Operando Raman-online FTIR investigation of ceria, vanadia/ceria and gold/ceria catalysts for toluene elimination

Article abstract of DOI:10.1016/j.jcat.2018.05.001

Toluene oxidation on three ceria catalysts, CeO₂, VO_x/CeO₂ and Au/CeO₂, was investigated by an operando Raman-online FTIR reactor cell. The reactive surface oxygen sites were preferred sites for vanadium and gold deposition. The deposited vanadium existed as V⁵⁺, while most of the gold was in Au⁺ state and roughly a third of ceria was in the reduced Ce³⁺ state. The Au/CeO₂ and CeO₂ catalysts (T₅₀^? = 260 and 290 °C) were more active and selective toward the complete oxidation of toluene than VO_x/CeO₂ catalyst (T₅₀^? = 370 °C), where T₅₀^? refers to the temperature for 50% of CO₂ yield. The η²-peroxide O_2ads^2? was detected on Au/CeO₂ and CeO₂ catalysts, where toluene molecules preferentially adsorbed parallel to the surface via π-bonding. Au/CeO₂ gave complete combustion producing mainly CO₂. On the other hand, η¹-superoxide O_{2 ads}^? was found on VO_x/CeO₂ catalyst and the toluene molecule adsorbed via σ-bonding forming carbenium ions on the vanadia Br?nsted acid sites. This catalyst produced significant amount of benzaldehyde as partial oxidation byproduct and CO (<25%). The nature of the active sites, configurational adsorption of toluene and the reactive oxygen species play important roles in the catalyst activity and selectivity leading to a large contrast in the catalytic behavior of the CeO₂, VO_x/CeO₂ and Au/CeO₂ catalysts.

Full text of DOI:10.1016/j.jcat.2018.05.001

Journal of Catalysis 364 (2018) 80–88
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Operando Raman-online FTIR investigation of ceria, vanadia/ceria
and gold/ceria catalysts for toluene elimination
b,
Qingyue Wang a, King Lun Yeung a, , Miguel A. Bañares
⇑
⇑
a Department of Chemical and Biological Engineering, Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong Special Administrative Region
b Spectroscopy and Industrial Catalysis Group, Instituto de Catálisis y Petroleoquímica, ICP-CSIC, Marie Curie 2, E-28049 Madrid, Spain
a r t i c l e i n f o
a b s t r a c t
Article history:
Toluene oxidation on three ceria catalysts, CeO2, VOx/CeO2 and Au/CeO2, was investigated by an operando
Raman-online FTIR reactor cell. The reactive surface oxygen sites were preferred sites for vanadium and
gold deposition. The deposited vanadium existed as V5+, while most of the gold was in Au+ state and
roughly a third of ceria was in the reduced Ce3+ state. The Au/CeO2 and CeO2 catalysts (T5⁄0 = 260 and
Received 19 January 2018
Revised 4 April 2018
Accepted 1 May 2018
290 °C) were more active and selective toward the complete oxidation of toluene than VOx/CeO2 catalyst
2ꢀ
(T⁄50 = 370 °C), where T5⁄0 refers to the temperature for 50% of CO2 yield. The
g
2-peroxide O2 ads was detected
Keywords:
on Au/CeO2 and CeO2 catalysts, where toluene molecules preferentially adsorbed parallel to the surface via
-bonding. Au/CeO2 gave complete combustion producing mainly CO2. On the other hand,
1-superoxide
Oꢀ2 ads was found on VOx/CeO2 catalyst and the toluene molecule adsorbed via
-bonding forming carbe-
nium ions on the vanadia Brønsted acid sites. This catalyst produced significant amount of benzaldehyde
as partial oxidation byproduct and CO (<25%). The nature of the active sites, configurational adsorption of
toluene and the reactive oxygen species play important roles in the catalyst activity and selectivity leading
to a large contrast in the catalytic behavior of the CeO2, VOx/CeO2 and Au/CeO2 catalysts.
Ó 2018 Elsevier Inc. All rights reserved.
Ceria catalyst
VOCs oxidation reaction
Operando reactor
Vanadium oxides
Gold
p
g
r
1. Introduction
dispersion (<2.5 nm) to yield catalysts of high activity and selectiv-
ity. Transition metal oxide catalysts including CuO, V2O5, MnOx,
Volatile organic compounds (VOCs) are important indoor air
pollutants. Most VOCs are irritants that cause discomfort and
psychological stresses to the occupants. Many VOCs can trigger
autoimmune response (e.g. asthma, eczema) and airway inflamma-
tion causing respiratory distress [1,2], while others such as
formaldehyde, benzene and toluene are known or suspected car-
cinogens [3,4]. Studies have shown that catalytic oxidation is effec-
tive in removing airborne VOCs [5] and ceria is one of the most
promising catalysts [6]. Ceria has a large capacity for oxygen stor-
age and displays good redox property [7]. Ceria itself is active
enough to catalyze the oxidation of hydrocarbons including
toluene. The numerous defects in ceria in form of oxygen vacancies
or polarons are believed to be responsible for its unique catalytic
properties [7–10].
ZrO2, Co3O4 and related mixed oxides have been shown to be active
for VOCs abatement [5,13]. Vanadia in particular is an attractive
catalyst for VOC oxidation not only because of its high turnover
frequency, but also its tolerance to SO2, H2S, NOx, organosulfur
compounds as well as organochloride compounds that are ubiqui-
tous in the environment [14,15]. Dai’s group [16] reported that
VOx/CeO2 can deliver complete conversion of 1,2-dichloroethane
oxidation at 250 °C with no observable deactivation over 360 h.
There are still considerable debates as to the exact reaction
mechanism for VOCs oxidation on ceria catalysts [17,18]. The pre-
sent work employed an operando Raman reactor cell with an online
FTIR to investigate toluene oxidation on ceria catalysts (CeO2, VOx/
CeO2, Au/CeO2) to gain a better understanding of the ‘‘working”
catalyst and identify the surface species during reaction, in order
to establish the structure-activity relationship at a molecular level.
Noble metals supported on ceria are reported to be more active
and selective for catalytic oxidation of VOCs [11]. Ousmane and
coworkers [12] reported Au/CeO2 catalyst is more active than gold
supported on TiO2 and Al2O3. Ceria could stabilize gold at high
2. Experimental
2.1. Preparation of ceria catalysts
⇑
Corresponding authors.
E-mail addresses: kekyeung@ust.hk (K.L. Yeung), miguel.banares@csic.es
Ceria (CeO2) was prepared from 0.01 M Ce(NO3)3 solution
(M.A. Bañares).
obtained by dissolving Ce(NO3)3ꢁ6H2O salt (99%, Sigma-Aldrich)
https://doi.org/10.1016/j.jcat.2018.05.001
0021-9517/Ó 2018 Elsevier Inc. All rights reserved.

Q. Wang et al. / Journal of Catalysis 364 (2018) 80–88
81
in warm deionized distilled (DDI) water. The pH of the solution
was adjusted gradually from pH 4.5 to pH 9 with the addition of
0.5 M NH4OH (BDH) solution. The original clear solution changed
to light pink and then to light yellow following vigorous stirring
(ca. 1000 rpm). The yellow slurry obtained overnight, was washed
repeatedly with warm, distilled water (60 °C) and centrifuged in
JA-14 rotor (14,000 rpm) until the supernatant’s pH was neutral.
The solids were vacuum dried at 65 °C for 4 h, followed by calcina-
tion in flowing air (1000 sccm) at 300 °C for 2 h to produce the
CeO2 catalyst (Ce1).
The 2.5 wt% vanadia on ceria catalyst (2.5VCe) was obtained by
an impregnation method. Briefly, 0.5 g ceria (Ce1) powder was first
dispersed in 250 ml DDI by ultrasound for 30 min to obtain a sus-
pension. The vanadia precursor, (NH4)3[VO2(C2O4)2]ꢁ2H2O, was
obtained by boiling an equal volume mixture of 0.02 M NH4VO3
(98%, Nacalai Tesque) and 0.04 M (NH4)2C2O4 (99%, Sigma-
Aldrich) solutions. Afterward, a 0.024 g (NH4)3[VO2(C2O4)2]ꢁ2H2O
crystal was dissolved in 50 ml DDI water and added to the Ce1 sus-
pension and the mixture was placed in a rotary evaporator kept at
65 °C under a rotation speed of 120 rpm. The resulting paste was
then dried in vacuum at 65 °C for 2 h, before calcining in 1000
sccm flowing air at 300 °C for 2 h to produce a bright yellow
2.5VCe catalyst.
operated at 200 kV with a resolution of 0.29 nm. A selected area
electron diffraction (SAED) and interplanar d-spacing measure-
ment were carried out by Gatan Digital Micrograph program. The
catalyst powder was dispersed in ethanol (Absolute, Merck) and
deposited on a holey carbon TEM grid. The catalysts were also ana-
lyzed by the PANalytical X’pert model X-ray diffractometer (XRD)
and Physical Electronics 5600 X-ray photoelectron spectroscopy
(XPS). The XRD patterns was obtained from powder X-ray diffrac-
tometer with X-ray source of Cu anode, wavelength of 1.5406 Å
and graphite monochromator. The samples were grounded into
fine powder and then placed in a glass sample holder. The XPS
was equipped with a 150 W monochromatic Al K
a X-ray source
(h = 1486.6 eV) and Kratos Axis Ultra DLD spectrometer. The
m
spectra were calibrated with the C1s peak at 285.0 eV. The catalyst
was placed on an indium foil (99.99%) and outgassed under ultra-
high vacuum. Nitrogen physisorption was carried out in a Beckman
Coulter SA 3100 surface area analyzer. The 0.1 g catalyst was out-
gassed in vacuum at 200 °C before nitrogen adsorption at ꢀ196 °C.
The adsorption and desorption isotherm data were used to calcu-
late the Brunauer-Emmett-Teller (BET) surface area, pore volume
and the Barrett-Joyner-Halenda (BJH) pore size distribution.
The 2.5 wt% gold on ceria catalyst (2.5AuCe) was prepared by
the following procedure. 0.5 g Ce1 was dispersed by ultrasound
in 250 ml DDI water before the drop-wise addition of a 0.001 M
gold solution. The solution was prepared from HAuCl4ꢁxH2O
(99.999%, Sigma-Aldrich) after adjusting the pH to 8.5 by adding
0.05 M NH4OH solution. After vigorous mixing at 1200 rpm at 80
°C for 2 h, the mixture was cooled to room temperature and stirred
overnight. The resulting slurry was washed repeatedly with warm,
distilled water (60 °C) and centrifuged (14,000 rpm) to remove
ammonium and chloride ions. The paste was then vacuum dried
at 65 °C for 4 h, and calcined in 1000 sccm flowing air at 300 °C
for 2 h. The Ce1, 2.5VCe and 2.5AuCe catalysts were stored in
sealed containers under dry atmosphere before use.
2.3. Operando reaction study
The toluene oxidation reaction was carried out in an operando
fixed bed reactor cell made of optical-grade quartz. The reactor fit-
ted snuggly into an insulated heating block as shown in Fig. 1 [19].
The reaction temperature was monitored by a type 2 thermocouple
(thermocoax) inserted into the catalyst bed and controlled by a
temperature programmer unit (PID Eng&Tech). The catalyst was
observed in situ by a Renishaw micro-Raman System 1000 using
a LEICA DMILM microscope with a 20ꢂ long-distance objective
lens and the spectra were obtained under Argon laser (514 nm)
excitation at 4 scans and 30 s acquisition time. The composition
of the reaction gases were analyzed by an on-line ThermoNicolet
6700 Fourier transformed infrared (FTIR) spectrometer fitted with
a thermostated (120 °C) gas cell. The spectra were collected from
650 to 4000 cmꢀ1 with a resolution of 4 cmꢀ1 and sampling
interval of 9.85 s. FTIR calibrations were made and the details are
discussed in the Supplementary Information Part 1 (SI. Part 1).
2.2. Characterization of ceria catalysts
The Ce1, 2.5VCe, 2.5AuCe catalysts were examined by JEOL 2100
high resolution transmission electron microscopy (HRTEM)
Fig. 1. Schematic diagram of the operando reactor setup equipped with in situ Raman and online FTIR spectrometers.

82
Q. Wang et al. / Journal of Catalysis 364 (2018) 80–88
Fig. 2. High-resolution transmission electron microscopy (HRTEM) images of (a) 2.5VCe and (b) 2.5AuCe with the selected area electron diffraction (SAED) in the figure insets.
Fig. 4. Nitrogen physisorption isotherms of 2.5VCe, Ce1 and 2.5AuCe catalysts with
their corresponding pore size distributions shown in the figure inset.
Fig. 3. Powder X-ray diffraction patterns of 2.5VCe, Ce1 and 2.5AuCe catalysts.
Table 1
Particle sizes of CeO2 particles in 2.5VCe, Ce1 and
2.5AuCe catalysts as determined from TEM and XRD
measurements.
Table 2
Textural properties of 2.5VCe, Ce1 and 2.5AuCe catalysts.
Sample
BET surface area
Pore volume
Pore size
(nm)
(m2ꢁgꢀ1
)
(cm3ꢁgꢀ1
)
Catalyst
CeO2 Particle size (nm)
TEM
XRD
2.5VCe
Ce1
2.5AuCe
123.8
121.9
133.9
0.17
0.17
0.20
5.1
5.1
5.1
2.5VCe
Ce1
2.5AuCe
11
12
11
4
4
4
12.6
12.8
10.6
(PolyScience 9106A12E), then concentrated toluene vapor was
blended downstream with synthetic air (UHP, Air Liquide) using
a pair of mass flow controllers (PID Eng&Tech) to obtain 1000
ppm toluene in 18.8% O2, 5.8% Ar and 75.4% N2, and a gas hourly
speed volume (GHSV) of 34,500 hꢀ1. The reaction was carried out
at temperatures between 25 and 450 °C.
Detailed study of the catalyst stability under humid conditions
(SI. Part 2) was carried out at 26,000 ppm H2O (19.8 Torr) with cat-
alyst subjected to temperature from 100 to 350 °C. Another mode
with 692,000 ppm H2O (634.0 Torr) representing extreme humid
conditions with a relative humidity level of 5.44% at 200 °C was
The catalyst was ground in an agate mortar and sieved to pro-
duce 0.25–0.50 mm powder (i.e. 32–60 mesh). A 0.150 g catalyst
was packed into the operando reactor cell. The catalyst bed was
sandwiched between two inert SiC (Aldrich) beds to facilitate heat
transfer and minimize the void volume. The ensemble was held in
place by quartz wool. The catalyst was pretreated at 300 °C in flow-
ing 200 sccm synthetic air (20% O2/80% N2, Air Liquide) for 2 h
before reaction. The toluene vapor was generated in argon (UHP,
Air Liquide) by bubbler kept at 10 °C in a circulating oil bath

Q. Wang et al. / Journal of Catalysis 364 (2018) 80–88
83
Fig. 5. X-ray photoelectron spectra of (a) Ce3d (b) O1s (c) V2p (d) Au4f from Ce1, 2.5VCe and 2.5AuCe catalysts.
also tested. A typical catalytic stability on-stream experiment was
run for 48 h under continuous catalytic process.
of the catalysts are summarized in Table 2. In general, all three cat-
alysts display similar type IV mesoporous adsorption-desorption
isotherms with a type H2 hysteresis loop [21]. The three catalysts
have comparable surface area of 128 8 m2 gꢀ1. The pore size dis-
tribution was calculated from adsorption branch of the isotherm
using BJH model [22] and plotted as an inset in Fig. 4. All catalysts
have the same average pore size of 5.1 nm and pore volume of
3. Results and discussion
3.1. Characterizations
0.17–0.20 cm3ꢁgꢀ1
.
The transmission electron micrographs of 2.5VCe and 2.5AuCe
catalysts are shown in Fig. 2a and b respectively with high-
resolution transmission electron microscopy (HRTEM) images
and selected area electron diffraction (SAED) patterns. For refer-
ence, the micrograph for their support Ce1 is included in SI. Part
3. The powder X-ray diffraction patterns (XRD) are presented in
Fig. 3. Ceria displays predominantly CeO2 (1 1 1), (2 0 0), (2 2 0),
(3 3 1) and (2 2 2) crystallographic planes, which is observed on
both HRTEM/SAED and XRD analysis. Particle sizes estimated from
the TEM micrographs are consistent with the values determined
from X-ray diffraction patterns (Table 1). The detailed calculations
based on Debye-Scherrer equation are given in SI. Part 4. The vana-
dia species are highly dispersed and crystalline V2O5 could not be
detected by SAED (Fig. 2a) and XRD (Fig. 3). Gold is not evident
in the XRD of 2.5AuCe (Fig. 3), but it can be clearly observed in
the micrographs (Fig. 2b). The Au (1 1 1) was detected by SAED
and can be identified in the HRTEM images from its characteristic
d-spacing of 0.23 nm. The gold nanoparticles on ceria was also con-
firmed by energy dispersive X-ray spectroscopy (EDX) analysis,
shown as SI. Part. 3. The gold has a mean particle size of 4 1 n
m. It appears to be supported mainly on CeO2 (1 1 1), which agrees
with the literature report [20].
The XPS spectra of Ce3d, O1s, V2p and Au4f are plotted in Fig. 5
for Ce1, 2.5VCe and 2.5AuCe catalysts. Fig. 5a presents five peaks
belonging to Ce3d3/2 with two main bands (u0, u) and three satel-
lite bands (u0, u00, u000), and another five peaks belonging to Ce3d5/2
with two main bands (v0, v) and three satellite bands (v0, v00, v000).
The Ce3+ shows Ce3d3/2 bands, u0 (900.3 eV) and u0 (902.2 eV),
and Ce 3d5/2 bands v0 (880.2 eV) and v0 (884.3 eV), while Ce4+ dis-
plays Ce3d3/2 bands, u (901.2 eV), u00 (907.9 eV), u000 (917.0 eV), and
Ce3d5/2 bands, v (882.5 eV), v00 (889.3 eV), v000 (898.6 eV) [23,24].
The plots of O1s spectra in Fig. 5b indicates that in addition to
the lattice oxygen O (529.5 eV), there is a separate band Ob
a
(531.6 eV) belonging to the oxygen near vacancy sites [8,25,26].
The summary data in Table 3 shows that oxygen vacancies and
reduced Ce3+ are abundant in all three catalysts. The deposition
of vanadium and gold leads to 0.4 eV upward binding energy shift
of O and Ob (Fig. 5b) and a decrease of Ob/(O + Ob). This is partic-
a
a
ularly evident for 2.5VCe, suggesting that vanadium bonds on the
oxygen sites near vacancies (Ob). Gold adatom appears to nucleate
on the oxygen vacancy site of ceria [27]. Table 3 also shows catalyst
deposition caused a significant increase (ca. 10%) in the fraction of
Ce3+ from around 0.3 to 0.33.
Fig. 5c shows vanadia in 2.5VCe exists mainly as V5+ (517.4 eV).
The characteristic V4+ (516.4 eV) and V3+ (515.5 eV) were not
The nitrogen physisorption isotherms of Ce1, 2.5VCe and
2.5AuCe catalysts are plotted in Fig. 4 and the textural properties
Table 3
XPS analysis of surface species concentrations. (calculated based on peak area).
Sample
Ce3+/(Ce4+ + Ce3+
)
Ob/(O + Ob)
V5+, V4+, V3+ & Au3+, Au+, Au0
a
2.5VCe
Ce1
2.5AuCe
32.4%
29.8%
33.7%
21.7%
45.7%
36.8%
V5+ 99.0%
V4+ 1.0%
V3+ 0%
Au3+ 10.0%
Au+ 73.2%
Au0 16.7%

84
Q. Wang et al. / Journal of Catalysis 364 (2018) 80–88
detected [28]. Nearly all the vanadium formed V5+AOACe3+ with
the sacrificial reduction of ceria at the interface [29]. Bañares and
coworkers [19,29] provided experimental evidence that a strong
interaction between vanadia and ceria would reduce ceria sites
at the interface. Discussion with Sauer’s group delivered a DFT
calculation that provided insight on the nature of the strong
interaction between vanadia and ceria responsible for the partial
reduction of Ce [30], of which the experimental evidence has been
reported recently [31]. The gold in 2.5AuCe catalyst exists mainly
as Au+ (88.6 eV and 85.0 eV) followed by Au0 (88.4 eV and 84.4
eV), and Au3+ (90.2 eV and 86.3 eV) as shown by Fig. 5d [32]. The
Au+ (73.2%) is reported to promote the reducibility of ceria and
enhance its catalytic activity [33,34].
3.2. Operando study of toluene oxidation reaction
The toluene oxidation was carried out in an operando fixed-bed
reactor cell. The reaction over the ceria catalyst was observed
Fig. 6. Operando Raman spectra of (a) 2.5VCe, (b) Ce1, (c) 2.5AuCe catalysts and Raman contour projection of (d) 2.5VCe, (e) Ce1, (f) 2.5AuCe. [Notes: The color on Raman
contour maps is qualitative relative intensity.]
Fig. 7. Operando Raman bands intensity of ring breathing mode v1 of
p
-bonding (at 646 cmꢀ1) and
-bonding,
r
-bonding (at 786 cmꢀ1) for (a) 2.5VCe, (b) Ce1, (c) 2.5AuCe, and intensity
-bonding, N-C7H+9.]
of carbenium ion [C7H+9] (at 1346 cmꢀ1) for (d) 2.5VCe, (e) Ce1, (f) 2.5AuCe. [Notes:
-
r
-p

Q. Wang et al. / Journal of Catalysis 364 (2018) 80–88
85
Fig. 8. Relative intensity ratio ID/IF2g Raman bands of CeO2 for (a) 2.5VCe (b) Ce1 (c) 2.5AuCe; Operando Raman peak intensity of superoxide Oꢀ2 ads (at 1121 cmꢀ1) for (d)
2ꢀ
2.5VCe and peroxide O2 ads (at 838 cmꢀ1) for (e) Ce1 and (f) 2.5AuCe. [Notes: h-ID/IF2g
,
-Oꢀ2 ads-superoxide, -O2 a2dꢀs-peroxide.]
in situ by micro-Raman spectroscopy, while the gas composition at
the reactor outlet was monitored in real-time by an online FTIR.
Fig. 6a–c plot the operando Raman spectra of the catalysts at repre-
sentative reaction temperatures ranging from 25 to 450 °C with the
corresponding contour maps shown in Fig. 6d–f. The Raman bands
intensity are normalized to the intensity of the ceria F2g mode at
457 cmꢀ1. The other Raman peaks of CeO2 include 590 and 1179
cmꢀ1 have been respectively assigned to defect-related (D) band
and the second longitudinal optical (2LO) modes [35,36]. The
Raman bands for vanadia at 1019 and 1033 cmꢀ1 belong to the
vanadyl mode of ceria-supported vanadia species in 2.5VCe cata-
lyst [30,37]. These bands are obscured by toluene during reaction,
but reappear following complete conversion of toluene at temper-
atures above 350 °C. Gold usually does not display any characteris-
tic Raman signals itself, but its presence causes broadening of the
CeO2 peak at 457 cmꢀ1, implying smaller particle size following
gold deposition [38] which is consistent with the XRD results
(Table 1).
ion is formed at the Brønsted acid sites of vanadia [44,45]. Further-
more, the absence of C6H5-CH3 stretching vibration mode m13
(1209 cmꢀ1) and CH3 bending mode (1380 cmꢀ1) suggests a distor-
tion of methyl group following adsorption [40]. Both 786 cmꢀ1 and
1346 cmꢀ1 Raman bands disappear at high temperatures with the
conversion of toluene over 2.5VCe catalyst. In contrast, the toluene
is adsorbed by
p-bonding on both Ce1 and 2.5AuCe catalysts as
shown in Fig. 7b and c, respectively. The Raman signals for carbe-
nium are absent in these two catalysts (Fig. 7e and f). The contour
maps in Fig. 6d–f provide an overview of the effects of reaction
temperature on the catalyst and surface adsorbates. It can be seen
from the data that carbonaceous species build-up is most severe on
2.5VCe followed by Ce1 and 2.5AuCe, and their disappearance
occurs at temperatures in order of 2.5VCe > Ce1 > 2.5AuCe. The
carbon balance of 2.5VCe, Ce1, and 2.5AuCe is shown in Fig. S10,
and thermogravimetric analysis confirms 2.5VCe has the most sev-
ere carbonaceous deposits among the three catalysts. (Fig. S11).
Besides the adsorbed toluene molecules, reactive oxygen
species were detected on the catalysts. Observation of reactive
oxygen species on ceria surface is rare under reaction conditions
particularly at high temperatures [36,46]. The Raman signals for
The adsorbed toluene displays red shifted m
8a (1590 cmꢀ1), m19a
(1495 cmꢀ1) and m12 (1001 cmꢀ1) modes from the aromatic ring
and the m9a (1161 cmꢀ1) and m18a (1027 cmꢀ1) from the in plane
CAH bending modes [39–42]. There is a noticeable suppression
of m2 CAH stretching vibration at 3068 cmꢀ1 in 2.5AuCe and Ce1
compared to 2.5VCe. The phenyl ring adsorbed in flat orientation
is reported to cause weaker m2 CAH stretching vibration and an
observable frequency downshift and broadening of m1 ring-
breathing vibration compared to adsorption of toluene perpendic-
ular to the surface [41,42]. The m1 ring-breathing mode indicates
g
1-superoxide Oꢀ2 ads (1121 cmꢀ1) on 2.5VCe (Fig. 8d) and g2
-
peroxide O2 ads (838 cmꢀ1) on Ce1 and 2.5AuCe catalysts (Fig. 8e
and f) [36,46] were detected due to the reducing effects of toluene.
Superoxide species was observed by Long et al. [47] during alkane
oxidations (i.e. methane and ethane) up to temperatures of 750 °C.
The spectral assignments for these reactive oxygen species have
been confirmed in a recent study on polycrystalline ceria [48].
2ꢀ
-bonding (786 cmꢀ1) on 2.5VCe catalyst (Fig. 6a) and
p-bonding
The Oꢀ2 ads and O2 ads species not only participate in the oxidation
2ꢀ
r
(646 cmꢀ1, belongs to benzoate intermedia) on Ce1 and 2.5AuCe
catalysts (Fig. 6b and c) [41,42]. Plotting these characteristic
of toluene, they also play an important role in the re-oxidation of
2ꢀ
2ꢀ
CeO2 via Oꢀ2 ads ? O2 ads ? 2Oꢀads ? 2O
[7,36]. The presence of
lattice
Raman bands in Fig. 7 shows that toluene is adsorbed via the
bonding on all three ceria catalysts at low temperatures (<100
°C). Above 100 °C, the adsorbed toluene forms -bonding on
p
-
Oꢀ2 ads on 2.5VCe is consistent with the reducing effects of vanadia
on ceria and an indication of the degree of reduction of the ceria
support [46].
r
2.5VCe catalyst (Fig. 7a) with a concomitant appearance of Raman
band at 1346 cmꢀ1 (along with 1239 cmꢀ1) that belong to carbe-
nium ion (Fig. 7d) [43,44]. Studies suggest that the carbenium
Oxygen vacancies account for most of the defects found in CeO2
and the ceria Raman intensity ratio ID/IF2g shows there are less oxy-
gen vacancies in 2.5VCe and 2.5AuCe than in Ce1(Fig. 8a–c). This is

86
Q. Wang et al. / Journal of Catalysis 364 (2018) 80–88
in CeO2 2LO Raman signal (1179 cmꢀ1) (Fig. 8f. inset), which indi-
cates that gold species facilitates the re-oxidation of CeO2. In short,
the Raman spectra suggest that vanadia hinders re-oxidation of
reduced ceria, while gold promotes its re-oxidation.
The composition of the reaction gas was monitored by an online
FTIR, representative spectra are plotted in Fig. 9 for Ce1, 2.5VCe and
2.5AuCe catalysts. The online concentrations of toluene, CO2 and
H2O are determined from gas calibration and plotted in Fig. 10a–c
for the three catalysts. There is a significant toluene conversion over
2.5VCe at low temperatures, but the main products are benzalde-
hyde and CO (Fig. 10a). The selectivity for benzaldehyde is up to
40%. (SI. Part. 7). Ignition occurs at 300 °C with a substantial CO2
generation, but CO concentration remains considerable. In contrast,
Ce1 and 2.5AuCe catalysts produce less byproducts (Fig. 10b and c).
Also, both catalysts have significantly lower ignition temperatures
of 200 °C. 2.5AuCe performs better than Ce1, delivering nearly com-
plete toluene oxidation to CO2 and H2O at 300 °C.
Fig. 11 plots the conversion, yields of CO2 and byproducts (CO &
C6H5CHO) at different reaction temperatures. Fig. 11a shows that
2.5AuCe is most active catalyst with conversion at 300 °C (C300
)
of 96%, followed by Ce1 (70%) and 2.5VCe (23%). It is also the best
catalyst for deep oxidation of toluene according to CO2 yield with
only a trace amount of CO produced (Fig. 11b). Ce1 has similar
T*50 of 290 °C, but toluene conversion never exceeded 90% with par-
tial oxidation products (ca. 25%). On these two catalysts, toluene is
adsorbed parallel to the surface via
p-bonding (Fig. 7b and c). A
strong -interaction destabilizes the aromatic ring resulting in ring
p
decomposition and oxidation. Toluene oxidation over Ce1 catalyst
is believed to proceed Mars-van Krevelen reaction mechanism
[11,13]. The role of gold on the redox cycle (Ce3+ to Ce4+) of ceria
has been proposed by S. Scirè’s group [49] and they suggested that
gold on ceria weaken the CeAO bond and further increase the
mobility of the surface lattice oxygen, thus promoting toluene
oxidation.
Low temperature toluene conversion over 2.5VCe produces
mostly partial oxidation products, like benzaldehyde and CO. Sig-
nificant toluene conversion to CO2 does not occur until 370 °C
(T*50), and yield of byproducts remains high even at 450 °C. Toluene
is adsorbed on 2.5VCe catalyst via the methyl group forming
r-
Fig. 9. Representative online FTIR spectra of toluene oxidation outgas released over
bonding with the surface. The clear observation of carbenium ion
on operando Raman spectra suggests that adsorption is forming
by adsorption on Brønsted acid site of V5+AOACe3+ in 2.5VCe. This
leads to partial oxidation and production of benzaldehyde. Vana-
dium also titrates the re-oxidation of ceria via surface oxygen
vacancies and hinders access to active Ce4+ sites for complete oxi-
dation of toluene.
(a) 2.5VCe (b) Ce1 (c) 2.5AuCe catalysts in the operando reactor.
consistent with the XPS results in Table 3. Along with increasing
temperature, the defects sites in CeO2 lattice (indicated by ID/IF2g
decrease with increasing adsorbed oxygen species (Fig. 8d–f). The
plot of
1-superoxide O2ꢀ ads on 2.5VCe (Fig. 8d) suggests that the
)
g
Dispersion of gold and vanadium on ceria improves the stability
of the catalyst to water vapor (SI. Part 2). Toluene oxidation
reaction in the presence of 26,000 ppm H2O has led to a significant
decrease in Ce1 activity particularly in terms of CO2 yield (Fig. S5
strong interactions between ceria and vanadia was hindering the
re-oxidation of the reduced ceria [29,30]. On the other hand, g2
-
2ꢀ
peroxide O2 ads on 2.5AuCe (Fig. 8f) increases with temperature
up to 300 °C, before a rapid decline with a concomitant increase
Fig. 10. Integration of characteristic online FTIR peaks for (a) 2.5VCe, (b) Ce1 and (c) 2.5AuCe catalysts. [Notes: j-C6H5CH3, -CO2, -H2O, -CO, -C6H5CHO.]

Q. Wang et al. / Journal of Catalysis 364 (2018) 80–88
87
Fig. 11. (a) Conversion and (b) CO2 yield and byproduct yield of 2.5VCe, Ce1 and 2.5AuCe catalysts in the operando reactor.
and Table S2). The effect is reversible and the catalyst regained its
original activity once the water was turned off, as observed from a
48 h experiment. (Fig. S6). Water adsorption particularly on the
defect sites of Ce1 is believed to play a role in decrease of catalytic
performance under humid condition. 2.5AuCe and 2.5VCe with Au
and V occupying the defect sites were less affected by the water
vapor. This is particularly true for 2.5AuCe which maintained its
activity even under 692,000 ppm H2O (Table S3).
different reaction behavior. Vanadium and gold nucleate on oxy-
gen vacancy sites of CeO2 and have a reducing effect on the ceria
with concomitant increase in Ce3+. Vanadium and gold mainly
exist in the oxidized states of V5+ (99.0%) and Au+ (73.2%). Toluene
molecule is adsorbed with its aromatic ring parallel to the surface
of Ce1 and 2.5AuCe catalysts via p-bonding that appears to desta-
bilize the aromatic ring resulting in a more rapid and complete oxi-
2ꢀ
dation to CO2 and H2O. For both catalysts,
g
2-peroxide O2 ads shows
The Raman spectra in Fig. 6 indicate that toluene adsorption on
up on operando Raman spectra, but gold appears to catalyze the
2-peroxide O2 ads in the re-oxidation of CeO2.
2ꢀ
2.5VCe catalyst is via
r
-bonding of methyl group on the catalyst
transformation of
g
surface. Similar toluene adsorption configuration on V2O5/TiO2 cat-
alysts was observed by van Hengstum and coworkers during
toluene selective oxidation to benzaldehyde/benzoic [50]. On
2.5VCe catalyst, benzyl carbenium ion is formed when toluene
adsorbed on the Brønsted acid sites of vanadia (Fig. 7), which fur-
ther leads to formation of carbonaceous deposits on surface
(Fig. S11) and high concentrations of partial oxidation products
(Figs. 10 and 11). On ceria surface, toluene adsorbs parallel to the
This could explain the better performance of 2.5AuCe for complete
oxidation of toluene. Toluene is adsorbed as carbenium ion on
Brønsted acid sites of 2.5VCe and produces benzaldehyde as the
main partial oꢀxidation byproduct. 2.5VCe appears to stabilize g1
-
superoxide O2 ads preventing the re-oxidation of CeO2, which
explained its poorer oxidation performance with significant CO
production.
surface via
p-bonding, destabilizing the aromatic ring and result-
Acknowledgments
ing in higher oxidation rates on Ce1 and 2.5AuCe catalysts. In the
latter catalyst, gold is reported to enhance the oxidation reaction
by increasing the reducibility of the lattice oxygen of ceria and pos-
sibly their mobility [11]. López et al. [8] suggested that the preva-
lence of surface oxygen vacancies play an important role in toluene
oxidation over ceria catalysts. They found that catalysts with a high
surface oxygen-to-bulk oxygen ratio is more active for toluene oxi-
dations, as measured by temperature programmed reduction (TPR)
experiments [8]. Similar positive correlation was observed in this
This work was financial supported by the Spanish Ministry
[CTQ2014-57578-R] and [CTM2017-82335-R]; NSFC/RGC Joint
Research Scheme [N-HKUST626/13]; European Commission Grant
760928 BIORIMA and Guangzhou Collaborative Innovation Key
Program [201704030074]. The authors express their gratitude to
Dr. Laura Pascual, Dr. Wei Han, Ms. Mei Leng Liu, Mr. Hoi Yau
Cheng and Mr. Nick K C Ho for technical supports. Also, thanks to
the Oversea Research Award from School of Engineering, HKUST
for supporting the exchange program at the ICP-CSIC, in Spain.
2ꢀ
work between the surface oxygen species O2 ads (or Oꢀ2 ads) in
Fig. 8 and the catalytic activity in Fig. 11. The strong adsorption
of benzyl alcohol and styrene on Au(1 1 1) with phenyl-ring paral-
lel to surface was indicated by C.M. Friend’s group [51,52], in line
with the presence of benzoate intermediates [53], which would
be readily burnt without intermediate oxidation products [49].
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in
the online version, at https://doi.org/10.1016/j.jcat.2018.05.001.
Furthermore, small size gold nanoparticles (ca. 2–4 nm) is said to
2ꢀ
have influence on the surface peroxide species (O
) [54], as also
2 ads
References
observed in this study (Fig. 8). The experimental results of this
study indicate that gold is responsible for 2.5AuCe catalyst’s high
selectivity for toluene oxidation to carbon dioxide, either directly
as a catalyst or indirectly by facilitating the re-oxidation of ceria
(Fig. 8). The 2.5AuCe catalyst has the added advantage of being
stable even at high humidity conditions (SI. Part. 2).
[1] U. Nurmatov, N. Tagiyeva, S. Semple, G. Devereux, A. Sheikh, Eur. Respir. Rev.
24 (2015) 92–101.
[2] M. Hulin, M. Simoni, G. Viegi, I. Annesi-Maesano, Eur. Respir. J. 40 (2012) 239–
254.
[3] S. Burgaz, O. Erdem, G. Çakmak, N. Erdem, A. Karakaya, A.E. Karakaya,
Biomarkers 7 (2002) 151–161.
[4] J.E. Huff, J.K. Haseman, D.M. DeMarini, S. Eustis, R.R. Maronpot, A.C. Peters, R.L.
Persing, C.E. Chrisp, A.C. Jacobs, Environ. Health Perspect. 82 (1989) 125–163.
[5] M.S. Kamal, S.A. Razzak, M.M. Hossain, Atmos. Environ. 140 (2016) 117–134.
[6] L. Vivier, D. Duprez, ChemSusChem 3 (2010) 654–678.
[7] A. Trovarelli, Catal. Rev. – Sci. Eng. 38 (1996) 439–520.
[8] J.M. López, A.L. Gilbank, T. García, B. Solsona, S. Agouram, L. Torrente-
Murciano, Appl. Catal. B 174–175 (2015) 403–412.
4. Conclusions
The CeO2, VOx/CeO2 and Au/CeO2 catalysts present different
active sites for toluene adsorption and reaction, resulting in very

88
Q. Wang et al. / Journal of Catalysis 364 (2018) 80–88
[9] Y.M. Chiang, E.B. Lavik, I. Kosacki, H.L. Tuller, J.Y. Ying, Appl. Phys. Lett. 69
[33] H. Bake, S.F. Zaman, Y.A. Alhamed, A.A. Al-Zahrani, M.A. Daous, S.U. Rather, H.
Driss, L.A. Petrov, RSC Adv. 6 (2016) 22555–22562.
[34] R. Si, M. Flytzani-Stephanopoulos, Angew. Chem. Int. Ed. 47 (2008) 2884–
2887.
(1996) 185–187.
[10] A. Trovarelli, Catalysis by Ceria and Related Materials, Imperial College Press,
London, 2002, pp. 15–50.
[11] S. Scirè, P.M. Riccobene, C. Crisafulli, Appl. Catal. B 101 (2010) 109–117.
[12] M. Ousmane, L.F. Liotta, G. Di Carlo, G. Pantaleo, A.M. Venezia, G. Deganello, L.
Retailleau, A. Boreave, A. Giroir-Fendler, Appl. Catal. B 101 (2011) 629–637.
[13] W.B. Li, J.X. Wang, H. Gong, Catal. Today 148 (2009) 81–87.
[14] C. Cho, S. Ihm, Environ. Sci. Technol. 36 (2002) 1600–1606.
[15] J.J. Spivey, J.B. Butt, Catal. Today 11 (1992) 464–500.
[16] Q. Dai, S. Bai, H. Li, W. Liu, X. Wang, G. Lu, Appl. Catal. B 168–169 (2015) 141–
155.
[35] Z. Wu, A.J. Rondinone, I.N. Ivanov, S.H. Overbury, J. Phys. Chem C. 115 (2011)
25368–25378.
[36] Z. Wu, M. Li, J. Howe, H.M. Meyer, S.H. Overbury, Langmuir 26 (2010) 16595–
16606.
[37] M.V. Bosco, M.A. Bañares, M.V. Martínez-Huerta, A.L. Bonivardi, S.E. Collins, J.
Mol. Catal. A 408 (2015) 75–84.
[38] J.E. Spainer, R.D. Robinso, F. Zhang, S. Chan, I.P. Herman, Phys. Rev. B 64 (2001)
245407.
[17] D.M. Gómez, V.V. Galvita, J.M. Gatica, H. Vidal, G.B. Marin, Phys. Chem. Chem.
Phys. 16 (2014) 11447–11455.
[18] N. Burgos, M. Paulis, M. Antxustegi, M. Montes, Appl. Catal. B 38 (2002) 251–
1158.
[19] M.V. Martínez-Huerta, G. Deo, J.L.G. Fierro, M.A. Bañares, J. Phys. Chem. C 112
(2008) 11441–11447.
[39] D. Lin-Vien, N.B. Colthup, W.G. Fateley, J.G. Grasselli, The Handbook of Infrared
and Raman Characteristic Frequencies of Organic Molecules, Academic Press,
San Diego, 1991, p. 412.
[40] M. Tasumi, T. Urano, M. Nakata, J. Mol. Struct. 146 (1986) 383–396.
[41] Y.J. Kwon, D.H. Son, S.J. Ahn, M.S. Kim, K. Kim, J. Phys. Chem. 98 (1994) 8481–
8487.
[20] T. Akita, M. Okumura, K. Tanaka, M. Kohyama, M. Haruta, J. Mater. Sci. 40
(2005) 3101–3106.
[21] G. Leofanti, M. Padovan, G. Tozzola, B. Venturelli, Catal. Today 41 (1998) 207–
219.
[42] X. Gao, J.P. Davies, M.J. Weaver, J. Phys. Chem. 94 (1990) 6858–6864.
[43] A.A. Davydov, Mater. Chem. Phys. 19 (1988) 97–112.
[44] A.A. Davydov, A.A. Budneva, E.A. Paukshtis, React. Kinet. Catal. Lett. 39 (1989)
419–424.
[22] K.S.W. Sing, Pure Appl. Chem. 57 (1985) 603–619.
[23] A. Pfau, K. Schierbaum, Surf. Sci. 321 (1994) 71–80.
[24] M.M. Natile, A. Glisenti, Surf. Sci. Spectra 13 (2006) 17–30.
[25] H. Huang, Y. Gu, J. Zhao, X. Wang, J. Catal. 326 (2015) 54–68.
[26] X. Wang, B. Chen, G. Chen, X. Sun, RSC Adv. 6 (2016) 87978–87987.
[27] Z. Liu, S.J. Jenkins, D.A. King, Phys. Rev. Lett. 94 (2005) 196102.
[28] A. Shi, X. Wang, T. Yu, M. Shen, Appl. Catal. B 106 (2011) 359–369.
[29] M.V. Martínez-Huerta, J.M. Coronado, M. Fernández-García, A. Iglesias-Juez, G.
Deo, J.L.G. Fierro, M.A. Bañares, J. Catal. 225 (2004) 240–248.
[30] M. Baron, H. Abbott, O. Bondarchuk, D. Stacchiola, A. Uhl, S. Shaikhutdinov, H.
Freund, C. Popa, M.V. Ganduglia-Pirovano, Angew. Chem. Int. Ed. 48 (2009)
8006–8009.
[45] M. Badlani, I.E. Wachs, Catal. Lett. 75 (2001) 137–149.
[46] V.V. Pushkarev, V.I. Kovalchuk, J.L. D’Itri, J. Phys. Chem. B 108 (2004) 5341–
5348.
[47] R. Long, Y. Huang, H. Wan, J. Raman Spectrosc. 28 (1997) 29–32.
[48] C. Schilling, A. Hofmann, C. Hess, M.V. Ganduglia-Pirovano, J. Phys. Chem. C
121 (2017) 20834–20849.
[49] S. Scirè, S. Minicò, C. Crisafulli, C. Satriano, A. Pistone, Appl. Catal. B 40 (2003)
43–49.
[50] A.J. Hengstum, J. Pranger, S.M. Hengstum-Nijhuis, J.G. Ommen, P.J. Gellings, J.
Catal. 101 (1986) 323–330.
[51] J.C.F. Rodríguez-Reyes, C.M. Friend, R.J. Madix, Surf. Sci. 506 (2012) 1129–
1134.
[31] A. Iglesias-Juez, M.V. Martínez-Huerta, E. Rojas-García, J.-M. Jehng, M.A.
Bañares, J. Phys. Chem. C 122 (2018) 1197–1205.
[32] A. Karpenko, R. Leppelt, V. Plzak, R.J. Behm, J. Catal. 252 (2007) 231–242.
[52] R.G. Quiller, X. Liu, C.M. Friend, Chem. Asian J. 5 (2010) 78–86.
[53] X.E. Verykios, F.P. Stein, R.W. Coughlin, Catal. Rev. Sci. Eng. 22 (1980) 197–234.
[54] B.K. Min, C.M. Friend, Chem. Rev. 107 (2007) 2709–2724.