You are on page 1of 15

Catalysis

Science &
Technology
Published on 03 January 2022. Downloaded by Pontifícia Universidade Catolica do Rio de Janeiro on 1/21/2022 5:12:09 PM.

View Article Online


PAPER View Journal

The tendency of supports to generate oxygen


Cite this: DOI: 10.1039/d1cy01915e
vacancies and the catalytic performance of Ni/
ZrO2 and Ni/Mg(Al)O in CO2 methanation†
O. E. Everett Espino,ab P. C. Zonetti,b R. R. Celin,c L. T. Costa, c
O. C. Alves, c

J. C. Spadotto,a L. G. Appel b and R. R. de Avillez *a

Ni/ZrO2, Ni/Mg(Al)O, and Ni/SiO2 catalysts were employed in CO2 methanation. The catalysts were
characterized by XPS, XRF, XRD (Rietveld refinement method), TPR, EPR, BET, CO2 + H2-TPSR, CO + H2-
TPSR, CO2-TPD, CO-TPD, S/TEM-XEDS, and DFT calculations. The catalytic performance of these catalysts
in CO2 methanation was analyzed employing a conventional microreactor. CO2 consumption and
formation rates of the products were obtained under differential conditions. The Ni/ZrO2 catalyst exhibited
the highest activity and selectivity for the methanation of CO2 compared to Ni/Mg(Al)O and Ni/SiO2. The
Mg-based catalyst reaches the Ni/ZrO2 performance at high temperatures. Some authors have proposed
that CO rupture or formate decomposition (the rate-limiting step of CO2 methanation) occurs on pairs of
oxygen vacancy–cus sites. DFT calculations showed that oxygen vacancies improve CO adsorption and
dissociation over the Ni/ZrO2 catalyst. For Ni/ZrO2, the vacancies are generated by the insertion of Ni in
the ZrO2 lattice (DRX) and by the defects of ZrO2 oxide (EPR), whereas Mg(Al)O shows intrinsic vacancies
(EPR). Ni/ZrO2 exhibits the same metallic area compared with Ni/Mg(Al)O. However, the former is more
active and selective than Ni/Mg(Al)O at a large temperature range, showing that H2 dissociation is not the
rate-limiting step. Indeed, the Zr-based catalyst generates H2O or eliminates oxygen, generating vacancies
Received 21st October 2021, at much lower temperatures than Ni/Mg(Al)O (CO2 + H2-TPSR and CO + H2-TPSR). However, at higher
Accepted 3rd January 2022
temperatures (>400 °C), the catalytic behavior of these two catalysts is similar. Indeed, the elimination of
H2O and oxygen vacancy formation are easier for both catalysts under these conditions. Thus, this work
DOI: 10.1039/d1cy01915e
shows that the catalytic performance of Ni-based catalysts is associated with the support's facility to
rsc.li/catalysis release O (reducibility).

1. Introduction energies. Thus, storage is mandatory always to deliver energy


when there is demand. Among the many technologies
Currently, it is widely accepted that climate changes generated proposed for energy storage, “power-to-gas” (PtG) stands out.
by the greenhouse effect have damaged the flora, fauna, and Briefly, PtG can be described as follows. H2 is produced by H2O
humankind. In this context, the increase in CO2 concentration electrolysis employing the energy surplus when the energy
in the atmosphere is associated with global warming. Since the demand is lower than the generation. Then, synthetic natural
contribution of fossil fuels to this phenomenon is very relevant, gas (SNG) can be generated from the stored CO2 and H2.
considerable efforts have been made to replace these fuels with Methane is widely used as fuel, especially in the domestic
renewable ones. Nowadays, the most promising ones, i.e., wind heating, transport, and industrial sectors. According to
and solar energies, are already used worldwide. However, Ducamp et al.,1 the PtG technology enables the possible
intermittence is one of the most critical features of these interconnection of electrical and gas networks, bringing energy
management benefits. Nowadays, the SNG production cost is
a
Departamento de Engenharia Química e de Materiais, Pontifícia Universidade higher than that of natural gas (NG), according to Guilera
Católica do Rio de Janeiro, Rua Marquês de São Vicente, 225, Gávea, 22451-900, et al.'s2 cost analysis of SNG production in a PtG facility. Their
Rio de Janeiro, RJ, Brazil. E-mail: avillez@puc-rio.br data showed that the methanation process capital expenditure
b
Divisão de Catálise, Biocatálise e Processos Químicos, Instituto Nacional de Tecnologia, (CAPEX) is relevant for SNG feasibility. Therefore, studies on
Av Venezuela 82, sala 518, Praça Mauá, 20081-312, Rio de Janeiro, RJ, Brazil
c improving the methanation catalyst performance are essential
Instituto de Química, Universidade Federal Fluminense, Campos de Valonginho s/n,
Centro, 24020-141, Niterói, RJ, Brazil to the PtG technology deployment.
† Electronic supplementary information (ESI) available. See DOI: 10.1039/ Many metals like Ru, Mn, Rh, Fe, Ni, and Co are employed in
d1cy01915e CO2 methanation. Due to its low cost and good reactivity, nickel

This journal is © The Royal Society of Chemistry 2022 Catal. Sci. Technol.
View Article Online

Paper Catalysis Science & Technology

is the most used and studied. Many supports are proposed for concerning their catalytic behavior and physicochemical
this reaction, for example, Al2O3, CeO2, La2O3, MgO, MgAlO, properties. The Ni/SiO2 catalyst is employed as a reference.
SiO2, ZrO2, zeolites, TiO2, and CeO2–ZrO2.3 Our group's previous This work generates additional information associated with
work has shown that Ni supported on monoclinic ZrO2 is a very CO2 methanation by using DFT calculations.
Published on 03 January 2022. Downloaded by Pontifícia Universidade Catolica do Rio de Janeiro on 1/21/2022 5:12:09 PM.

promising catalyst for CO and CO2 methanation.4


Nowadays, the key role of the O vacancies in CO2 2. Experimental procedures
methanation has different understandings. It was proposed
2.1 Preparation of the catalyst
that the metal–support interaction in the presence of O
vacancies increases the CO2 and CO methanation rates.5,6 The catalysts Ni/ZrO2, Ni/SiO2, and Ni/Mg(Al)O were
Taking into account that the reaction occurs on the support, synthesized by a wet impregnation method using
it was suggested that the oxygen vacancies enhance the CO2 Ni(NO3)2·6H2O as a precursor. Norpro Saint Gobain supplied
adsorption7–11 by CO generation and O deposition on the O ZrO2 (monoclinic) and SiO2 (amorphous) and SASOL the
vacancies or by carbonate generation. Alternatively, the O Mg(Al)O mixed oxide (MgO : Al2O3 = 70 : 30%). The Ni
vacancies are associated with CO dissociation or formate concentrations for Ni/ZrO2, Ni/SiO2, and Ni/Mg(Al)O catalysts
decomposition to methoxy species, both suggested as the were 5 wt%, 10 wt%, and 5 wt%, respectively. After the
rate-limiting step of the methanation reaction. The CO2 impregnation, the solids were dried overnight and calcined
methanation mechanisms employing vacancy–cus pairs12 are at 500 °C for 5 h in an air flow (20 mL min−1).
described in detail in the Results and discussion section.
Indeed, new catalyst preparation procedures have been 2.2 Catalyst passivation procedure
employed to promote the generation of O vacancies and the
Firstly, the catalysts were treated at 500 °C for 30 min under
catalytic behavior and also to better describe the role of these
a N2 flow (30 mL min−1). Afterward, the Ni/ZrO2 and Ni/SiO2
defects in CO2 methanation.13–16
catalysts were reduced at 500 °C for 30 min under a pure H2
Using Ni/ZrO2 doped with Ca, our group observed that
flow (30 mL min−1), whereas Ni/Mg(Al)O was treated at 700
both mechanism suggestions, via CO or formates, seem to be
°C for 30 min under a N2 flow (30 mL min−1) and reduced at
acceptable. After doping ZrO2 with Ca, the amount of CO2
700 °C for 60 min under a pure H2 flow (30 mL min−1). After
adsorbed increased. Moreover, Ca generates extrinsic O
the reduction step, the catalysts were cooled to room
vacancies on ZrO2. Therefore, more O vacancy–cus pairs were
temperature under a N2 flow. The passivation procedure was
created, increasing the number of sites for CO dissociation or
carried out around −60 °C using a cryogenic cooling bath
formate decomposition to methoxide. Thus, the activity of
under a 4% O2/N2 flow (30 mL min−1) for 1 h. Finally, the
the catalyst increases.16
solids remained at room temperature for 1 h under a 4% O2/
It is worth noting that O vacancies should be recovered to
N2 flow.23
close the catalytic cycle. In other words, O should be removed
from the crystal defects. This subject is rarely discussed in
the literature. Nonetheless, Rodríguez et al.17 observed that 2.3 Characterization of the catalysts
adding Pr to CeO2 increases the rate of vacancy recovery. The specific surface area of the catalysts and supports was
Ni supported on Mg(Al)O has been proposed as a catalyst evaluated with N2 physisorption and the Brunauer–Emmett–
for CO2 methanation by many authors. Mg(Al)O is a mixed Teller (BET) method using a Micromeritics ASAP 2010
oxide derived from hydrotalcite-like precursors. This catalyst analyzer. The samples were oven-dried at 100 °C for 24 h.
shows a large metallic and surface area, excellent stability,18 The solids were positioned inside the apparatus and heated
and high activity and selectivity at high temperatures. under vacuum at 350 °C for 2 h. N2 adsorption was carried
However, it exhibits a poor performance at low out at −196 °C.
temperatures.19 The addition of La, Ce, Zr, Co, Fe, and V as The chemical composition analyses were carried out using
promoters to Ni/Mg(Al)O improves its catalytic behavior at a Bruker S8 Tiger wavelength dispersive X-ray fluorescence
low temperatures.19–22 According to some authors, the spectrometer (WDXRF) equipped with a rhodium tube
promoters increase the catalyst's reducibility.20 Rodrigues operated at 30–60 kW. A semi-quantitative method (QUANT-
et al.,23 studying the reverse water gas shift (RWGS) reaction EXPRESS/Bruker) determined the metal content incorporated
employing electron paramagnetic resonance (EPR), observed into the prepared catalysts.
the presence of oxygen vacancies in the Ni/Mg(Al)O catalyst. CO2 temperature-programmed desorption (CO2-TPD)
However, the literature does not associate the tendency to experiments employed a microreactor coupled to a Dycor
generate oxygen vacancies with the catalytic behavior of Ni/ mass spectrometer (Ametek Process Instruments). The Ni/
Mg(Al)O in CO2 methanation. ZrO2 and Ni/SiO2 catalysts (500 mg) were treated at 500 °C (5
This work aims to improve the knowledge related to the °C min−1) for 30 min under a He (30 mL min−1) flow before
role of the oxygen vacancies in CO2 methanation, specifically being reduced with 10% H2/N2 (30 mL min−1) for 30 min. Ni/
to describe the correlation between catalyst support Mg(Al)O was treated at 700 °C (5 °C min−1) for 30 min under
reducibility and the O vacancy recovery step of this reaction. a He (30 mL min−1) flow and reduced at the same
Hence, the catalysts Ni/ZrO2 and Ni/Mg(Al)O are compared temperature under 10% H2/N2 (30 mL min−1) for 60 min.

Catal. Sci. Technol. This journal is © The Royal Society of Chemistry 2022
View Article Online

Catalysis Science & Technology Paper

After the reduction, all samples were cooled under a He flow Ni/SiO2 samples were heated at 500 °C (5 °C min−1) for 30
(30 mL min−1) until they reached room temperature. Then, min under a N2 flow rate of 30 mL min−1. Then, the
the samples were exposed to CO2 (30 mL min−1) for 1 h at reduction step used the same temperature with pure H2 for
room temperature. Next, CO2-TPD measurements were another 30 min. The Ni/Mg(Al)O catalyst was heated at 700
Published on 03 January 2022. Downloaded by Pontifícia Universidade Catolica do Rio de Janeiro on 1/21/2022 5:12:09 PM.

carried out by heating the samples at 10 °C min−1 up to 500 °C for 30 min under a N2 flow and reduced with pure H2 at
°C under a He flow (80 mL min−1). The fragment m/z = 44 700 °C for 1 hour. This reaction was carried out at
(CO2) was continuously monitored during the analysis. In atmospheric pressure using a conventional system with a
order to classify basic sites according to their strength (weak, fixed-bed microreactor. It was monitored by online gas
medium, and strong), the TPD profiles were deconvoluted in chromatography equipped with a flame ionization detector
Gaussian curves. The weak basic sites are assigned to a curve (FID). The cyclohexane vapors were generated in a saturator
that exhibits a maximum at a temperature lower than 127 °C, (12 °C) using H2 as carrier gas (90 mL min−1). The H2/C6H12
medium basic sites between 127 °C and 307 °C, and strong ratio was 13.
basic sites above 307 °C. The electron paramagnetic resonance (EPR) spectra of the
Temperature-programmed surface reaction of CO2 and H2 reduced supports of ZrO2 and Mg(Al)O were obtained using an
(CO2 + H2-TPSR) and CO + H2 methanation (CO + H2-TPSR) ESP 300 Bruker spectrometer at a frequency of 9.7 GHz. The
experiments employed a microreactor coupled to a Dycor microwave power, field scan, and amplitude modulation were
mass spectrometer (Ametek Process Instruments). The 20 mW, 0–6000 G, and 5 G, respectively. Before the analyses,
samples (500 mg) were pretreated using the same procedure the samples (150 mg) were submitted to the following
described in the methodology of the CO2-TPD. They were pretreatment: firstly, the ZrO2 oxide was treated at 500 °C for
then exposed to CO2 (20 mL min−1) or CO (20 mL min−1) for 30 min under a He flow (30 mL min−1). Next, it was reduced at
30 min at room temperature. After that, the TPSR 500 °C for 30 min under a pure H2 flow (30 mL min−1) and then
measurements were carried out by heating the samples up to cooled to room temperature under an Ar flow (30 mL min−1).
500 °C (10 °C min−1) under a pure H2 flow (20 mL min−1). The same process was applied to Mg(Al)O, except for the
The fragments m/z = 15, m/z = 28, m/z = 44, and m/z = 18, following conditions. It was oxidized at 700 °C under a He flow
assigned to CH4, CO, CO2, and H2O, respectively, were (30 mL min−1) for 30 min and reduced at the same temperature
continuously monitored. for 60 min under a pure H2 flow (30 mL min−1). Finally, the
The temperature-programmed reduction (TPR) analyses reactors containing the samples with Ar were carefully sealed
were performed using Micromeritics AutoChem II and transported to the EPR equipment.
equipment. Firstly, the Ni/ZrO2 and Ni/SiO2 catalysts were X-ray diffraction (XRD) patterns of the calcined and
dried at 500 °C under a N2 flow (30 mL min−1) for 30 min. reduced passivated catalysts were obtained using a D8-
After that, they were reduced at 500 °C for 30 min with a Discover diffractometer equipped with a copper tube, a nickel
10% H2/N2 flow and purged for 15 min under a N2 flow. filter, and a Lynxeye solid-state detector with Cu Kα radiation
Then, they were calcined at 500 °C for 30 min under a (λ = 0.15406 nm, tube tension 40 kV and current 40 mA). The
synthetic air flow (30 mL min−1), and the samples were angular range varied from 10° to 90°, with 0.02° and 1 s per
cooled under a N2 flow (30 mL min−1) until they reached step. Rietveld refinement technique was employed using
room temperature. The same process was applied to Ni/ TOPAS software to quantitatively identify the different
Mg(Al)O, except for the reduction conditions of 700 °C for 1 crystalline phases present in the samples, their cell volumes,
h. The analysis was conducted under a 10% H2/N2 mixture lattice parameters, and site occupancies. The reduced
(50 mL min−1) from 30 to 900 °C (10 °C min−1). The ZrO2 passivated catalysts were pretreated using the procedure
oxide shows strong basic sites which react with CO2 of the described above.
air, generating carbonate species which can be reduced X-ray photoelectron spectroscopy (XPS) of the fresh
during the TPR procedure, consuming H2 and changing the samples employed a Specs Phoibos-150 hemispherical
TPR profiles. Thus, the TPR pretreatment comprises two spectrometer equipped with a monochromatic source of
steps, reduction and reoxidation of the samples. AlKα. The C1s photoelectron spectroscopy line at 284.6 eV
The CO temperature-programmed desorption (CO-TPD) binding energy was used as a reference.
analyses employed a microreactor coupled to a QMG 220 Scanning and transmission electron microscopy (S/TEM)
Prisma Plus mass spectrometer from Pfeiffer Vacuum GmbH. analyses were performed on a S/TEM JEOL JEM 2100F
The ZrO2 and Mg(Al)O samples (400 mg) were oxidized at 500 microscope, under diffraction, with an X-ray energy
°C under synthetic air (30 mL min−1) for 60 min. CO dispersive spectroscopy (XEDS) detector (Thermo Scientific)
adsorption occurred at 25 °C for 30 min (60 mL min−1) under and NSS 2.3 software. The Ni/Al(MgO) or NiZrO2 reduced
a 3% CO/He mixture. Desorption was carried out from 25 °C passivated nanoparticle samples were dissolved in
to 500 °C, at a flow rate of 80 mL min−1, maintained at 500 isopropyl alcohol and disaggregated for 15 min employing
°C for 1 h. Fragment m/z = 44 (CO2) was continuously ultrasound equipment; then, one drop of the solution was
monitored during desorption. placed on a copper grid with a carbon film. The reduced
The rate of cyclohexane dehydrogenation was used as an passivated catalysts were pretreated using the procedure
indirect measure of the Ni surface area.24,25 The Ni/ZrO2 and described above.

This journal is © The Royal Society of Chemistry 2022 Catal. Sci. Technol.
View Article Online

Paper Catalysis Science & Technology

2.4 DFT calculations and to study the adsorption of CO. The model contains 36
DFT calculations were carried out using the Quantum atoms; there are 24 O atoms and 12 Zr atoms with three
Espresso package26 under periodic boundary conditions. multilayers in sequence (O–Zr–O–O–Zr–O–O–Zr–O), leaving a
10 Å wide vacuum region in the direction perpendicular to
Published on 03 January 2022. Downloaded by Pontifícia Universidade Catolica do Rio de Janeiro on 1/21/2022 5:12:09 PM.

Projector augmented wave (PAW)27 was employed to describe


core electrons. Relaxation calculations were used to obtain the surface to prevent interactions between the periodic
the optimized geometries with the convergence criteria for images, as shown in Fig. 1b. One oxygen atom from the
energy and force of 2.0 × 10−5 Ry and 5.0 × 10−2 Ry/bohr, zirconia slab was removed to account for vacancy formation
respectively. The generalized gradient approximation (GGA) (denoted as ‘vac.’ in Fig. 1c). The oxygen vacancy formation
proposed by Perdew, Burke, and Ernzerhof (PBE)28 was used energy is defined as:
to treat the electronic exchange and correlation effects. 1
EOv ¼ Eslab‐VO þ EO2 − Eslab (1)
Grimme's semi-empirical method was used to describe the 2
long-range dispersion interactions.29 An energy cutoff of 30 Eslab-VO is the total energy of the zirconia slab with a vacancy, EO2
Ry and Gaussian smearing of 0.01 Ry were used for the is the calculated total energy for an O2 molecule, and Eslab is the
monoclinic ZrO2. The Brillouin zone was sampled using a total energy of the zirconia slab without a vacancy. A negative
Monkhorst–Pack 3 × 3 × 1 k-point mesh.30 EOv value indicates that formation of the vacancy is favorable.
The monoclinic phase shown in Fig. 1a was built from the For all calculations, the atoms in the lower and middle layers
18190 ICSD crystallographic information files. The cell were fixed, but the atoms at the surface layers and the atomic
structure has four 4-fold O and three 3-fold O atoms around positions of the adsorbed CO molecule were optimized.
one Zr atom. The optimized lattice parameters, a = 5.17 Å, b For the Ni atom adsorption on the ZrO2(111) surface
= 5.23 Å, c = 5.36 Å, and θ = 99.28°, are in good agreement study, the Ni atom was firstly located 2 Å above the oxygen
with the theoretical31–33 and experimental results.34 The slab- atom. Then, the system was optimized, leaving the top layer
building technique was applied to create the ZrO2 surfaces to relax together with the Ni atom, maintaining all the other
layers fixed. The binding energy (Eb) was determined as the
difference between the energy of the slab with the adsorbed
Ni atom ENi/ZrO2 and the sum of the energies of the clean
zirconium oxide EZrO2 and the Ni metal (ENi), as in eqn (2):

Eb = ENi/ZrO2 − (EZrO2 + ENi) (2)

Following similar studies,35,36 a negative adsorption


energy suggests that Ni atoms are more likely to spread
around the surface rather than aggregate. A positive
adsorption energy indicates that the Ni atoms prefer to
remain aggregated in the metal phase. The adsorption energy
for the systems was defined by:

Eads = E(CO/surface) − E(CO) − E(surface) (3)

E(CO/surface) is the total energy of the surface interacting with the


gas molecule of CO, and E(CO) and E(surface) are the energies of
the isolated gas and clean surface, respectively. Therefore, a
negative value means exothermic adsorption. The deformation
energy of the slabs can be obtained by eqn (4):

Edeform-slab = Eslab/ads − Eslab/isolated (4)

Eslab/ads and Eslab/isolated are the total energy of the optimized


and isolated slabs at the geometry after adsorption,
respectively. The adsorbate interaction energy on solid
surfaces was calculated by eqn (5):

Fig. 1 The bulk crystal structure of (a) monoclinic ZrO2, (b) the Eint = ECO-slab − Eslab − ECO (5)
monoclinic ZrO2 (111) surface model and (c) monoclinic ZrO−x with an
oxygen vacancy. The yellow spheres represent the oxygen vacancy.
The red spheres represent the oxygen atom, and the blue spheres ECO-slab is the energy of a slab model of the chemisorbed CO
represent the zirconium atom. molecule on the surfaces; Eslab and ECO are the energies

Catal. Sci. Technol. This journal is © The Royal Society of Chemistry 2022
View Article Online

Catalysis Science & Technology Paper

obtained by a single-point calculation after the final


structure. Finally, the projected density of states (PDOS) was
generated by self-consistent calculation using the pwscf
package.26 The charge transfer between the surfaces and the
Published on 03 January 2022. Downloaded by Pontifícia Universidade Catolica do Rio de Janeiro on 1/21/2022 5:12:09 PM.

carbon dioxide molecule was achieved using Bader's analysis


software implemented by Henkelman.37,38 Charge–density
plots were prepared using the Crystalline Structures and
Densities (XCrySDen) program.39

2.5 The methanation of CO2 and CO


Fig. 2 CO2 conversion versus temperature. The blue, yellow, and
The treatment conditions of the catalysts before reactions are black symbols represent Ni/ZrO2, Ni/Mg(Al)O, and Ni/SiO2,
described as follows: the Ni/ZrO2 and Ni/SiO2 catalysts were respectively. The experimental conditions are as follows: catalyst mass
dried at 500 °C (5 °C min−1) for 30 min under a flow of He of 200 mg, H2/CO2 ratio 4 : 1, flow rate 80 mL min−1, and pressure 1 atm.
30 mL min−1 and then at the same temperature they were
reduced with pure H2 at a flow rate of 30 mL min−1 for
another 30 min. The Ni/Mg(Al)O catalyst was dried at 700 °C
for 30 min under a He flow of 30 mL min−1, and it was catalyst shows low activity at low temperatures. This result is
reduced with pure H2 (flow rate of 30 mL min−1) at 700 °C for in line with the literature.19
1 hour; all samples were cooled under He at a flow rate of 30 Fig. 3 shows the selectivity of the Ni-based catalysts to
mL min−1 until it reached the reaction temperature. CH4 and CO at the same temperature range. The Ni/ZrO2
CO2 methanation was carried out in a fixed bed reactor catalyst exhibits selectivity toward methane of around 98%
monitored by an online Agilent GC-6880 gas chromatograph. and very low to CO from 240 to 420 °C. Due to low CO2
Catalytic tests were carried out using catalyst mass 200 mg, conversions at low temperatures, selectivity to CH4 and CO of
flow rate 80 mL min−1, GHSV 4800 h−1, temperature 350 °C, Ni/Mg(Al)O and Ni/SiO2 are only observed above 300 °C and
pressure 1 atm, and gas mixture composition CO2 : H2 : He = 270 °C, respectively. As the temperature increases, the Ni/
10 : 40 : 50. Silicon carbide (800 mg) was used as a diluent to Mg(Al)O catalyst selectivity to CH4 increases as well, from
avoid hot spots in the catalyst bed. The reactor diameter and 60% to 97%, whereas its selectivity to CO decreases. This
catalytic bed height were 22 mm and 1 cm, respectively. The catalyst reaches the Ni/ZrO2 CH4 selectivity at 390 °C. As the
grain size was standardized by employing a strainer with an temperature increases, the Ni/SiO2 selectivity toward CO
opening of 0.053 mm (ABNT 270). The catalytic tests were decreases, and to CH4 increases. The selectivity of the Ni/SiO2
run for 14 h. catalyst is the lowest to CH4 and the highest to CO.
The CO and CO2 methanation rates were measured under The CO conversion and the selectivity of Ni/ZrO2, Ni/
differential conditions (conversion <10%) at 350 °C, 1 atm. Mg(Al)O, and Ni/SiO2 catalysts to CH4 at 350 °C are shown in
Differential conditions were achieved employing the Fig. 4 and 5 as a function of time. The Ni/ZrO2 and Ni/Mg(Al)
following conditions (process of trial and error): Ni/ZrO2 O catalysts are quite stable for 14 hours on stream, but Ni/
catalyst (25 mg and 113 mL min−1), Ni/SiO2 catalyst (35 mg SiO2 exhibits some deactivation (Fig. 5). This time span was
and 82 mL min−1), and Ni/Mg(Al)O catalyst (27 mg and 44 employed to generate a first observation related to the
mL min−1). The gas mixture molar composition was CO : He = deactivation of the catalyst during CO2 methanation.
10 : 90 and CO2 : H2 : He = 10 : 40 : 50. However, this phenomenon should be studied at a higher
The CO2 conversion is defined by the ratio of the moles time on stream.
CO2 consumed to the moles CO2 introduced in the feed. The The Ni/ZrO2 catalyst exhibits the highest CO2
definition of the selectivity to one specific compound is the consumption and CH4 formation rate and the lowest CO
ratio of the number of moles CO2 consumed to synthesize
this compound to the total number of moles CO2 consumed.

3. Results and discussion


3.1 Catalytic tests
Fig. 2 exhibits the CO2 conversion as a function of
temperature for the three prepared catalysts. The Ni/ZrO2
catalyst shows higher CO2 conversions than the Mg- and Si-
based catalysts, whereas the Ni/Mg(Al)O catalyst exhibits
almost the same conversion as Ni/SiO2 at temperatures
between 200 and 360 °C. At higher temperatures, the Ni/ Fig. 3 Selectivity to CH4 (squares) and CO (triangles) versus
Mg(Al)O shows higher conversions than Ni/SiO2, reaching the temperature. The catalyst colors and the experimental conditions are
Ni/ZrO2 catalyst conversion above 400 °C. The Mg-based the same as in Fig. 2.

This journal is © The Royal Society of Chemistry 2022 Catal. Sci. Technol.
View Article Online

Paper Catalysis Science & Technology

Table 2 Specific area (S, m2 g−1), chemical analysis (wt% Ni), number of
basic sites according to their strength (μmolCO2 gcat−1), and cyclohexane
consumption rate (–rC6H12, mmol gcat−1 min−1)

Ni/ZrO2 Ni/Mg(Al)O Ni/SiO2


Published on 03 January 2022. Downloaded by Pontifícia Universidade Catolica do Rio de Janeiro on 1/21/2022 5:12:09 PM.

Specific area 66 143 67


Ni 4.84 4.52 8.15
Number of weak basic sites 47 19 2
Number of medium basic sites 73 30 4
Number of strong basic sites 84 33 —
–rC6H12 3.9 3.9 3.1

Fig. 4 Conversion versus time on stream at 350 °C. The experimental


conditions and the catalyst colors are the same as in Fig. 2.
dehydrogenation rates. Considering that this rate is directly
proportional to the Ni metallic area,24 it can be inferred that
the three catalysts show almost the same metallic area.
Indeed, the Ni content of the prepared catalysts was tuned to
obtain the same metallic area.
Fig. 6 shows the TPR profiles of the catalysts. The Ni/ZrO2
catalyst displays a peak at 329 °C, assigned to NiO reduction,
with two shoulders. According to Zonetti et al.,47 the shoulder
at a higher temperature can be associated with reduction of the
Ni species in the ZrO2 bulk. The Ni/Mg(Al)O catalyst exhibits a
very broad peak at 850 °C, which can be assigned to the
reduction of Ni2+ in the mixed oxide Mg(Al, Ni)O.23,48 It is well
Fig. 5 Selectivity to CO (dashed lines) and CH4 versus time on stream
described in the literature; Ni2+ can replace Mg2+ in the
at 350 °C. The experimental conditions and the catalyst colors are the periclase lattice.18,23,49,50 The reduction in the 250–600 °C range
same as in Fig. 2. might be attributed to NiO out of the mixed oxide. Only 75%
(Table S1†) of the Ni2+ in the Ni/Mg(Al)O catalyst was reduced,
whereas 93% was reduced for the Zr-based catalyst. Taking into
formation rate (Table 1). The Ni/Mg(Al)O catalyst shows a account the TPR profile, the pretreatment temperature
higher CH4 formation rate than Ni/SiO2, which seems to be (reduction) employed in the catalytic tests of the Ni/Mg(Al)O
slightly more active according to the CO2 consumption rate. catalyst was 700 °C (see methodology).
These results corroborate the data presented in Fig. 2 and 3 Fig. 7 shows the EPR spectra of the ZrO2 and Mg(Al)O
and show that the supports' characteristics are relevant for reduced oxides. The ZrO2 spectrum shows paramagnetic
CO2 and CO methanation.1,40–46 signals related to Zr3+ (g = 1.978 and 1.963), corresponding to
the coordinatively unsaturated (cus) sites. The ZrO2 and Mg(Al)
O spectra depict narrow signals at g = 2.06 and g = 2.05,
3.2 Characterization of the catalysts respectively, associated with ferromagnetic ordering, i.e.,
Table 2 shows the number of weak, medium, and strong isolated vacancies.23,51,52 The EPR spectra were deconvoluted
basic sites. As verified, Ni/ZrO2 shows the highest number of employing Lorentz curves. It was inferred that the vacancy
weak, medium, and strong basic sites, whereas Ni/SiO2 is the concentration of the ZrO2 oxide is 1.6 times that of the Mg(Al)O
lowest. Fig. S1† shows the CO2-TPD spectra. Table 2 shows mixed oxide. Five equally spaced lines at the g = 2.00 region of
that the oxide support defines the basicity of the catalyst. the Mg(Al)O EPR spectrum are probably due to Mn2+, which
This table also exhibits the specific area of the catalysts and may be a Mg(Al)O impurity. The Ni catalysts reduced were not
their Ni concentration obtained by X-ray fluorescence
analysis. The Mg-based catalyst shows the highest specific
area and the Si one the highest Ni concentration. Finally,
Table 2 displays very similar values of the cyclohexane

Table 1 CO2 consumption rate (−rCO2, mmol gcat−1 min−1), CH4 formation
rate (rCH4, mmol gcat−1 min−1), CO formation rate (rCO, mmol gcat−1 min−1)
for CO2 methanation at 350 °C

Catalyst Ni/ZrO2 Ni/Mg(Al)O Ni/SiO2


CO2 methanation
−rCO2 131 40 62
rCH4 122 23 14
rCO 9 17 48 Fig. 6 TPR profiles of the oxidized catalysts.

Catal. Sci. Technol. This journal is © The Royal Society of Chemistry 2022
View Article Online

Catalysis Science & Technology Paper

Fig. S3† displays the XRD patterns of the calcined


catalysts. The Rietveld refinement of the oxidized Ni/ZrO2
indicates more Ni2+ in the ZrO2 monoclinic lattice than in
the reduced catalyst (see the Ni2+ occupancy factors of the
Published on 03 January 2022. Downloaded by Pontifícia Universidade Catolica do Rio de Janeiro on 1/21/2022 5:12:09 PM.

monoclinic ZrO2 phase); as a result, the O2− occupancy


factors decreased. Thus, oxygen vacancies are created after
the calcination of the catalyst. The reduction pretreatment
favors the migration of Ni2+ to the ZrO2 surface, where it is
reduced to Ni0, leaving the ZrO2 lattice. After the reduction
Fig. 7 EPR spectra of ZrO2 and Mg(Al)O reduced oxides. process, some oxygen vacancies and Ni2+ are still present in
the ZrO2 lattice.
The XRD pattern of the Ni/Mg(Al)O reduced catalyst is also
shown in Fig. S2.† The peak related to the (0 0 2) face-
analyzed by EPR owing to the very intense signal generated by centered-cubic nickel, Ni0 (space group Fm3̄m), is visible (2θ
Ni0, which hides bands related to other species. = 51.7°). The other peaks are partially superposed with the
Fig. S2 and S3 (ESI†) depict the XRD patterns of reduced peaks diffracted by the cubic periclase (MgO), which also
and oxidized Ni/SiO2 as references of Ni0 and NiO, respectively. shows the Fm3̄m space group. The lattice parameter of the
Fig. S2† displays the XRD patterns of the reduced MgO is 0.41867 nm which is much smaller than that of
passivated catalysts. The Rietveld refinement (Table 3) normal nanocrystalline MgO (0.4212 nm)56,57 but identical to
suggests that some Ni2+ ions substitute Zr4+ ions of the ZrO2 the value found by Roelofs et al. who studied the thermal
lattice (see occupancy factors of the monoclinic ZrO2 lattice). decomposition of hydrotalcite at 723 K.58 This lattice
The smaller ionic radius of Ni2+ (0.069 nm) compared to Zr4+ parameter reduction may be attributed to the partial
(0.084 nm) favors the substitution of the latter by the former substitution of Mg2+ ions (ionic radius = 0.072 nm) by Al3+
in the ZrO2 lattice.53 Moreover, the Ni2+ and Zr4+ charge (ionic radius = 0.054 nm).18,22,23,59 The Rietveld refinement
difference creates oxygen vacancies.54,55 Table 3 also shows does suggest the presence of Al3+ ions substituting the Mg2+
the composition of the catalysts. It was observed that the ions in the periclase for both the oxidized and the reduced
total percentage of metallic Ni plus Ni present in NiO is Ni/Mg(Al)O catalysts. This substitution gives rise to a
consistent with the X-ray fluorescence analysis (Table 2). periclase solid solution with Mg2+ vacancies.60,61 However,
these vacancies are not observed by the Rietveld refinement.
According to Wang,61 the cationic vacancy concentration in
Table 3 Microstructural and crystallographic parameters determined by
periclase decreases after calcination. Thus, this mixed oxide
Rietveld refinement of XRD data
might show a very low concentration of cationic vacancies.
Catalysts Ni/ZrO2 Ni/Mg(Al)O Ni2+ also seems to be present in the periclase structure of the
Oxidized Reduced Oxidized Reduced oxidized Ni/Mg(Al)O catalyst (Table 3). However, it does not
(passivated) (passivated) generate vacancies due to its charge. The reduction
pretreatment also, in this case, favors the migration of Ni2+
ZrO2 (monoclinic)
from the mixed oxide lattice to the Mg(Al)O surface, where it
Occupancy O−2 (1) 0.92 0.94 is reduced to Ni0. XRD also shows a broad peak of MgAl2O4
O−2 (2) 0.97 0.99
Zr4+ 0.97 0.99
spinel with a tiny crystallite size.
Ni2+ 0.03 0.01 Table 3 also shows the mass percentage of the observed
nanocrystalline phases in the Ni/Mg(Al)O catalyst. The spinel-
MgO like phase could not be included due to its low crystallinity
Occupancy Mg2+ 0.44 0.6 and lack of an appropriate description.
Al3+ 0.52 0.4 Crystallite sizes determined by Rietveld fitting, integral
O−2 (1) 1 1
Ni2+ 0.04 0 breadth method, showed that the pure ZrO2 monoclinic is
about 10 nm and did not change with the Ni/ZrO2 catalyst's
Phases heat treatment. The oxidized Ni/ZrO2 catalyst showed a face-
Phase/wt% m-ZrO2 92.2 92.5 centered-cubic NiO phase with 14 nm and the reduced
t-ZrO2 2.3 2.1 catalyst showed a metallic nickel phase with 21 nm. Thus,
Ni0 0 4.2 0 2.6 the reduction treatment allowed the metallic crystallites to
NiO 5.4 1.2 0 0
MgO 99.1 97.2 grow. The crystallite size of the periclase formed during the
Al2O3 0.9 0.3 calcination of pure Mg(Al)O is about 3.3 nm. However, the
Cell volume/Å3 m-ZrO2 139.91 140.22 Ni/Mg(Al)O catalyst's crystallite size is about 4.6 nm for both
t-ZrO2 67.12 67.40
the reduced and the oxidized conditions. The crystallite size
Ni0 43.65 43.95
NiO 54.40 55.00 of the Ni0 formed during the reduction treatment of this
MgO 75.5010 73.39 catalyst is about 4.6 nm. For Ni/Mg(Al)O and Ni/ZrO2

This journal is © The Royal Society of Chemistry 2022 Catal. Sci. Technol.
View Article Online

Paper Catalysis Science & Technology

Table 4 XPS analysis of the oxidized Ni/Mg(Al)O and Ni/ZrO2 catalysts of crystalline nanoparticles, MgO and Ni0. However, an Al
Ni 2p3/2 binding Atomic
compound was not observed. Indeed, this element is in the
Catalyst energy/eV Species ratio MgO crystal forming the Mg(Al)O mixed oxide and, probably,
an amorphous phase component. XEDS analysis confirms
Published on 03 January 2022. Downloaded by Pontifícia Universidade Catolica do Rio de Janeiro on 1/21/2022 5:12:09 PM.

Ni/Mg(Al)O 853.2 NiO (ref. 63) Ni/Mg 0.027


855.6 Ni2+ in the MgO lattice23 the presence of Al, Mg, O, and Ni. Fig. 8c and d present the
861.5 Satellite corresponding centered dark-field (CDF) images of Fig. 8a
Ni/ZrO2 852.9 NiO (ref. 63) Ni/Zr 0.03 obtained from two different diffraction ring spots of the
855.5 Ni2+ in the ZrO2 lattice62
861.6 Satellite
SAED pattern (Fig. 8b). Fig. 8c shows the CDF image
Pure NiO 853.3 NiO (ref. 63) generated with the (220) diffraction planes of MgO crystals.
(reference) Fig. 8d is the CDF image generated with the (200) diffraction
planes of Ni0 crystals of the catalyst. It is worth noting that
the nanoparticles of Ni/Mg(Al)O catalyst are well distributed
catalysts, the Ni0 and NiO phases, respectively, show and not agglomerated (this observation was consistent in
crystallite sizes of about the same dimension as the supports. several TEM analyses). The average Mg(Al)O particle size is
Fig. S4† displays the Ni 2p3/2 spectra of the oxidized approximately 4.23 ± 2.27 nm, whereas that of Ni0 is 5.45 ±
catalysts. Table 4 shows the Ni 2p3/2 binding energies and 2.57 nm, both agreeing with the XRD results. This result
their assigned species. It is interesting to stress that both suggests that Ni0 and Mg(Al)O particles might be
catalysts show the same following species: Ni2+ cation in the nanocrystals. See histograms (Fig. S5 and S6†).
Mg(Al)O and ZrO2 lattice and NiO.23,62 These species are in Fig. 9 shows the BF image of the passivated reduced Ni/
line with the XRD results, except for NiO of the Mg-based ZrO2 catalyst and its corresponding STEM/XEDS maps of Ni,
catalyst, which was not observed using XRD. It can be Zr, and O. It is verified that this catalyst is constituted mostly
suggested that this oxide shows small particles not detectable by spherical-like nanoparticles. STEM/XEDS results confirm
by this technique. the presence of Ni, Zr, and O in this catalyst. However, unlike
Table 4 also shows the XPS atomic ratios of Ni/Mg and Ni/ what was observed for the Ni/Mg(Al)O catalyst, Ni seems to
Zr. It can be inferred that Ni/ZrO2 and Ni/Mg(Al)O present be concentrated in specific regions of the support (this
similar Ni atomic concentrations on their surfaces. This result observation was consistent in several STEM/XEDS analyses).
is in line with cyclohexane dehydrogenation rates (Table 2), The ZrO2 average particle size is in the range of 13–14 nm.
which suggest that these catalysts show similar metallic areas. This result is in line with the Rietveld fitting, which showed
The bright-field (BF) TEM image in Fig. 8a shows the that monoclinic ZrO2 was about 10 nm (this ZrO2 particle
passivated reduced Ni/Mg(Al)O catalyst. The selected area seems to be a nanocrystal). The size of the Ni0 particle was
electron diffraction (SAED) pattern in Fig. 8b confirms that also measured. The value obtained is 25.03 ± 6.90 nm, which
the Ni/Mg(Al)O catalyst observed in Fig. 8a contains two types is again in line with the XRD data.

Fig. 8 (a) BF-TEM image of Ni/Mg(Al)O catalyst; (b) SAED pattern obtained from (a); (c) and (d) centered dark-field (CDF)-TEM images obtained
from the spot marked with circles 1 and 2 in (b), respectively.

Catal. Sci. Technol. This journal is © The Royal Society of Chemistry 2022
View Article Online

Catalysis Science & Technology Paper


Published on 03 January 2022. Downloaded by Pontifícia Universidade Catolica do Rio de Janeiro on 1/21/2022 5:12:09 PM.

Fig. 11 Adsorption of CO onto the Ni/ZrO2 (111) surface. Red, cyan,


green, and gray spheres represent O, Zr, Ni, and C atoms, respectively.

distance, 1.07 Å, is slightly shortened with respect to the


free molecule (1.14 Å); the C atom in CO is adsorbed on the
surface of the Zr atom, generating a new surface behaving
as a Brønsted acid site.
The results for this CO–Ni/ZrO2 system can be compared
with similar calculations made recently.36 Important
electronic states of the CO molecule before and after
Fig. 9 BF-STEM image of the Ni/ZrO2 catalyst and corresponding
STEM-XEDS of Ni, Zr, and O.
adsorption are shown in Fig. 12. The distance between the
carbon atom and zirconium for this system is greater, with a
value of 2.04 Å and Eads of −1.83 eV.
Fig. 12 shows the density of states (DOS) for a single
3.3 Computational study molecule of CO (Fig. 12a) and the PDOS of the carbon atom
DFT calculations were carried out to support the of CO adsorbed on the surface of the NiZrO2. Fig. 12a and b
experimental data and shed light on the possible mechanism clearly show that the electronic states of CO at approximately
of the CO2 methanation surface from a microscopic point of −12.8, −10.5, and −7.8 eV (Fig. 12a) are reduced considerably
view. The ZrO2 (111) surface structure model, as shown in in energy as the CO binds to the Ni/ZrO2 surface (Fig. 12b).
Fig. 1b, exposes three multilayers, each of which includes The CO adsorption mechanism on the top site of Zr of Ni/
one zirconium layer enclosed between two oxygen layers in ZrO2 can be qualitatively explained as follows. The CO
the sequence O–Zr–O–O–Zr–O–O–Zr–O. The surface Zr–Zr molecule prefers to be adsorbed on the top site due to the
bond distance is 3.36 Å, and the Zr–O bond distance is 2.05 interaction of HOMO-5σ from CO, and this interaction
Å. The lattice constant of bulk ZrO2 is 5.17 Å. The CO was results in orbitals shifted to lower energies. Similarly, the 2π*
initially positioned at 2.00 Å above the surface, and different antibonding CO molecule orbitals that interact with Zr at the
orientations have been compared for each initial adsorption
site. On the ZrO2 (111) surfaces, four initial adsorption sites
were tested for the CO molecule. The most favorable position
is on top of the Zr atom (Fig. 10).
As shown in Fig. 11, after full structural optimization,
the CO gas molecule preferred to adsorb on the Zr top site
of the Ni/ZrO2 (111) surface, showing the same behavior
observed with the ZrO2 (111) support. The adsorption energy
for CO on ZrO2 (111) is −0.35 eV. Luo et al.64 found a
similar behavior in their investigation of the adsorption of
CO on ZrO2 (111) and doped with Ca. The C–O bond

Fig. 12 (a) Total density of states of the free CO (blue) and (b) the
Fig. 10 Adsorption of CO onto the ZrO2 (111) surface. Red, cyan, and partial density of states, PDOS, onto the Bader charge of the CO
gray spheres represent the O, Zr, and C atoms, respectively. (orange) molecule placed above the Ni/ZrO2 (111) surface.

This journal is © The Royal Society of Chemistry 2022 Catal. Sci. Technol.
View Article Online

Paper Catalysis Science & Technology


Published on 03 January 2022. Downloaded by Pontifícia Universidade Catolica do Rio de Janeiro on 1/21/2022 5:12:09 PM.

Fig. 13 (a) Doped Ni2/ZrO−x (111) represented by a nickel slab model


with one oxygen vacancy on the top surface. (b) Electron localization
function (ELF) plot for the Ni2/ZrO−x (111) surface with one fourfold
oxygen vacancy.
Fig. 14 Adsorption of CO onto the Ni2/ZrO−x (111) surface. Red, cyan,
green, and gray spheres represent the O, Zr, Ni, and C atoms,
respectively.

surface also expand and move towards the Fermi level, which
is slightly occupied, as shown in Fig. 12b.
The experimental results showed that zirconia oxide Table 5 Adsorption energies, distances, interaction energy, deformation
doped with nickel generates vacancies responsible for for CO adsorbed on the ZrO2, Ni/ZrO2, and Ni2/ZrO−x surfaces
preferential CO adsorption.10 The effect of oxygen vacancy Total Bader charge (e−)
Adsorption Eads Eint Edeform-slab D
and surface Ni substituting the Zr4+ ions was also
system (eV) (eV) (eV) (Å) Sum(CO) = 0.01(free)
investigated. The Ni ion was placed near a fourfold oxygen
CO–ZrO2 −0.35 −0.89 0.11 1.07 −0.07
vacancy position (denoted as Ni2/ZrO−x), as shown in Fig. 13.
CO–Ni/ZrO2 −1.28 −3.13 0.21 1.13 −0.19
The vacancy formation energy for this environment was −5.35 CO–Ni2/ZrO−x −3.80 −5.89 2.90 1.19 −0.60
eV; the Bader's charge analysis indicates a charge of 2.51 e−
on the Zr atom and 2.23 e− on each Zr atom neighboring
state upon the formation of the oxygen vacancy, indicating (LUMO) is the degenerate 2π antibonding orbital. Fig. 15b
that Zr atoms change their oxidation state from Zr4+ to Zr3+. shows the PDOS and Bader charge of the adsorbed CO
The removal of an oxygen ion from the ZrO2 (111) surface molecule in Ni2/ZrO−x. Compared to the free CO molecule,
leads to the localization of d orbital electrons in the vacancy the peak of 5σ and 4σ is significantly depressed in the
region. Hence, the highly favorable formation of a vacancy adsorbed CO due to the transfer of electrons from the CO to
can be attributed to the introduction of a localized electronic the Zr atoms. The 2π* isolated CO molecule orbitals move to
charge of 2 e− with two unpaired electrons situated on two d the low-energy region below the Fermi level and hybridize
orbitals nearest to the formed vacancy, as in the case of a strongly with the Ni2/ZrO−x surface's higher valence band.
vacancy in Ni/ZrO2 doped with nickel (denoted as Ni2/ZrO−x). The significant elongation of the 2π orbital indicates a
Fig. 13b shows the electron localization function (ELF) strong interaction of the adsorbed molecule with the Ni2/
mapped on the a plane that contains the vacancy. The ELF is ZrO−x. The charge transfer of −0.60 e− induces the broadening
high in the oxygen vacancy region caused by doping with nickel, of the 2π* orbital and experiences an increase in electron
showing that the electrons remain trapped at the defect site. density upon adsorption, resulting from interacting with the
Calculations performed on the Ni2/ZrO−x (111) surface d orbitals of the Ni2ZrO−x surface, which results in an
with an oxygen O2− vacancy showed that CO is also adsorbed. elongation of the C–O bond (from 1.14 Å to 1.19 Å) due to
Fig. 14 shows the fourfold coordinated oxygen vacancy, which the antibonding nature of the 2π* orbital. Therefore, such a
changes the oxidation state from +4 to +3. The detected CO process favors the dissociation of the CO molecule and
interaction was a strong chemisorption type, with Eads,CO = greatly influences the absorbed molecule.
−3.80 eV, and the C atom bonds to two Zr atoms. Table 5 The calculated adsorption energies are listed in Table 5. We
shows that the adsorbate CO molecule's Bader charges are can find that adsorption on ZrO2 (111) (Ead = −0.35 eV) is less
higher than that for the free molecule. The charge transfer of stable than at Ni/ZrO2 (Ead = −1.28 eV). The adsorption of CO
−0.60 e− implies that the interaction between the surface and over Ni2/ZrO−x is more favored than over Ni/ZrO2 and ZrO2 sites.
the adsorbed molecule is very strong. It is important to note The present calculation is consistent with the literature
that in the reduced Ni2/ZrO−x (111) surface, the molecule only reports on the behavior of CO on the surface with vacancies.65
interacts with two atoms of the Zr3+ bridge type (located at In this sense, it has been indicated that the interaction of CO
the top of the anionic vacancy generated by the insertion of on the Ni2/ZrO−x (111) surface is very prominent when oxygen
Ni into the ZrO2 lattice as shown in Fig. 14). vacancies are present.66 Indeed, Metiu et al.67 have shown that
The mechanism of CO bridge adsorption over the Ni2/ low valence dopants turn the oxide surface into a Lewis base
ZrO−x nickel-doped surface can be explained by inspecting that strongly favors CO adsorption. The calculated interaction
the CO molecule and the surface electronic structure before energies (Eint) for the system with CO are shown in Table 5.
and after adsorption. The calculated partial density of states This strong molecule (CO)/surface coupling was confirmed by
(PDOS) of CO projected onto the C and O are plotted in the interaction energy of −3.13 eV and −5.89 eV for the Ni/ZrO2
Fig. 15. For the free CO molecule, as observed in Fig. 15a, the and Ni2/ZrO−x (111) surfaces, respectively, showing that the
highest occupied molecular orbital (HOMO) is the degenerate molecule was strongly bonded to the surface by the presence of
5σ orbital, and the lowest unoccupied molecular orbital oxygen vacancy. The deformation energy of the most stable Ni2/

Catal. Sci. Technol. This journal is © The Royal Society of Chemistry 2022
View Article Online

Catalysis Science & Technology Paper

with the cus–oxygen vacancy pair, where it further


dissociates, freeing the carbon to be hydrogenated to CH4.
The CO dissociation, according to some authors, can be H-
assisted.68 The rate-limiting step is the CO dissociation on
Published on 03 January 2022. Downloaded by Pontifícia Universidade Catolica do Rio de Janeiro on 1/21/2022 5:12:09 PM.

the cus–oxygen vacancy pair. Again, the O will be removed


from the vacancy by the H species and will produce H2O.
Both mechanisms considered C–O bond dissociation as
the rate-limiting step, involving the cus–oxygen vacancy
pairs.7,8,10,11,16 Thus, the elimination of the O, i.e., the
recovery of the vacancies, is a critical step.
Fig. 3 and Table 1 show that all these catalysts generate
CO. This means that all these solids might synthesize CH4
according to the CO mechanism. The formate might be
related to the catalysts with active supports, i.e., Ni/ZrO2 and
Fig. 15 (a) Total density of states of free CO (blue) and (b) PDOS onto Ni/Mg(Al)O. Indeed, in the case of Ni/ZrO2 and Ni/Mg(Al)O,
the Bader charge of the CO (red) molecule placed above the nickel- these mechanisms might occur concomitantly.
doped Ni2/ZrO−x (111) surface with one oxygen vacancy. The Ni/SiO2 catalytic behavior shows that Ni can generate
CH4. However, its lower activity and selectivity to CH4 (Fig. 2
and 3) are associated with silica's inert character.70,71
ZrO−x was slightly higher than the deformation energy for the Fig. 16 and 17 display the CO2 + H2-TPSR profiles for Ni/
ZrO2 and Ni/ZrO2 surfaces, according to Table 5. The weak ZrO2 and Ni/Mg(Al)O, respectively. Comparing the CH4 and
bond could explain the low deformation energy for ZrO2 CO2 desorption, the Zr-based catalyst seems to adsorb more
between the carbon atom and the zirconium atoms, unlike Ni/ CO2 than Ni/Mg(Al)O. Moreover, H2O desorption of Ni/ZrO2
ZrO2 and Ni2/ZrO−x surfaces. These data suggest a continuous (261 °C) occurs at much lower temperatures compared with
weakening of the CO bond and consequently leads to an Ni/Mg(Al)O (361 °C). It is observed that the maximum
increase of the adsorption energy due to oxygen vacancy. The evolution of CH4 from the Mg-based catalyst occurs at 235
fourfold oxygen vacancy on the surface adjacent to the Zr atom °C, whereas that of Ni/ZrO2 at 261 °C, suggesting that the
increased the stability and further improved CO molecule Mg-based catalyst is slightly more active. Thus, the
adsorption. This feature could be attributed to oxygen leaving elimination of water and recovery of vacancies is more
two excess electrons on Zr4+ cations, filling the empty 4d difficult for the Ni/Mg(Al)O catalyst, which might impact the
orbitals of the Zr4+ cations to form Zr3+. Such a process activity and selectivity of this catalyst in CO2 methanation.
positively influences the dissociation of the CO molecule. Fig. S7† exhibits the CO2 + H2-TPSR spectra related to Ni/
SiO2. The amount of CH4 generated is minimal compared
with Zr- and Mg-based catalysts. This result shows that the
3.4 Further results and discussion primary role of Ni is H2 dissociation.
Two mechanisms seem to describe this reaction better when The CO + H2-TPSR spectra (Fig. 18) shows similar CH4
active supports are employed. The first assumes that the bands (intensity and maximum temperature) for both
adsorbed CO2 generates carbonate species that react with catalysts. However, the H2O spectra are again different. Their
monoatomic H species, formed by the parallel H2 maxima are around 313 °C and 418 °C for Ni/ZrO2 and Ni/
dissociation on the metal, which spill over towards the Mg(Al)O, respectively, confirming again that it is difficult to
support. The CO2 adsorption occurs on the basic sites and remove the oxygen of the Mg-based catalyst.
also on the oxygen vacancies. The carbonate species are Fig. 19 displays the CO-TPD of ZrO2 and Mg(Al)O. Both
hydrogenated to formate species; these species are reduced catalysts generate CO2. However, the CO2 band related to
to methoxide species and after that decompose to CH4. ZrO2 occurs at a lower temperature, confirming that CO
Finally, H reduces the surface recovering the vacancies, and
H2O is generated.8,68 The rate-limiting step is the dissociation
of the C–O bond of the formate species, which occurs on the
cus–oxygen vacancy pair. The second mechanism16,40,46,68,69
assumes that the adsorbed CO2 gives one of its O ions to an
anionic vacancy of the support and generates CO, which is
adsorbed on the oxide surface, allowing more time for the
following reactions. The H2 molecule dissociates on the
metal, and the monoatomic H species spill over to the oxide,
where they react with the O ion, generating H2O and
recovering the oxygen vacancy. This first step is the reverse
water gas shift (RWGS) reaction. The CO molecule interacts Fig. 16 CO2 + H2-TPSR profile of reduced Ni/ZrO2.

This journal is © The Royal Society of Chemistry 2022 Catal. Sci. Technol.
View Article Online

Paper Catalysis Science & Technology


Published on 03 January 2022. Downloaded by Pontifícia Universidade Catolica do Rio de Janeiro on 1/21/2022 5:12:09 PM.

Fig. 17 CO2 + H2-TPSR profile of reduced Ni/Mg(Al)O. Fig. 19 CO-TPD profile of oxidized Mg(Al)O (yellow) and oxidized
ZrO2 (blue).

oxidation is easier on this catalyst. Considering the Mars–Van


Krevelen mechanism, it can be inferred that ZrO2 is reduced Thus, the ranking of the number of vacancies of the catalysts
more easily than Mg(Al)O. is the following: Ni/Mg(Al)O < Ni/ZrO2.
The XRD Rietveld refinement of the Ni/Mg(Al)O catalyst Table 1 (CO2 methanation rates) depicts that increasing the
does not show anionic (oxygen) vacancies, whereas the EPR vacancy concentration of the catalysts decreases the CO
spectrum exhibits these vacancies. The samples analyzed by formation rates and increases the CH4 formation rates. As
XRD were reduced and passivated. This procedure might described above, C–O rupture is considered the rate-limiting
eliminate the O superficial vacancies. The EPR data were step of the CO mechanism. According to the DFT calculations
obtained from samples reduced but not passivated, which (section 3.3), the vacancies facilitate the C–O dissociation,
might keep these vacancies. The Mg(Al)O particle and crystal weakening this bond and keeping this molecule adsorbed on
sizes are around 4.2 nm (TEM and XRD). Thus, the anionic the surface. Thus, vacancies on the oxide surface promote the
vacancies can be associated with the nanoparticle's intrinsic CO and CO2 methanation reaction, in line with Table 1 results.
defects. The Ni0 particle size of the Ni/ZrO2 reduced and passivated
Further, formation of cationic vacancies might be sample, observed by TEM, is much larger than that of the Mg-
expected due to the charge of Al3+ and Mg2+. However, these based catalyst. Considering that both catalysts show the same
vacancies were not observed. According to Wang et al.,61 Ni concentration (Table 2), the metallic surface area of Ni/
these vacancies vanish due to high-temperature treatments. Mg(Al)O should be higher than that of the Zr-based catalysts.
XRD Rietveld refinement of oxidized and reduced Ni/ZrO2 However, the XPS data and cyclohexane dehydrogenation rates
catalysts does show Ni2+ in the ZrO2 lattice and O vacancies. showed that both catalysts have similar Ni0 metallic areas.
The EPR spectrum of ZrO2 (monoclinic) also exhibits O Indeed, the Ni/Mg(Al)O reduction process temperature is not
vacancies. XPS analysis of the oxidized sample also shows the able to reduce all Ni (see TPR profiles, Fig. 6). It is proposed that
presence of Ni2+ in the ZrO2 lattice (surface). Silva-Calpa et al. Ni substitutes Mg in the MgO structure, which forms a NiO–
showed that adding even small amounts of Zn, a lower- MgO solid solution74 (see Table 3) and therefore is not available
valence dopant, to ZrO2 generates O vacancies and increases as a metallic nickel. Moreover, the XRD data (Table 3) of the
O mobility.72 Thus, the reducibility of the ZrO2 is improved reduced and passivated samples show that the amount of Ni0 of
by Ni in the ZrO2 lattice. the Mg-based catalyst is much lower than that of Ni/ZrO2. Thus,
The SiO2 substrate is amorphous, so it does not allow the the smaller particle size of Ni0 of the Mg-based catalyst would
classical definition of vacancies in its structure. Moreover, generate a higher metallic area than Ni/ZrO2 but is
recent DFT calculations show that the diffusion of neutral O compensated for by the lower concentration of Ni0 on the Ni/
vacancies in amorphous SiO2 is unfeasible.73 Therefore, this Mg(Al)O surface, which leads to a lower number of Ni0 particles.
mechanism will not be available during the catalytic process. The Ni/ZrO2 is more active than Ni/Mg(Al)O (Table 1) in the
CO2 methanation. Considering that both catalysts show the
same metallic area, it can be inferred that the dissociation of
H2 in the CO2 methanation is not kinetically relevant.
The XRD and EPR data show that after reduction
(pretreatment), much more vacancies are observed for the Zr-
based catalyst than for the Ni/Mg(Al)O catalyst. Table 1
exhibits that the ratio between the rate of CH4 formation of
Ni/ZrO2 and Ni/Mg(Al)O is 5.3 at 350 °C. However, on
increasing the temperature the Mg- and Zr-based catalysts
exhibited similar behavior, as observed in Fig. 2 and 3. Thus,
the number of vacancies generated after the reduction
Fig. 18 CO + H2-TPSR profile of reduced Ni/Mg(Al)O (yellow) and process (pretreatment of the catalysts) seems not to be the
reduced Ni/ZrO2 (blue). main factor for describing the catalytic behavior of these

Catal. Sci. Technol. This journal is © The Royal Society of Chemistry 2022
View Article Online

Catalysis Science & Technology Paper

catalysts. The CO2 + H2-TPSR spectra of Ni/ZrO2 and Ni/ Conflicts of interest
Mg(Al)O show that the former generates vacancies more
easily (H2O elimination). The mechanisms described above There are no conflicts to declare.
suggest that the rate of O elimination from the supports is
Published on 03 January 2022. Downloaded by Pontifícia Universidade Catolica do Rio de Janeiro on 1/21/2022 5:12:09 PM.

intrinsically linked to the catalytic behavior. Thus, the Acknowledgements


number of vacancies and the reducibility of the supports are The authors are grateful to the Brazilian Center for Research in
both relevant for catalyst activity and selectivity to CH4. Physics (CBPF-Brazil) for the access to the LABNANO Electron
It is interesting to observe that the catalytic behavior of Microscopy facilities and RPE laboratory for the EPR analysis.
Ni/Mg(Al)O changes with the temperature (Fig. 2 and 3). As We are also thankful to the LACAT/INT staff for the
the temperature increases, the activity and selectivity to CH4 characterization of the catalysts. The authors acknowledge
of the Mg-based catalyst increases and are very similar to that Saint-Gobain NorPro and also Sasol for supplying the supports.
of the Zr-based catalyst for temperatures higher than 400 °C. This study was financed in part by the Coordenação de
It can be suggested that as the temperature increases, the O Aperfeiçoamento de Pessoal de Nível Superior – Brasil, (CAPES)
elimination, i.e., vacancy generation, is easier. Therefore, the – Finance Code 001, and Conselho Nacional de
performance of the Mg-based catalyst improves. It is worth Desenvolvimento Científico e Tecnológico – CNPq, Brazil, CNPq
noting that the behavior of Ni/SiO2 improves in the CO2 305095/2015-3. LT Costa is grateful for the funding provided by
methanation as the temperature of the reaction increases. FAPERJ, PRINT-CAPES, and CNPq fellowship 7.310459/2018-00.
However, it did not reach the performance of the Mg- and Zr-
based catalysts. References
According to Metiu,75 the higher the electrophilic
character of the surface, the easier the oxygen vacancies are 1 J. Ducamp, A. Bengaouer, P. Baurens, I. Fechete, P. Turek
generated. Zirconium is much more electrophilic than Mg and F. Garin, C. R. Chim., 2018, 21, 427–469.
and Al at low concentrations in Ni/Mg(Al)O. The rate-limiting 2 J. Guilera, J. R. Morante and T. Andreu, Energy Convers.
step of the CO2 methanation occurs on the oxygen vacancies. Manage., 2018, 162, 218–224.
Thus, the Ni/ZrO2 catalyst is more active than Ni/Mg(Al)O 3 K. Ghaib and F. Z. Ben-Fares, Renewable Sustainable Energy
due to the electrophilic character of its surface. At high Rev., 2018, 81, 433–446.
temperatures, the behavior of these catalysts becomes similar 4 D. C. D. da Silva, S. Letichevsky, L. E. P. Borges and L. G.
due to the increased oxygen mobility under these conditions. Appel, Int. J. Hydrogen Energy, 2012, 7, 1–6.
5 Z. Hao, J. Shen, S. Lin, X. Han, X. Chang, J. Liu, M. Li and X.
4. Conclusion Ma, Appl. Catal., B, 2021, 286, 119922.
6 S. Chen, A. M. Abdel-Mageed, M. Li, S. Cisneros, J.
The Zr- and Mg-based catalysts are active and selective catalysts Bansmann, J. Rabeah, A. Brückner, A. Groß and R. J. Behm,
for CO2 methanation. However, the latter should be employed J. Catal., 2021, 400, 407–420.
at high temperatures. This work shows that the activity and 7 J. Ashok, M. L. Ang and S. Kawi, Catal. Today, 2017, 281,
selectivity of the Ni-based catalysts are determined by the 304–311.
facility of the support to release O (reducibility). 8 F. Wang, S. He, H. Chen, B. Wang, L. Zheng, M. Wei, D. G.
The catalytic behavior of Ni/SiO2 is very poor compared Evans and X. Duan, J. Am. Chem. Soc., 2016, 138, 6298–6305.
with that of the Mg- and Zr-based catalyst. Indeed, Ni0 can 9 H. Takano, H. Shinomiya, K. Izumiya, N. Kumagai, H.
generate CH4 from CO2 and contribute to the Ni/Mg(Al)O and Habazaki and K. Hashimoto, Int. J. Hydrogen Energy,
Ni/ZrO2 catalytic behaviors. The Ni/ZrO2 and Ni/Mg(Al)O 2015, 40, 8347–8355.
catalysts show the same metallic area. However, the former is 10 X. Jia, X. Zhang, N. Rui, X. Hu and C. J. Liu, Appl. Catal., B,
more active and selective than Ni/Mg(Al)O at a broad 2019, 244, 159–169.
temperature range, showing that H2 dissociation is not 11 J. Lin, C. Ma, Q. Wang, Y. Xu, G. Ma, J. Wang, H. Wang, C.
kinetically relevant. Dong, C. Zhang and M. Ding, Appl. Catal., B, 2019, 243,
The intrinsic vacancies of the Mg(Al)O nanocrystal seem 262–272.
to be very important for the Ni/Mg(Al)O catalyst performance. 12 Y. Liu, C. Xia, Q. Wang, L. Zhang, A. Huang, M. Ke and Z.
In the case of Ni/ZrO2, the vacancies are both intrinsic and Song, Catal. Sci. Technol., 2018, 8, 4916–4924.
extrinsic, generated by the Ni insertion in the ZrO2 lattice. 13 X. Xu, L. Liu, Y. Tong, X. Fang, J. Xu, D. E. Jiang and X.
These vacancies are also certainly related to the ZrO2 Wang, ACS Catal., 2021, 11, 5762–5775.
reducibility and also relevant for the catalytic behavior. 14 M. Zhu, P. Tian, X. Cao, J. Chen, T. Pu, B. Shi, J. Xu, J.
DFT calculations pointed out that oxygen vacancies Moon, Z. Wu and Y. F. Han, Appl. Catal., B, 2021, 282,
promote CO adsorption and also C–O rupture, the rate- 119561.
limiting step of the CO2 methanation. Thus, the oxygen 15 J. Zeng, Y. Sun, J. Zhang, Z. Chang, J. Yang, J. Liu, J. Li and
vacancies are directly associated with the rate-limiting step of F. Yu, Fuel, 2022, 309, 122099.
the CO2 methanation. The DFT calculations are in line with 16 O. E. Everett, P. C. Zonetti, O. C. Alves, R. R. de Avillez and
the catalytic behavior. L. G. Appel, Int. J. Hydrogen Energy, 2020, 45, 6352–6359.

This journal is © The Royal Society of Chemistry 2022 Catal. Sci. Technol.
View Article Online

Paper Catalysis Science & Technology

17 A. Bueno-Lopez, S. L. Rodríguez, A. Davo-Quinonero, J. Juan- 45 Z. L. M. Cai, J. Wen, W. Chu and X. Cheng, J. Nat. Gas
Juan, E. Bailon-García and D. Lozano-Castello, J. Phys. Chem. Chem., 2011, 20, 318–324.
C, 2021, 125, 12038–12049. 46 X. Su, J. Xu, B. Liang, H. Duan, B. Hou and Y. Huang,
18 N. Bette, J. Thielemann, M. Schreiner and F. Mertens, J. Energy Chem., 2016, 25, 553–565.
Published on 03 January 2022. Downloaded by Pontifícia Universidade Catolica do Rio de Janeiro on 1/21/2022 5:12:09 PM.

ChemCatChem, 2016, 8, 2903–2906. 47 P. C. Zonetti, S. Letichevsky, A. B. Gaspar, E. F. Sousa-Aguiar


19 Y. Wang, Y. Xu, Q. Liu, J. Sun, S. Ji and Z. J. Wang, J. Chem. and L. G. Appel, Appl. Catal., A, 2014, 475, 48–54.
Technol. Biotechnol., 2019, 94, 3780–3786. 48 X. Yu, N. Wang, W. Chu and M. Liu, Chem. Eng. J., 2012, 209,
20 Q. Liu, J. Sun, Q. Feng, S. Ji and Z. J. Wang, Catal. Today, 623–632.
2020, 339, 127–134. 49 A. F. Lucrédio, J. D. A. Bellido and E. M. Assaf, Appl. Catal.,
21 K. Świrk, P. Summa, D. Wierzbicki, M. Motak and P. Da A, 2010, 388, 77–85.
Costa, Int. J. Hydrogen Energy, 2021, 46, 17776–17783. 50 B. Rebours, J. B. D. E. de la Caillerie and O. Clause, J. Am.
22 C. Mebrahtu, F. Krebs, S. Perathoner, S. Abate, G. Centi and Chem. Soc., 1994, 116, 1707–1717.
R. Palkovits, Catal. Sci. Technol., 2018, 8, 1016–1027. 51 N. Guskos, G. J. Papadopoulos, V. Likodimos and S. Patapis,
23 M. T. Rodrigues, P. C. Zonetti, O. C. Alves and E. F. Sousa- Mater. Res. Bull., 2002, 37, 1051–1061.
aguiar, Appl. Catal., A, 2017, 543, 98–103. 52 B. M. Maoz, E. Tirosh, B. Sadan, I. Popov and G. Markovich,
24 D. C. R. M. Santos, L. Madeira and F. B. Passos, Catal. J. Mater. Chem., 2011, 21, 9532–9537.
Today, 2010, 149, 401–406. 53 F. Sun, C. Yan, Z. Wang and C. Guo, Int. J. Hydrogen Energy,
25 W. J. Conner, G. Pajonk and S. Teichner, in Advances in 2015, 40, 15985–15993.
catalysis, ed. W. P. D. D. Eley and H. Pines, Academic Press, 54 W. Shan, M. Luo, P. Ying, W. Shen and C. Li, Appl. Catal., A,
New York, 1986, pp. 1–79. 2003, 246, 1–9.
26 S. Internazionale and S. Avanzati, J. Phys.: Condens. Matter, 55 N. Yisup, Y. Cao, W. Feng, W. Dai and K. Fan, Catal. Lett.,
2009, 21, 1–19. 2005, 99, 207–213.
27 M. Torrent, N. A. W. Holzwarth, F. Jollet, D. Harris, N. Lepley 56 H. Zhou, J. Wang, M. Mai, X. Ma, S. Hu, M. Xu, L. Bai and S.
and X. Xu, Comput. Phys. Commun., 2010, 181, 1862–1867. Yan, Thin Solid Films, 2020, 709, 138074.
28 J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 57 A. J. Logsdail, D. O. Scanlon, C. R. A. Catlow and A. A. Sokol,
1996, 77, 3865–3868. Phys. Rev. B: Condens. Matter Mater. Phys., 2014, 90, 1–8.
29 A. Allouche, J. Comput. Chem., 2012, 32, 174–182. 58 J. C. A. A. Roelofs, J. A. van Bokhoven, A. J. van Dillen,
30 J. D. Pack and H. J. Monkhorst, Phys. Rev. B: Solid State, J. W. Geus and K. P. de Jong, Chem. – Eur. J., 2002, 8,
1977, 16, 1748–1749. 5571–5579.
31 X. Xia, R. J. Oldman and C. R. A. Catlow, J. Mater. Chem., 59 M. Xu, E. Iglesia, C. R. Apestegu, D. I. Cosimo and E. T. Al,
2011, 21, 14549–14558. J. Catal., 1998, 510, 499–510.
32 D. A. Daramola, M. Muthuvel and G. G. Botte, J. Phys. Chem. 60 R. Da Silva Alvim, I. Borges and A. A. Leitao, J. Phys. Chem.
B, 2010, 114, 9323–9329. C, 2018, 122, 21841–21853.
33 J. Li, S. Meng, J. Niu and H. Lu, J. Adv. Ceram., 2017, 6, 61 J. A. Wang, A. Morales, X. Bokhimi, O. Novaro, T. López and
43–49. R. Gómez, Chem. Mater., 1999, 11, 308–313.
34 K. Pokrovski, K. T. Jung and A. T. Bell, Langmuir, 2001, 17, 62 M. Romero-sáez, A. B. Dongil, N. Benito, R. Espinoza-
4297–4303. gonzález, N. Escalona and F. Gracia, Appl. Catal., B,
35 I. Beltrán, S. Gallego, J. Cerdá, S. Moya and C. Muñoz, Phys. 2018, 237, 817–825.
Rev. B: Condens. Matter Mater. Phys., 2003, 68, 1–5. 63 C. D. Wagner, W. M. Riggs, L. E. Davis, J. F. Moulder and
36 A. Cadi-Essadek, A. Roldan and N. H. de Leeuw, Surf. Sci., G. E. Muilenberg, Handbook of X-ray photoelectron
2016, 653, 153–162. spectroscopy, Minn, Phys. Electron. Div., Perkin-Elmer Corp.,
37 R. Bader, Bader Charge Analysis, http://theory.cm.utexas.edu/ Eden Prairie, 1979, pp. 1–190.
henkelman/code/bader/. 64 H. Luo, C. Zeng, D. Tian, H. Wang and Y. Fu, Advances in
38 G. Henkelman, A. Arnaldsson and H. Jónsson, Comput. Engineering Research, 2015, 101–105.
Mater. Sci., 2006, 36, 354–360. 65 Y. Hinuma, T. Toyao, T. Kamachi, Z. Maeno, S. Takakusagi,
39 A. Kokalj, J. Mol. Graphics Modell., 1999, 17, 176–179. S. Furukawa, I. Takigawa and K. I. Shimizu, J. Phys. Chem. C,
40 P. Frontera, A. Macario, M. Ferraro and P. Antonucci, 2018, 122, 29435–29444.
Catalysts, 2017, 7, 1–28. 66 X. Han, J. Yang, B. Han, W. Sun, C. Zhao, Y. Lu, Z. Li and J.
41 W. Li, X. Nie, X. Jiang, A. Zhang, F. Ding, M. Liu and Z. Liu, Ren, Int. J. Hydrogen Energy, 2017, 42, 177–192.
Appl. Catal., B, 2018, 220, 397–408. 67 H. Metiu, S. Chrétien, Z. Hu, B. Li and X. Sun, J. Phys. Chem.
42 W. Gac, W. Zawadzki, M. Rotko, M. Greluk, G. Słowik and G. C, 2012, 116, 10439–10450.
Kolb, Catal. Today, 2019, 1–15. 68 J. Gao, Q. Liu, F. Gu, B. Liu, Z. Zhong and F. Su, RSC Adv.,
43 G. Garbarino, P. Riani, L. Magistri, G. Busca, I. Civile and A. 2015, 5, 22759–22776.
Dicca, Int. J. Hydrogen Energy, 2014, 39, 11557–11565. 69 A. Karelovic and P. Ruiz, J. Catal., 2013, 301, 141–153.
44 P. Panagiotopoulou and X. E. Verykios, J. Phys. Chem., 70 Y. Schuurman, C. Marquez-Alvarez, V. C. H. Kroll and C.
2017, 121, 5058–5068. Mirodatos, Catal. Today, 1998, 46, 185–192.

Catal. Sci. Technol. This journal is © The Royal Society of Chemistry 2022
View Article Online

Catalysis Science & Technology Paper

71 P. Ferreira-Aparicio, C. Márquez-Alvarez, I. Rodríguez-Ramos, 73 M. S. Munde, D. Z. Gao and A. L. Shluger, J. Phys.: Condens.


Y. Schuurman, A. Guerrero-Ruiz and C. Mirodatos, J. Catal., Matter, 2017, 29, 1–10.
1999, 184, 202–212. 74 V. Prostakova, J. Chen, E. Jak and S. A. Decterov, CALPHAD:
72 L. R. Silva-Calpa, P. C. Zonetti, C. P. Rodrigues, O. C. Alves, Comput. Coupling Phase Diagrams Thermochem., 2012, 37, 1–10.
Published on 03 January 2022. Downloaded by Pontifícia Universidade Catolica do Rio de Janeiro on 1/21/2022 5:12:09 PM.

L. G. Appel and R. R. De Avillez, J. Mol. Catal. A: Chem., 75 E. W. Mcfarland and H. Metiu, Chem. Rev., 2013, 113,
2016, 425, 166–173. 4391–4427.

This journal is © The Royal Society of Chemistry 2022 Catal. Sci. Technol.

You might also like