Complete photocatalytic reduction of CO2 to methane by H 2 under solar light irradiation

Article abstract of DOI:10.1021/ja500924t

Nickel supported on silica-alumina is an efficient and reusable photocatalyst for the reduction of CO₂ to methane by H₂, reaching selectivity above 95% at CO₂ conversion over 90%. Although NiO behaves similarly, it undergo

Full text of DOI:10.1021/ja500924t

Communication
pubs.acs.org/JACS
Complete Photocatalytic Reduction of CO2 to Methane by H2 under
Solar Light Irradiation
Francesc Sastre, Alberto V. Puga, Lichen Liu, Avelino Corma,* and Hermenegildo García*
̀ ̀
Instituto Universitario de Tecnología Química CSIC-UPV, Universitat Politecnica de Valencia, Av. De los Naranjos s/n, 46022
Valencia, Spain
S
* Supporting Information
16 nm (Figures S2 and S3); furthermore, both Ni(0) and NiO
were found, based on lattice fringes. The role of silica−alumina
in this type of heterogeneous catalyst is to increase metal
dispersion and stability of the active NPs.
ABSTRACT: Nickel supported on silica−alumina is an
efficient and reusable photocatalyst for the reduction of
CO2 to methane by H2, reaching selectivity above 95% at
CO2 conversion over 90%. Although NiO behaves
similarly, it undergoes a gradual deactivation upon reuse.
About 26% of the photocatalytic activity of Ni/silica−
alumina under solar light derives from the visible light
photoresponse.
Under the experimental conditions, Ni/SiO2·Al2O3 gave the
highest photocatalytic activity without an apparent relationship
with the bandgap of the photocatalyst (see values in Table 1).
The products in the gas phase and their distribution depend on
CO2 conversion. It was observed that CO appears as a primary
but unstable product that converts in a subsequent step into
CH4 as a secondary and stable product. Trace amounts of
ethane were observed in some cases (see Table 1). It was also
noted that the mass balances of carbon, considering exclusively
the products in the gas phase, are lower than 100%, especially at
higher CO2 conversions.
During the photocatalytic CO2 reduction by H2, the
temperature measured at the photocatalyst bed was always
lower than 150 °C. We notice that the initial ambient
temperature increases rapidly when using Ni/SiO2·Al2O3 as
photocatalyst as CO2 is consumed, reaching a maximum at 140
°C in 15 min when CO2 disappears completely. This should
reflect the exothermicity in eq 1. Control experiments in an
autoclave showed that, in the dark, using Ni/SiO2·Al2O3 as
catalyst, CO2 reduction by H2 starts to occur at 180 °C, and at
this temperature a methane yield of 0.15% at 2.5 h was
observed (Figure S4). No temperature increase was recorded in
the autoclave at 180 °C, and CO2 conversion was negligible.
Notice that the results in Table 1 for Ni/SiO2·Al2O3
correspond to 15 min. Precedents in the literature report the
catalytic thermal reduction of CO2 on Ni−Pt and Ni−Pd alloys
performing the reaction at 300 °C, achieving high selectivity
toward CO with minor CH4 formation.14 Thus, all these data
support that CH4 formation is a photoactivated and not a
thermal process, although it could be accompanied by a thermal
catalytic contribution. Additionally, experiments using labeled
13CO2 as substrate lead to the formation of 13CH4, showing that
CH4 is the product of CO2 reduction (see Figure S5 for the
relative intensity of the peaks in the m/z range from 10 to 20
amu of the mass spectrum).
ow-temperature selective reduction of CO2 to methane
using solar light is a highly important process that could
L
alleviate the current dependence on fossils fuels and could be
neutral from the point of view of climate change and
atmospheric CO2 balance.1−10 Recently we have reported
that CO can be either oxidized to CO2 or reduced to CH4 using
n-type or p-type metal oxide semiconductors, respectively.11,12
In particular, and more related to the present work, p-type
semiconductors such as NiO, FeOx, CuO, and other transition
metal oxides are able to reduce CO to CH4 by water or, more
efficiently, by H2 using solar light.12 Working with 3.5 molar
excess of H2, selectivity to CH4 as high as 97% was achieved, at
essentially complete CO conversion, using bulk NiO as
photocatalyst. Considering the high importance of CH4 as a
potential solar fuel and that CO2 is a good feedstock from the
environmental point of view and in terms of availability, we
have conducted a study on the photocatalytic reduction of CO2
to CH4 by H2 with p-type semiconductors using visible light,
according to eq 1.
CO2 + 4H2 → CH4 + 2H2O ΔHr = −165 kJ/mol
(1)
In a preliminary catalyst study, a series of transition metal
oxides were tested for the reduction of CO2 by H2 with
simulated sunlight as the source of energy. The results achieved
after 1 h reaction time with a conventional solar simulator are
presented in Table 1. As can be seen there, while very low
conversions were observed for micrometric Fe3O4 or CuO,
other photocatalysts and particularly NiO gave much higher
conversions. The series of samples also included Ni supported
on silica−alumina (Ni/SiO2·Al2O3). According to X-ray
diffraction (XRD; Figure S1, Supporting Information), this
sample contains a mixture of Ni(0) and NiO nanoparticles
(NPs). The size distribution of these NPs based on
transmission electron microscopy (TEM) is between 7 and
Previously we found that, in the photocatalytic reduction of
CO, some elemental C can deposit on the photocatalyst
surface. In the present case, combustion chemical analyses of
the photocatalysts before and after their use reveal that, while
the fresh material does not contain a detectable percentage of C
Received: January 27, 2014
Published: April 11, 2014
© 2014 American Chemical Society
6798
dx.doi.org/10.1021/ja500924t | J. Am. Chem. Soc. 2014, 136, 6798−6801

Journal of the American Chemical Society
Communication
Table 1. Photocatalytic Activity for CO2 Conversion, Including Molar Balances and Selectivities of CH4, CO, and C2H6, in the
a
Presence of Metal Oxide Semiconductors as Photocatalysts
c
selectivity (%)
b
c
d
catalyst
particle size (nm)
conversion (%)
molar balance (%)
C (%)
CH4
CO
C2H6
TiO2 (3.2 eV)
Ni/SiO2·Al2O3
NiO (3.5 eV)
20−30
7−16
7−30
12−50
120−200
<50
0
99.0
89.1
91.9
89.4
98.2
99.9
99.9
100
0.1
1.32
1.6
−
−
2.8
−
−
−
−
−
−
0.7
2.8
−
e
94.9
89.8
15.0
1.9
97.2
100
88.1
100
4.3
f
NiO
1.1
11.9
f
NiO
0.47
0.5
−
Fe2O3 (2.2 eV)
Fe3O4
51.0
12.3
0
95.0
96.7
−
<50
0.3
0.5
CuO (1.7 eV)
CoO (2.6 eV)
<50
0.04
0.11
−
38.7
<50
27.4
99.1
60.5
0.8
a
Irradiation conditions: mixture of CO2 (3.7 mmol, 15%), H2 (17 mmol, 70%), and N2 (3.7 mmol, 15%); irradiation time, 1 h; photocatalyst mass,
250 mg; irradiation source, solar simulator. Blank controls illuminated in the absence of any solid or contacting the solid and gas phase in the dark
b
did not lead to any product. See Supporting Information for the origin and characterization data of these commercially available photocatalysts. The
c
values in parentheses correspond to the reported bandgap energy of the semiconductor in ref 13. Calculated by considering exclusively the products
d
in the gas phase. Weight percentage of carbon on the solid after the photocatalytic experiment determined by combustion chemical analyses.
e
f
Irradiation time, 0.25 h. Newly synthesized materials; see preparation procedures in the Experimental Details.
by combustion chemical analysis, after photocatalytic reaction a
few percent of C is present in the used catalyst (see Table 1).
Considering the amounts of deposited carbon, mass balances
become satisfactory (>96.6%). In analogous experiments,
whereby CO2 and water were submitted to irradiation under
otherwise identical conditions, only trace amounts of CO and
CH4 were detected after very long irradiation times, indicating
that essentially no photocatalytic reforming of carbon takes
place under the present conditions.
The benefits of using Ni/SiO2·Al2O3, combining the
photocatalytic activity of NiO with a large surface area support
that stabilizes the semiconductor NPs, are revealed by the fact
that Ni/SiO2·Al2O3 exhibits a remarkable reusability that was
not achieved with NiO, which undergoes a gradual deactivation
upon reuse. Studies on the recyclability of the photocatalysts
are presented in Table 2. It can been seen there that Ni/SiO2·
Al2O3 was stable for six uses. Reuses were done by removing
the reaction gases and adding fresh CO2 and H2 into the
photoreactor at the end of each irradiation time (0.25 h). In
contrast, in the case of NiO, a significant decay in the activity
was observed. Notice that, in this latter case, regeneration of the
catalyst by calcination under air at 450 °C results in a gradual
decrease of activity in the first three consecutive cycles,
followed by an abrupt deactivation during the fourth use. This
is due to the sintering of NiO NPs during the reaction/
regeneration cycle, as shown by the decrease in surface area. To
support this hypothesis, larger particle sizes of NiO were
synthesized and tested (12−50 nm NPs or 120−200 nm
hexagonal nanoplates; see the Experimental Details for
synthesis and TEM micrographs in Figures S6 and S7,
respectively). They led to noticeably lower conversions (15.0
and 1.9% for NiO of 12−50 and 120−200 nm, respectively, see
Table 1), thus confirming the influence of particle size on
activity. As commented earlier, low CO2 conversion is
characterized by product distributions containing a large
percentage of CO to the detriment of CH4.
Table 2. Photocatalytic Activity for CO2 Conversion,
The above experiments were carried out with simulated
sunlight that is known to contain a small percentage of UV light
(about 4%). It has been frequently observed that, even though
the percentage of UV light in the solar light is low, it can
contribute largely or even completely (in the case of TiO2) to
the observed overall photocatalytic activity.15 In order to
increase the photocatalytic efficiency, there is an ongoing
interest in searching for photocatalysts that are active with
visible light. Since NiO has been reported previously to exhibit
some visible light photoresponse,12,16−18 and Ni-based NPs
have been shown to display activity for electrocatalytic CO2
reduction,18,19 it is of interest to assess if the photocatalytic
CO2 reduction by H2 can take place to some extent exclusively
with visible irradiation on NiO. To investigate this point, we
performed a set of irradiations using UV-free light (λ > 450 nm;
see transmittance spectrum in Figure S8). It was noticed that
NiO exhibits some residual photocatalytic activity with visible
light irradiation, requiring, however, much longer reaction
times. In contrast, as we can see in Table 3, Ni/SiO2·Al2O3
exhibited a much better visible light photoresponse than NiO,
although longer times were needed as compared to UV-
containing irradiations. In any case, conversions above 90%
with very high selectivity toward CH4 could be obtained using
Including Molar Balances and Selectivities of CH4, CO, and
C2H6, for Reuses of the Catalyst
a
d
selectivity
conversion molar balance
d
catalyst
use
(%)
(%)
CH4
CO C2H6
b
NiO
1
2
3
4
89.8
85.1
73.9
14.4
91.9
89.5
75.3
92.3
100
−
−
−
−
0.8
99.2
99.7
46.8
0.3
1.0
52.2
Ni/SiO2·
c
1
2
3
4
5
6
94.9
96.3
96.8
93.9
93.0
94.2
89.1
78.2
87.0
86.7
86.4
99.9
97.2
99.9
99.2
97.3
97.1
98.4
2.8
0.1
0.8
2.7
2.9
1.6
−
Al2O3
−
−
−
−
−
a
Irradiation conditions: mixture of CO2 (3.7 mmol, 15%), H2 (17
mmol, 70%), and N2 (3.7 mmol, 15%); irradiation time, 1 h;
photocatalyst mass, 250 mg; irradiation source, solar simulator.
b
c
Activation of NiO at 450 °C with air for 5 h. Irradiation time,
d
0.25 h, adding fresh feed after each run. Calculated by considering
exclusively the products in the gas phase.
6799
dx.doi.org/10.1021/ja500924t | J. Am. Chem. Soc. 2014, 136, 6798−6801

Journal of the American Chemical Society
Communication
Table 3. Photocatalytic Activity for CO2 Conversion,
Including Molar Balances and Selectivities of CH4, CO, and
C2H6, under Irradiation with a Solar Simulator Equipped
a
with a Cutoff Filter (λ > 450 nm)
selectivity (%)
molar
irradiation conversion
balance
(%)
catalyst
time (h)
(%)
CH4
98.6
CO C2H6
Ni/SiO2·
Al2O3
1
90.5
89.9
1.3
−
NiO
3
7.7
99.5
55.6
44.3
1.8
a
Irradiation conditions: mixture of CO2 (3.7 mmol, 15%), H2 (17
mmol, 70%), and N2 (3.7 mmol, 15%); photocatalyst mass, 250 mg;
irradiation source, solar simulator.
Ni/SiO2·Al2O3 exclusively with visible light. Comparison of the
conditions of Table 3 with those of Tables 1 and 2 indicate that,
although Ni/SiO2·Al2O3 has visible light photocatalytic activity,
about 74% of the response under simulated sunlight irradiation
derives from the small UV portion of the spectrum, as
estimated by comparing the conversion at 0.25 h with 450 nm
cutoff filter (25%) with that obtained using the full wavelength
range provided by the solar simulator (94.9%).
The photoaction spectrum of Ni/SiO2·Al2O3 and the
apparent quantum yields of the process were determined in
the range 350−500 nm using monochromatic light and
monitoring the disappearance of CO2 (see Figure S9). The
intensity of the monochromatic light was much lower than that
under simulated solar light irradiations, and for this reason, the
selectivity of the reaction changed and ethylene became the
main product. The results show that the photoresponse of the
process follows the absorption spectrum of Ni/SiO2·Al2O3 (see
also Figure S10), with increasing efficiency going to the UV
region, reaching a maximum value of 6%.
With regard to the photocatalytic mechanism, preliminary
transient absorption experiments under 355 or 532 nm
excitation in acetonitrile suspensions allowed the detection of
transients decaying in a few microseconds for Ni/SiO2·Al2O3
(see Figure S11). These transients could be assigned to the
charge-separated state, based on quenching by O2 and CH3OH,
showing the photoresponse of the material (see Figure S12).
We also observed that the temporal profile of the signal was
influenced by H2 and, particularly, CO2 purging (see Figure
S13), indicating that the monitored transient is sensitive to the
presence of these gases.
Considering also the high reduction potential needed for
direct electron transfer to CO2 and previous reports on the
photocatalytic activity of NiO for water splitting,20 it seems
more reasonable to assume that the process occurs through H2
activation with the formation of Ni−H. Attempts to detect a
Ni−H phase by XRD have met with failure, probably due to the
high reactivity of NiH with the ambient. Ni−H will be the
active CO2 reducing agent, rather than conduction band
electrons. Our proposal is summarized in eqs 2 and 3. Ni could
activate hydrogen either by reducing H2 to H− and then
reacting with Ni+ or by oxidizing H2 to H+ and then reacting
with Ni−. This proposal, based on the formation of metal
hydrides, is compatible with the observation of CH4 formation
in cases, such as Fe2O3, where the reduction potential of
conduction band electrons is known to be too low to reduce
water.
namely the photocatalytic reduction of CO2 to CH4 with
sunlight. High conversion and selectivity were achieved in short
times using H2 in 4.6 molar excess. An important observation is
that the use of a high-surface-area insulating support such as
silica−alumina increases the efficiency and durability intrinsic to
the photocatalyst, due to an increase in metal dispersion and
stability. The reported process can also take place under UV-
free visible light.
Experimental Details. Photocatalysts were commercial
samples, with the exception of NiO samples of 12−50 or 120−
200 nm particles. The main textural and analytical data for the
commercial photocatalysts are provided in Table S1.
NiO NPs (diameter 12−50 nm) were prepared by a sol−gel
method. In a typical synthesis, nickel acetate tetrahydrate (0.5
g) was dissolved in ethanol under constant stirring for 2 h at
40−50 °C, yielding a light greenish sol (50 mL, 0.5 mol/L). An
oxalic acid solution (100 mL, 0.23 M in ethanol) was then
added slowly to the warm sol to yield a thick greenish gel. The
gel product was subsequently dried/digested at 80−110 °C for
24 h and subjected to calcination at 500 °C for 3 h.
NiO nanoplates (diameter 120−200 nm, thickness 20−50
nm) were prepared by a hydrothermal method. Nickel acetate
tetrahydrate (0.5 g) was dissolved in water (50 mL). An
aqueous NaOH solution (5 mL, 1.25 mol/L) was added
dropwise to the nickel acetate solution. After the solution was
stirred for 0.5 h, the precipitate thus formed was separated by
centrifugation and decantation. The resulting green solid was
transferred to an autoclave for hydrothermal treatment at 175
°C for 16 h.
Irradiations in the presence of photocatalysts were carried
out spreading uniformly a thin bed of powdered semiconductor
previously pelletized into the photoreactor, covering a surface
of 1 × 1.5 cm2. The reactor body was made of aluminum to
favor heat transfer to the environment, and no special cooling
system was used. The temperature and pressure inside the
photoreactor were measured by an internal thermocouple and a
manometer, respectively. Figure S14 presents two photographs
of the photoreactor used and the placement of the photo-
catalyst and thermocouple. The photoreactor was located 5 cm
below the solar simulator beam, and the maximum temperature
never increased above 150 °C. CO2 gas diluted with N2 and H2
was introduced into the reactor. Simulated sunlight irradiations
of the solids acting as photocatalysts were carried out using a
solar simulator (Newport, Oriel Instruments, model 69921)
coupled with an AM1.5 filter that provides simulated
concentrated sunlight.
In conclusion, in the present Communication we report a
reaction that could potentially impact solar fuel production,
6800
dx.doi.org/10.1021/ja500924t | J. Am. Chem. Soc. 2014, 136, 6798−6801

Journal of the American Chemical Society
Communication
The course of the reaction was followed by periodically
analyzing the gas phase. At the final time, the possibility of the
presence of elemental carbon or organic compounds adsorbed
onto the solid was also considered, and the solids were analyzed
by combustion chemical analysis (PerkinElmer CHNOS
analyzer) or submitted to solid−liquid extraction using
dichloromethane as solvent. No products were detected in
the extract.
ACKNOWLEDGMENTS
■
Financial support by the Spanish Ministry of Economy and
Competitiveness (Severo Ochoa and CTQ2012-32315) and by
the Generalidad Valenciana (Prometeo 2012/013) is gratefully
acknowledged. A.V.P. is grateful to both the Consejo Superior
́
de Investigaciones Cientificas (CSIC) and the European Social
Fund (ESF) for a JAE-Doc postdoctoral grant. Mr. Herme G.
́
Baldovi is thanked for recording the transient spectra and
Reproducibility of the data was checked by performing
independent experiments in quadruplicate, whereby consistent
results were obtained with a dispersion of the conversion and
selectivity values less than 5%.
quenching experiments of Ni/SiO2·Al2O3.
REFERENCES
■
(1) Bensaid, S.; Centi, G.; Garrone, E.; Perathoner, S.; Saracco, G.
ChemSusChem 2012, 5, 500−521.
(2) Centi, G.; Quadrelli, E. A.; Perathoner, S. Energy Environ. Sci.
2013, 6, 1711−1731.
The gas products were analyzed using a Rapid Refinery Gas
Analyzer from Bruker that consists of a three-channel gas
chromatograph. The first channel analyzes H2 using a
micropacket HayeSep Q and Molsieve 5Å column with Ar as
carrier gas and TC detector. The second channel analyzes CO,
CO2, N2, and O2 with a combination of micropacket Haysep Q,
H−N, and Molsieve 13× columns using He as carrier gas and a
thermal conductivity detector. The third channel analyzes
hydrocarbons from C1 to C5 in an Al2O3 column with He as
carrier gas and a flame ionization detector. The mass balance of
each experiment was determined by summing all the moles of
the products in the gas phase and dividing by the moles of CO2
converted. Quantification of the moles of the products in the
gas phase was carried out considering that N2 remains constant
during the experiment and using this gas as internal standard.
The response factor of the products with respect to N2 was
determined by independent calibrations. Conversion values
refer to CO2 and were calculated by dividing the difference
between the initial and final moles of CO2 by the initial CO2
moles and then multiplying by 100. Selectivity toward CH4 was
calculated by dividing the moles of CH4 formed by the sum of
moles of carbon of all the carbon-containing products and then
multiplying by 100. Deposition of elemental carbon or the
presence of nonvolatile compounds on the solid photocatalyst
after irradiation was determined by combustion elemental
analysis. The maximum carbon content on the photocatalyst
was lower than 2 wt%.
(3) Dey, G. R. J. Nat. Gas Chem. 2007, 16, 217−226.
(4) Habisreutinger, S. N.; Schmidt-Mende, L.; Stolarczyk, J. K.
Angew. Chem., Int. Ed. 2013, 52, 7372−7408.
(5) Indrakanti, V. P.; Kubicki, J. D.; Schobert, H. H. Energy Environ.
Sci. 2009, 2, 745−758.
(6) Inoue, T.; Fujishima, A.; Konishi, S.; Honda, K. Nature 1979,
277, 637−638.
(7) Izumi, Y. Coord. Chem. Rev. 2013, 257, 171−186.
́
́
(8) Navalon, S.; Dhakshinamoorthy, A.; Alvaro, M.; Garcia, H.
ChemSusChem 2013, 6, 562−577.
(9) Roy, S. C.; Varghese, O. K.; Paulose, M.; Grimes, C. A. ACS Nano
2010, 4, 1259−1278.
(10) Wang, C.; Xie, Z.; deKrafft, K. E.; Lin, W. J. Am. Chem. Soc.
2011, 133, 13445−13454.
(11) Sastre, F.; Oteri, M.; Corma, A.; Garcia, H. Energy Environ. Sci.
2013, 6, 2211−2215.
(12) Sastre, F.; Corma, A.; García, H. Angew. Chem., Int. Ed. Engl.
2013, 52, 12983−12987.
(13) Xu, Y.; Schoonen, M. A. A. Am. Mineral. 2000, 85, 543−556.
(14) Porosoff, M. D.; Chen, J. G. J. Catal. 2013, 301, 30−37.
́
(15) Gomes Silva, C.; Juarez, R.; Marino, T.; Molinari, R.; García, H.
J. Am. Chem. Soc. 2010, 133, 595−602.
(16) Jin, Q.; Ikeda, T.; Fujishima, M.; Tada, H. Chem. Commun.
2011, 47, 8814−8816.
(17) Ni, L.; Tanabe, M.; Irie, H. Chem. Commun. 2013, 49, 10094−
10096.
(18) Shrestha, N. K.; Yang, M.; Nah, Y.-C.; Paramasivam, I.;
Schmuki, P. Electrochem. Commun. 2010, 12, 254−257.
(19) Perez-Cadenas, A. F.; Ros, C. H.; Morales-Torres, S.; Perez-
Cadenas, M.; Kooyman, P. J.; Moreno-Castilla, C.; Kapteijn, F. Carbon
2013, 56, 324−331.
(20) Tong, L.; Iwase, A.; Nattestad, A.; Bach, U.; Weidelener, M.;
Gotz, G.; Mishra, A.; Bauerle, P.; Amal, R.; Wallace, G. G.; Mozer, A. J.
Energy Environ. Sci. 2012, 5, 9472−9475.
ASSOCIATED CONTENT
■
S
* Supporting Information
Textural data for the commercial oxides; XRD and TEM of
commercial Ni/SiO2·Al2O3; catalytic thermal reaction at
various temperatures; mass spectra of the gaseous product
from the reaction of 13CO2 and H2 on Ni/SiO2·Al2O3 as
compared to that of CH4; TEM of the newly synthesized NiO
NPs or nanoplates; photoaction spectrum; absorption spectra
of NiO and Ni/SiO2·Al2O3; transmission spectra for the cutoff
filter used to obtain UV-free visible light; transient spectra
recorded for Ni/SiO2·Al2O3; and photographs of the solar
simulator and the reactor used. This material is available free of
charge via the Internet at http://pubs.acs.org.
NOTE ADDED AFTER ASAP PUBLICATION
■
The Supporting Information file was replaced in order to
include figures that were inadvertently omitted during the
revision process. The correct file reposted April 29, 2014
AUTHOR INFORMATION
■
Corresponding Authors
acorma@itq.upv.es
hgarcia@qim.upv.es
Notes
The authors declare no competing financial interest.
6801
dx.doi.org/10.1021/ja500924t | J. Am. Chem. Soc. 2014, 136, 6798−6801