Insights into the Oxygen Vacancy Filling Mechanism in CuO/CeO
2
Catalysts: A Key Step Toward High Selectivity in Preferential CO
Oxidation
Arantxa Davo
-Quin
onero,* Esther Bailo
n-García, Sergio Lo
pez-Rodríguez, Jero
nimo Juan-Juan,
Dolores Lozano-Castello
, Max García-Melchor, Facundo C. Herrera, Eric Pellegrin, Carlos Escudero,
and Agustín Bueno-Lo
pez*
Cite This: ACS Catal. 2020, 10, 6532−6545
Read Online
ACCESS
Metrics & More Article Recommendations
*
sı
Supporting Information
ABSTRACT: The preferential CO oxidation (CO-PROX) re-
action is paramount for the purification of reformate H
2
-rich
streams, where CuO/CeO
2
catalysts show promising opportunities.
This work sheds light on the lattice oxygen recovery mechanism on
CuO/CeO
2
catalysts during CO-PROX reaction, which is critical
to guarantee both good activity and selectivity, but that is yet to be
well understood. Particularly, in situ Raman spectroscopy reveals
that oxygen vacancies in the ceria lattice do not form in significant
amounts until advanced reaction degrees, whereas pulse O
2
isotopic tests confirm the involvement of catalyst oxygen in the
CO and H
2
oxidation processes occurring at all stages of the CO-
PROX reaction (Mars−van Krevelen). Further mechanistic insights are provided by operando near-ambient pressure X-ray
photoelectron spectroscopy (NAP−XPS) and near edge X-ray absorption fine structure (NEXAFS) experiments, which prove the
gradual CuO reduction and steady oxidized state of Ce ions until the very surface reduction of CeO
2
at the point of selectivity loss.
Experiments are complemented by density functional theory (DFT) calculations, which reveal a more facile oxygen refill according
to the trend CuO > CeO
2
>Cu
2
O. Overall, this work concludes that the oxygen recovery mechanism in CO-PROX switches from a
direct mechanism, wherein oxygen restores vacancy sites in the partially reduced CuO particles, to a synergistic mechanism with the
participation of ceria once Cu
x
O particles reach a critical reduction state. This mechanistic switch ultimately results in a decrease in
CO conversion in favor of the undesired H
2
oxidation, which opens-up future research on potential strategies to improve oxygen
recovery.
KEYWORDS: CO-PROX reaction, ceria, copper, operando NAP−XPS, DFT calculations, oxygen vacancies, reaction mechanism
1. INTRODUCTION
The preferential CO oxidation (CO-PROX) involves the
selective oxidation of the low-content CO impurities (0.5−2%
vol.) present in reformate streams after processing in water−
gas shift reactors.
1
This catalytic strategy efficiently allows for
exhaustive CO removal from H
2
-rich streams below the 10−
100 ppm of CO-tolerance level accepted for proton exchange
membrane fuel cells, whose performance and durability are
strongly affected by CO poisoning.
2−4
In particular, CO-PROX
brings promising opportunities in the implementation of on-
board and portable H
2
-dependent technologies, where light-
ness is a requirement.
5
In the search for active and cost-effective catalysts, copper
oxide and cerium oxide binary mixtures have demonstrated
noteworthy activity and near-optimal features.
6−9
The catalytic
performance of CuO/CeO
2
materials relies on the synergistic
metal oxide/support interactions arising from complex redox
effects induced between the CuO and CeO
2
phases at the
interfacial contact points.
6,10
These redox features include
labile electron exchange between the Cu
2+
/Cu
+
and Ce
4+
/Ce
3+
redox pairs, ease of formation of surface oxygen vacancies in
ceria, and the promotion and stabilization of Cu
+
sites.
11−15
Notably, many precedent studies based on in situ and
operando advanced spectroscopic studies have pinpointed
the stable surface Cu
+
species as active sites for CO oxidation,
whereas the reduced metal Cu entities would favor the
undesired H
2
oxidation.
11,16−18
In turn, the oxidation state of
copper species is deemed to determine CO selectivity, while
the eventual reduction of CuO leads to selectivity losses by
Received: February 6, 2020
Revised: May 8, 2020
Published: May 11, 2020
Research Articlepubs.acs.org/acscatalysis
© 2020 American Chemical Society
6532
https://dx.doi.org/10.1021/acscatal.0c00648
ACS Catal. 2020, 10, 6532−6545
Downloaded via UNIV NACIONAL DE LA PLATA on August 18, 2021 at 14:43:40 (UTC).
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
virtue of boosting the competing H
2
oxidation at high
temperatures.
19−21
Since cationic Cu
+
species are the result
of the interfacial redox interactions between the CuO and
CeO
2
phases,
22,23
many studies have been devoted to
promoting the formation of Cu
+
by means of a rational
catalyst nanodesign.
14,15,24−28
Accordingly, highly dispersed
CuO
x
particles provide the highest reducibility to copper
oxide/cerium oxide mixtures. Conversely, larger CuO
x
bulklike
clusters with a weaker interfacial interaction and a kinetically
limited interaction help to prevent further reduction to Cu
0
and improve the maintenance of the CO selectivity.
29,30
Therefore, the catalytic performance is influenced by the
balance of small Cu
δ+
particles and disperse bulk CuO
microstructures, which are tunable by means of catalyst
nanodesign controlling size, shape, composition, and electronic
effects.
Recent studies based on advanced in situ t ransient
techniques have presented valuable mechanistic insights in
the copper-catalyzed CO-PROX reaction,
31−33
and it is a
general consensus that reaction takes place mainly following a
Mars−van Krevelen (MvK) mechanism.
32,34,35
Since the MvK
mechanism involves the direct participation of lattice oxygen
species in the reaction, oxygen vacancies created in the vicinity
of active sites must be replenished by molecular O
2
from the
gas phase. Thus, the catalyst reoxidation capacity given by the
surface oxygen exchange ability and oxygen mobility has a
direct impact on the catalytic performance.
36,37
However, in
contrast with the current deep knowledge in molecular CO
and H
2
oxidation reactions, the mechanism of reoxidation steps
is not yet well understood.
32,38,39
In copper oxide−cerium
oxide catalysts, two complementary mechanisms of O vacancy
filling have been proposed. Namely, the direct mechanism,
where O
2
replenishes oxygen vacancies directly in the CuO
x
sites, and the synergistic mechanism, where O
2
uptake takes
place via the CeO
2
support and subsequent transfer to the
active O-deficient CuO
x
phase.
40
Although the prevalence of
each mechanism is known to be dependent on the oxygen
storage capacity and catalyst interfacial interactions, the
assessment of their specific contributions during CO-PROX
conditions remains unclear.
Herein we report a detailed mechanistic insight on the
CuO/CeO
2
activity toward the CO-PROX reaction with a
particular focus on the redox processes occurring at the
individual catalytic phases in reduction and reoxidation steps.
With this aim, CO-PROX operando near-ambient pressure X-
ray photoelectronic spectroscopy (NAP−XPS) and near edge
X-ray absorption fine structure (NEXAFS) experiments with
tunable incident soft X-ray photon energies were conducted,
obtaining XPS spectra which allow us to discern with high
sensitivity small variations in the redox processes at different
catalyst depths. Particularly, while Cu ions are gradually
reduced along with the CO-PROX reaction course, Ce ions
remain in a steady oxidized state up to a critical point where
the finest surface of ceria shows an incipient reduction. In
addition, DFT calculations indicate that ceria facilitates oxygen
to the surrounding Cu species at the triple phase boundary,
assisting in the oxygen recovery process once Cu reaches a
certain reduced state. In situ Raman spectroscopy of the CO-
PROX mixture confirms that ceria reduction becomes very
significant when further increasing the temperature above such
point, assigned to the total conversion of inlet O
2
. On the
contrary, O
2
pulse isotopic experiments demonstrate the
involvement of catalyst oxygen in the CO and H
2
oxidations
along the entire CO-PROX reaction range, which overall
suggests the participation of lattice oxygen from different
sources, from CuO in first instance and, second, from ceria. In
summary, this work presents evidence of the transition from a
direct O vacancy filling mechanism on CuO, to a synergistic
O
2
uptake via ceria, which determines the CO-PROX
selectivity of the CuO/CeO
2
catalyst.
2. EXPERIMENTAL METHODS
2.1. Catalyst Preparation and Characterization. The
CeO
2
support was prepared by thermal decomposition of
cerium(III) nitrate following a flash calcination procedure,
introducing the precursor in a preheated mu ffle furnace at 200
°C and then heating up to 500 °C in a ramp of 10 °C/min.
The CuO/CeO
2
catalyst was synthesized via incipient wetness
impregnation of copper(II) nitrate aqueous solution into the
ceria support, followed by flash calcination with the same
protocol as for the support preparation. The target nominal
composition was set to 5% w/w Cu.
The general physicochemical characterization results,
including N
2
adsorption at −196 °C (Figure S1, Table S1),
XRD (Figure S2, Table S2), Raman spectroscopy (Figure S3),
temperature-programmed reduction with H
2
(Figure S4), and
transmission electron microscopy (Figure S5), are described in
the Supporting Information.
2.2. CO-PROX Catalytic Tests. Fixed-bed CO-PROX
catalytic tests were conducted with 150 mg of catalyst placed in
a U-shaped quartz reactor (16 mm inner diameter) and 100
mL/min (GHSV: 30000 h
−1
) of the flowing reactant mixture,
i.e., 2% CO, 30% H
2
, and 2% O
2
balance N
2
, leading to a
stoichiometric O
2
:CO excess (λ) of 2. To test the effect of the
oxygen partial pressure, experiments in O
2
:CO stoichiometric
conditions with λ = 1 were also carried out. Catalytic tests were
performed with a heating rate of 2 °C/min up to 250 °C, and
the exhaust gases were analyzed using a gas chromatograph
(HP model 6890 Plus Series) equipped with two columns:
Porapak Q 80/100 for CO
2
and H
2
O separation and Molecular
Sieve 13X for O
2
and CO separation, coupled to a thermal
conductivity detector (TCD). The effect of CO
2
and H
2
O
inhibitors in the catalytic activity was studied by adding 10%
CO
2
,5%H
2
O, and 10% CO
2
+5%H
2
O to the reactant CO-
PROX gas mixture in 2 °C/min ramp experiments with λ =2
(Figure S9).
2.3. Isotopic Experiments with
36
O
2
. Isotopic experi-
ments were performed with
36
O
2
using an injection valve with
a loop (100 μL) and two high sensitivity pressure transducers.
These experiments were carried out in a fixed-bed tubular
quartz reactor with 80 mg of catalyst in a constant feeding
mixture consisting of 20 mL/min of 1% CO, 30% H
2
, and He
balance. The outlet gas composition was monitored with a
mass spectrometer (MS) Pfei ff er vacuum (model OmniStar).
The reactor was heated using a furnace controlled by a
temperature regulator at selected temperatures representative
for diff erent CO selectivity regimes along the CO-PROX
reaction, namely 75, 100, and 150 °C. Once MS signals were
stab ilized at the desired t emperature under the flowing
mixture, three
36
O
2
pulses (Isotec, 99%; 100 μL and 620
mbar) were injected. The results obtained were reproducible at
all temperatures, and the reproducibility of the method was
further confirmed by the injection of Ar pulses (100 μL and
620 mbar) prior to the
36
O
2
pulses.
2.4. In Situ Raman Spectroscopy Experiments. In situ
Raman spectra were recorded in a LabRam Jobin Ivon Horiba
ACS Catalysis pubs.acs.org/acscatalysis Research Article
https://dx.doi.org/10.1021/acscatal.0c00648
ACS Catal. 2020, 10, 6532−6545
6533
instrument with a laser excitation source of He:Ne (632.8 nm).
Experiments were performed in a high temperature chamber
fed with a regular flow of 100 mL/min of He or CO-PROX gas
mixture (i.e., 2% CO, 2% O
2
, 30% H
2
, and He balance).
Raman spectra were recorded in both atmospheres at selected
temperatures (i.e., 50, 75, 100, 150, 200, and 250 °C) to study
structural changes upon exposure to the CO-PROX reactant
mixture. A monocrystalline Si reference (521 cm
−1
) was used
to calibrate the position of the bands.
2.5. NAP−XPS and NEXAFS Experiments under CO-
PROX Operando Conditions. NAP−XPS spectra were
recorded under CO-PROX reaction conditions at the NAPP
branch of the CIRCE beamline at the ALBA Synchrotron Light
Source.
41
For each analysis, two different sets of photon
energies were used, namely 1082 and 1372 eV for the Ce 3d
and Cu 2p regions and 972 and 722 eV for the O 1s and C 1s
regions. These energies provide a variability in the surface
sensitivity according to the estimations of the mean free paths
(MFP) in the Cu and Ce oxide structures (see Table S3 for
details).
The CuO/CeO
2
catalyst was pelletized with a gold mesh to
prevent surface charging while providing a Au 4f reference for
the peak position during XPS analysis. Catalytic activity of the
gold mesh was experimentally ruled out. A Puregas gas inlet
system (SPECS) was used to keep the total pressure in the
XPS chamber constant at 1 mbar and to control the gas feed.
The pelletized catalyst was pretreated in O
2
/He atmosphere at
250 °C for 1 h and then cooled down to 50 °C. Subsequently,
the CO-PROX reacting mixture containing 1% CO, 1% O
2
,
30% H
2
, and balance N
2
was dosed at 30 mL/min, and exhaust
gases were monitored with a MS installed in the second stage
of the differential pumping system of the XPS electron energy
analyzer. The CO-PROX reaction progress was controlled at
temperature intervals of 50 °C with corresponding stabilization
at each point up to total O
2
conversion until reaching the final
temperature of 450 °C. For each temperature, XPS spectra
were recorded once stationary state was achieved based on MS
signals stabilization.
CO-PROX operando NEXAFS measurements at the Cu L-
edge (930−950 eV) were performed in total electron yield
mode measuring the sample current at each temperature after
the series of NAP−XPS scans using the same experimental
conditions.
2.6. Computational Methods. Theoretical calculations
reported in this work were conducted by means of periodic
density functional theory (DFT) using the Perdew−Burke−
Ernzenhof (PBE) exchange-correlation functional,
42
as im-
plemented in the Vienna ab initio simulation package (VASP)
code, version 5.4.1.
43,44
The core electrons of Ce, Cu, and O
ions were described using projector augmented wave (PAW)
potentials,
45
while their valence states were represented by
plane-waves with a kinetic cutoff energy of 500 eV. In the case
of Ce ions, an effective Hubbard U term (U
eff
) of 4.5 eV was
also added to the DFT calculated energies (DFT+U) as an on-
site correction for the electrons localized in the 4f orbital,
following Dudarev’s approach.
46
The choice of this U
eff
value is
based on the satisfactory results obtained for a wide range of
reactions catalyzed by ceria.
47−50
In the case of Cu
2+
, the
analogue treatment of the d
9
electron was conducted using an
U
eff
of 7, as recommended in the literature.
51
The equilibrium lattice constant for the Cu and Ce bulk
oxides was optimized with a Γ-centered k-point grid of 5 × 5 ×
5 and 7 × 7 × 7, respectively, and using the Birch−Murnaghan
equation of state. Starting from the optimized bulk structures,
the most abundant facets were modeled by their corresponding
surface slabs, namely CuO(111), Cu
2
O(111), and CeO
2
(111)
displaying different periodicities in order to expose an equal
number of surface oxygens. These slabs were built thick
enough to ensure there is minimal interaction between the top
and the bottom (3 metal layers for CeO
2
, 4 for CuO and Cu
2
O
slabs), with a sufficiently large vacuum gap (ca. 15 Å)
perpendicular to the surface to minimize the interaction
between periodic slabs in that direction. The geometry of the
surface slabs was optimized using a Γ-centered k-point grid
mesh of 3 × 3 × 1. Oxygen vacancy formation energies, E
O‑vac
,
on the various slabs were calculated as
EE E E/
O vac vac slab slab O2
1
2
=−[+
]
‐‐
where E
slab
is the energy of the stoichiometric slab, E
O2
is the
energy of an oxygen gas molecule, and E
vac‑slab
is the energy of
the slab with a lattice oxygen vacancy with the two electrons
left behind in the most favorable configuration. All the E
O‑vac
values featuring different electron distributions are presented in
Table S4.
3. RESULTS AND DISCUSSION
3.1. CO-PROX Catalytic Tests in Fixed-Bed Reactor.
Figure 1 shows the CO-PROX activity profile of the prepared
CuO/CeO
2
catalyst in a first reaction cycle at two different
oxygen partial pressures. Regardless of the oxygen inlet, CuO/
CeO
2
exhibits an exceptional behavior in terms of CO
conversion and CO selectivity, in agreement with previous
Figure 1. CO-PROX catalytic performance of CuO/CeO
2
in terms of
CO conversion (X
CO
, diamonds), O
2
conversion (X
O2
, circles), and
CO selectivity (Sel, triangles) profiles for (a) λ = 2 and (b) λ =1.
Region 1 refers to the CO selective regime, whereas Region 2
corresponds to the nonselective regime.
ACS Catalysis pubs.acs.org/acscatalysis Research Article
https://dx.doi.org/10.1021/acscatal.0c00648
ACS Catal. 2020, 10, 6532−6545
6534
studies.
52−54
Because of the competitive H
2
oxidation reaction,
two different regions must be discerned in Figure 1, namely the
CO selective (ca. <110 ° C) and nonselective regime (ca. >110
°C). Such critical temperature is defined by the H
2
oxidation
onset, a process which becomes more predominant as
temperature increases because of its higher activation energy
compared to CO oxidation.
55
As a consequence, the selectivity
regime transition is relevant since it dictates the optimum
operating temperature for the optimum CO activity and
selectivity for a given experimental CO-PROX reaction
condition. According to Figures 1a and 1b, neither CO
oxidation nor H
2
oxidation onset are significantly affected by
the inlet O
2
pressure as the transition between both regimes
remains unaltered.
However, as H
2
gains relevance, CO oxidation is hampered
due to the limited O
2
supply, as illustrated in the selectivity
profile. When X
O2
is total, CO conversion decreases in favor of
H
2
oxidation, which occurs near the regime transition point in
λ = 1 conditions, while at higher temperatures when λ =2.
Besides, the O
2
:CO excess (λ = 2) allows us to reach higher
CO conversions compared to the stoichiometric conditions (λ
= 1) since H
2
oxidation onset is lower than the temperature
required for total X
CO
. Hence, setting λ > 1 is beneficial in CO-
PROX, though only moderate values are practicable in order to
avoid an excessive residual H
2
oxidation.
56,57
Therefore, λ =2
will be set as the default CO-PROX conditions for the
operando analyses presented henceforward in this study.
Overall, these results confirm that CO-PROX is a competitive
process where selectivity is determined by the remaining
partial pressure of O
2
, although CO and H
2
oxidation reactions
are not affected independently.
Catalytic tests using the CeO
2
and CuO bare phases were
carried out as control experiments and both showed negligible
individual activity within the CO-PROX temperature window
(Figure S6) in contrast to the binary CuO/CeO
2
catalyst.
These e xperiments point out the synergistic Cu− Ce
interactions at the CuO/CeO
2
interface as the main factor
responsible for the improved performance of the combined
catalyst, as well reported.
27,28,58,59
In this regard, character-
ization by Raman spectroscopy (Figure S3) and H
2
-TPR
(Figure S4) proves the existing strong interaction between
CuO and CeO
2
in the CuO/CeO
2
catalyst. Altogether the
general characterization results indicate that the 5% w/w Cu
catalyst prepared by f lash calcination is composed of both
finely disperse CuO
x
particles and bigger CuO bulklike clusters
in weaker interaction with the ceria carrier. XRD results also
reveal that a minor portion of the Cu present (ca. 0.56%) is
inserted in the c eria lattice (Table S2), presumably
concentrated on the outer surface layers.
The robustness and recyclability of the CuO/CeO
2
catalyst
were confirmed by running fou r consecutive CO-PROX
reaction cycles (Figure S7) followed by a fifth 10 h time-on-
stream isothermal experiment (Figure S8). Finally, the
suitability of the CuO/CeO
2
catalyst in the presence of CO
2
and H
2
O inhibitors was demonstrated, whose resulting impact
in the catalytic activity followed the trend: CO
2
<H
2
O<CO
2
+H
2
O(seeFigure S9), in agreement with previous
studies.
57,60
Overall, the CO-PROX activity results demon-
strate excellent performance of the CuO/CeO
2
catalyst.
3.2.
36
O
2
Pulse Isotopic Experiments. To investigate the
participation of lattice oxygen from the CuO/CeO
2
catalyst in
the CO-PROX reaction mechanism, a series of
36
O
2
pulse
isotopic experiments were next performed at different
temperatures. According to the gas profiles measured after
the
36
O
2
pulses (Figure 2), only CO
2
and H
2
O species were
detected with no sign of O
2
being released. This observation
can be rationalized with the strongly reducing conditions of
these experiments, leading to a highly O-deficient CuO/CeO
2
catalyst that captures the incoming
36
O
2
molecules to restore
the O vacant sites. In addition, Figure 2 shows that the area of
the
18
H
2
O peak increases with temperat ure due to the
promoted H
2
oxidation reaction, while the area of the CO
2
peaks decreases due to the selectivity loss in the CO-PROX
reaction (Figure S10), in agreement with the catalytic
experiments described in the previous section. The effect of
temperature is also reflected in the sharpening of the profiles of
the evolved products, which can be attributed to a faster
desorption. The most relevant insight, however, is the evident
delay in H
2
O release compared to CO
2
, as well as the large
broadening of the H
2
O signal. This is consistent with an
increased retention of water molecules at the surface compared
to CO
2
. The potential accumulation of H
2
O on the catalyst
surface also relates with the stronger inhibition by H
2
O than by
CO
2
, as the corresponding catalytic results show (Figure
S9).
57,60
In addition, complementary temperature-pro-
grammed experiments (TPD) from Figure S10 in the
Supporting Information evidence a significant H
2
O and CO
2
Figure 2. Normalized MS signals measured after
36
O
2
pulses in H
2
+
CO flow with the CuO/CeO
2
catalyst at different temperatures: (a)
75 °C, (b) 100 ° C, and (c) 150 °C. The zero-time was set after
36
O
2
was pulsed.
ACS Catalysis pubs.acs.org/acscatalysis Research Article
https://dx.doi.org/10.1021/acscatal.0c00648
ACS Catal. 2020, 10, 6532−6545
6535
retention capacity in the CuO/CeO
2
catalyst. Interestingly,
H
2
O surface saturation leads to important CO
2
corelease, and
vice versa. In fact, CO
2
and H
2
O coaddition maximizes
chemisorption capacity, which also relates with the much
stronger inhibition by CO
2
+H
2
O copresence (Figure S9).
The signals detected after the
36
O
2
pulses correspond to
44
CO
2
,
46
CO
2
, and
18
H
2
O, where most oxygen atoms come
from CO and the catalyst O atoms (
16
O); the only
18
O-
containing molecule was
46
CO
2
, and neither
48
CO
2
nor
20
H
2
O
were detected. Notably, the formation of
44
CO
2
(
16
OC
16
O;
nonisotopic) involves catalyst
16
O abstraction and anionic
vacancy formation, which is indicative of CO oxidation taking
place via a Mars−van Krevelen (MvK) mechanism. On the
other hand,
46
CO
2
formation (
18
OC
16
O; scramble of non-
isotopic and isotopic) may involve an adsorbed
18
O species in
the vicinity of CO or occur via direct oxidation of adsorbed
CO by
36
O
2
. The relative areas of the
44
CO
2
and
46
CO
2
peaks,
however, indicate that the former reaction pathway is much
more relevant than the latter regardless of the temperature.
Analogously, H
2
oxidation involves a catalyst oxygen (
16
O) to
yield
18
H
2
O via a MvK mechanism. In contrast with CO
oxidation, other alternative H
2
oxidation mechanisms involving
pulsed
36
O
2
can be ruled out since
20
H
2
O is not detected. In
summary, isotopic pulse experiments allow us to unequivocally
confirm that both CO and H
2
oxidation reactions on the CuO/
CeO
2
catalyst in a CO-PROX environment occur via a MvK
mechanism all along the temperature profile. Besides, the
global is otop ic yield of products shown in Figure S11
demonstrates there is not a significant effect of temperature
in the oxygen exchange capacity of the catalyst within the 75−
150 °C tested range.
3.3. In Situ Raman Spectroscopy Experiments. Figure
3 compiles the Raman spectra of the CuO/CeO
2
catalyst
recorded under flowing He and CO-PROX gas mixture
atmospheres at different temperatures, which relate with
ceria crystalline changes. The spectra show a main band
centered around 464 cm
−1
, attributed to the F
2g
symmetric
vibration mode of oxide anions around their equilibrium
positions in tetrahedral sites within the cubic crystal structure
of ceria.
7,61,62
The position of the F
2g
band is highly sensitive to
small changes in the crystalline features of CeO
2
, and it is well
reported to respond upon lattice dilation with a proportional
lower frequency (red) shift.
62
Additionally, Raman spectra of
ceria-based materials typically display minor bands around 540
and 600 cm
−1
, so-called D bands, D
1
and D
2
respectively,
which are ascribed to the presence of lattice defects.
61,63
Hence, the area band ratio D/F
2g
is widely used as a measure
of oxygen defect concentration in ceria, though the discern-
ment between the D
1
and D
2
band modes is still a matter of
debate.
64
Recent experimental/computational Raman studies
in ceria-doped materials have assigned D
1
bands to the
presence of oxygen vacancy defects, whereas D
2
is attributed to
the presence of defective elements in solid solution within the
ceria crystal.
63−65
Since pulse isotopic experiments confirm a MvK mechanism
in CO-PROX for both CO and H
2
oxidation reactions, an
increased population of oxygen vacancies is expected in the
catalyst alongside the reaction course. The formation of an O
vacancy results in a charge imbalance in the CeO
2
lattice that
must be compensated by the reduction of two Ce
4+
cations to
Ce
3+
. The distribution of the reduced Ce
3+
ions is determined
by the most favorable arrangement around the vacancy site,
which has been established as the nearest neighbor (NN) and
next nearest neighbor (NNN) positions relative to the defect
position.
66
As a result, the Ce
3+
−NNN cation formed upon
oxygen vacancy formation would remain in an 8-fold
coordination contributing to the D
2
mode, as a sign of
incipient reduction of ceria with highly dispersed surface
oxygen vacancies. Assuming this hypothesis, the D
2
band at
600 cm
−1
attributed to isolated vacancies cannot be formed in
independence of the D
1
band at 540 cm
−1
in nondoped ceria
materials, as is the case of this study. For this reason, the in situ
Raman spectra presented herein show near-equal D
1
and D
2
contributions, which are gathered together as a broad D band
that acquires a flat profile in the 540−600 cm
−1
range. The
influence of temperature on the D and F
2g
Raman bands in He
and CO-PROX atmospheres is compiled in Figures 4a and 4b.
The comparison between the trends observed under He and
CO-PROX conditions allows us to discern the effect of
reactant gases from the inherent lattice thermal expansion in
the CuO/CeO
2
catalyst. According to the He-recorded Raman
spectra shown in Figure 3, the ceria lattice is expanded by the
effect of temperature above 150 °C resulting in a proportional
F
2g
red shift. Another contribution to such expansion is the
presence of Ce
3+
cations, with larger radii than Ce
4+
, that
balance the charge deficit left upon the formation of oxygen
vacancies in ceria induced by temperature. However, the
associated defect D band barely increases within 50−250 °C,
so this contribution is modest.
On the other hand, the recorded CO-PROX Raman spectra
are more complex and respond to the different CO-PROX
reaction regimes. In particular, the evolution of the D band in
the CO-PROX mixture (Figure 4a) encompasses that recorded
in He up to 100 °C. Beyond this temperature, the CuO/CeO
2
catalyst presents a gradually higher population of oxygen
vacancy defects under CO-PROX conditions, which experi-
Figure 3. In situ Raman spectra for the CuO/CeO
2
catalyst recorded
at different temperatures in 100 mL/min of He (dotted lines) and
CO-PROX mixture (solid lines). We note that spectra have been
normalized to the maximum intensity of the F
2g
band.
ACS Catalysis pubs.acs.org/acscatalysis Research Article
https://dx.doi.org/10.1021/acscatal.0c00648
ACS Catal. 2020, 10, 6532−6545
6536
