Catalysis Today 215 (2013) 201–207

Contents lists available at ScienceDirect

Catalysis Today
journal homepage: www.elsevier.com/locate/cattod

Catalytic CO2 valorization into CH4 on Ni-based ceria-zirconia.

Reaction mechanism by operando IR spectroscopy
P.A. Ussa Aldana a , F. Ocampo b , K. Kobl b , B. Louis b , F. Thibault-Starzyk a , M. Daturi a ,
P. Bazin a , S. Thomas a , A.C. Roger b,∗
a
Laboratoire Catalyse et Spectrochimie, UMR CNRS 6506, ENSICAEN, Université de Caen, Caen, France
b
Laboratoire des Matériaux, Surfaces et Procédés pour la Catalyse, Groupe Energie et Carburants pour un Environnement Durable, UMR CNRS 7515,
Université de Strasbourg, Strasbourg, France

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

Article history: Despite several studies, the use of CO2 as a chemical reactant is currently limited to a few processes. With
Received 12 November 2012 the aim to formulate an efficient catalyst for CO2 conversion into methane, nickel was selected, being
Received in revised form 14 February 2013 well known to be active for the catalytic COx methanation. Ceria-zirconia mixed oxide was chosen as a
Accepted 26 February 2013
support because of its properties to activate CO2 . Yields close to those predicted by thermodynamic were
Available online 10 April 2013
obtained with weak deactivation even after 90 h on stream. Interactions between Ni and the ceria-zirconia
support seem to be a key parameter for the methanation reaction. The use of IR operando spectroscopy
Keywords:
revealed that the main mechanism for CO2 methanation does not require CO as reaction intermediate.
CO2
Methanation
The better activity observed compared to Ni-silica can be explained by the importance of weak basic sites
Ceria-zirconia the adsorption of CO2 .
operando FTIR © 2013 Elsevier B.V. All rights reserved.

1. I Introduction Nickel is well known to be active in catalytic COx methanation

[3–6] and for its low capacity of growing chain in Fischer-Tropsch
The use of CO2 as a reactant could contribute to the decrease reaction [7]. The most common catalyst for the methanation reac-
of greenhouse gas emissions through chemical recycling. Its use tion is Ni/Al2 O3 for which the activity is thought to be partially due
as chemical reactant (in parallel to its sequestration and storage) is to a nickel aluminate spinel phase located at the metal-support
limited at the moment to a few processes [1] although several stud- interface which may stabilize the metal particles.
ies on its catalytic conversion into liquid hydrocarbons have been Zirconium oxide has also been often studied as support and
performed [2]. CO2 methanation remains the most advantageous the tetragonal phase has been reported to be more active than the
reaction with respect to thermodynamics when compared to the monoclinic one [8,9]. Here again the role of a specific interaction
production of other hydrocarbons or alcohols. The methane pro- between nickel and the zirconia support is pointed out. In addition,
duction can be inserted into a global scheme where decarbonized some studies have shown that the addition of Ce, Sm or Pr to zir-
electricity, from solar or wind power, would be used in delocalized conia increased the catalytic activity [8,10] due to the stabilization
flexible units to perform water electrolysis to produce hydrogen of Zr in a cubic or tetragonal structure. In fact, ceria-zirconia mixed
which would react with CO2 capture from emission sources. Hence, oxide can be seen as an adequate support for the CO2 methanation
the methane produced can be used directly as a fuel, indirectly as an because of its properties to activate CO2 . The results obtained on Ni-
energy source to generate electricity in a power plant or be injected ceria-zirconia systems for CO2 methanation [11–13] are promising,
into the existing natural gas network. This recycling concept con- indeed.
siders CO2 as a carbon source for energy, CO2 being a raw material The mechanism of methanation has been principally studied on
and not only a waste. This kind of process would partially solve the noble metal-based catalysts. It is generally admitted that CO2 is
problem of the stabilization of the electric grids which arises from first dissociated into CO either directly [14–16] or through carbon-
the increasing electricity production from renewable and inter- ate and/or formate intermediates [17,18]. The formed CO would be
mittent sources. This global process can then be considered as a then dissociated into C and O atoms on the metal sites and further
chemical storage of excess of electricity. hydrogenated into methane by dissociated H2 still on the metal
particles [5,19]. Only a few works, based on kinetic studies [20] or
DFT calculations [21], assume that CO2 methanation does not occur
∗ Corresponding author. Tel.: +33 3 68 85 27 69; fax: +33 3 68 85 27 68. through the formation of CO but via carbonates and formates which
E-mail address: [email protected] (A.C. Roger). are further directly hydrogenated into CH4 .

0920-5861/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.cattod.2013.02.019
202 P.A.U. Aldana et al. / Catalysis Today 215 (2013) 201–207

This work is devoted to the study of the mechanism of the (50 mL min−1 ) and for another 1.5 h in pure Ar flow (50 mL min−1 )
methanation reaction from CO2 /H2 under stoichiometric con- to desorb any H2 that might have been spilled over the support
ditions to evaluate and understand the catalytic behaviour of [23]. The catalyst was then cooled down to room temperature and
Ni-ceria-zirconia catalysts in comparison with conventional Ni- H2 chemisorption pulses were performed. The sample was then
silica systems. In particular, the role of CO in the global reaction purged with pure Ar to remove the reversibly adsorbed H2 . The
scheme will be discussed. TPD analysis was carried out from room temperature to 800 ◦ C at a
heating rate of 15 ◦ C min−1 under pure Ar flow.
2. Experimental
2.3. Catalytic activity
2.1. Catalyst synthesis
The catalysts were reduced in situ before reaction in a 80% H2 /N2
The catalyst Ni-CZsol–gel was synthesized by a pseudo sol–gel stream for 6 h with a total gas flow of 45 mL min−1 at 400 ◦ C with a
method, based on the thermal decomposition of propionate precur- heating ramp of 2 ◦ C min−1 .
sors. Ni loading was 5 wt%, the molar ratio cerium/zirconium was Carbon dioxide methanation was conducted at atmospheric
1. The starting materials were cerium (III) acetate sesquihydrate, pressure in a fixed-bed down-flow reactor at 350 ◦ C. A thermocou-
zirconium (IV) acetylacetonate and nickel (II) acetate tetrahydrate. ple was inserted in the furnace to measure the pretreatment and
These starting salts were separately dissolved in boiling propionic reaction temperatures in situ. The reactor was heated in a tubular
acid at a cation concentration of 0.12 mol L−1 , maintained at 100 ◦ C furnace monitored by a temperature controller.
for 1 h, leading exclusively to propionate precursors. The boiling The flow of reactants was regulated by calibrated mass flow
solutions were mixed for 2 h under reflux and then the solvent controllers (Brooks). H2 and CO2 were mixed at a stoichiometric
was evaporated until a resin was obtained. Finally, the resin was H2 :CO2 ratio of 4:1 and N2 was added as an internal standard
calcined in air at 500 ◦ C for 6 h, with a heating ramp of 2 ◦ C min−1 . (PH2 = 0.654 bar; PCO2 = 0.164 bar; PN2 = 0.182 bar). The total flow
The catalyst Ni-CZimp was synthesized by wet impregnation rate was set to 55 mL min−1 with 150 mg of catalyst loaded into
of 5 wt%Ni on a stoichiometric ceria-zirconia support (noted CZ) the reactor corresponding to a space velocity around 43 000 h−1
prepared as described in the pseudo sol–gel method. The nickel pre- for CZ-based catalysts and around 11 000 h−1 for silica-based sam-
cursor (Ni (II) acetate tetrahydrate) was dissolved in ethanol. The ples. The feed and products were analyzed on-line by a micro gas
latter solution and the support were mixed in a rotary evaporator, chromatograph (Hewlett Packard, Quad Series Micro GC, alumina,
ethanol was evaporated and nickel impregnated on the support. poraplot and molecular sieve 5 Å columns) equipped with a TCD.
The solid was then dried at 120 ◦ C for 1 h and calcined under air at Values of turnover frequencies (TOF) have been calculated on
500 ◦ C for 6 h, with a heating ramp of 2 ◦ C min−1 . the basis of the results of metal surface areas given by H2 -TPD. The
The catalyst Ni-SiO2 was synthesized by wet impregnation of TOF (s−1 ) represents the number of CO2 molecules converted per
5 wt%Ni on a silica support prepared according to the method devel- Ni◦ surface atom per second.
oped by Jansen et al. [22]. The gel was dried for 12 h at 90 ◦ C and
was then submitted to a hydrothermal treatment for 3 h at 200 ◦ C 2.4. Operando IR measurements
before being calcined under air at 800 ◦ C for 10 h.
Operando IR measurements were performed in a sandwich IR
2.2. Catalyst characterization reactor-cell [24]. The IR operando system is composed of four main
parts: the infrared spectrometer, the IR reactor-cell, the gas flow
X-ray diffraction patterns (XRD) were recorded on a Bruker D8 set-up and the exhaust gas analyzers [25]. The cell is connected to
Advance diffractometer with a VANTEC detector side and Ni filtered the operando gas-system including mass flow controllers for the
Cu K␣ radiation (1.5418 Å) over a 2 range of 10–90◦ and a posi- introduction of gases into the lines. The system allows investigat-
tion sensitive detector using a step size of 0.05◦ and a step time ing the composition at the outlet of the cell by a Quadrupole Mass
of 1 s. The crystallite size of the NiO phase was evaluated from X- Spectrometer (Pfeiffer Omnistar GSD 301). IR spectra (64 scans per
ray broadening of the most intense reflection (200) of cubic NiO by spectrum, optical resolution of 4 cm−1 ) were collected every 50 ◦ C
using the Debye-Scherrer equation. with a Thermo Scientific Nicolet 6700 spectrometer, equipped with
The specific surface area (SSA) of the different catalysts was a MCT detector. Spectra were subtracted from the reduced fresh
determined by N2 adsorption–desorption measurements at 77 K sample at the corresponding temperature. Samples were pressed as
using the Brunauer–Emmet–Teller (BET) method (Micrometrics self supported wafer of around 20 mg and reduced in situ at 400 ◦ C
sorptometer Tri Star 3000). Prior to N2 adsorption, the sample was for 3 h under a flow of 20 mL min−1 of 80% H2 /Ar. Samples were
outgassed at 200 ◦ C overnight to desorb moisture adsorbed on the then cooled under Ar at 150 ◦ C before introducing the reaction mix-
surface and inside the porous network. ture containing 40% H2 , 10% CO2 (or 13.3% CO) with Ar as carrier
Temperature-programmed reduction (TPR) was made on a gas (total flow rate 20 mL min−1 ) with a gas hourly space velocity of
Micromeritics AutoChem II analyzer to study the reducibility of 120 000 h−1 for CZ-based catalysts and around 30 000 h−1 for silica-
the catalysts. TPR measurements were carried out with 100 mg of based samples. These compositions are different from those used
catalyst, loaded in a quartz U-tube and heated from room temper- for catalytic activity tests in order to avoid high water concentra-
ature to 900 ◦ C at a heating rate of 15 ◦ C min−1 under 10% H2 /Ar tion in the operando cell. The temperature was kept at 150 ◦ C for
with a total gas flow of 50 mL min−1 . TPR profiles are presented as 30 min before heating up to 400 ◦ C in 3 h under reaction flow.
a function of temperature herein. However, it is worthy to mention
that H2 consumptions were calculated using the profiles drawn as 3. Results and discussions
a function of time for which the baseline returns to its initial value.
The Ni metal surface area (m2 gcat −1 ) was calculated from the The diffractograms of the three fresh catalysts and of the CZ
amount of chemisorbed hydrogen, measured by H2 temperature- support are shown in Fig. 1. Ni-SiO2 pattern shows small peaks
programmed desorption (H2 -TPD) on a Micromeritics AutoChem II. of NiO at 37.3◦ and 43.2◦ , in addition to a broad peak around
Prior to H2 -TPD, the catalyst (200 mg) was pre-reduced at 400 ◦ C at 22.8◦ assigned to amorphous silica. NiO phase is also detected in
a heating rate of 2 ◦ C min−1 and held for 1 h under a 10% H2 /Ar flow Ni-CZsol–gel and Ni-CZimp , with a higher intensity in the catalyst
 P.A.U. Aldana et al. / Catalysis Today 215 (2013) 201–207 203

Table 1
Textural characterization of the catalyst.

Intensity (a.u.) d) Sample SBET (m2 g−1 ) a

dNiO (nm) b
SNi (m2 gcat −1 )

Ni-SiO2 560 15 8.1

CZ 88 - -
c) Ni-CZsol–gel 62 21 3.1
Ni-CZimp 67 22 3.1
b) a
From XRD on fresh samples.
a) b
Measured by H2 -TPD on reduced samples.
10 20 30 40 50 60 70 80 90
Bragg Angle (2 )

H2 consumption (a.u.)
Fig. 1. XRD patterns: (a) Ni-SiO2 , (b) CZ, (c) Ni-CZsol–gel , (d) Ni-CZimp .

d)

c)
prepared by impregnation. The crystalline phase of ceria-zirconia
(Fig. 1b) does not seem to be modified by the presence of Ni, which b)

was either impregnated (Fig. 1d) or directly added in the prepara- a)

tion method (Fig. 1c). 50 250 450 650 850

The good crystallinity of Ni-CZsol–gel and Ni-CZimp can also be Temperature (°C)
observed by TEM (Fig. 2). The two catalysts are mainly composed
of monodisperse particles of ceria-zirconia of a size below 10 nm. Fig. 3. TPR profiles: (a) Ni-SiO2 , (b) CZ, (c) Ni-CZsol–gel , (d) Ni-CZimp .
Larger particles assigned to NiO are also observed.
The particle size of NiO calculated from XRD measurements is 100 a)

CO 2 conversion (%)
around 20 nm for the ceria-zirconia based catalysts. The specific
surface areas of the two materials are of the same order, 62 and 80 c)

67 m2 g−1 for Ni-CZsol–gel and Ni-CZimp respectively. The particle

60
size of NiO is only 15 nm for Ni-SiO2 , due to the high specific surface d)
area of the silica support (560 m2 g−1 for Ni-SiO2 ). 40
Fig. 3 shows the hydrogen consumption profiles of the three b)
catalysts, as well as that of CZ, for comparison. For Ni-SiO2 , the 20
main reduction occurs at 360 ◦ C, typical of supported NiO [26]. The
0
hydrogen consumption continues up to temperatures as high as
150 200 250 300 350 400
600 ◦ C, accounting for the reduction of Ni2+ in strong interaction
with the silica support, most probably nickel silicates [27]. Temperature (°C)
In agreement with the previous characterization results, Ni-
Fig. 4. Activity vs temperature (a) thermodynamic, (b) Ni-SiO2 , (c) Ni-CZsol–gel , (d)
CZsol–gel and Ni-CZimp present very similar reduction profiles. The Ni-CZimp for CO2 hydrogenation.
hydrogen consumption of the two materials begins slowly around
250 ◦ C and ends around 600 ◦ C. The reduction peak located between
380 and 400 ◦ C is attributed to NiO and the broad zone between 250 and Ni-CZimp is equal to 3.1 m2 gcat −1 , much lower than that calcu-
and 600 ◦ C corresponds to the reduction of the ceria-zirconia sup- lated for Ni-SiO2 (8.1 m2 gcat −1 ). This confirms a better dispersion
port. The reduction of the support is facilitated by the presence of of nickel in the latter catalyst.
nickel. The reduction temperatures are lower than those observed The catalytic results obtained on the three catalysts in the reac-
for the reduction of bare CZ (Fig. 3a). It should be added that in tion of carbon dioxide methanation are presented in Fig. 4, which
the case of Ni-CZsol–gel , when compared to the impregnated cata- shows the CO2 conversion between 150 and 400 ◦ C. The CO2 con-
lyst, species which are more difficult to reduce are evidenced by version becomes significant from 250 ◦ C upwards and then rapidly
the presence of more intense broad reduction zone around 500 ◦ C. increases with temperature to tend towards the thermodynamic
This accounts for a stronger Ni-support interaction, initiated by the value. In spite of their comparable physico-chemical characteris-
one-step preparation method. tics, Ni-CZsol–gel and Ni-CZimp do not exhibit the same catalytic
The metal surface area of nickel generated by a reducing pre- behaviour. Over the whole range of temperature the CO2 conver-
treatment at 400 ◦ C has been evaluated by H2 -TPD, the results are sion remains higher on Ni-CZsol–gel than on Ni-CZimp , for instance
given in Table 1. The value of the metallic surface of Ni-CZsol–gel 62% compared to 17% at 300 ◦ C. The performance of Ni-SiO2 is

Fig. 2. TEM images: (a) Ni-CZsol–gel , (b) Ni-CZimp .

204 P.A.U. Aldana et al. / Catalysis Today 215 (2013) 201–207

100
C
h)
80 b)
CO2 conversion (%)

g)
f)
60 e)
c) d)
c)
40 b)
a)
a)
2200 2000 1800 1600 1400
20 Wavenumber (cm-1)

0
0 30 60 90
B
Time on stream (h) h)
g)
f)
Fig. 5. Activity vs time at 350 ◦ C (a) Ni-SiO2 , (b) Ni-CZsol–gel , (c) Ni-CZimp for CO2
e)
hydrogenation. d)
c)
b)
a)
lower, despite the fact that it possesses the highest initial metal 2200 2000 1800 1600 1400
surface area. Wavenumber (cm-1)
To better discriminate the catalysts, isothermal tests were per- A
formed for 90 h at 350 ◦ C. The plots of CO2 conversion over time are h)
g)
shown in Fig. 5. The detailed catalytic results are given in Table 2. f)
The initial conversion of CO2 follows the order of activity estab- e)
lished from the tests with increasing temperature. The values are d)
79.7, 59.8 and 35.0% for Ni-CZsol–gel , Ni-CZimp and Ni-SiO2 , respec-
tively, but they do not follow at all the order of available metal c)
b)
surface area. This is illustrated by the values of TOF given in Table 2:
a)
0.429, 0.313 and 0.076 s−1 for Ni-CZsol–gel , Ni-CZimp and Ni-SiO2 ,
2200 2000 1800 1600 1400
respectively. Ceria-zirconia is therefore a better support than silica
Wavenumber (cm-1)
for Ni in the CO2 methanation reaction.
The conversion rapidly decreases with time for CZimp whereas Fig. 6. FTIR spectra for CO2 hydrogenation–A: Ni-SiO2 , B: CZ, C: Ni-CZsol–gel ; (a)
it remains much more stable for the two other catalysts. With 150 ◦ C–1 min, (b) 150 ◦ C–30 min, (c) 200 ◦ C, (d) 250 ◦ C, (e) 300 ◦ C, (f) 350 ◦ C, (g) 400 ◦ C
deactivation, the selectivity to methane, initially higher than 97.3%, – 1 min, (h) 400 ◦ C – 20 min.
decreases down to 84.7% after 90 h. At the same time, the forma-
tion of by-products increases. The selectivity to ethane rises from The species formed on the catalytic surface during the metha-
0.1 to 0.8%, the selectivity to carbon monoxide increases from 2.6 nation reaction were followed between 150 and 400 ◦ C for CO
to 14.5%. For the two other catalysts, even if the global tendency and CO2 hydrogenation on CZ, Ni-CZsol–gel and Ni-SiO2 samples
is comparable, the selectivities are rather stable. For Ni-SiO2 , the (Figs. 6 and 7). After reduction at 400 ◦ C, reaction flow was
selectivity to CO is always higher than 10%, to the detriment of the introduced at 150 ◦ C for 30 min.
formation of methane. For all the catalysts of the present study, the In the case of CO2 methanation, carbonates (essentially mono
lower the CO2 conversion, the higher is the tendency to produce and bidentates) are present at 150 ◦ C in a large amount for CZ
CO instead of CH4 . and Ni-CZsol–gel samples with a broad signal between 1250 and
In order to explain the better activity of Ni-CZsol–gel with respect 1700 cm−1 (Fig. 6B and C) [28–31]. Hydrogen carbonates are also
to Ni-SiO2 despite its much lower metallic surface, IR operando present on the surface of these samples as features at 3616, 1594,
measurements have been carried out under the conditions of 1220 and 1060 cm−1 [32] were observed after 1 min contact at
CO2 methanation. As carbon monoxide is the main by-product 150 ◦ C with a decrease upon time on stream. In addition, on CZ
and is discussed in literature as a reaction intermediate in the and Ni-CZsol–gel samples, formate species are not initially detected
CO2 methanation, the reaction of CO methanation was also stud- as no contribution is observed in the (C-H) region (not shown)
ied. but are well evidenced on the latter catalyst after 30 min at 150 ◦ C

Table 2
Catalytic results at 350 ◦ C.

Catalyst Time (h) XCO2 (%) SCH4 (%) SCO (%) SC2H6 (%) TOF (s−1 )

Ni-SiO2 Initial 35.0 88.3 11.6 0.1 0.076

50 28.4 86.7 13.2 0.1 –
90 27.6 85.5 14.3 0.2 –

Ni-CZsol–gel Initial 79.7 99.3 0.6 0.1 0.429

50 70.6 98.6 1.2 0.2 –
90 67.9 98.4 1.4 0.2 –

Ni-CZimp Initial 59.8 97.3 2.6 0.1 0.313

50 30.7 88.4 10.9 0.7 –
90 25.4 84.7 14.5 0.8 –
 P.A.U. Aldana et al. / Catalysis Today 215 (2013) 201–207 205

temperature with contributions at 1479, 1378 and 1058 (not

C shown) cm−1 [28,30]. This is in agreement with the higher ther-
h) mal stability of such species. For Ni-containing samples, a sharp
g) band (or a shoulder, for Ni-SiO2 ) at 1574 cm−1 and two compo-
nents at 1368 and 1380 cm−1 are present starting from 200 ◦ C and
f)
are assigned to the formation of formates (Fig. 6A and C). These
e) species were found in a larger amount for the Ni-CZsol–gel catalyst
d)
with the band intensities undergoing important modifications with
c)
reaction temperature: their intensity increases until 250 ◦ C before
b)
a) decreasing drastically. This temperature corresponds to the begin-
2200 2000 1800 1600 1400 ning of CO2 conversion (Fig. 8c) with the formation of CH4 and CO
Wavenumber (cm-1)
in significant amount as well as ethane in smaller proportion (not
shown). This may indicate that, on Ni-CZsol–gel catalyst, formates
are involved in the formation of these products, either directly
B
h) being intermediates, or indirectly being inhibitors of the active sites
g) for CO2 reduction, so that their departure from the surface leads
to CO and CH4 formation. It should also be noted that CO forma-
f) tion on the surface occurs from 150 ◦ C upwards and can be clearly
characterized by the contribution around 2010 cm−1 attributed to
e)
d) monocarbonyls on nickel surface [33]. The intensity of this band
c) remains unchanged with temperature for Ni-CZsol–gel catalyst. A
b) partial re-oxidation of ceria can be evidenced with a decrease of
2200 2000 1800 1600 1400
a) the electronic transition band of Ce3+ at 2125 cm−1 [ 34]. In par-
Wavenumber (cm-1) allel, the bridged OH on Ce3+ at ∼3680 cm−1 decrease, giving rise
to bridged hydroxyls on Ce4+ at ∼3660 cm−1 [35]. This oxidation
A h) phenomenon was more pronounced for the CZ sample which was
g) found less reducible than Ni-CZsol–gel by TPR.
f) The situation is quite different on the Ni-SiO2 sample as less
e) carbonates are formed at 150 ◦ C because of the weaker basicity
d) of the support. However, formate species are observed already at
c) 150 ◦ C with a band centred at 1590 cm−1 and a doublet at 1370
and 1383 cm−1 [36]. In this case, formate contribution decreases
upon temperature without any correlation with catalytic activity.
b) Moreover, a strong contribution of the bands related to carbonyls
a) on Ni◦ are present above 150 ◦ C (Fig. 6A), together with bridged
2200 2000 1800 1600 1400
(1880–1950 cm−1 ), mono- (2010 cm−1 ), di- (2045 cm−1 ) and tri-
Wavenumber (cm-1)
carbonyls (2050–2080 cm−1 ) [33]. This may be an indication of CO2
Fig. 7. FTIR spectra for CO hydrogenation – A: Ni-SiO2 , B: CZ, C: Ni-CZsol–gel ; (a) dissociation on the metallic surface as reported in the literature
150 ◦ C – 1 min, (b) 150 ◦ C – 30 min, (c) 200 ◦ C, (d) 250 ◦ C, (e) 300 ◦ C, (f) 350 ◦ C, (g) [37,38]. These last two species are reported to be responsible for
400 ◦ C – 1 min, (h) 400 ◦ C – 20 min. the sintering of Ni [33] and were observed on Ni-CZsol–gel sample
in negligible amount (Fig. 6C). Increasing the temperature leads
which may indicate a reduction of hydrogenate carbonates into to a decrease of the carbonyl coverage of nickel whereas formate
formate species at 150 ◦ C. On bare CZ, no significant activity was species are present only as traces above 250 ◦ C.
measured (Fig. 8b) and increasing the temperature led to a pro- In the case of CO hydrogenation experiments, formates were
gressive desorption monodentate at 1519 and 1328 cm−1 , and detected on CZ and Ni-CZsol–gel samples in larger amount and
bidentate carbonates at 1575 and 1292 cm−1 [28,30,32] whereas already at 150 ◦ C (Fig. 7B and C). No formate bands were evidenced
bands related to polydentate carbonates clearly appear at high on silica based catalysts (Fig. 7A). Bands of carbonyls on the nickel
surface were logically found to be much more intense compared
40 to experiments led with CO2 (Fig. 7A and C compared with Fig. 6A
and C). Comparison with in situ experiments of CO chemisorption
35 a) (not shown) revealed that saturation of Ni by carbonyls is almost
occurring on Ni-CZsol–gel even at 400 ◦ C. This may limit the access of
CO2 conversion (%)

30
hydrogen to the metal and explain the lower activity measured for
25 c)
methane formation under CO than under CO2 . For Ni-SiO2 catalyst,
20 the decrease of the carbonyl band upon temperature increase may
be partially related to the sintering of Ni due to the large presence
15 of di- and tri-carbonyls on nickel at 150 ◦ C.
10 The comparison of the CO2 and CO conversions (Figs. 8 and 9)
b) clearly shows that the activity of Ni-SiO2 is comparable in both
5 cases (around 35% of conversion at 400 ◦ C) whereas it is strongly
0
decreased under CO for Ni-CZsol–gel (10% of CO conversion com-
150 200 250 300 350 400 pared to 30% of CO2 conversion at 400 ◦ C). This tends to show
that the mechanism of CO2 conversion with a ceria-zirconia sup-
Temperature ( °C)
port would be different from the one with a silica support. For
Fig. 8. Activity vs time at different temperatures on (a) Ni-SiO2 , (b) CZ, (c) Ni-CZsol–gel Ni-CZsol–gel , it seems to indicate that CO on Ni would not be an
for CO2 hydrogenation (operando IR). intermediate of CO2 methanation.
206 P.A.U. Aldana et al. / Catalysis Today 215 (2013) 201–207

35 production of CO in the gas phase still continues for a few minutes

(Fig. 10), strengthening the assumption that it is not an interme-
30 diate species, as aforementioned. The amount of CO adsorbed on
Ni0 metal sites decreases very slowly (Fig. 10) compared to the
CO conversion (%)

25
a) CO outlet gas phase concentration confirming the strong interac-
20 tion with nickel surface. This may inhibit the dissociation of H2
on nickel surface due to a competitive adsorption with CO and
15 decrease the activity of the catalyst. CO is probably formed through
c) a redox cycle in which CO2 is dissociated on reduced ceria of the
10 support. Without hydrogen in the feed, formation of CO continues
b) as long as Ce3+ cations are available. This is in good agreement with
5
the evolution of the transition electronic band of Ce3+ (Fig. 10) and
0 to quantitative data: in the present experiment 20 ␮mol of CO were
150 200 250 300 350 400 formed for 80 ␮mol of total cerium in the sample, cerium being only
Temperature (°C) partially reduced after a reducing treatment at 400 ◦ C [12]. This is
also in accordance with the negative peak observed at 2125 cm−1
Fig. 9. Activity vs time at different temperatures on (a) Ni-SiO2 , (b) CZ, (c) Ni-CZsol–gel attributed to ceria oxidation during CO2 exposure.
for CO hydrogenation (operando IR).
After the reintroduction of hydrogen in the flow the initial state
is immediately recovered, both in the gas phase and at the surface.
Operando experiments were also performed at 400 ◦ C during When CO2 flow is stopped, the band of adsorbed CO at
transient periods where the reactant gas flow was modified by 2013 cm−1 quickly disappears as well as CO in the gas phase, while
substituting alternatively one of the reactants by argon. the decrease of carbonates and formate species is much slower and
The evolution of gas phase composition as well as the inten- the formation of CH4 continues for a short period. This means that
sity of the bands of formates, carbonates and monocarbonyls are most of the CO is not formed by a mechanism involving formate
presented in Fig. 10 for Ni-CZsol–gel during the transient periods. or carbonate hydrogenation nor does it appear in the sequence of
When H2 is turned off in the feed, methane formation stops methane formation. At the same time, these observations confirm
immediately (Fig. 10) which accounts for the poor hydrogen that on Ni-CZsol–gel CO is only formed by a redox cycle associated
adsorption capacity of the sample (low metal surface area). Car- with CO2 reduction and Ce3+ oxidation.
bonates accumulate on the surface (Fig. 10), being no longer These transient experiments clearly show that on Ni-CZsol–gel
hydrogenated. Without hydrogen, formates are not formed and catalyst CO and CH4 arise from different reaction pathways. The
their contribution decrease quickly (Fig 10). In contrast to that, the sequence of methane formation involves the ceria-zirconia support
in the formation of covalent carbonates and formates.
In the case of Ni-SiO2 , analysis of the outlet gas phase composi-
tion revealed very fast response to feed gas variations. In fact both
CH4 and CO formation stopped immediately after turning off CO2 or
H2 in the flow (not shown). The same behaviour was evidenced for
surface species i.e. carbonyl bands completely disappeared in less
than 2 min after turning off CO2 whereas no major changes were
observed after turning off H2 . Here no C-storage species are then to
be considered at the surface.
For this catalyst, the silica support would have no other role
than to disperse the Ni◦ active phase. The competition between
CO2 reduction and hydrogen dissociation on the Ni◦ surface would
explain the much lower TOF values (Table 2) of Ni-SiO2 , compared
to CZ-containing catalysts for which H2 would be dissociated on
Ni0 and CO2 would be preferentially adsorbed on the support.
In addition to the lower values of CO2 conversion observed on
Ni-SiO2 during the isothermal test at 350 ◦ C (Fig. 5), the high specific
surface of silica is thought to be responsible for the low deactiva-
tion rate of Ni-SiO2 , preventing Ni◦ from sintering by increasing
the inter-particle distances. Moreover, the Ni-silica interaction
evidenced by TPR may be an additional characteristic for the cat-
alytic stability, as it was already discussed for Ni-CZsol–gel [13].
To summarize the previous observations, a mechanism for CO2
methanation on the studied Ni-CZsol–gel catalyst is proposed and
presented in Fig. 11. The main reaction pathway (Fig. 11a) does
not involve CO as an intermediate. Carbon dioxide adsorbs on sites
of mild basicity to form covalent carbonates, hydrogen carbonates
and then bidentate carbonates. These species are further reduced
and hydrogenated by H atoms formed by dissociation of hydrogen
at the surface of Ni0 particles to form formates and then proba-
bly formaldehyde-type and methoxy species (not observed) and
to finally release methane. A parallel pathway may exist (Fig. 11b)
where CO is formed by a redox cycle on reduced ceria. This path-
Fig. 10. Evolution of gas phase composition and surface species concentration dur- way may decrease activity by CO-H2 competitive adsorption on
ing transient experiment. metal particles and enhance deactivation by Ni sintering through
 P.A.U. Aldana et al. / Catalysis Today 215 (2013) 201–207 207

Fig. 11. reaction mechanism proposed on Ni-CZsol–gel sample for: (a) CO2 methanation and (b) CO formation.

the formation of di- or tri-carbonyls and by the formation of stable [10] Q. Ge, A. Kiennemann, A.C. Roger, W. Li, H. Xu, Journal of Natural Gas Chemistry
polycarbonates which block the CO2 adsorption sites. 13 (1) (2004) 41–44.
[11] F. Ocampo, B. Louis, A.C. Roger, Applied Catalysis A: General 369 (2009) 90–96.
The difference in catalytic behaviour between Ni-CZsol–gel and [12] F. Ocampo, B. Louis, L. Kiwi-Minsker, A.C. Roger, Applied Catalysis A-General
Ni-CZimp has been already discussed elsewhere [13] and has been 392 (2011) 36–44.
ascribed to the particular interaction between Ni2+ cations and CZ [13] F. Ocampo, B. Louis, A. Kiennemann, A.C. Roger, Materials Science and Engi-
neering 19 (1) (2011) (article number 012007).
support in the case of Ni-CZsol–gel which prevents the Ni◦ parti- [14] I.A. Fisher, A.T. Bell, Journal of Catalysis 162 (1996) 54–65.
cles from sintering. The strong deactivation observed for Ni-CZimp [15] C. de Leitenburg, A. Trovarelli, J. Kaspar, Journal of Catalysis 166 (1997) 98–107.
(Fig. 5) can be explained by the second pathway which will become [16] A. Beuls, C. Swalus, M. Jacquemin, G. Heyen, A. Karelovic, P. Ruiz, Applied Catal-
ysis B 113–114 (2012) 2–10.
predominant when the metal-support interface is decreased.
[17] F. Solymosi, A. Erdöhelyi, T. Bansagi, Journal of Catalysis 68 (1981) 371–382.
[18] M. Marwood, R. Doepper, A. Renken, Applied Catalysis A 151 (1997)
4. Conclusions 223–246.
[19] G.D. Weatherbee, C.H. Bartholomew, Journal of Catalysis 77 (1982)
460–472.
From the different results obtained by IR spectroscopy it appears [20] Z.A. Ibraeva, N.V. Nekrasov, B.S. Gudkov, V.I. Yakerson, Z.T. Beisembaeva, E.Z.
that on the Ni-ceria-zirconia catalysts, the main mechanism for CO2 Golosman, S.L. Kiperman, Theoretical and Experimental Chemistry 26 (1990)
620–624.
methanation does not require CO as reaction intermediate. Hence,
[21] H.Y. Kim, H.M. Lee, J.N. Park, Journal of Physical Chemistry C 114 (2010)
H2 would be dissociated on Ni◦ sites while CO2 would be activated 7128–7131.
on the ceria-zirconia support to form carbonates which would be [22] J.C. Jansen, Z. Shan, L. Marchese, W. Zhou, N.v.d. Puil, T. Maschmeyer, Chemical
hydrogenated into formates and further into methoxy species. This Communications (2001) 713–714.
[23] S. Bernal, J.J. Calvino, M.A. Cauqui, J.M. Gatica, C. Larese, J.A. Pérez Omil, J.M.
mechanism involves weak basic sites of the support for the adsorp- Pintado, Catalysis Today 50 (1999) 175–206.
tion of CO2 and implies a stable metal-support interface. It explains [24] S. Wuttke, P. Bazin, A. Vimont, C. Serre, Y.K. Seo, Y.K. Hwang, J.S. Chang, G. Férey,
the much better activity of these catalysts compared to Ni-silica on M. Daturi, Chemistry - A European Journal 18 (2012) 11959–11967.
[25] T. Lesage, C. Verrier, P. Bazin, J. Saussey, M. Daturi, Physical Chemistry Chemical
which both CO2 and H2 are activated on Ni◦ particles. Physics 5 (2003) 4435–4440.
[26] R. Takahashi, S. Sato, T. Sodesawa, S. Tomiyama, Applied Catalysis A 286 (2005)
Acknowledgment 142–147.
[27] F. Basile, P. Benito, G. Fornasari, D. Gazzoli, I. Pettiti, V. Rosetti, A. Vaccari,
Catalysis Today 142 (2009) 78–84.
The authors want to acknowledge the CNRS for the financial [28] C. Binet, M. Daturi, J.C. Lavalley, Catalysis Today 50 (1999) 207–225.
support through the PIE program no. PR10-2.2.4-2. [29] C. Li, Y. Sakata, T. Arai, K. Domen, K. Maruya, T. Onishi, Journal of the Chemical
Society, Faraday Transactions 1 85 (1989) 929–943.
[30] M. Daturi, C. Binet, J.C. Lavalley, A. Galtayries, R. Sporken, Physical Chemistry
References Chemical Physics 1 (1999) 5717–5724.
[31] G. Busca, V. Lorenzelli, Mat. Chem. 7 (1982) 89–126.
[1] G. Centi, S. Perathoner, Catalysis Today 148 (2009) 191–205. [32] M. Daturi, C. Binet, J.C. Lavalley, G. Blanchard, Surface and Interface Analysis 30
[2] E.E. Benson, C.P. Kubiak, A.J. Sathrum, J.M. Smieja, Chemical Society Reviews 38 (2000) 273–277.
(2009) 89–99. [33] M. Agnelli, H.M. Swaan, C. Marquez-Alvarez, G.A. Martin, C. Mirodatos, Journal
[3] J. Sehested, S. Dahl, J. Jacobsen, J.R. Rostrup-Nielsen, Journal of Physical Chem- of Catalysis 175 (1998) 117–128.
istry B 109 (6) (2005) 2432–2438. [34] C. Binet, A. Badri, J.C. Lavalley, Journal of Physical Chemistry 98 (1994)
[4] S.I. Fujita, N. Takezawa, Chemical Engineering Journal 68 (1997) 63–68. 6392–6398.
[5] S.I. Fujita, M. Nakamura, T. Doi, N. Takezawa, Applied Catalysis A 104 (1993) [35] M. Daturi, E. Finocchio, C. Binet, J.C. Lavalley, F. Fally, V. Perrichon, Journal of
87–100. Physical Chemistry B 103 (1999) 4884–4891.
[6] A.E. Aksoylu, Z. Misirli, I. Önsan, Applied Catalysis A 168 (1998) 385–397. [36] G.C. Cabilla, A.L. Bonivardi, M.A. Baltanás, Applied Catalysis A 255 (2003)
[7] B. Ernst, L. Hilaire, A. Kiennemann, Catalysis Today 50 (1999) 413–427. 181–195.
[8] M. Yamasaki, M. Komori, E. Akiyama, H. Habazaki, A. Kawashima, K. Asami, K. [37] D.E. Peebles, D.W. Goodman, J.M. White, Journal of Physical Chemistry 87
Hashimoto, Materials Science and Engineering A 267 (1999) 220–226. (1983) 4378–4387.
[9] M.Yamasaki, H. Habazaki, K. Asami, K. Izumiya, K. Hashimoto, Catalysis Com- [38] S.J. Choe, H.J. Kang, S.J. Kim, S.B. Park, D.H. Park, D.S. Huh, Bulletin of the Korean
munications 7 (2006) 24–28. Chemical Society 26 (2005) 1682–1688.