Calcium titanate photocatalyst prepared by a flux method for reduction of carbon dioxide with water

Article abstract of DOI:10.1016/j.cattod.2014.10.039

Several calcium titanate samples were prepared by a flux method with various flux salts and various substrate concentration and also by a solid state reaction method. The prepared calcium titanate of various morphologies were loaded with Ag cocatalyst, an

Full text of DOI:10.1016/j.cattod.2014.10.039

G Model
CATTOD-9329; No. of Pages8
ARTICLE IN PRESS
Catalysis Today xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Catalysis Today
journal homepage: www.elsevier.com/locate/cattod
Calcium titanate photocatalyst prepared by a flux method for
reduction of carbon dioxide with water
Hisao Yoshidaa,b,∗, Like Zhangc, Masumi Satoa, Takeshi Morikawad, Tsutomu Kajinod,
Takeshi Sekitoe, Shinichi Matsumotoe, Hirohito Hiratae
a Kyoto University, Graduate School of Human and Environmental Studies, Kyoto 606-8501, Japan
b Kyoto University, Elements Strategy Initiative for Catalysts and Batteries (ESICB), Kyoto 615-8520, Japan
c Nagoya University, Graduate School of Engineering, Nagoya 464-8603, Japan
d Toyota Central R&D Labs., Inc., Nagakute, Aichi 480-1192, Japan
e Toyota Motor Corporation, Toyota 471-8572, Japan
a r t i c l e i n f o
a b s t r a c t
Article history:
Several calcium titanate samples were prepared by a flux method with various flux salts and various
substrate concentration and also by a solid state reaction method. The prepared calcium titanate of
various morphologies were loaded with Ag cocatalyst, and examined for the photocatalytic reduction
of carbon dioxide with water. A sample prepared with a NaCl flux with a moderate concentration of
solute exhibited the highest and stable photocatalytic activity for carbon monoxide production among the
prepared samples. It is revealed that the large and flat facets could stabilize the Ag cocatalyst nanoparticles
on the surface to enhance the photocatalytic activity for the carbon monoxide production.
© 2014 Elsevier B.V. All rights reserved.
Received 11 August 2014
Received in revised form 17 October 2014
Accepted 20 October 2014
Available online xxx
Keywords:
Heterogeneous photocatalyst
CO2 reduction
Artificial photosynthesis
Flux method
1. Introduction
water to promote various reactions without consuming additional
energy or reagents by using photoenergy, i.e., solar energy.
Since we will probably continue to release much carbon dioxide
into the atmosphere, we must establish a new technology for the
reduction of carbon dioxide in consideration of the carbon cycle on
the earth. In order to reduce carbon dioxide chemically, we need
a suitable reductant and a suitable catalyst. Although methane or
other organic compounds are known as reducing agents, the use of
them would substantially involve the emission of carbon dioxide,
which seems ineffective. Although hydrogen is a better reductant,
at present hydrogen is usually produced from natural gas, which
method also produces carbon dioxide eventually [1,2]. The most
effective reductant would be water that only releases oxygen after
the reaction although the activation of water is usually not easy
without consuming energy. It is known that photocatalysis can
decompose water molecule to form hydrogen and oxygen at mild
condition [3–6], which suggests that the photocatalyst can activate
As one of the possible and attractive methods, heterogeneous
photocatalytic reduction of carbon dioxide with water has been
widely studied so far [7–16]. In the 1970s many kinds of photo-
catalytic systems employing SrTiO3 single crystal, Zn-doped GaP
photocathode and predominant semiconductor powders such as
TiO2, ZnO, CdS, GaP and SiC were examined to reduce carbon diox-
ide with water and some kinds of products were detected [7–9]. In
the most of these and subsequent studies, however, the formation
of molecular oxygen was scarcely mentioned or much less than the
expected quantity although oxygen must be formed as an oxidized
product from water [10].
In 1993, Sayama and Arakawa found that ZrO2 photocatalysts
produced hydrogen, oxygen and carbon monoxide from aqueous
solution of NaHCO3 and a Cu-loaded ZrO2 photocatalyst exhib-
ited higher selectivity for the carbon monoxide [11]. In 2011, Kudo
and co-workers reported that Ag-loaded BaLa4Ti4O15 photocata-
lysts produced these products from aqueous solution in a carbon
dioxide flow system [12]. Recently, Ag-loaded Ga2O3 photocata-
lysts have been also studied as similar reaction systems [13–16]. In
these photocatalytic systems, the following reactions would take
place:
∗
Corresponding author at: Course of Studies on Material Science, Department of
Interdisciplinary Environment, Graduate School of Human and Environmental Stud-
ies, Kyoto University, Yoshida-nihonmatsu-cho, Sakyo-ku, Kyoto 606-8501, Japan.
Tel.: +81 75 753 6594; fax: +81 75 753 2988.
CO2 + 2H+ + 2e− → CO + H2O
(1)
E-mail address: yoshida.hisao.2a@kyoto-u.ac.jp (H. Yoshida).
http://dx.doi.org/10.1016/j.cattod.2014.10.039
0920-5861/© 2014 Elsevier B.V. All rights reserved.
Please cite this article in press as: H. Yoshida, et al., Calcium titanate photocatalyst prepared by a flux method for reduction of carbon
dioxide with water, Catal. Today (2014), http://dx.doi.org/10.1016/j.cattod.2014.10.039

G Model
CATTOD-9329; No. of Pages8
ARTICLE IN PRESS
2
H. Yoshida et al. / Catalysis Today xxx (2014) xxx–xxx
2H+ + 2e− → H2
2H2O + 4h+ → O2 + 4H+
(2)
(3)
In these systems, two reactions, photocatalytic reduction of car-
bon dioxide to carbon monoxide (Eq. (4)) and photocatalytic water
splitting (Eq. (5)), competitively proceeded, where the reaction
selectivity varied with the photocatalyst:
CO2 → CO + 1/2O2
H2O → H2 + 1/2O2
(4)
(5)
Among the reported photocatalysts, the Ag/BaLa4Ti4O15 photo-
catalyst consisted of plate-like particles with anisotropic structure
[12], where the reduction to produce carbon monoxide occurred
on the Ag cocatalyst located on the edges of the plates while the
oxidation would progress on the BaLa4Ti4O15 surfaces, that is, it
is considered that the reaction fields for reduction and oxidation
would be separated to contribute to the high photocatalytic per-
formance.
Fig. 1. Photocatalytic reactor for reduction of carbon dioxide with water.
prepared by a solid state (SS) reaction method: the same starting
materials, CaCO3 and TiO2, were dried at 383 K, mechanically mixed
in a stoichiometric ratio by a wet ball-milling method (CaCO3 23 g,
TiO2 18.6 g, alumina balls of 1 g and acetone of 80 mL were used) for
24 h, dried in an oven at 343 K overnight, and heated in air atmo-
sphere at 1273 K or 1373 K for 10 h in an alumina crucible. These
samples are referred to as CTO(SS, calcination temperature), e.g.,
CTO(SS,1273 K). Ag cocatalyst of 0.1 or 0.5 wt% (y wt%) was loaded
on the surface of the CaTiO3 photocatalysts by a photodeposition
method, where methanol of 25 vol% was used as the reductant. The
samples are referred to as Ag(y)/CTO, e.g., Ag(0.5)/CTO(NaCl,40).
Powder X-ray diffraction (XRD) pattern was recorded at room
temperature on a Rigaku diffractometer MiniFlexII/AP using Ni-
filtered Cu K␣ radiation (30 kV, 15 mA). We used the diffraction
line at 2ꢀ = 33.1◦ to estimate the mean crystallites size of the
CaTiO3 samples with Scherrer equation, where a diffraction line
of Si powder (Kishida) at 28.5◦ was used as a reference for the
determination of the half maximum full-width. Diffuse reflectance
(DR) UV–Visible spectrum was recorded at room temperature on a
JASCO V-570 equipped with an integrating sphere covered with
BaSO4, where BaSO4 was used as the reference. The bandgap
was estimated from the spectra according to Tauc plot [32]. The
Brunauer–Emmett–Teller (BET) specific surface area was calculated
from the amount of N2 adsorption at 77 K, which was measured
by a Quantachrome Monosorb. Field emission scanning electron
microscopy (FE-SEM) images were recorded by a JEOL JSM-7500FA
or a Hitachi S-5200. The sizes of particles were measured for around
50 particles by a scale on the SEM images, and the average particle
size was determined for each sample.
In the present study, in order to discuss the relationship between
the morphologic structure of the photocatalyst and the photocat-
alytic activity for the reduction of carbon dioxide, we employed
calcium titanate (CaTiO3) as a semiconductor photocatalyst and
a flux method (molten salt method) for the preparation of the
photocatalyst. The CaTiO3 photocatalyst have both a conduction
band with available potential for reduction of carbon dioxide and
a valence bands with enough potential for oxidation of water to
oxygen [17]. As for the activation of water, it has already been
reported to be photocatalytically active for water splitting [18,19]
and photocatalytic steam reforming of methane [19,20]. In addi-
tion, the point of zero charge of the CaTiO3 in water is reported
as pHPZC = 3 [21], i.e., the surface is negatively charged in neu-
tral water, which may be advantageous for adsorption of carbon
dioxide also in an aqueous solution. The flux method is expected
to provide fine single crystals and thus this method has been
applied to synthesize various micro or nano-sized photocatalysts
in recent years, for example, many kinds of titanate photocatalysts
were synthesized such as Na2Ti6O13 [22], Na2Ti3O7 [23], K2Ti6O13
[24,25], SrTiO3 [26], La2Ti2O7 [27,28], PbTiO3 [29], CaZrTi2O7 [30]
and AgLi1/3Ti2/3O2 [31], although CaTiO3 has not been reported yet.
In the present study, we prepared several CaTiO3 samples by a
flux method and a conventional solid state reaction method, and
examined their photocatalytic performances for the carbon dioxide
reduction with water.
Photocatalytic reactions were carried out in a specially designed
reactor of gas–liquid–solid three phases (Fig. 1) under a flow of CO2
(Taiyo Nippon Sanso, 99.999%) gas upon photoirradiation from the
bottom (16 cm2) by a 300 W xenon lamp without passing any opti-
cal filters, which entirely emitted from UV to visible light, under
ambient temperature and pressure. In a quartz cell, 0.2 g of the
photocatalyst powder was dispersed in 10 mL of water saturated
with 11 mmol of NaHCO3 (Wako, >99.5%). The pH of the solu-
tion was 8.3. The mixture of the solution and the photocatalyst
powder was photoirradiated with magnetically stirring, where the
reaction occurred. The incident light intensity measured in the
2. Experimental
Most of the CaTiO3 samples were synthesized by a flux method
from CaCO3 (Kojundo 99.99%) and TiO2 (rutile, Kojundo 99.9%) as
solutes by using CaCl2, KCl or NaCl (Kishida 99.5%) as a flux in the
same manner as our previous work [25]. The molar ratio of CaCO3
to TiO2 was unity, and various concentrations of the solute (x mol%
as CaTiO3) in the molten salt mixture were examined, where x was
defined as: x [mol%] = (amount of CaTiO3 [mol])/(amount of CaTiO3
[mol] + amount of a flux [mol]) × 100. The mixed starting materials
were put into a platinum crucible, which was loosely covered by a
lid, heated at a rate of 200 K h−1 to 1373 K, held at this temperature
for 10 h, and then slowly cooled at a rate of 100 K h−1 to 773 K, fol-
lowed by being naturally cooled to room temperature in an electric
furnace. It is considered that the decarbonation of CaCO3would take
place to form CaO species during heating, the mixture of the starting
materials would be molten, and then both CaO and TiO2 clusters in
the molten salt would react with each other to form CaTiO3 crystal-
lites during the cooling step. The products were well washed with
hot water (353 K) four times to remove the flux. These samples
are referred to as CTO(flux, x), e.g., CTO(NaCl,5). Two samples were
range of 254 10 nm at the center of the cell was 22 mW cm−2
,
and the temperature of the reaction cell became 323 K during the
photoirradiation. At regular interval (typically every 1 h), the out-
let gas was collected in a sampling column and introduced to
an online gas chromatograph with a thermal conductivity detec-
tor to determine the amount of H2, O2 and CO. The production
rate of H2 and CO was precisely determined in this method. The
amount of O2 was determined by the subtraction of the amount
of air leak into the system, which sometimes contained large error
unfortunately. Other products were not observed in this method.
Selectivity for CO in the reduced products, SCO was defined as
Please cite this article in press as: H. Yoshida, et al., Calcium titanate photocatalyst prepared by a flux method for reduction of carbon
dioxide with water, Catal. Today (2014), http://dx.doi.org/10.1016/j.cattod.2014.10.039

G Model
CATTOD-9329; No. of Pages8
ARTICLE IN PRESS
H. Yoshida et al. / Catalysis Today xxx (2014) xxx–xxx
3
atomic ratio of the starting materials, i.e., CaCO3 and TiO2. In the
present case, the atomic ratio of Ca and Ti was unity, which thus
provided CaTiO3 crystallites. In other words, the CaTiO3 crystal-
lites would originate from TiO2 and CaCO3. It is considered that
CaO clusters dissolved in the molten salt, which could be formed
from CaCO3 by thermal decomposition, would predominantly react
with dissolved TiO2 clusters to form the CaTiO3 crystallites, while
Ca2+ cation of the CaCl2 flux could not react with TiO2 species in
the present condition, as previously discussed [25]. It is also noted
that, even the crystallization of CaTiO3 took place in other molten
salts, such as KCl and NaCl, no compounds containing K and Na
would be formed. Thus, it is clear that these alkali chloride fluxes
such as CaCl2, KCl and NaCl could not react with TiO2 and could not
exchange with the Ca cation in the crystallites of CaTiO3 in this con-
dition. The average crystallites size of the samples were estimated
from XRD profiles (Table 1), which clarified that the CTO(CaCl2,5)
sample obviously had larger crystallites than others.
The FE-SEM images revealed that the use of CaCl2, KCl and NaCl
as a flux gave various morphologies. In some cases, various facets
could be found on their surfaces. The CaCl2 flux gave clear and large
cubic-like crystals of 2–30 ␮m on a side (Fig. 3a), which consisted
of relatively larger crystals than others. By using the KCl flux, a mix-
ture of small roundish particles of 0.1–0.5 ␮m in diameter and long
rod-like crystals of 3–8 ␮m in length was obtained in this condition
(Fig. 3b). The NaCl flux provided polyhedral particles of 1–7 ␮m in
diameter (Fig. 3c), where some facets were found on the surfaces.
The CTO(SS,1273) sample showed tightly connected roundish and
polyhedral particles (Fig. 3d). The size of the particles observed in
these SEM images was in the range of 0.1–30 ␮m. On the other hand,
the average crystallites sizes of these samples determined from
XRD were ranged between 25 and 100 nm (Table 1). This discrep-
ancy indicates that these particles of unique morphology observed
in the SEM images would not be single crystals but polycrystals.
Although the color of the prepared CTO sample slightly varied
with the preparation condition, most of them seemed pale purple or
pale pink. Fig. 4 shows DR UV–Visible spectra of the representative
CTO samples. The adsorption edge is observed around 350 nm in
wavelength for each samples. The estimated bandgap of these sam-
ples was around 3.6 eV (Table 1). Among these samples, it would
be said that the sample consisting of larger crystallites tended to
exhibit smaller bandgap. The CTO(SS,1273) sample exhibited also a
shoulder band around 350–400 nm in wavelength (Fig. 4e), which
might originate from electron transition from the valence band to
shallow trap sites below the conduction band due to oxygen vacan-
cies, or from unreacted TiO2. Since no TiO2 phase was detected in
the XRD profiles, it should be amorphous TiO2 species if existed.
Anyway, judging from the color of the sample, there would be some
kinds of defects on the CTO(SS,1273) sample. The CTO(CaCl2,5)
sample exhibited also a similar band in the range of 350–420 nm in
wavelength as well as a broad absorption band in the visible light
region of 420–800 nm (Fig. 4a). The latter would originate from
some kinds of lattice defects or surface defects like as color cen-
ters. This means that the CTO(CaCl2,5) sample had some kinds of
defects although it consisted of the large cubic-like crystals shown
Fig. 2. XRD patterns of the CTO(CaCl2,5) sample (a), the CTO(KCl,5) sample (b), the
CTO(NaCl,5) sample (c), the CTO(NaCl,40) sample (d), the CTO(NaCl,90) sample (e)
and the CTO(SS,1273) sample (f), and a calculated one from a database ICSD #74212
(g).
follows: SCO (%) = 100 × (production rate of CO)/(sum of production
rates of CO and H2). The ratio of the electrons and holes con-
sumed, e−/h+ was calculated from the production rate as follows:
e−/h+ = (sum of production rates of CO and H2)/(production rate of
O2 × 2).
3. Results and discussion
3.1. Characterization
Fig. 2 shows XRD patterns of representative CaTiO3 samples
and a calculated one from a database (ICSD #74212). All the CTO
samples exhibited clear diffraction lines assignable to CaTiO3 crys-
tal and no other diffraction lines from impurity was observed in
these profiles, indicating that these prepared CTO samples predom-
inantly consisted of the CaTiO3 crystallites. It is noted that any kind
of flux such as CaCl2, KCl and NaCl provided CaTiO3 crystallites.
Even when excess amount of molten CaCl2 was used as a flux, no
other calcium-rich compounds such as Ca3Ti2O7 and Ca4Ti3O10 [33]
were formed. This result revealed that the composition of calcium
titanate would be determined by not the solute concentration (x
mol%) in the molten salt mixture and the kind of the flux but the
Table 1
Properties of the representative CaTiO3 samples.
Entry
Sample
Crystallite diametera (nm)
Bandgapb (eV)
BET specific surface areac (m2 g−1
)
1
2
3
4
5
CTO(CaCl2,5)
CTO(KCl,5)
CTO(NaCl,5)
CTO(NaCl,40)
CTO(SS,1273)
99.6
25.3
29.0
26.7
26.2
3.54
3.64
3.62
3.60
3.60
19.1
3.62
2.04
1.10
1.67
a
Estimated from XRD.
Estimated from DR UV–Vis spectra.
Estimated from N2 adsorption.
b
c
Please cite this article in press as: H. Yoshida, et al., Calcium titanate photocatalyst prepared by a flux method for reduction of carbon
dioxide with water, Catal. Today (2014), http://dx.doi.org/10.1016/j.cattod.2014.10.039

G Model
CATTOD-9329; No. of Pages8
ARTICLE IN PRESS
4
H. Yoshida et al. / Catalysis Today xxx (2014) xxx–xxx
Fig. 3. FE-SEM images of the CTO(CaCl2,5) sample (a), the CTO(KCl,5) sample (b), the CTO(NaCl,5) sample (c), and the CTO(SS,1273) sample (d). The images (a)–(c) were
recorded by a JEOL JSM-7500FA and the image (d) was recorded by a Hitachi S-5200.
in the SEM image (Fig. 3a). In Table 1, BET specific surface area of the
samples were listed, and the CTO(CaCl2,5) sample exhibited a large
value. These facts suggest that the large cubic-like crystals might
be microporous although it could not be recognized on the SEM
image. The CTO(NaCl,5) and CTO(NaCl,40) samples showed only a
very week and broad absorption band in this visible light region
(Fig. 4c and d), implying that these samples had less defects than
the CTO(CaCl2,5) and CTO(SS,1273) samples. The CTO(KCl,5) sam-
ple showed no absorption in the range of longer than 350 nm in
wavelength (Fig. 4b), suggesting that this sample had the lowest
amount of such kind of defects among these samples.
3.2. Photocatalytic performance
Photocatalytic tests for CO2 reduction with H2O as a reduc-
tant without using any sacrificial reagents were carried out in a
flow reactor, where the formation rates for each product were
determined. Fig. 5 shows the time course of the production rates
on the Ag(0.5)/CTO(NaCl,40) sample. The products were CO, H2,
and O2, where CO was the reduction product from CO2 and
H2 was produced by water splitting as a competitive reaction
[11–14]. The production rates of CO and H2 were initially high
but decreased steeply in the initial period for 5 h. After this unsta-
ble period, the production rate of CO and H2 became constant.
On the other hand, evolution of O2 was not observed at the ini-
tial stage, where oxygen might be adsorbed on the surface of
photocatalyst [34–38] or consumption for oxidizing the photo-
catalyst or Ag cocatalyst. After the initial period, the production
rate of O2 also became constant. In the present reaction system,
it is suggested that the above mentioned reactions shown in Eqs.
Fig. 4. DR UV–Visible spectra of representative CaTiO3 samples; the CTO(CaCl2,5)
sample (a), the CTO(KCl,5) sample (b), the CTO(NaCl,5) sample (c), the CTO(NaCl,40)
(1)–(3) would take place. Without Ag cocatalyst, or without the
photoirradiation, the reaction did not proceed. The reaction con-
tinuously lasted in the constant rate even after 24 h. Thus, it is
sample (d), and the CTO(SS,1273) sample (e).
Please cite this article in press as: H. Yoshida, et al., Calcium titanate photocatalyst prepared by a flux method for reduction of carbon
dioxide with water, Catal. Today (2014), http://dx.doi.org/10.1016/j.cattod.2014.10.039

G Model
CATTOD-9329; No. of Pages8
ARTICLE IN PRESS
H. Yoshida et al. / Catalysis Today xxx (2014) xxx–xxx
5
0.8
0.6
(b)
0.4
(a)
0.2
(c)
0
0
5
10
15
20
25
Time on stream / h
Fig. 5. Time course of the production rate of CO (a), H2 (b) and O2 (c) in the photo-
catalytic reduction of CO2 with water over the Ag(0.5)/CTO(NaCl,40) sample.
proposed that the reaction would photocatalytically proceed in the
constant rate, at least at the steady state after the initial period.
The Ag cocatalyst would receive the photoexcited electrons
from the conduction band of the CaTiO3 semiconductor and func-
tion as the reduction sites for production of CO and H2 shown in
Eqs. (1) and (2), respectively. Since other cocatalysts such as Pt,
Rh, and Pd did not promote the production of CO and Cu cocata-
lyst promoted CO formation at very low rate (0.016 ␮mol h−1), the
Ag cocatalyst is especially important for CO2 reduction to produce
CO on the present photocatalyst. This would be related to the less
adsorption ability of Ag metal for CO molecule.
Fig. 6. [A] Variation of CO production rate (a,b) and CO selectivity (c,d) in the pho-
tocatalytic CO2 reduction with H2O on the Ag(0.1)/CTO(NaCl, x) samples (a, c) and
those on the Ag(0.5)/CTO(NaCl, x) samples (b, d). [B] The average particle size of the
CTO(NaCl, x) samples estimated by the SEM images versus the solute concentration
during the preparation of the CTO(NaCl, x) samples by the flux method.
Table 2 shows the production rates after 5 and 20 h from the start
of the photocatalytic reaction. It is obvious that the photocatalytic
reduction of CO2 proceeded over all the Ag-loaded CaTiO3 samples.
The selectivity to CO, i.e., the reaction selectivity for CO2 reduction,
was also listed in Table 2, which much varied with the sample. The
consumed ratio of the electrons and holes (e−/h+) was also listed in
Table 2, which should be unity since photocatalysis must consume
equivalent amounts of photoexcited electrons and holes, and most
of the samples actually showed the e−/h+ ratio close to unity at
the steady state (20 h later). As for the samples of low activity, the
oxygen evolution rate was often not determined correctly.
The three samples prepared by the different flux with the same
solute concentration, the Ag(0.1)/CTO(CaCl2,5), Ag(0.1)/CTO(KCl,5)
and Ag(0.1)/CTO(NaCl,5) samples, exhibited stable production rates
at 20 h later, and the e−/h+ values became closed to unity (Table 2,
entries 1–3), suggesting that the photocatalytic reaction clearly
proceeded on each sample. However, the selectivity was differ-
ent from each other. The Ag(0.1)/CTO(CaCl2,5) sample of unique
cubic-like morphology exhibited high and selective photocatalytic
activity for not CO2 reduction but water splitting to produce H2 and
O2. This sample had larger size of crystallites and larger surface area
(Table 1, entry 1), which might contribute to the high and selec-
tive photocatalytic activity for water splitting although this sample
showed some absorption band probably due to defects (Fig. 4a).
On the other hand, the Ag(0.1)/CTO(NaCl,5) sample showed the
highest CO selectivity among the three samples (Table 1, entry 3).
This Ag(0.1)/CTO(NaCl,5) sample prepared by the flux method was
higher than the Ag(0.1)/CTO(SS,1273) and Ag(0.1)/CTO(SS,1373)
sample prepared by the solid state reaction method (Table 1, entries
10 and 11).
the Ag(0.5)/CTO(NaCl, x) samples presented each volcano plot upon
the solute concentration, where the Ag(0.1)/CTO(NaCl,50) sample
and the Ag(0.5)/CTO(NaCl,40) sample corresponded to the peaks
of the volcano plots for the Ag(0.1)/CTO(NaCl, x) samples and the
Ag(0.5)/CTO(NaCl, x) samples, respectively. The most active one
in the present study was the Ag(0.5)/CTO(NaCl,40) sample, which
exhibited the CO production rate of 0.35 ␮mol h−1 and the selectiv-
ity for the CO2 reduction was as high as 45% (Table 2 entry 7). It is
noted that the CO production rates over these samples prepared by
the flux method (Table 2, entries 4 and 7) were 10 times higher than
that over the samples prepared by the solid state reaction method
(Table 2, entries 11 and 12). The CO selectivity was also obviously
higher for these samples prepared by the flux method than those
by the solid state reaction method.
Fig. 7 shows some representative FE-SEM images of these CTO
samples before loading Ag cocatalyst. All the samples consisted of
the polyhedral particles covered with many facets. The particle size
observed in the images varied with the solute concentration and
it is clear that the CTO(NaCl,40) sample had the larger polyhedral
polycrystals than others. The average particle sizes of the polycrys-
tals estimated from the SEM images for the CTO(NaCl,x) samples
were plotted upon the solute concentration in Fig. 6B. It is interest-
ing that the similar volcano plot appeared as Fig. 6A. Thus, the CO
production rates were plotted versus the average size of the poly-
crystals as shown in Fig. 8. As clearly shown, the CO production
rate and the average size of the polycrystals showed a good corre-
lation. It is proposed that the larger polyhedral polycrystals loaded
with Ag cocatalyst would be advantageous for the higher photo-
catalytic activity for the CO2 reduction. As shown in Fig. 4d, the
CTO(NaCl,40) sample would have less defects than the CTO(NaCl,5)
sample. These results suggest that the large polyhedral polycrystals
with the flat facets and with fewer defects would be suitable for the
photocatalytic reduction of CO2 with water to produce CO.
Thus, the effect of the solute concentration was examined on the
Ag(y)/CTO(NaCl, x) samples, which results were listed in Table 2,
entries 3–5 for the Ag(0.1)/CTO(NaCl, x) samples and entries 6–10
for the Ag(0.5)/CTO(NaCl, x) samples. The CO production rate and
CO selectivity were plotted in Fig. 6A. Both the CO production rates
and the CO selectivity on both the Ag(0.1)/CTO(NaCl, x) samples and
In order to elucidate the reason why the CO production rate
decreased with time at the initial stage as shown in Fig. 5, the
Please cite this article in press as: H. Yoshida, et al., Calcium titanate photocatalyst prepared by a flux method for reduction of carbon
dioxide with water, Catal. Today (2014), http://dx.doi.org/10.1016/j.cattod.2014.10.039

G Model
CATTOD-9329; No. of Pages8
ARTICLE IN PRESS
6
H. Yoshida et al. / Catalysis Today xxx (2014) xxx–xxx
Table 2
Results of the reaction tests for the photocatalytic reduction of CO2 with H2O on the Ag/CTO samples.
Entry
Samplea
Production rate (5 h later) (␮mol h−1
)
Production rate (20 h later) (␮mol h−1
)
CO
H2
O2
SCO (%)b
e−/h+c
CO
H2
O2
SCO (%)b
e−/h+c
1
2
3
4
5
6
7
8
9
Ag(0.1)/CTO(CaCl2,5)
Ag(0.1)/CTO(KCl,5)
0.014
0.022
0.080
0.24
0.083
0.17
0.35
0.45
0.20
0.10
1.27
0.69
0.40
0.60
0.57
0.32
0.48
0.56
0.46
0.38
0.11
0.42
0.12
0.25
–
0.53
0.13
–
0.25
0.35
0.21
0.24
–
0.78
2.8
17
29
13
34
42
44
29
5.3
1.4
–
0.80
2.5
–
1.7
1.5
1.5
1.0
–
0.006
0.010
0.036
0.24
0.060
0.093
0.35
0.29
0.21
0.13
0.023
0.024
0.93
0.47
0.36
0.67
0.40
0.19
0.43
0.53
0.40
0.28
0.13
0.49
0.40
0.21
0.16
0.52
0.26
–
0.36
0.39
0.28
0.20
–
0.6
2.1
9.1
26
13
33
45
35
35
29
1.2
1.1
1.2
0.87
0.90
–
1.1
1.1
1.1
1.1
–
Ag(0.1)/CTO(NaCl,5)
Ag(0.1)/CTO(NaCl,50)
Ag(0.1)/CTO(NaCl,90)
Ag(0.5)/CTO(NaCl,20)
Ag(0.5)/CTO(NaCl,40)
Ag(0.5)/CTO(NaCl,50)
Ag(0.5)/CTO(NaCl,60)
Ag(0.5)/CTO(NaCl,70)
Ag(0.1)/CTO(SS,1273)
Ag(0.1)/CTO(SS,1373)
10
11
12
21
69
6.8
0.25
0.030
15
4.6
–
–
–
–
a
Loading amount of Ag cocatalyst was 0.1 or 0.5 wt%.
Selectivity to CO, see text.
The ratio of the consumed electrons to the consumed holes, see text.
b
c
Fig. 8. Variation of the CO production rate on the Ag(0.1)/CTO(NaCl, x) samples (a)
and the Ag(0.5)/CTO(NaCl, x) samples (b) versus the average particle size of the
CTO(NaCl, x) samples.
state of the Ag particles on the Ag(0.1)/CTO(NaCl,50) sample before
and after the reaction test were examined. This photocatalyst also
exhibited the constant activity for producing CO and H2 after initial
period for 5 h until 24 h later (not shown but similar to Fig. 5). In
the DR UV–Visible spectra (Fig. 9a), the sample before use exhib-
ited a broad band with the maximum at 470 nm in wavelength
assignable to the localized surface plasmon resonance (LSPR) of
the Ag nanoparticles. After the use in the photocatalytic reaction,
Fig. 7. FE-SEM images of the CTO(NaCl,20) sample (a), the CTO(NaCl,40) sample (b)
and the CTO(NaCl,70) sample (c), recorded by a Hitachi S-5200.
Fig. 9. DR UV–Visible spectra of the Ag(0.1)/CTO(NaCl,50) sample before (a) and
after (b) use in the photocatalytic CO2 reduction with H2O.
Please cite this article in press as: H. Yoshida, et al., Calcium titanate photocatalyst prepared by a flux method for reduction of carbon
dioxide with water, Catal. Today (2014), http://dx.doi.org/10.1016/j.cattod.2014.10.039

G Model
CATTOD-9329; No. of Pages8
ARTICLE IN PRESS
H. Yoshida et al. / Catalysis Today xxx (2014) xxx–xxx
7
Fig. 10. FE-SEM images of the Ag(0.1)/CTO(NaCl,50) sample before (a) and after (b) use and the Ag(0.1)/CTO(SS,1273) sample before (c) and after (d) use in the photocatalytic
CO2 reduction with H2O. Circles highlight the Ag nanoparticles.
the band maximum shifted to shorter wavelength around 420 nm
(Fig. 9b), suggesting the variation of the state of the Ag nanopar-
ticle, i.e., the particle size of the Ag nanoparticle would become
slightly smaller. Fig. 10 shows the FE-SEM images recorded in a
reflection mode, where the Ag particles could be observed as while
dots (highlighted by the circles). On the Ag(0.1)/CTO(NaCl,50) sam-
ple, various size of Ag nanoparticles were scattered before the
reaction (Fig. 10a), while the particles were located on some spe-
cific large and flat facets after the reaction test for 24 h (Fig. 10b).
Generally, it is considered that the photoformed holes and elec-
trons can oxidize and reduce both deposited nanoparticles and
adsorbed molecules and ions on the photocatalyst surface. In the
present study, since the Ag nanoparticles were deposited by the
photodeposition method under air atmosphere and Ag might be
photoreduced before adsorbed by the photocatalyst surface, the
Ag nanoparticles might be partially oxidized or deposited even on
the oxidative sites. This means that the metallic Ag atom in the Ag
nanoparticles on the oxidative sites can be oxidized to Ag+ cation
and released from the metallic particles while the released Ag+
cation can be reduced to form metallic Ag nanoparticles on the
reductive sites. If the facets on the polyhedral polycrystals could
be divided into oxidative and reductive facets, Ag nanoparticles
on the oxidative facets would disappear and Ag nanoparticles on
the reductive facets would be stabilized to function as the reduc-
tive sites for CO2 reduction. On the CTO(NaCl,50) sample, several
large polygonal facets on the polyhedral particles were observed
in the SEM images (Fig. 10a and b). Before the reaction the Ag
nanoparticles were dispersed on many facets. After the photocat-
alytic reaction, many Ag nanoparticles were found on some specific
facets, where the Ag cocatalyst did not aggregate. This suggests that
the Ag nanoparticles would be well stabilized on the large facets
even after the photocatalytic reaction test. These facets would be
the reductive facets. This dispersed and stabilized state of the Ag
nanoparticles on the large and flat reductive facets would con-
tribute to the constant and high photocatalytic activity in the steady
state.
On the other hand, the activity for the photocatalytic CO2 reduc-
tion of the Ag(0.1)/CTO(SS,1273) sample gradually decreased for
20 h while the activity for water splitting maintained (Table 2, entry
11). The CTO(SS,1273) sample consisted of roundish particles con-
nected together, where no large facets were observed (Fig. 10c and
d). Although the Ag nanoparticles were dispersed on the surface
before the reaction test (Fig. 10c), they aggregated so much after
the reaction test probably due to the lack of the large flat facets
(Fig. 10d). Thus, it is proposed that the aggregation of the Ag cocat-
alyst would be the reason for the decrease of the photocatalytic
activity for CO2 reduction. These facts suggest that the CO2 reduc-
tion to CO would take place on the Ag cocatalyst surface and the
water splitting to produce H2 would dominantly occur on the sur-
face of the CaTiO3 photocatalyst. The high and stable photocatalytic
activity of the Ag/CaTiO3 photocatalysts prepared with NaCl flux for
CO2 reduction would be achieved with the large flat facets stabiliz-
ing the Ag nanoparticles as the specific cocatalyst for CO formation.
Although the positive effect of the large size of particles has been
also reported and explained by the reduction of the recombination
of the photoexcited electron and hole pairs, for example, for WO3
photocatalyst [39], the reason for the size effect in the present study
would be different from the reported case.
Please cite this article in press as: H. Yoshida, et al., Calcium titanate photocatalyst prepared by a flux method for reduction of carbon
dioxide with water, Catal. Today (2014), http://dx.doi.org/10.1016/j.cattod.2014.10.039

G Model
CATTOD-9329; No. of Pages8
ARTICLE IN PRESS
8
H. Yoshida et al. / Catalysis Today xxx (2014) xxx–xxx
4. Conclusions
[2] K. Shimura, H. Yoshida, Catal. Surv. Asia 18 (2014) 24–33.
[3] A. Kudo, Y. Miseki, Chem. Soc. Rev. 38 (2009) 253–278.
[4] Y. Inoue, Energy Environ. Sci. 2 (2009) 364–386.
Each flux, CaCl2, KCl and NaCl, provided unique morphology
of CaTiO3 particles in the flux method (the molten salt method).
Among them, by using the NaCl flux with the moderate concentra-
tion of solute, the CaTiO3 samples consisting of the large polyhedral
particles covered with large flat facets and with fewer defects could
be prepared.
[5] R. Abe, J. Photochem. Photobiol. C 11 (2010) 179–209.
[6] K. Maeda, K. Domen, J. Phys. Chem. Lett. 1 (2010) 2655–2661.
[7] J.C. Hemminger, R. Carr, G.A. Somorjai, Chem. Phys. Lett. 57 (1978) 100–104.
[8] M. Halmann, Nature 275 (1978) 115–116.
[9] T. Inoue, A. Fujishima, S. Konishi, K. Honda, Nature 277 (1979) 637–638.
[10] H. Yoshida, in: Z. Ling (Ed.), Energy Efficiency and Renewable Energy Through
Nanotechnology, Green Energy and Technology, Springer-Verlag, London,
2011, pp. 531–559 (Chapter 15).
All the samples promoted the photocatalytic activity for CO2
reduction with water to produce CO and O2 and water splitting to
produce H2 and O2, where the two reactions competitively pro-
ceeded. The polyhedral CaTiO3 photocatalyst with Ag cocatalyst
exhibited a high and stable activity for the photocatalytic reduc-
tion of CO2, which was 10 times higher than the Ag-loaded CaTiO3
sample prepared by the solid state reaction method. It was revealed
that the CO2 reduction for CO formation take place especially on
the Ag cocatalyst, while the water splitting can proceed on the sur-
face of CaTiO3 photocatalyst. The large flat facets on the polyhedral
CaTiO3 particles of polycrystals would stabilize the Ag nanopar-
ticles to promote the photocatalytic CO2 reduction to CO. The
most active photocatalyst prepared in the present study was the
Ag(0.5)/CTO(NaCl,40) sample, which exhibited the CO production
rate of 0.35 ␮mol h−1 (0.2 g of the photocatalyst with the light of
0.35 W measured in the range of 254 10 nm) and the selectivity
for the CO2 reduction was 45% at the steady state.
[11] K. Sayama, H. Arakawa, J. Phys. Chem. 97 (1993) 531–533.
[12] K. Iizuka, T. Wato, Y. Miseki, K. Saito, A. Kudo, J. Am. Chem. Soc. 133 (2011)
20863–20868.
[13] Z. Like, E. Tanabe, K. Morikawa, T. Kazino, T. Sekihuzi, S. Matumoto, Y. Hirata,
H. Yoshida, 108th Symposium of Catal. Soc., Kitami, Japan, 2011, 1D15.
[14] K. Teramura, Z. Wang, S. Hosokawa, Y. Sakata, T. Tanaka, Chem. Eur. J. 20 (2014)
9906–9909.
[15] N. Yamamoto, T. Yoshida, S. Yagi, Z. Like, T. Mizutani, S. Ogawa, H. Nameki, H.
Yoshida, J. Surf. Sci. Nanotechnol. 12 (2014) 263–268.
[16] M. Yamamoto, T. Yoshida, N. Yamamoto, H. Yoshida, S. Yagi, J. Surf. Sci. Nan-
otechnol. 12 (2014) 299–303.
[17] J.S. Jang, P.H. Borse, J.S. Lee, K.T. Lim, O.S. Jung, E.D. Jeong, J.S. Bae, H.G. Kim, Bull.
Korean Chem. Soc. 32 (2011) 95–99.
[18] H. Mizoguchi, K. Ueda, M. Orita, S.C. Moon, K. Kajihara, M. Hirano, H. Hosono,
Mater. Res. Bull. 37 (2002) 2401–2406.
[19] K. Shimura, H. Yoshida, Energy Environ. Sci. 3 (2010) 615–617.
[20] K. Shimura, H. Miyanaga, H. Yoshida, Stud. Surf. Sci. Catal. 175 (2010) 85–92.
[21] J. Coreno, O. Coreno, J. Biomed. Mater. Res. A 75 (2005) 478–484.
[22] K. Teshima, K. Yubuta, S. Sugiura, T. Suzuki, T. Shishido, S. Oishi, Bull. Chem.
Soc. Jpn. 79 (2006) 1725–1728.
[23] C.Y. Xu, J. Wu, P. Zhang, S.P. Hu, J.X. Cui, Z.Q. Wang, Y.D. Huang, L. Zhen, Crys-
tEngComm 15 (2013) 3448–3454.
[24] S. Suzuki, K. Teshima, M. Kiyohara, H. Kamikawa, K. Yubuta, T. Shishido, S. Oishi,
CrystEngComm 14 (2012) 4176–4180.
[25] H. Yoshida, M. Takeuchi, M. Sato, L. Zhang, T. Teshima, M.G. Chaskar, Catal.
Today 232 (2014) 158–164.
Acknowledgements
We would like to thank Dr. Junya Ohyama for his help of
XRD measurements. The SEM measurements were carried out by
the help of Nano-fabrication platform at VBL, Nagoya University,
under Nanotechnology platform, The Ministry of Education, Cul-
ture, Sports, Science and Technology (MEXT), Japan. This work
was financially supported by Toyota Motor Corporation, a Grant-
in-Aid for Scientific Research (B), (No. 25289285), a Grant-in-Aid
for Scientific Research on Innovative Areas “All Nippon Artificial
Photosynthesis Project for Living Earth (AnApple)” (No. 25107515)
from the Japan Society for the Promotion of Science (JSPS), and the
Program for Element Strategy Initiative for Catalysts & Batteries
(ESICB), commissioned by the MEXT of Japan.
[26] H. Kato, M. Kobayashi, M. Hara, M. Kakihana, Catal. Sci. Technol. 3 (2013)
1733–1738.
[27] D. Arney, B. Porter, B. Greve, P.A. Maggard, J. Photochem. Photobiol. 199 (2008)
230–235.
[28] Q. Wang, T. Hisatomi, Y. Moriya, K. Maeda, K. Domen, Catal. Sci. Technol. 3
(2013) 2098–2103.
[29] D. Arney, T. Watkins, P.A. Maggardw, J. Am. Ceram. Soc. 94 (2011) 1483–1489.
[30] Y. Miseki, K. Saito, A. Kudo, Chem. Lett. 38 (2009) 180–181.
[31] Y. Hosogi, H. Kato, A. Kudo, J. Mater. Chem. 18 (2008) 647–653.
[32] J. Tauc, R. Grigorovici, A. Vancu, Phys. Status Solidi 15 (1966) 627–637.
[33] M.M. Elcombe, E.H. Kisi, K.D. Hawkins, T.J. White, P. Goodman, S. Matheson,
Acta Cryst. B 47 (1991) 305–314.
[34] J. Kiwi, M. Grätzel, J. Phys. Chem. 88 (1984) 1302–1307.
[35] J. Kiwi, C. Morrison, J. Phys. Chem. 88 (1984) 6146–6152.
[36] K. Yamaguchi, S. Sato, J. Phys. Chem. 89 (1985) 5510–5513.
[37] G. Muneura, V. Rives-Arnau, A. Saucedo, J. Chem. Soc. Faraday Trans. 1 (75)
(1979) 736–747.
References
[38] A. Gonzalez-Elipe, G. Muneura, J. Soria, J. Chem. Soc. Faraday Trans. 1 (75) (1979)
748–761.
[1] L. Yuliati, H. Yoshida, Chem. Soc. Rev. 37 (2008) 1592–1602.
[39] F. Amano, E. Ishinaga, A. Yamanaka, J. Phys. Chem. C 117 (2013) 22584–22590.
Please cite this article in press as: H. Yoshida, et al., Calcium titanate photocatalyst prepared by a flux method for reduction of carbon
dioxide with water, Catal. Today (2014), http://dx.doi.org/10.1016/j.cattod.2014.10.039