled water and dried at 333 K for 12 h.
performed using a Thermo VG Scientific ESCALAB 250 spectrome-
Pt/FeOx and Pd/FeOx were also prepared with a similar procedure,
ter with Al Ka radiation (1486.6 eV). Binding energies were cor-
using H2PtCl6 and PdCl2 as the precursors, respectively. The Au,
rected for surface charging by referencing them to the energy of
Pt, and Pd loadings, determined by inductively coupled plasma
the C1s peak of the contaminant carbon at 284.6 eV.
atomic emission spectrometry (ICP-AES), are 2.1 wt%, 2.5 wt%,
X-ray absorption near edge structure (XANES) spectra at the Au
and 4.1 wt%, respectively. In some cases, the as-synthesized Au/
LIII-edge were obtained on the BL14W1 beamline at the Shanghai
FeOx was further subject to calcination in air for 6 h at 473, 573,
Synchrotron Radiation Facility (SSRF), Shanghai Institute of
and 673 K, respectively.
Applied Physics (SINAP), China. A Si(1 1 1) double-crystal mono-
chromator was used to reduce the harmonic component of the
2.2. Characterization monochrome beam.
57
Fe Mössbauer spectra of the as-synthesized Au/FeOx were
IR spectra were collected in a diffuse reflectance mode (DRIFTS) recorded using a Topologic 500A spectrometer and a proportional
using a Bruker EQUINOX 55 FT-IR spectrometer with an MCT counter at room temperature. 57Co(Rh) moving in a constant accel-
detector. Before CO adsorption, the as-synthesized Au/FeOx sample eration mode was used as a radioactive source. 57Fe Mössbauer
was pretreated at 333 K in a flowing He gas stream for 0.5 h in a spectral parameters such as isomer shift (IS), electric quadrupole
DRIFTS cell (HC-500, Pike technologies) and then cooled to 313, splitting (QS), and relative spectral areas of the different compo-
258, and 213 K, respectively. After each pretreatment, a back- nents of the absorption patterns were evaluated.
92 L. Li et al. / Journal of Catalysis 299 (2013) 90–100

In situ XRD patterns were collected on a PW3040/60 X’Pert PRO 3.3. In situ DRIFT of CO adsorption
(PANalytical) diffractometer equipped with an in situ cell that al-
lows the introduction of gases at room temperature. The diffrac- To uncover the underlying reason for the exceptionally high
tometer was operated at 40 kV and 40 mA using a CuKa activity of the as-synthesized Au/FeOx catalyst for CO oxidation,
radiation source (k = 0.15432 nm). we first employed in situ DRIFT to study CO adsorption on the cat-
alyst surface at or below RT. As shown in Fig. 3A, when the Au/FeOx
catalyst was exposed to CO at RT, two bands appeared immediately
2.3. Activity test at 2171 and 2120 cm1 after 37 s. The former can be ascribed to
the overlap of gaseous CO and adsorbed CO on Au3+, while the lat-
The catalytic activity for CO oxidation was measured using a ter is due to gaseous CO [26]. With extended exposure time, the
continuous-flow fixed-bed reactor system. Both the as-synthesized band at 2120 cm1 grew gradually and became higher than the
and the calcined Au/FeOx catalysts samples were used directly band at 2171 cm1, indicating that a new band was developed at
without further reduction treatment, while the as-synthesized 2120 cm1. Since the adsorption band of Au0–CO is usually located
Pd/FeOx and Pt/FeOx were reduced before the reaction with H2 around 2110 cm1 [26–31], we can ascribe the band at 2120 cm1
for 0.5 h at 323 and 473 K, respectively. A feed stream containing to CO adsorbed on Aud+ (0 < d < 1) [32]. This result indicates that
1.0 vol% CO and 1.0 vol% O2 balanced with He was allowed to pass Au3+ has been gradually reduced to Aud+ (0 < d < 1) upon exposure
through the catalyst sample. The inlet and outlet gas compositions to CO at RT. The presence of Aud+ after exposure to CO was also re-
were analyzed on line by a gas chromatograph (HP 6890, TDX-01 ported on Au/FeOx [33] and Au/FePO4 catalysts [31]. On the other
column). hand, concomitant with the CO adsorption, CO2 was produced with
the characteristic bands at 2360 and 2339 cm1, along with the
formation of carbonate species between 1200 and 1800 cm1
3. Results (see Fig. S4 in the Supporting information). The production of
CO2 in the absence of gaseous oxygen suggests that the surface lat-
3.1. Structural characterization of Au/FeOx tice oxygen of the ferrihydrite support participates in CO oxidation.
If this is true, the reduction of Fe3+ to Fe2+ on the surface of ferrihy-
The as-synthesized Au/FeOx catalyst has a large surface area of drite support must be occurring as a result of CO oxidation. Here,
289 m2/g. The XRD patterns of the Au/FeOx samples, which were the IR spectra can provide powerful information about the redox
calcined at different temperatures, are shown in Fig. S2. The as- of iron oxide. We noted that together with CO adsorption and
synthesized sample shows essentially an amorphous structure CO2 production, there was a strong increase in IR absorbance at
with only one broad peak at 2h = 35.7°, which can be indexed as high frequencies (3600–4000 cm1), and this increase could be
a reflection of ferrihydrite [24,25]. A crystalline hematite phase maintained in subsequent exposure to He. According to Boccuzzi
evolves as a result of calcination treatment, which is associated et al. [29], the increase in IR absorbance at high frequencies can
with the reduction of surface areas to 189, 100, and 59 m2/g for be attributed to the formation of a magnetite-like (Fe3O4) surface
473, 573, and 673 K-calcined samples. Either for the as-synthe- phase whose optical band gap is lower than that of Fe2O3 [34].
sized or for the calcined samples, there is no gold species that could To confirm this point, we replaced CO with H2 and observed the
be detected by XRD, indicating that gold species are highly dis- same change in the region of high frequencies (Fig. S5). More inter-
persed on the support. However, one can see from the TEM images estingly, when the exposure gas was switched from He to O2, the IR
(Fig. 1) that Au NPs are not uniform; a few aggregated NPs in 10– absorbance at high frequencies (3600–4000 cm1) decreased along
20 nm are clearly observed even in the as-synthesized Au/FeOx with the disappearance of adsorbed CO and CO2 bands (at 198 s),
sample. Except for these large NPs, most areas of the as- indicating that the surface Fe2+ is again oxidized to Fe3+ by gas oxy-
synthesized sample are absent for any discernible Au NPs within gen. Evidently, the change in the IR absorbance at 3600–4000 cm1
the resolution limit of our TEM, suggesting most of the gold is can be regarded as an indicator of the redox between Fe3+ and Fe2+.
highly dispersed and below 1 nm in the as-synthesized Au/FeOx Therefore, these IR results have provided strong evidence that the
catalyst. In contrast, the sample calcined at 673 K showed many ferrihydrite support participates directly in CO oxidation at RT by
Au NPs of 2–5 nm, indicating that significant sintering of Au NPs providing surface lattice oxygen.
occurred during calcination. The 57Fe Mössbauer spectrum of the To investigate whether the redox mechanism can still work be-
as-synthesized Au/FeOx at 293 K shows a paramagnetic doublet low 273 K, we conducted the IR experiment at 258 K. As shown in
(Fig. S3), and the corresponding hyperfine parameters (Table S1) Fig. 3B, the IR spectra of the Au/FeOx at 3600–4000 cm1 upon
are completely consistent with a ferrihydrite phase [13–17]. sequential exposure to CO and O2 at 258 K are very similar to those
at RT, except for a longer time of CO exposure required to reduce
Fe3+ (227 s at 258 K versus 52 s at RT), indicating that redox be-
3.2. Catalytic activities tween Fe3+ and Fe2+ of the Au/FeOx can also take place well at
258 K. More intriguingly, it was found that the Au/FeOx treated
The catalytic activities of Au/FeOx catalysts that were calcined with first CO, and then, O2 gas flow at 258 K became so activated
at different temperatures are shown in Fig. 2. It is evident that that the redox between Fe3+ and Fe2+ was able to take place fast
the catalytic activities decreased with increasing calcination tem- at an even lower temperature, for example, at 213 K, as shown in
perature, which is in accordance with earlier reports [7,14–18]. Fig. 3C. A similar activation process was also reported by Widmann
In particular, for the as-synthesized Au/FeOx sample, CO conver- et al. with Au/CeO2 catalyst [35] and by Daniells et al. with Au/FeOx
sion attained 100% even at 273 K, demonstrating its exceptionally catalyst [16] and was interpreted as the requirement for a surface
high activity for low-temperature CO oxidation. In comparison oxygen vacancy to obtain high activity for CO oxidation. Clearly,
with the as-synthesized Au/FeOx, the as-synthesized Pd/FeOx and the surface lattice oxygen of the FeOx support in the Au/FeOx cata-
Pt/FeOx are less active and follow the order Au/FeOx > Pd/FeOx > lyst was very reactive toward CO, and the redox between Fe3+ and
Pt/FeOx. It should be stressed that the Pd/FeOx catalyst is signifi- Fe2+ proceeded quite easily even below RT. This is in good agree-
cantly more active than the Pt/FeOx catalyst; it could convert CO ment with the high activity of Au/FeOx for low-temperature CO
completely at RT, showing its high potential for CO oxidation as oxidation. In addition, compared with CO adsorption at RT, it was
an alternative to more expensive Au/FeOx [22]. noted that the band of CO adsorption blue-shifted to 2123 cm1
L. Li et al. / Journal of Catalysis 299 (2013) 90–100 93

Fig. 1. Low-magnification HAADF-STEM images of the as-prepared Au/FeOx (a and b) and Au/FeOx calcined at 673 K (c and d).

at 258 K and 2126 cm1 at 213 K, suggesting that the gold particles in situ Raman spectroscopy to detect the changes occurring on
are more positively charged with a decrease in the temperature for the surface of the catalyst in different atmospheres.
CO adsorption.
To gain further understanding of the redox reaction occurring
on the as-synthesized Au/FeOx catalyst, we performed kinetic stud- 3.4.1. Au/FeOx
ies with FT-IR to determine the redox reaction rate and activation The in situ Raman spectra of the as-synthesized Au/FeOx in dif-
energy. The amounts of CO2 produced by the redox reaction path- ferent atmospheres are shown in Fig. 5. The Raman spectrum of
way were determined by the integrated absorbance of CO2 in the the ferrihydrite support presents four features at 380, 510, 710,
FT-IR spectra, which was then calibrated by CO2 pressure. Fig. 4 and 1341 cm1, which are typical for ferrihydrite [39]. Upon depo-
shows the CO2 pressures against time at different temperatures sition of gold, the intensities of these bands decrease greatly, possi-
(213–233 K) when the Au/FeOx was exposed to 1333 Pa CO. bly due to the masking effect of gold NPs on the ferrihydrite surface.
Clearly, the CO2 produced via the redox reaction pathway can be When the Au/FeOx sample was exposed to CO gas flow at RT, all the
well fitted to the first-order kinetics. The corresponding initial features of ferrihydrite disappeared and simultaneously a new band
rates were determined to be 33.7, 28.7, and 21.9 Pa s1 at 233, appeared at 653 cm1. This band can be assigned to the A1g mode of
223, and 213 K, respectively. Compared with CO reaction with Fe3O4 [40], indicating that surface Fe3+ has been reduced to Fe2+
the lattice oxygen of Fe2O3 without the presence of gold [12], the upon exposure to CO. When the gas flow was switched from CO to
redox reaction rates over the Au/FeOx were much higher due to a O2, the characteristic band of Fe3O4 at 653 cm1 disappeared, while
very low apparent energy (9.0 kJ/mol; see inset in Fig. 4) [36]. This those of the ferrihydrite appeared again, indicating that surface Fe2+
is also in good agreement with the exceptionally high activity of has been oxidized to Fe3+ upon exposure to O2. Associated with the
the Au/FeOx catalyst at low temperatures. redox between Fe3+ and Fe2+, the color of the Au/FeOx changed from
brown to gray and then recovered to brown (Fig. 6) when the Au/
3.4. In situ Raman spectra FeOx was exposed to CO gas flow and then to O2. However, in con-
trast to the spectra and color change of the Au/FeOx, the ferrihydrite
Raman spectroscopy is a powerful technique to determine the support remained unchanged when it was exposed to CO gas flow
surface change of oxides [37,38]. To further confirm the redox (Fig. S6). This result indicates that the presence of gold on the FeOx
reaction between Fe3+ and Fe2+ of the catalyst, we employed support promotes greatly the redox reaction between Fe3+ and Fe2+.
94 L. Li et al. / Journal of Catalysis 299 (2013) 90–100

Fig. 2. Curves of CO conversion with reaction temperature over (a) Au/FeOx


catalysts that were calcined at different temperatures and (b) Pd/FeOx and Pt/FeOx
1 1
catalysts. Space velocity: 30; 000 ml g1
cat h for Au/FeOx and 15; 000 ml g1
cat h for
Pd/FeOx and Pt/FeOx catalysts.

3.4.2. Pd/FeOx and Pt/FeOx


It is often noticed that the presence of metal NPs on a reducible
oxide can lead to the reduction in the oxide support shifting to-
ward a much lower temperature, which is often interpreted as
hydrogen spillover from the metal to the support due to strong me-
tal–support interaction (SMSI) [41,42]. Since gold plays a key role
in assisting the redox reaction between Fe3+ and Fe2+, one may
ask if gold is unique in activating the surface lattice oxygen of
the FeOx support. To answer this question, we prepared Pd/FeOx
and Pt/FeOx by a co-precipitation method similar to that for
Au/FeOx and conducted in situ Raman experiments. As shown in
Fig. 7, quite similarly to Au/FeOx, the Pd/FeOx showed a typical
Raman feature of Fe3O4 upon exposure to 2% CO/He at RT, while
the Pt/FeOx still maintained the Raman feature of ferrihydrite,
indicating that the reduction of Fe3+ to Fe2+ occurred smoothly
on Pd/FeOx but not on Pt/FeOx at RT. Only when the CO treatment
temperature was raised to 393 K or above could the reduction of
Fe3+ to Fe2+ take place on the Pt/FeOx. Clearly, the redox activity Fig. 3. IR spectra of Au/FeOx sample as a function of time after the gas flow was
of ferrihydrite is closely correlated to the metal on it. Both Au switched (a) from He to 2% CO/He; (b) from 2% CO/He to He; and (c) from He to 2%
O2/He at (A) RT, (B) 258 K, and (C) 213 K. Before being cooled to 213 K, the sample
and Pd could activate the surface Fe–O bond to the extent that
had been treated under the conditions in (B). It takes ca. 33 s for the gas flow to
the surface lattice oxygen is easily removed by CO at RT. reach the IR cell after each switch.

3.5. In situ XRD and XANES structure remained unchanged during pretreatment with 2%
CO/He at 313 K for 180 min. This result indicates that the redox
Since both DRIFT and Raman can only penetrate a very limited between Fe3+ and Fe2+ which was observed with in situ DRIFT
depth of the catalyst, we subsequently performed in situ XRD to and Raman techniques indeed occurred only on the surface or sub-
see if the redox of Fe3+/Fe2+ of the Au/FeOx catalyst can occur surface, and more specifically, at the interface between gold and
through the whole bulk catalyst upon exposure to CO. The in situ ferrihydrite. According to Behm and co-workers [20], the amount
XRD experiment was conducted under 2% CO/He gas flow at of removable oxygen accounts for only a small percent of the total
313 K. As shown in Fig. 8, the XRD pattern of the as-synthesized surface oxygen, one can imagine what a very tiny amount of reduc-
Au/FeOx catalyst is typical of amorphous structure; neither crystal- ible Fe3+ is present in comparison with bulk Fe3+, which is why the
line iron oxide nor gold could be observed. This amorphous redox behavior could not be detected by XRD.
L. Li et al. / Journal of Catalysis 299 (2013) 90–100 95

Fig. 4. Kinetics for the oxidation of CO over the Au/FeOx catalyst as measured by the
change in the absorbance of CO2 as a function of time and temperature. The inset
shows the Arrhenius plot.

Fig. 5. RT Raman spectra of (a) ferrihydrite support; (b) Au/FeOx; (c) exposure of
Au/FeOx to 2% CO/He gas flow; and (d) exposure of Au/FeOx to 2% O2/He following
CO treatment.

Similar to XRD, XANES is also a bulk technique and therefore


could not detect the change of surface Fe species either. However,
it was able to detect the change of the oxidation state of gold under
Fig. 6. White light illuminations in Raman experiments for (a) Au/FeOx; (b)
different atmospheres because the total gold content was only exposure of Au/FeOx to 2% CO/He; and (c) exposure of Au/FeOx to 2% O2/He after CO
2.1 wt%. The XANES spectra of Au/FeOx at the Au LIII-edge under exposure. The scale bar corresponds to 4 lm.
different atmospheres are shown in Fig. 9. The as-synthesized
Au/FeOx catalyst presents an intensive white line that is indicative
of Au3+. According to Finch et al. [14], the as-synthesized Au/FeOx redox of the iron oxide support is involved in the CO oxidation.
was composed of ferrihydrite and a hydrated gold oxyhydroxide With the pulse reactor system, we can determine quantitatively
phase (AuOOHxH2O). Upon exposure to 2% CO/He gas flow at RT the consumption of CO as well as the production of CO2; further-
for 5 min, the white line intensity decreased greatly and resembled more, by combining with the microcalorimeter, we can also deter-
that of metallic gold, indicating that Au3+ has been reduced almost mine the heat released during each pulse.
to Au0 by CO. This is slightly different from our in situ DRIFT result,
which mainly shows the presence of Aud+. A probable reason is that 3.6.1. Pulse reaction of CO oxidation on as-synthesized Au/FeOx
the Aud+ is only slightly positively charged in reference to Au0 [26] Fig. 10 illustrates the consumption of CO, the production of CO2,
and therefore cannot be distinguished from metallic gold by and the heat evolved during pulses of CO on the Au/FeOx at 313 K.
XANES. Moreover, the Aud+ remained unchanged even when it It can be seen clearly that pulsing CO led to the production of CO2
was further subjected to treatment with 2% O2/He gas flow at RT. accompanied by a heat release of about 100 kJ/mol. Moreover, both
the CO consumption and the CO2 production decreased with pulse
3.6. Pulse reaction of CO oxidation number and eventually were close to zero at the 18th pulse. The
total amount of CO consumed during the 18 pulses was measured
The above in situ DRIFT and Raman spectra have provided com- to be 50 lmol. Assuming that all Au3+ species are reduced to Au0
pelling evidence that under the promotion of Au (or even Pd), the during CO-pulsing, the amount of CO consumed for gold reduction
surface redox of the ferrihydrite support took place at low temper- would be only 3.3 lmol, which is less than 1/15 of the measured
atures. Subsequently, we used microcalorimetry in combination value. Evidently, most of the CO was consumed for the reaction
with a pulse reactor system to investigate whether and how the with the surface lattice oxygen of the ferrihydrite support, which
96 L. Li et al. / Journal of Catalysis 299 (2013) 90–100

Fig. 9. XANES spectra of Au/FeOx after different treatments in comparison with the
reference Au foil (a). (b) As-synthesized sample; (c) sample exposed to 2% CO/He
gas for 5 min at RT; and (d) sample exposed to 2% O2/He gas for 5 min at RT
following step (c).

Fig. 7. Raman spectra of Pd/FeOx at room temperature (A) and Pt/FeOx at 393 K (B)
obtained after different treatments: (a) Pd/FeOx or Pt/FeOx; (a0 ) exposure of Pt/FeOx
to 2% CO/He at room temperature; (b) exposure of Pd/FeOx or Pt/FeOx to 2% CO/He;
(c) exposure of Pd/FeOx or Pt/FeOx to 2% O2/He after CO exposure.

Fig. 10. Amounts of CO reacted and CO2 produced, as well as heat evolved, during
pulses of CO on Au/FeOx at 313 K.

dence with the stoichiometric ratio of O2/CO in the CO oxidation


reaction. As a result of the O2 pulsing, a heat of about 400 kJ/mol
was produced, which can be ascribed to the oxidation of Fe2+ to
Fe3+ by gaseous oxygen [43]. In this way, the surface lattice oxygen
of the ferrihydrite was replenished by gas oxygen, and then, the
reaction cycle for CO oxidation was finished. From Fig. 11, it can
be seen that no deactivation occurred during 19 pulses.
In the sequential pulses of CO and O2, CO2 is produced only via
the redox reaction pathway; that is, CO reacts first with the surface
lattice oxygen of the ferrihydrite, resulting in the reduction in sur-
Fig. 8. In situ XRD patterns of the as-synthesized Au/FeOx sample as a function of
time after the gas flow was switched from He to 2% CO/He at 313 K.
face Fe3+ to Fe2+, and the consumed surface lattice oxygen is subse-
quently replenished by gaseous oxygen. To give an estimate of
what percentage such a redox reaction pathway accounts for in
is in agreement with our IR and Raman results. With increasing the overall reaction, we performed a quantitative comparison of
pulses of CO, the concentration of reactive surface lattice oxygen the amount of CO2 produced by the sequential pulses of CO and
decreases, which leads to the eventual cessation of the reaction O2 with that produced by the co-pulsing of CO and O2. In contrast
at the 18th pulse. In order for the CO oxidation to proceed contin- to the sequential pulses of CO and O2, the co-pulsing CO and O2 will
uously, the surface lattice oxygen must be replenished with gas produce CO2 in all possible reaction pathways, including the Lang-
oxygen. Therefore, sequential pulses of CO and O2 were conducted. muir–Hinshelwood reaction pathway, where CO adsorbed onto
Fig. 11 illustrates the consumption of CO and O2 as well as the pro- gold reacts with oxygen adsorbed onto gold or at the interface of
duction of CO2 and heat during sequential pulses of CO and O2. The gold and support, the Eley–Rideal reaction pathway, where CO ad-
molar amount of CO2 produced in each pulse is lower than the con- sorbed onto gold reacts with gaseous oxygen, and the redox (Mars–
sumption of CO, since a part of the CO2 is adsorbed onto the fer- van Krevelen) reaction pathway. Fig. 12 shows the amounts of CO2
rihydrite support in the form of carbonate species, as indicated produced respectively by sequential pulsing and co-pulsing of CO
by the DRIFT spectra (Fig. S4 in the Supporting information). Mean- and O2 on the Au/FeOx. One can see that the amount of CO2 pro-
while, the consumption of O2 was just half that of CO, in coinci- duced by the sequential pulsing of CO and O2 accounted for up
L. Li et al. / Journal of Catalysis 299 (2013) 90–100 97

Fig. 13. Amount of CO adsorbed/reacted (a) and CO2 produced (b) when the Au/
Fig. 11. Sequential pulses of CO and O2 on Au/FeOx at 313 K. (a) Amount of O2 or CO FeOx samples calcined at different temperatures were exposed to pulses of CO at
reacted; (b) amount of CO2 produced; (c) heat evolved. 313 K.

to 70% of that produced by co-pulsing, either at 313 K or at 213 K.


This result indicates that the CO oxidation over Au/FeOx proceeds
predominantly via the redox reaction pathway.

3.6.2. Pulse reaction of CO oxidation on Au/FeOx calcined at different


temperatures
To study the effect of calcination on the redox properties of the
Au/FeOx catalyst, we calcined the as-synthesized Au/FeOx samples
at 473, 573, and 673 K, respectively, and investigated their redox
behaviors upon exposure to CO. As shown in Fig. 13, with elevation
of the calcination temperature, the surface lattice oxygen of the
support on the Au/FeOx samples became more difficult to take
away by exposure to CO at 313 K, and the redox of FeOx could
not occur at all for the 673 K-calcined Au/FeOx catalyst. Therefore,
calcination has a detrimental effect on the redox activity of the Au/
FeOx catalyst. Evidently, this result is in good agreement with the
activity result shown in Fig. 2, further confirming that CO oxidation
over Au/FeOx proceeds via a redox mechanism.

4. Discussion

4.1. Reaction pathway for low-temperature CO oxidation over Au/FeOx

Two main reaction pathways are proposed for CO oxidation over


supported gold catalysts. One is the Langmuir–Hinshelwood reac-
tion pathway, where CO adsorbed onto gold reacts with oxygen
activated either on gold (competitive L–H mechanism) or on sup-
port (noncompetitive L–H mechanism) at the perimeter sites of
gold nanoparticles [44]. The other is the Mars–van Krevelen (redox)
reaction pathway, where CO reacts with the lattice oxygen of the
support. The major difference between the two reaction pathways
lies in the origin of reactive oxygen. Although reducible oxides usu-
ally lead to more active gold catalysts, and they are proposed to play
Fig. 12. Amount of CO2 produced during sequential pulses of CO and O2 in
important roles in activating oxygen [11,12,20,45], it is unclear yet
comparison with that in steady state reaction on Au/FeOx: (a) 313 K; (b) 213 K. The whether the lattice oxygen participates directly in the CO oxidation
steady state reaction is accomplished by co-pulsing of CO and O2. reaction, especially below RT. Morgan et al. studied CO oxidation
98 L. Li et al. / Journal of Catalysis 299 (2013) 90–100

over Au/CuMnOx catalyst using a TAP reactor [46]. In their studies, or below RT, and this reaction pathway contributed up to 70% to
the catalysts were first pretreated at 573 K by regular 16O2 followed low-temperature CO oxidation. In other words, the exceptionally
by evacuation, and were then exposed to consecutive pulses of iso- high activity of the Au/FeOx catalyst for low-temperature CO oxida-
tope-labeled 18O2 followed by 13CO at 373 K. They found that the tion can be attributed to the outstanding redox activity of FeOx
major CO2 isotope produced in the reaction was 13C16O16O, fol- with the aid of gold nanoparticles.
lowed by 13C18O16O, indicating a larger contribution of the Mars– It is surprising to find that the redox of FeOx in the Au/FeOx
van Krevelen mechanism and a smaller contribution of the Lang- could occur at a temperature as low as 213 K because the reduction
muir–Hinshelwood mechanism. Moreover, they found that the in iron oxide usually occurs above 473 K [52]. The FeOx support it-
presence of gold did not contribute to the Langmuir–Hinshelwood self did not show any change in either the sample color or the Ra-
mechanism but promoted the Mars–van Krevelen mechanism man spectrum when exposed to CO flow (Fig. S6). Therefore, the
greatly. Olea et al. draw a different conclusion, based on their TAP facile reduction in the support at low temperatures must be due
experiments on Au/TiO2 catalyst at or above RT. In their TAP exper- to the presence of gold nanoparticles, and this is in accordance
iments, isotope-labeled C16O and 18O2 were pulsed simultaneously with the Au-assisted redox mechanism proposed by Behm et al.
on the catalyst at both 298 and 423 K. They observed that the [20]. Herein, we would like to stress that not all of the surface lat-
production of C16O2 was far greater than that expected tice oxygen atoms are reactive toward CO. According to the
(C16O2:C16O18O:C18O2 = 4:1.5:1), which was explained by the rapid amount of removable oxygen (Fig. 11), we calculated the percent-
exchange of C16O18O with lattice oxygen atoms. In contrast to the age of the activated surface lattice oxygen (see the Supporting
oxygen scrambling in CO2 product, they did not observe oxygen ex- Information). It shows that the reactive oxygen accounts for only
change between 18O2 and lattice oxygen of the support in the 18O2 5.7% of the total surface lattice oxygen. This result implies that
pulsing. Accordingly, they made the claim that CO adsorbed revers- the Au-assisted redox of FeOx support should occur only at the sur-
ibly and weakly onto the catalyst surface while O2 adsorbed molec- face, more specifically at the interface between gold nanoparticles
ularly; CO oxidation occurred between adsorbed CO and O2 at the and the FeOx support. Due to the very limited amount of sites for
interface of gold and support, and lattice oxygen was only active redox, this behavior could not be detected by characterization
in oxygen exchange with CO2 [47]. By applying the TAP technique, techniques for bulk materials such as XRD, XAFS, and 57Fe
Behm and co-workers studied the nature of active oxygen species in Mössbauer spectroscopy. This also leads us to conclude that
CO oxidation over Au/CeO2 [35] and Au/TiO2 [20,21,45,48]. In all in situ surface-sensitive techniques are powerful to detect those
cases, they observed that (1) CO2 was produced solely during the changes that might be negligible for bulk material but pivotal for
CO pulses over the oxidized catalyst but not during the subsequent catalysis.
O2 pulses, indicative of reversible adsorption of CO; (2) the con-
sumption of CO is higher during the first sequence of CO pulses 4.2. Activation of the FeOx support by metal nanoparticles
attributed to the irreversible removal of active oxygen; (3) in the
subsequent pulses of CO and O2, the removal and replenishment It is noticed that the Au-assisted redox of the support is ubiqui-
of oxygen is totally reversible; (4) the oxygen removal and replen- tous in reducible oxide supported gold catalysts, such as Au/Mn2O3
ishment occurred only with the assistance of Au NPs and the [53], Au/CeO2 [35], Au/CuMnOx [46], and Au/TiO2 [20,21,45,48].
amount of reversibly deposited oxygen increased greatly with the Moreover, this redox mechanism is not only limited to gold but
reaction temperature; and (5) most importantly, the reversibly also works quite well on other metals supported on the FeOx sup-
deposited oxygen species was highly stable even at 673 K. Based port. Depending on the nature of metals on the FeOx support, the
on these observations, they proposed an Au-assisted Mars–van surface redox activity may vary to some extent, which in turn af-
Krevelen mechanism where CO reacts with the surface lattice oxy- fects their catalytic activity for low-temperature CO oxidation.
gen of the support, which is activated by gold nanoparticles. How- For example, we found that the surface redox of the FeOx support
ever, the spectroscopic evidence for the surface lattice oxygen is took place easily at RT with the assistance of Au or Pd (Figs. 5–
still lacking because the amount of reactive surface lattice oxygen 7). In accordance, both Au/FeOx and Pd/FeOx catalysts could
is very low [20]. Moreover, all their TAP experiments were per- convert CO to CO2 completely at RT (Fig. 2). In contrast, the surface
formed at or above 353 K; thus, one cannot draw a conclusion as redox of the FeOx support with the assistance of Pt occurred only at
to whether this reaction mechanism is valid for low-temperature elevated temperatures, which resulted in the low activity of Pt/
CO oxidation, which is the most important feature of gold catalysts FeOx catalyst for CO oxidation at RT. The totally consistent trends
as distinguished from other noble metal catalysts. between the redox activity of the FeOx in different metal/FeOx cat-
In the present work, by using surface-sensitive DRIFT and Ra- alysts and the activity evaluated by a fixed-bed reactor suggest
man, we observed redox between Fe3+ and Fe2+ associated with that the CO oxidation reaction over all these FeOx-supported metal
oxygen removal by CO and replenishment by O2, thus providing catalysts follows the Mars–van Krevelen mechanism. In our previ-
the first unambiguous spectroscopic evidence that the surface lat- ous study, we also found that the CO oxidation over Ir/CeO2 obeys
tice oxygen of the FeOx support indeed participates in the CO oxi- the Mars–van Krevelen mechanism [42].
dation occurring on the Au/FeOx catalyst. Participation of support Since low-temperature CO oxidation over FeOx-supported metal
oxygen as superoxide and peroxide species has been reported on catalysts follows the redox mechanism, the difference in CO oxida-
Fe-doped TiO2- [49] and nanocrystalline CeO2-supported Au cata- tion activity reflects, in principle, the difference in the activation
lysts [11]. However, our Raman results did not present evidence degree of the surface lattice oxygen by the neighboring metal
of the formation of superoxide or peroxide species after O2 expo- NPs. The number of the reactive surface lattice oxygen is closely
sure. Furthermore, the very high adsorption heat of O2 can exclude correlated with the perimeter sites of metal NPs. For instance, we
this possibility, since the calculated adsorption energy for molecu- found that the ratio of reactive surface lattice oxygen at RT to the
larly adsorbed oxygen was very low [50,51]. All our characteriza- perimeter gold atoms was between 5 and 6 (see the Supporting
tion results lead to a consistent conclusion that CO oxidation information). This result strongly suggests that only those surface
over Au/FeOx obeys the Mars–van Krevelen (redox) mechanism, lattice oxygen atoms in direct contact with or in the vicinity of me-
with the involvement of surface lattice oxygen. This conclusion is tal NPs can be activated and become highly reactive toward CO.
in good agreement with that reported by Li et al. for CO oxidation Theoretical studies on Au/Mg(OH)2 indicate strong interaction of
over an Au/FePO4 catalyst [31]. More importantly, we found that gold nanoparticles with the support by forming a short bond be-
the gold-assisted Mars–van Krevelen reaction could take place at tween edge gold atoms and oxygen atoms of the hydroxyl groups
L. Li et al. / Journal of Catalysis 299 (2013) 90–100 99

[54]. Maria and co-workers proposed that Au or Pt cations bonded into crystalline Fe2O3 when it was calcined at an increased
with surface lattice oxygen groups functioned as active sites for the temperature. Associated with the phase transformation,
water-gas shift reaction [55]. Investigations on an inverse model the hydroxyl groups also had a large decrease (Fig. S7). We
catalyst system, FeO/Pt, reveal that the metal–cation (Pt–Fe2+) found that even without the presence of gold, the ferrihy-
ensemble might be the active site for CO oxidation [56,57]. In drite was much easier to reduce than crystalline Fe2O3
our recent study on Pt1/FeOx single atom catalysts, we also pro- (Fig. S8). Similarly, Li et al. reported that FeOOH lost lattice
vided evidence that single Pt atoms are stabilized by forming a oxygen more easily than Fe2O3 when exposed to CO [60].
Pt–O–Fe bond, which activates the surface Fe–O bond and signifi- Although we cannot currently determine whether the active
cantly decreases the energy required for forming oxygen vacancies oxygen in the redox arises from the surface lattice hydroxyl
[58]. Accordingly, in this work, we propose that the metal-assisted groups, we believe that the rich hydroxyl groups on the sup-
redox of FeOx may originate from the formation of metal–O–Fe port surface play significant roles in stabilizing gold nano-
bonding, which weakens the Fe–O bond and allows the easy re- particles [61,62] and assisting CO oxidation via opening a
moval of lattice O by exposure to CO, and low-coordinated atoms new reaction channel of CO + OH ? COOH [63,64]. Actually,
in the metal nanoparticles will probably contribute to the bonding there have been some reports to show that gold supported
with Fe–O bonds. The presence of Aud+ detected by our in situ on metal hydroxides is more active than that on the
DRIFT also supports the formation of Au–O–Fe bonding. Neverthe- corresponding oxides for low-temperature CO oxidation
less, more direct evidence for metal–O–Fe bonding is yet to be [65–69]. Moreover, gold-catalyzed CO oxidation is rather
found. sensitive to the presence of water, and the moisture effect
is closely correlated with the promotional effect of OH
4.3. Calcination effect on Au/FeOx groups on O2 adsorption on the support [70,71]. Recently,
we prepared ferrihydrite-supported Ir, Pt, and Rh catalysts
There are consensual observations that calcination treatment using a similar co-precipitation method and found that they
has a detrimental effect on the activity of Au/FeOx in low- all exhibited high activities for preferential oxidation of CO
temperature CO oxidation, as we observed in Fig. 2. Actually, under rich H2 [72], demonstrating the predominant role of
Au/FeOx is rather sensitive to the history of heat treatment. It ferrihydrite in CO oxidation.
was reported that the dried sample without calcination could be
one order of magnitude more active than the calcined one [7,14– In summary, calcination treatment results in loss of hydroxyl
18]. Even subtle changes in the heat treatment conditions, for groups of the ferrihydrite support, which in turn leads to the sin-
example, heating in static air versus flowing air, could result in dra- tering of gold nanoparticles. Both effects lower the surface redox
matic differences in the activity for CO oxidation [7]. To explain the activity of the support and therefore lower its catalytic activity
significant decrease of activity with increasing calcination temper- for low-temperature CO oxidation.
ature, cationic gold [17] or bilayer structures containing less than
10 gold atoms [7] have been proposed as the active sites for low-
temperature CO oxidation. In contrast, the very high activity of 5. Conclusion
Au/FeOx that was prepared by deposition of 2 nm colloid Au
NPs on the FeOx support suggested that the bilayer structure of By using a series of in situ and surface-sensitive techniques, we
subnanometer gold clusters was not mandatory [59]. Then, what provide unequivocal evidence that CO oxidation on gold-supported
is the true reason for the high activity of Au/FeOx catalysts? In iron oxide support proceeds predominantly through an Au-assisted
the present work, based on transient reactions in sequential puls- Mars–van Krevelen mechanism where the support supplies surface
ing CO, we found that the removal of surface lattice oxygen became lattice oxygen to react with CO adsorbed onto neighboring gold
more difficult with increasing calcination temperature (Fig. 13). In nanostructures, producing CO2 accompanied by the reduction of
other words, the activity change with the calcination temperature Fe3+ to Fe2+. Both the size of gold nanostructures and the nature
is closely correlated with the surface redox activity of the FeOx of the FeOx support affect greatly the redox activity of the Au/FeOx
support. There are at least two possible reasons for the decrease catalyst. The exceptionally high activity of the as-synthesized Au/
in the redox activity of the FeOx support with an increase in the FeOx support, the decreasing activity with increasing calcination
calcination temperature. temperature, as well as the activity order of Au/FeOx > Pd/FeOx >
Pt/FeOx, can all be explained satisfactorily by this redox mecha-
(i) The effect of gold particle size. Since the low-temperature nism. This study not only provides an insightful understanding of
redox of the FeOx support occurs only with the assistance the reaction mechanism for gold-catalyzed CO oxidation, but also
of gold nanoparticles, the change of gold nanoparticles will encourages researchers to reevaluate the important roles of sup-
inevitably affect the redox activity of the support. Fig. 1 port and metal–support interaction in gold catalysis, as exempli-
clearly shows the growth tendency of gold nanoparticles fied in recent studies [73,74].
with an increase of the calcination temperature, which was
also observed earlier by other researchers [7]. According to Acknowledgments
the study of Behm and co-workers with Au/TiO2 catalysts
[45], the amount of removable oxygen was almost linearly We thank Shanghai Synchrotron Radiation Facility for the
correlated with the perimeter sites between gold nanoparti- XANES measurements. Support from the National Natural Science
cles and the support. Large gold nanoparticles will have Foundation of China (NNSFC 20803079, 21173218, 21176235,
fewer such perimeter sites per gold atom, and thus lead to and 21203181) is gratefully acknowledged.
a decrease of redox activity.
(ii) The effect of support. Compared to the above-mentioned
size effect of gold nanoparticles, the change of the support Appendix A. Supplementary material
nature with calcination temperature may impose a greater
effect on the redox activity. As shown in Fig. S2, the as- Supplementary data associated with this article can be found, in
synthesized amorphous ferrihydrite transformed gradually the online version, at http://dx.doi.org/10.1016/j.jcat.2012.11.019.
100 L. Li et al. / Journal of Catalysis 299 (2013) 90–100

References [41] J. Jia, J. Shen, L. Lin, Z. Xu, T. Zhang, D. Liang, J. Mol. Catal. A Chem. 138 (1999)
177.
[42] Y.Q. Huang, A.Q. Wang, L. Li, X.D. Wang, D.S. Su, T. Zhang, J. Catal. 255 (2008)
[1] M. Haruta, Catal. Today 36 (1997) 153.
144.
[2] G.C. Bond, D.T. Thompson, Catal. Rev. Sci. Eng. 41 (1999) 319.
[43] D.R. Lide, CRC Handbook of Chemistry and Physics, 84th ed., CRC Press, Boca
[3] A.S.K. Hashmi, G.J. Hutchings, Angew. Chem. Int. Ed. 45 (2006) 7896.
Raton, FL, 2004.
[4] M. Haruta, S. Tsubota, T. Kobayashi, H. Kageyama, M.J. Genet, B. Delmon, J.
[44] K. Liu, A.Q. Wang, T. Zhang, ACS Catal. 2 (2012) 1165.
Catal. 144 (1993) 175.
[45] M. Kotobuki, R. Leppelt, D.A. Hansgen, D. Widmann, R.J. Behm, J. Catal. 264
[5] M. Valden, X. Lai, D.W. Goodman, Science 281 (1998) 1647.
(2009) 67.
[6] M. Chen, D.W. Goodman, Science 306 (2004) 252.
[46] K. Morgan, K.J. Cole, A. Goguet, C. Hardacre, G.J. Hutchings, N. Maguire, S.O.
[7] A.A. Herzing, C.J. Kiely, A.F. Carley, P. Landon, G.J. Hutchings, Science 321
Shekhtman, S.H. Taylor, J. Catal. 276 (2010) 38.
(2008) 1331.
[47] M. Olea, M. Tada, Y. Iwasawa, J. Catal. 248 (2007) 60.
[8] N. Lopez, T.V.W. Janssens, B.S. Clausen, Y. Xu, M. Mavrikakis, T. Bligaard, J.K.
[48] A. Tost, D. Widmann, R.J. Behm, J. Catal. 266 (2009) 299.
Norskov, J. Catal. 223 (2004) 232.
[49] S. Carrettin, Y. Hao, V. Aguilar-Guerrero, B.C. Gates, S. Trasobares, J.J. Calvino, A.
[9] J.C. Fierro-Gonzalez, B.C. Gates, Chem. Soc. Rev. 37 (2008) 2127.
Corma, Chem. Eur. J. 13 (2007) 7771.
[10] H.H. Kung, M.C. Kung, C.K. Costello, J. Catal. 216 (2003) 425.
[50] J.D. Stiehl, T.S. Kim, S.M. McClure, C.B. Mullins, J. Am. Chem. Soc. 126 (2004)
[11] J. Guzman, S. Carrettin, A. Corma, J. Am. Chem. Soc. 127 (2005) 3286.
1606.
[12] M.M. Schubert, S. Hackenberg, A.C. van Veen, M. Muhler, V. Plzak, R.J. Behm, J.
[51] J.D. Stiehl, T.S. Kim, S.M. McClure, C.B. Mullins, J. Am. Chem. Soc. 126 (2004)
Catal. 197 (2001) 113.
13574.
[13] F.E. Wagner, S. Galvagno, C. Milone, A.M. Visco, L. Stievano, S. Calogero, J.
[52] W.K. Jozwiak, E. Kaczmarek, T.P. Maniecki, W. Ignaczak, W. Maniukiewicz,
Chem. Soc. Faraday Trans. 93 (1997) 3403.
Appl. Catal. A 326 (2007) 17.
[14] R.M. Finch, N.A. Hodge, G.J. Hutchings, A. Meagher, Q.A. Pankhurst, M.R.H.
[53] L.C. Wang, L. He, Y.M. Liu, Y. Cao, H.Y. He, K.N. Fan, J.H. Zhuang, J. Catal. 264
Siddiqui, F.E. Wagner, R. Whyman, Phys. Chem. Chem. Phys. 1 (1999) 485.
(2009) 145.
[15] N.A. Hodge, C.J. Kiely, R. Whyman, M.R.H. Siddiqui, G.J. Hutchings, Q.A.
[54] D.E. Jiang, S.H. Overbury, S. Dai, J. Phys. Chem. Lett. 2 (2011) 1211.
Pankhurst, F.E. Wagner, R.R. Rajaram, S.E. Golunski, Catal. Today 72 (2002) 133.
[55] Q. Fu, H. Saltsburg, M. Flytzani-Stephanopoulos, Science 301 (2003) 935.
[16] S.T. Daniells, A.R. Overweg, M. Makkee, J.A. Moulijn, J. Catal. 230 (2005) 52.
[56] Q. Fu, W.-X. Li, Y. Yao, H. Liu, H.-Y. Su, D. Ma, X.-K. Gu, L. Chen, Z. Wang, H.
[17] G.J. Hutchings, M.S. Hall, A.F. Carley, P. Landon, B.E. Solsona, C.J. Kiely, A.
Zhang, B. Wang, X. Bao, Science 328 (2010) 1141.
Herzing, M. Makkee, J.A. Moulijn, A. Overweg, J.C. Fierro-Gonzalez, J. Guzman,
[57] D.P. Sun, X.K. Gu, R.H. Ouyang, H.Y. Su, Q. Fu, X.H. Bao, W.X. Li, J. Phys. Chem. C
B.C. Gates, J. Catal. 242 (2006) 71.
116 (2012) 7491.
[18] B. Qiao, Y. Deng, Chem. Commun. (2003) 2192.
[58] B.T. Qiao, A.Q. Wang, X.F. Yang, L.F. Allard, Z. Jiang, Y.T. Cui, J.Y. Liu, J. Li, T.
[19] A.K. Tripathi, V.S. Kamble, N.M. Gupta, J. Catal. 187 (1999) 332.
Zhang, Nat. Chem. 3 (2011) 634.
[20] D. Widmann, R.J. Behm, Angew. Chem. Int. Ed. 50 (2011) 10241.
[59] Y. Liu, C.-J. Jia, J. Yamasaki, O. Terasaki, F. Schüth, Angew. Chem. Int. Ed. 49
[21] D. Widmann, Y. Liu, F. Schuth, R.J. Behm, J. Catal. 276 (2010) 292.
(2010) 5771.
[22] B. Qiao, L. Liu, J. Zhang, Y. Deng, J. Catal. 261 (2009) 241.
[60] P. Li, D.E. Miser, S. Rabiei, R.T. Yadav, M.R. Hajaligol, Appl. Catal. B 43 (2003)
[23] J. Lin, L. Li, Y.Q. Huang, W.S. Zhang, X.D. Wang, A.Q. Wang, T. Zhang, J. Phys.
151.
Chem. C 115 (2011) 16509.
[61] M.A. Brown, E. Carrasco, M. Sterrer, H.-J. Freund, J. Am. Chem. Soc. 132 (2010)
[24] J.L. Jambor, J.E. Dutrizac, Chem. Rev. 98 (1998) 2549.
4064.
[25] L. Liu, F. Zhou, L. Wang, X. Qi, F. Shi, Y. Deng, J. Catal. 274 (2010) 1.
[62] K.F. Zhao, B.T. Qiao, J.H. Wang, Y.J. Zhang, T. Zhang, Chem. Commun. 47 (2011)
[26] M. Mihaylov, H. Knözinger, K. Hadjiivanov, B.C. Gates, Chem. Ing. Tech. 79
1779.
(2007) 795.
[63] L. Xu, Y. Ma, Y. Zhang, Z. Jiang, W. Huang, J. Am. Chem. Soc. 131 (2009) 16366.
[27] S. Minico, S. Scire, C. Crisafulli, A.M. Visco, S. Galvagno, Catal. Lett. 47 (1997)
[64] K. Liu, A.Q. Wang, W.S. Zhang, J.H. Wang, Y.Q. Huang, X.D. Wang, J.Y. Shen, T.
273.
Zhang, Ind. Eng. Chem. Res. 50 (2011) 758.
[28] M.M. Schubert, M.J. Kahlich, H.A. Gasteiger, R.J. Behm, J. Power Sources 84
[65] D.A.H. Cunningham, W. Vogel, H. Kageyama, S. Tsubota, M. Haruta, J. Catal. 177
(1999) 175.
(1998) 1.
[29] F. Boccuzzi, A. Chiorino, M. Manzoli, D. Andreeva, T. Tabakova, J. Catal. 188
[66] D.A.H. Cunningham, W. Vogel, M. Haruta, Catal. Lett. 63 (1999) 43.
(1999) 176.
[67] T. Takei, I. Okuda, K.K. Bando, T. Akita, M. Haruta, Chem. Phys. Lett. 493 (2010)
[30] J.C. Fierro-Gonzalez, B.C. Gates, Catal. Today 122 (2007) 201.
207.
[31] M. Li, Z. Wu, Z. Ma, V. Schwartz, D.R. Mullins, S. Dai, S.H. Overbury, J. Catal. 266
[68] G.M. Veith, A.R. Lupini, S.J. Pennycook, N.J. Dudney, ChemCatChem 2 (2010)
(2009) 98.
281.
[32] M. Mihaylov, E. Ivanova, Y. Hao, K. Hadjiivanov, H. Knözinger, B.C. Gates, J.
[69] J.A. Singh, S.H. Overbury, N.J. Dudney, M.J. Li, G.M. Veith, ACS Catal. 2 (2012)
Phys. Chem. C 112 (2008) 18973.
1138.
[33] G. Smit, N. Strukan, M.W.J. Craje, K. Lazard, J. Mol. Catal. A Chem. 252 (2006)
[70] M. Date, M. Okumura, S. Tsubota, M. Haruta, Angew. Chem. Int. Ed. 43 (2004)
163.
2129.
[34] T. Hashimoto, T. Yamada, T. Yoko, J. Appl. Phys. 80 (1996) 3184.
[71] L.M. Liu, B. McAllister, H.Q. Ye, P. Hu, J. Am. Chem. Soc. 128 (2006) 4017.
[35] D. Widmann, R. Leppelt, R.J. Behm, J. Catal. 251 (2007) 437.
[72] J. Lin, B.T. Qiao, J.Y. Liu, Y.Q. Huang, A.Q. Wang, L. Li, W.S. Zhang, L.F. Allard, X.D.
[36] V. Aguilar-Guerrero, B.C. Gates, Catal. Lett. 130 (2009) 108.
Wang, T. Zhang, Angew. Chem. Int. Ed. 51 (2012) 2920.
[37] I.E. Wachs, C.A. Roberts, Chem. Soc. Rev. 39 (2010) 5002.
[73] I.X. Green, W.J. Tang, M. Neurock, J.T. Yates, Science 333 (2011) 736.
[38] J. Zhang, M.J. Li, Z.C. Feng, J. Chen, C. Li, J. Phys. Chem. B 110 (2006) 927.
[74] X.Y. Liu, M.H. Liu, Y.C. Luo, C.Y. Mou, S.D. Lin, H.K. Cheng, J.M. Chen, J.F. Lee, T.S.
[39] L. Mazzetti, P.J. Thistlethwaite, J. Raman Spectrosc. 33 (2002) 104.
Lin, J. Am. Chem. Soc. 134 (2012) 10251.
[40] D.L.A. deFaria, S.V. Silva, M.T. deOliveira, J. Raman Spectrosc. 28 (1997) 873.

You might also like