Photocatalytic conversion of carbon dioxide into methanol in reverse fuel cells with tungsten oxide and layered double hydroxide photocatalysts for solar fuel generation

Article abstract of DOI:10.1039/c3cy00959a

The phenomena of the photocatalytic oxidation of water and photocatalytic reduction of CO₂ were combined using reverse photofuel cells, in which the two photocatalysts, WO₃ and layered double hydroxide (LDH), were separated by a polymer electrolyte (PE) film. WO₃ was used for the photooxidation of water, whereas LDH, comprising Zn, Cu, and Ga, was used for the photoreduction of CO₂. For this process, photocatalysts pressed on both sides of the PE film were irradiated with UV-visible light through quartz windows and through the space in carbon electrode plates and water-repellent carbon paper for both gas flow and light transmission. 45% of the photocatalyst area was irradiated through the windows. The protons and electrons, which were formed on WO₃ under the flow of helium and moisture, transferred to the LDH via the PE and external circuit, respectively. Methanol was the major product from the LDH under the flow of CO₂ and helium. The observed photoreduction rates of CO₂ to methanol accounted for 68%-100% of photocurrents. This supports the effectiveness of the combined photooxidation and photoreduction mechanism as a viable strategy to selectively produce methanol. In addition, we tested reverse photofuel cell-2, which consisted of a WO₃ film pressed on C paper and LDH film pressed on Cu foil. The photoelectrodes were immersed in acidic solutions of pH 4, with the PE film distinguishing the two compartments. Both the photoelectrodes were completely irradiated by UV-visible light through the quartz windows. Consequently, the photocurrent from the LDH under CO₂ flow to WO ₃ under N₂ flow was increased by 2.4-3.4 times in comparison to photofuel cell-1 tested under similar conditions. However, the major product from the LDH was H₂ rather than methanol using photofuel cell-2. The photogenerated electrons in the irradiated area of the photocatalysts were obliged to diffuse laterally to the unirradiated area of photocatalysts in contact with the C papers in photofuel cell-1. This lateral diffusion reduced the photocatalytic conversion rates of CO₂, despite the advantages of photofuel cell-1 in terms of selective formation and easy separation of gas-phase methanol. This journal is the Partner Organisations 2014.

Full text of DOI:10.1039/c3cy00959a

Catalysis
Science &
Technology
View Article Online
View Journal | View Issue
PAPER
Photocatalytic conversion of carbon dioxide into
methanol in reverse fuel cells with tungsten oxide
and layered double hydroxide photocatalysts for
solar fuel generation†
Cite this: Catal. Sci. Technol., 2014,
4, 1644
Motoharu Morikawa,a Yuta Ogura,b Naveed Ahmed,b Shogo Kawamura,b
b
b
b
*
Gaku Mikami, Seiji Okamoto and Yasuo Izumi
The phenomena of the photocatalytic oxidation of water and photocatalytic reduction of CO2 were
combined using reverse photofuel cells, in which the two photocatalysts, WO3 and layered double
hydroxide (LDH), were separated by
a polymer electrolyte (PE) film. WO3 was used for the
photooxidation of water, whereas LDH, comprising Zn, Cu, and Ga, was used for the photoreduction of
CO2. For this process, photocatalysts pressed on both sides of the PE film were irradiated with UV-visible
light through quartz windows and through the space in carbon electrode plates and water-repellent
carbon paper for both gas flow and light transmission. 45% of the photocatalyst area was irradiated
through the windows. The protons and electrons, which were formed on WO3 under the flow of helium
and moisture, transferred to the LDH via the PE and external circuit, respectively. Methanol was the major
product from the LDH under the flow of CO2 and helium. The observed photoreduction rates of CO2 to
methanol accounted for 68%–100% of photocurrents. This supports the effectiveness of the combined
photooxidation and photoreduction mechanism as a viable strategy to selectively produce methanol.
In addition, we tested reverse photofuel cell-2, which consisted of a WO3 film pressed on C paper and
LDH film pressed on Cu foil. The photoelectrodes were immersed in acidic solutions of pH 4, with the
PE film distinguishing the two compartments. Both the photoelectrodes were completely irradiated by
UV-visible light through the quartz windows. Consequently, the photocurrent from the LDH under CO2
flow to WO3 under N2 flow was increased by 2.4–3.4 times in comparison to photofuel cell-1 tested
under similar conditions. However, the major product from the LDH was H2 rather than methanol using
photofuel cell-2. The photogenerated electrons in the irradiated area of the photocatalysts were obliged
to diffuse laterally to the unirradiated area of photocatalysts in contact with the C papers in photofuel
cell-1. This lateral diffusion reduced the photocatalytic conversion rates of CO2, despite the advantages
of photofuel cell-1 in terms of selective formation and easy separation of gas-phase methanol.
Received 23rd November 2013,
Accepted 18th February 2014
DOI: 10.1039/c3cy00959a
www.rsc.org/catalysis
CO2 corresponding to photosystem II, and subsequent dark
reactions to incorporate CO2 into carbohydrates utilizing
Introduction
The photocatalytic conversion of CO2 into fuels has emerged
as an attractive option,1 in terms of both reducing the
increased concentration of atmospheric CO2 as well as gener-
ating renewable hydrocarbon fuels that can directly be sup-
plied to our present energy infrastructure. Analogous to
photosynthesis, the photocatalytic conversion of CO2 involves
(1) the photooxidation of water and (2) the photoreduction of
NADPH formed in photosystem I.2,3
Thus far, several researchers have reported photocatalytic
cells for water oxidation2 and proton reduction4–6 by using
Pt–WO3 and Pt–SrTiO3,7 as well as TiO2 and Pt–TiO2 sepa-
rated by a polymer electrolyte (PE) film.8 The combination
of steps 1 and 2 of the photocatalytic conversion of CO2
has been realized using metal complexes and enzymatic
photocatalysts.9 To the best of our knowledge, there are no
studies reporting on the combination of semiconductor
photocatalysts for the photocatalytic conversion of CO2. In
principle, the combination of steps 1 and 2 can be regarded
as reverse fuel cell operation. This process is considered to be
a reverse mechanism because the products generated by
a Department of Nanomaterial Science, Graduate School of Advanced Integration
Science, Chiba University, Yayoi 1-33, Inage-ku, Chiba 263-8522, Japan
b Department of Chemistry, Graduate School of Science, Chiba University,
Yayoi 1-33, Inage-ku, Chiba 263-8522, Japan. E-mail: yizumi@faculty.chiba-u.jp
† Electronic supplementary information (ESI) available. See DOI: 10.1039/
c3cy00959a
1644 | Catal. Sci. Technol., 2014, 4, 1644–1651
This journal is © The Royal Society of Chemistry 2014

View Article Online
Catalysis Science & Technology
Paper
conventional fuel cells, i.e., CO2 and H2O, react to form metha-
nol and O2, which are the reactants in the conventional fuel cell.
In fact, the original concept of reaction between CO2 and H2O
was proposed10–12 and demonstrated to be a reverse fuel cell
operation,13 rather than a photocatalytic conversion process.
In this paper, we report the fabrication of a polymer
electrolyte fuel cell (PEFC; Fig. 1) comprising tungsten oxide
(WO3) and layered double hydroxide (LDH) photocatalysts,
and a photofuel cell consisting of acid solutions separated by
a PE film (Fig. 2). In addition, we experimentally demonstrate
solar fuel generation,10,14–17 in which the photooxidation of
water results in the transfer of protons and electrons to form
fuel combined with CO2.
The primary advantage of using photofuel cells is the
band potential of two photocatalysts, which is tunable in
comparison with the oxidation and reduction potential of
independent reactions, by the choice of the elemental compo-
sition of the semiconductor catalysts. Traditionally, WO3 has
been the catalyst of choice for the photooxidation of water
(E° = 1.23 − 0.0591 × pH V, 298 K), primarily because of the rel-
atively positive band potential.18 Similarly, LDH is an unusual
Fig.
2
Fabricated photofuel cell-2. The figures show the different
components (A) and LDH1 mounted onto Cu foil and WO3 mounted
onto water-repellent C paper (B, left to right). The smaller cell-2 was also
used in which the LDH1 side (right) was circulated with CO2 gas (C).
and are relatively inexpensive. Furthermore, the products
formed as a result of photocatalysis can be readily separated
in the gas phase. This is a unique advantage of using a
photofuel cell based on PEFC (Fig. 1), as reported for the
electrocatalytic cell separated by Nafion.24
class of layered materials comprising positively charged cation
2− 19
layer and interlayer anionic species such as CO3
.
Recently,
researchers have demonstrated the photocatalytic conversion
of CO2 into CO and methanol in water20 or hydrogen21–23 by
using LDHs comprising Ni, In, Zn, Cu, and Ga. From these
reports, we chose LDH compounds comprising Zn, Cu, and Ga
for the photoreduction of CO2 in this study.
Experimental
Synthesis of WO3 and LDH consisting of Zn, Cu, and Ga
WO3 was synthesized by the calcination of ammonium
paratungstate pentahydrate (NH4)10+(W12O41)10−·5H2O (Aldrich,
>99.99%) in air at 973 K for 4 h.
In addition, the added advantage of using a photofuel cell
is that the photocatalysts used are earth-abundant elements
Nitrogen adsorption isotherm measurements were
performed at 77 K within the pressure range 1.0–90 kPa in a vac-
uum system connected to diffusion and rotary pumps (10−6 Pa)
and equipped with a capacitance manometer (Models CCMT-
1000A and GM2001, ULVAC). The Brunauer–Emmett–Teller
surface area (SA) was calculated on the basis of eight-point
measurements between 10 and 46 kPa (P/P0 = 0.10–0.45) on the
adsorption isotherm. The samples were evacuated at 383 K for
2 h before the measurements. The specific SA of WO3 thus
obtained was estimated to be 20 m2 g−1.
+
In this study, we synthesized LDH1 [Zn1.5Cu1.5Ga(OH)8]2
[Cu(OH)4]2−·mH2O starting from nitrates of Zn, Cu, and Ga
and ammonium tetracholorocupurate dihydrate at pH 8.
Details of the experimental procedure adopted for the synthe-
sis of LDH1 have been reported in the literature.22,23 The
exact molecular formula of LDH1, as determined by using
extended X-ray absorption fine structure analysis, was found
to be Zn3Cu3Ga2(OH)13[(μ-O)3Cu(OH)(H2O)2]·mH2O, with the
dehydration of three water molecules.21 Furthermore, the
specific SA of LDH1 was estimated to be 62 m2 g−1.
Fig. 1 Fabricated photofuel cell-1. The figures show the different
components (A) and a photocatalytic test with flowing moisture + He
(balance) (front side) and circulating CO2 + He (balance) (rear side) (B).
A1 depicts photofuel cell-1 at the center of B. A2 and A3 depict compo-
nents of the photocathode and photoanode, respectively. A4 is a view
of the window of photofuel cell-1. A5 depicts the assembled quartz
window, silicon gasket, C electrode plate, C paper, silicon gasket, and
WO3–Nafion–LDH1 assembly.
Design of photofuel cell-1
For this process, 95 mg of WO3 or 45 mg of LDH1 was
suspended in 5% of Nafion dispersion solution (DE521,
Wako Pure Chemical; 0.2 mL) and 1-propanol (0.1 mL) and
This journal is © The Royal Society of Chemistry 2014
Catal. Sci. Technol., 2014, 4, 1644–1651 | 1645

View Article Online
Paper
Catalysis Science & Technology
separately mounted onto water-repellent carbon paper
(C paper, TGP-H-060H, Chemix) in an area of 4 cm2. Subse-
quently, WO3 mounted onto the C paper was pressed with
LDH1 mounted onto the C paper and separated using a PE
film (Nafion, NR-212, Dupont) of thickness 50 μm. The layers
were pressed using a tabletop press (Model SA-302, Tester
Sangyo Co.) by applying a pressure of 2.0 MPa at 393 K for
10 min. Then the C papers on both sides were carefully
removed, resulting in the formation of the WO3–Nafion–
LDH1 assembly.
The electronic conductivity of WO3 and LDH1 was
improved by mixing WO3 (95 mg) and LDH1 (45 mg) with a
minimum amount of C black (0.2 mg; Vulcan XC72, specific
SA 250 m2 g−1) using a mortar and pestle. Subsequently, the
obtained WO3 and LDH1 were pressed together with a Nafion
film separator to form a WO3/C–Nafion–LDH1/C assembly, by
using a procedure similar to that adopted for the fabrication
of WO3–Nafion–LDH1.
Photofuel cell-1 designed in this study is shown in Fig. S1
(ESI†). The prepared WO3–Nafion–LDH1 assembly was first
sandwiched by a water-repellent C paper and then by C
electrode plates, both via a serpentine route (the space sec-
tion for gas flow and window for light; Fig. 1A2, A3). Subse-
quently, this assembly was sandwiched by quartz window
plates. It should be noted that the C paper and C plate were
fixed in an area of 2.2 cm2 and the area of the serpentine
route was 1.8 cm2 out of the total 4 cm2 of each quartz win-
dow (Fig. 1). Therefore, among a catalyst area of 4 cm2, the
area of 1.8 cm2 (45%) was irradiated through the serpentine
route whereas the area of 2.2 cm2 (55%) was blocked by the
C paper and C plate from light (Fig. S1†).
The gas circulating through LDH1 was analyzed by an
inline gas chromatograph equipped with a thermal conduc-
tivity detector (GC-TCD; Shimadzu, Model GC-8A), using
He as the carrier gas. The columns were packed with Molecu-
lar Sieve 13X-S and polyethylene glycol-6000 supported on
Flusin P (GL Science). The current between the two electrodes
(Fig. 1A, B) was monitored simultaneously.
Design of photofuel cell-2
We also designed photofuel cell-2, as shown in Fig. 2. Both
photofuel cells were compared for their performance.
For the fabrication of photofuel cell-2, 95 mg of WO3 was
suspended in 5% of Nafion dispersion solution (0.24 mL)
and 1-propanol (0.16 mL) and mounted onto water-repellent
C paper in an area of 4 cm2. WO3 mounted onto the C paper
was then covered with a Kapton film (200H, Dupont) of thick-
ness 50 μm and pressed using a tabletop press (SA-302) by
applying a pressure of 2.0 MPa at 393 K for 10 min. Then,
the Kapton film was carefully removed to obtain the WO3/C
photoelectrode (Fig. 2B, right). In a separate process, 45 mg
of LDH1 was suspended in 5% of Nafion dispersion solution
(0.24 mL) and 1-propanol (0.16 mL) and mounted onto Cu
foil (2.3 × 2.5 cm2) of thickness 30 μm in an area of 4 cm2.
LDH1 mounted onto the Cu foil was then covered with a
water-repellent C paper and pressed using a tabletop press
(SA-302) by applying a pressure of 2.0 MPa at 393 K for 10 min.
Following that, the C paper was carefully removed to obtain
the LDH1/Cu photoelectrode. In addition, the LDH1/Cu
photoelectrode was prepared by dispersing LDH1 in
1-propanol (0.4 mL) instead of Nafion dispersion solution in a
manner similar to that for LDH1/Cu. LDH1 is sticky and
attached to Cu foil even in the absence of Nafion (Fig. 2B, left).
Photofuel cell-1 offers the advantage of easy separation of
gas-phase products, including methanol. However, a disad-
vantage associated with the design of the photofuel cell is
that only 45% of the photocatalysts get irradiated.
CO2 conversion tests using photofuel cell-2
As the next step, we analyzed the CO2 conversion perfor-
mance of photofuel cell-2. For this, the prepared WO3/C and
LDH1/Cu photoelectrodes were immersed in aqueous
hydrochloric acid solution (50 mL at each electrode) at pH 4.0,
separated by a 50 μm-thick Nafion film at the center of the cell
(Fig. 2A). Subsequently, N2 and CO2 were bubbled at a rate of
50 mL min−1 through the sides of WO3 and LDH1, respectively.
Both photocatalysts were irradiated using a 500 W Xe arc lamp
via a two-way quartz fiber light guide. The photocatalysts
were irradiated from both sides (Fig. 2A) for 30 min and kept
in the dark for 30 min. The light on/off cycle was repeated five
times. The distance between the fiber light exit and the
photoelectrode was maintained at 50 mm. The light intensity
on the photocatalyst at this position was 33 mW cm−2.
Moreover, a smaller photofuel cell-2 was also fabricated
(Fig. 2C) to make the product concentration relatively higher
for the GC analysis. 15 mL of HCl solution was introduced at
each electrode. CO2 was circulated in a glass line at a rate of
450 mL min−1 through the side of LDH1. The other conditions
were similar to that for Fig. 2A. The gas circulated through
LDH1 was analyzed by the inline GC-TCD (Model GC-9A) using
CO2 conversion tests using photofuel cell-1
The fabricated photofuel cell-1 was further tested for its CO2
conversion efficiency. As the first step, all solvents included
in the Nafion dispersion solution were removed by flowing
N2 gas at a rate of 100 mL min−1 through a water bubbler
maintained at 343 K, independently to WO3 and LDH1 for
10 h. Subsequently, WO3 in the system was purged with He
gas at a rate of 50 mL min−1 through the water bubbler at
323 K. CO2 (3.5%) with the remaining He gas (total 101 kPa)
was circulated in a glass line at a rate of 450 mL min−1
through LDH1 (Fig. 1A1).
WO3 and LDH1 on the Nafion film were irradiated by
UV-visible light from a 500 W xenon arc lamp (Ushio, Model
SX-UID502XAM) via a two-way quartz fiber light guide of 1 m
(San-ei Electric Co., Model 5Φ-2B-1000 L) for 10 h (Fig. 1B).
The distance between fiber light exit and photofuel cell win-
dow was maintained at 50 mm. The light intensity on the
photocatalyst was 33 mW cm−2. With the onset of irradiation,
the photocatalyst temperature reached as high as 315 K.
1646 | Catal. Sci. Technol., 2014, 4, 1644–1651
This journal is © The Royal Society of Chemistry 2014

View Article Online
Catalysis Science & Technology
Paper
Ar as the carrier gas and also a GC-flame ionization detector
(FID; Shimadzu, Model GC-18A) equipped with a capillary
column Ultra ALLOY-5 (Frontier Laboratories; inner diameter
250 μm, length 30 m) using He as the carrier gas.
Results
CO2 to methanol conversion in photofuel cell-1
WO3 on one side of the Nafion film was irradiated with
UV-visible light under the flow of He and moisture, while
LDH1 on the other side of the Nafion film was irradiated with
UV-visible light under the flow of CO2 (3.5%) (Fig. 1B). The
main product of the reaction was methanol, whose amount
monotonously increased as a function of time irradiated
under light (Fig. 3, entry a). Other gases generated during the
reaction, including CO and methane, were not detected, proba-
bly because the amount was below the detection limit of GC.
Similarly, H2 generation could not be monitored, as He was
the carrier gas. In this process, the methanol formation rate
was estimated to be 0.045 μmol h−1 gcat−1, considering that the
LDH1 photoelectrode was irradiated through the serpentine
route and the effective illumination area is 45% (Fig. 1A4).
During the total illumination period of 10 h, the current
from the LDH1 electrode to the WO3 electrode was stabilized
to a constant value of 0.22 μA in 1.6 h (Fig. 4a). The current
Fig. 4 Time evolution of the photocurrent generated during CO2
conversion tests using photofuel cell-1 consisting of WO3 and LDH1 (a)
and WO3 and LDH1 both mixed with C black (b). The photocatalysts
were in the dark for 10 h and UV-visible light was irradiated in the next
10 h (a, b).
penetration depth into both photocatalyst layers mixed
with C rather than the effect of improved electron conductiv-
ity by mixing the photocatalysts with C. During the test,
the photocurrent from the LDH electrode to the WO3
electrode stabilized to a constant value of 0.10 μA in 1.6 h
(Fig. 4b). The constant current corresponds to an electron
flow rate of 0.18 μmol-e− h−1 gLDH from WO3/C to LDH1/C
−1
(Table 1b). Most of the current was accounted for by the six-
−1
corresponds to an electron flow rate of 0.40 μmol-e− h−1 gLDH
electron reduction reaction (6 × 0.029 μmol-CH3OH h−1 gcat
=
−1
from WO3 to LDH1 (Table 1a). Here, 68% of the photocurrent
was accounted for by the six-electron reduction reaction
0.17 μmol h−1 gcat−1) from CO2 to methanol.
The generated photocurrents decreased in the initial 1.6 h
of irradiation both in Fig. 4a and b. In comparison to the
rather constant formation of methanol (Fig. 3), a major part
of the initial photocurrents may not be due to photocatalysis.
In this study, we performed two blank tests. One blank
test was performed at 323 K, which is higher than the maxi-
mum temperature (315 K) of kinetic tests irradiated by light,
under dark conditions and flow of He and moisture to WO3
and flow of CO2 (3.5%) and the remaining He to LDH1. The
other blank test was performed with light irradiation under
the flow of He and moisture to WO3, and the flow of He to
LDH1. For both tests, we could not detect products above
the detection limit of GC. The photocurrent was negligible
after 2 h of reaction.
−1
−1
(6 × 0.045 μmol-CH3OH h−1 gcat = 0.27 μmol-e− h−1 gcat
)
from CO2 to methanol.
Furthermore, we analyzed the effects of mixing C black
with the photoelectrodes by comparing the CO2 conversion
performance of the WO3/C–Nafion–LDH1/C assembly with
that of the WO3–Nafion–LDH1 described above. The CO2 con-
version results of the WO3/C–Nafion–LDH1/C assembly in the
photofuel cell are plotted in Fig. 3, entry b. As seen from the
figure, the methanol formation rate is 0.029 μmol h−1 gcat
.
−1
This indicates a reduction in the methanol formation rate
by 36% with the addition of C black to LDH1 and WO3.
The lower formation rate should be due to lower light
CO2 to methanol conversion using photofuel cell-2
The CO2 photoconversion performance of photofuel cell-1 was
compared with that of photofuel cell-2 containing WO3 and
LDH1 immersed in acid solutions, and fully exposed to light.
For the photofuel cell containing 95 mg of WO3 and
45 mg of LDH1 separately mounted onto C paper and Cu foil,
respectively (Fig. 2A, B), the photocurrent gradually increased
within 15–20 min of irradiation under UV-visible light and
gradually decreased to background levels within 10–15 min
in the dark (Fig. 5a).
Fig. 3 Time evolution of the formation of methanol in photofuel cell-1
consisting of WO3 and LDH1 (a) and WO3 and LDH1 mixed with C black
(b). The photocatalysts were in the dark for 10 h and UV-visible light
was irradiated in the next 10 h (a, b).
The maximum photocurrent from the LDH1 electrode
to the WO3 electrode was 1.15 μA, as determined from five
on/off cycles of light irradiation. The photocurrent corresponded
This journal is © The Royal Society of Chemistry 2014
Catal. Sci. Technol., 2014, 4, 1644–1651 | 1647

View Article Online
Paper
Catalysis Science & Technology
Table 1 Comparison of photocurrents generated during the CO2 conversion tests using the photofuel cells designed in this study
e− flow rate
MeOH or H2 fo−r1m rate
LDH1b support
Current (μA) (μmol h
g
)
(μmol h−1 gLDH
)
a
−1
−1
LDH
Entry Cell type
WO3 support
a
b
Photofuel cell-1 Nafion film
Nafion film
Nafion film
Nafion film
0.22
0.10
0.40d
0.18d
0.045 (MeOH)d
0.029 (MeOH)d
(C black mixedc ) (C black mixedc )
c
d
Photofuel cell-2 C paper
Cu foil
1.15
1.63
0.96
1.36
0.49 (H2)
0.67 (H2)
C paper
Cu foil (no Nafion disp sol usede
)
a
b
c
d
e
95 mg. 45 mg. 0.2 mg. Divided by exposed LDH1 sample to light (45 mg × 0.45). LDH1 was dispersed in 1-propanol only and mounted
on Cu foil.
to an electron flow rate of 0.96 μmol h−1 gLDH−1, a factor of
2.4 times greater than that using photofuel cell-1 (Table 1c, a).
In this comparison, the electron flow rates were estimated by
the area of photocatalyst exposed to light irradiation, e.g., 45%
and 100% for photofuel cell-1 and photofuel cell-2, respectively.
In addition, we performed the CO2 photoconversion test
using photofuel cell-2 consisting of WO3 (95 mg) on C paper
and LDH1 (45 mg) on Cu foil, which was prepared without
the Nafion dispersion solution. The change in photocurrent
during the five on/off cycles of light is depicted in Fig. 5b. As
seen from the figure, the photocurrent increases gradually
under light and decreases gradually to background level in
the dark. This trend is very similar to that shown in Fig. 5a.
However, the maximum photocurrent (1.63 μA) was found to
increase by a factor of 1.42 times with the elimination of
Nafion in the LDH1 photoelectrode mounted on Cu foil.
The photocurrent corresponded to an electron flow rate of
1.36 μmol h−1 gLDH−1, which was 3.4 times greater than that
using photofuel cell-1 (Table 1d, a).
rate of 0.49 μmol-H2 h−1 gLDH−1. H2 formation was also
observed in dark cycles. This fact is related to the slow decrease
of photocurrent in dark cycles (Fig. 5a), suggesting diffusion
control. Most of the current (0.96 μmol-e− h−1 gLDH
)
−1
was accounted for by the two-electron reduction reaction
(2 × 0.49 μmol-H2 h−1 gLDH−1 = 0.98 μmol h−1 gLDH−1; Table 1c).
When WO3 on C paper was purged with N2 gas and
LDH1 on Cu foil was circulated with CO2, prepared without
the Nafion dispersion solution, the hydrogen formation
under UV-visible light irradiation became faster, at a rate
of 0.67 μmol-H2 h−1 gLDH (Fig. 6b). Most of the current
−1
(1.36 μmol-e− h−1 gLDH−1; Fig. 5b) was accounted for by the
−1
two-electron reduction reaction (2 × 0.67 μmol-H2 h−1 gLDH
1.34 μmol h−1 gLDH−1; Table 1d).
=
In both tests in Fig. 6a and b, other gases generated
during the reaction through LDH1 were not detected above
the detection limit of GC-TCD. The acid solution for LDH1
in photofuel cell-2 under the conditions of Fig. 6b (Fig. 2C)
irradiated under UV-visible light for 10 h was analyzed by
GC-FID, but methanol was not detected above the detection
limit (4.83 pmol in 1.0 μL injection). The detection limit
corresponded to the formation rate of 0.16 μmol-CH3OH h−1
gLDH−1. Taking the six-electron reduction into account, the
detection limit was 0.96 μmol-e− h−1 gLDH−1 versus the observed
Both the gas circulated through LDH1 and the acid
solution for LDH1 were analyzed in a separate test using a
smaller photofuel cell-2 (Fig. 2C). The time evolution is
depicted in Fig. 6. When WO3 on C paper was purged with
N2 gas and LDH1 on Cu foil was circulated with CO2, both
prepared using a Nafion dispersion solution, hydrogen was
a major product from LDH1 (Fig. 6, entry a) at a formation
current of 1.36 μmol-e− h−1 gLDH (Fig. 5b). Therefore metha-
−1
nol was not the major product in photofuel cell-2.
Fig.
5
Time evolution of photocurrent generated during CO2
Fig. 6 Time evolution of the formation of hydrogen during CO2
conversion tests using photofuel cell-2 comprising WO3 on C paper,
prepared using Nafion suspension solution (a, b) and LDH1 on Cu foil,
prepared using Nafion suspension solution (a) and prepared using
1-propanol without Nafion (b).
conversion tests using photofuel cell-2 comprising WO3 on C paper,
prepared using Nafion dispersion solution (a, b) and LDH1 on Cu foil,
prepared using Nafion suspension solution (a) and prepared using
1-propanol without Nafion (b).
1648 | Catal. Sci. Technol., 2014, 4, 1644–1651
This journal is © The Royal Society of Chemistry 2014

View Article Online
Catalysis Science & Technology
Paper
solutions between the WO3 side and LDH1 side. Alternatively,
the pH of the solution in the LDH1 side would have increased
with the increase in temperature from 290 to 310 K due to the
solubility dependence of CO2 during the five on/off irradiation
cycles. Consequently, to compensate the pH change, the back-
ground current might have increased gradually.
Discussion
It is well known that the mechanism underlying the photo-
catalytic splitting of water using WO3 under UV irradiation is
via the following equation.25,26
2H2O → O2 + 4H+ + 4e−
(1)
Reaction (1) in the presence of the WO3 catalyst is
followed by reaction (2) over LDH1.
In this study, WO3 exhibited constant oxygen generation
upon irradiation by UV-visible light in the presence of a sacri-
ficial oxidant (Ag+).27 The potentials for the photoreduction
reactions, e.g., H2 formation, are beyond the bandgap energy
of WO3.18 Thus, the protons generated in this process trans-
ferred to LDH1 via Nafion film in the case of photofuel cell-1,
and via the acid solution and Nafion film in the case of
photofuel cell-2. On the other hand, the electrons generated
in this process transferred to LDH1 via the external circuit
(Scheme 1).
On comparing the photocurrent generated in photofuel
cell-1 and photofuel cell-2 with the same types and amounts of
photocatalysts, it was observed that photofuel cell-2 generated
a higher photocurrent. This suggests that proton diffusion in
the HCl solutions (pH 4) of photofuel cell-2 (Table 1c, d) is suffi-
ciently high and that the diffusion of protons in 50 μm-thick
Nafion film is relatively critical. The conductivity of Nafion film at
a relative humidity of 10–100% was reported to be 79 mS cm−1
based on electrochemical impedance spectroscopy.28 If the
applied voltage was assumed to be 1 V, the conductivity
corresponded to proton conductivity of the order of 10−5 mol h−1
for Nafion film of thickness 50 μm. In contrast, the proton dif-
fusion rate in acid solution (pH) can be estimated based on the
catalytic rate of the diffusion-limited reaction.29 The hydrogen
formation rate from an Fe2S2(CO)4[P(OCH3)3]2 complex was of
the order of 10−1 mol h−1. Apparently, proton conduction in
Nafion film is more critical compared to that in acid solution,
but it is somewhat faster compared to the charge flow in the
cells of this study (3.6–61 nmol h−1; Table 1), suggesting rate
control primarily by surface reactions.
CO2 + 6H+ + 6e− → CH3OH + H2O
(2)
The mechanism underlying reaction (2) seems to be simi-
lar to that of reaction (2′), which has been reported for LDH1
and [Zn1.5Cu1.5Ga(OH)8]2 CO32−·mH2O photocatalysts.21–23
+
More specifically, the importance of protonation to the one-
electron-reduced CO2 species has been discussed while
explaining the formation of formic acid30 and methane31 in
the presence of TiO2 and Zn–Ge oxynitride.32 Specific reac-
tion pathways have been proposed for reaction (2) or reaction
(2′), which result in methanol (or CO/methane) via the forma-
tion of formaldehyde,21,23,33 carbene,33 and glyoxal33 by mul-
tiple protonation steps and electron supply.
CO2 + 3H2 → CH3OH + H2O
(2′)
Similar to the reports for reaction (2′),21–23 methanol was
the major product formed in photofuel cell-1. 68%–100% of
electron production/consumption during the steady-state
period, between 2 and 10 h, in photofuel cell-1 (Fig. 3 and 4)
can be attributed to the conversion of CO2 to methanol. We
evaluated the photocatalytic performance starting from H2O
and CO2 according to the photocurrent from LDH1 to WO3
electrodes in the photofuel cells (Table 1).
The excessive photocurrents observed in the initial 1.6 h
in photofuel cell-1 corresponded to 0.022 μmol of electrons
(Fig. 4a, b). A major part of the initial photocurrents are not
due to photocatalysis. One of the possibilities is that WO3
proceeded reaction (1) whereas the supplied electrons from
WO3 to the photocathode were consumed non-catalytically,
e.g. the partial reduction of CuII sites in LDH1 under irradia-
tion. In this study, the amount of Cu used in the cell was
180 μmol. We tried to detect reduced CuI sites among major
CuII sites in LDH1 used for the tests in Fig. 5b & 6b by syn-
chrotron Cu K-edge X-ray absorption spectroscopy at KEK-PF
& SPring-8.21,23,34 However, 120 ppm of CuI at maximum was
hard to detect experimentally.
Fig. 5a and b show a gradual increase in the background
current. This could primarily be attributed to the slow supply
of protons to LDH1, leading to a concentration gradient of
The major generation of hydrogen from the photocathode
(LDH1) was directly monitored in photofuel cell-2 (Fig. 6). In
contrast, the part of the photocurrent that was not attributed
to the conversion of CO2 to methanol, which is less than 32%
of total photocurrents (Table 1), may be ascribed to the for-
mation of hydrogen in photofuel cell-1. The reduction poten-
tial for reaction (2) (E° = −0.32 − 0.0591 × pH V), in which
methanol concentration is negligible compared to proton
concentration, is similar to that for reaction (3) (E° = 0 −
0.0591 × pH V).35 In the blank test with light irradiation
Scheme 1 Schematic illustration of the flow of materials and electrons
during photocatalytic CO2 conversion in reverse photofuel cell-1
consisting of WO3 and LDH1.
This journal is © The Royal Society of Chemistry 2014
Catal. Sci. Technol., 2014, 4, 1644–1651 | 1649

View Article Online
Paper
Catalysis Science & Technology
under the flow of He and moisture to WO3, and He to LDH1
in photofuel cell-1, the photocurrent was negligible, after 2 h
of reaction. Thus, H2 formation was negligible but further
photocatalytic reduction to methanol should proceed at the
interface of LDH1 and the gas (CO2 & He) in photofuel cell-1
whereas reaction (3) predominantly proceeded at the inter-
face of LDH1 and the acid solution of pH 4.0 in photofuel
cell-2. The observed clear contrast in selectivity is due to gas-
eous CO2 (3.5%) in photofuel cell-1 (methanol formation) ver-
sus the limited solubility of CO2 in water in photofuel cell-2
(hydrogen formation).
and a LDH, Zn3Cu3Ga2(OH)13[(μ-O)3Cu(OH)(H2O)2]·mH2O, as
the photoreduction catalyst of CO2. This cell could be used to
enable solar fuel generation if optimum photooxidation/
reduction catalysts are chosen independently.
Photofuel cell-1, which was designed on the basis of PEFC,
was advantageous in terms of the easy separation of metha-
nol gas as the major product. On the other hand, solar fuel
generation was limited, especially from the photoelectron
flow efficiency viewpoint. The electrons generated in WO3 by
the photooxidation of water were obliged to move laterally to
the unirradiated area of WO3 and then to carbon, and from
carbon to the unirradiated area of LDH1 and then laterally to
the irradiated area of LDH1.
2H+ + 2e− → H2
(3)
Photofuel cell-2, consisting of independent photoelectrodes
immersed in acid solution (pH 4), had higher photocurrents,
by 2.4–3.4 times. However, the reaction on LDH1 caused H2
formation and methanol was not found above the detection
limit of GC-FID. Future studies are required to improve
photofuel cell-1 and to control the potential of the harder step,
i.e., CO2 photoreduction at the photocathode by using sustain-
able sources such as solar cells.27
In summary, efficient CO2 conversion to methanol can be
observed over LDH1 in photofuel cell-1, which is analogous
to the CO2 photoconversion to methanol in CO2 + H2 mixed
gases,21–23 whereas hydrogen was preferably formed in
photofuel cell-2.
Although the major photocatalytic products (methanol or
H2) were different, the electron flow rates in photofuel cell-2
(0.96–1.36 μmol h−1 gLDH−1; Table 1c, d) were observed to be
significantly higher, by 2.4–3.4 times, when compared to
those in photofuel cell-1 (0.40 μmol h−1 gLDH−1; Table 1a). A
reason for this is the longer electron diffusion path from the
WO3 layer to the C electrode and that from the C electrode to
the LDH1 layer for photofuel cell-1. In particular, the proton
transfer in the serpentine area (Fig. 1A2–5) should be good,
while the electrons separated from holes in the irradiated ser-
pentine area of photocatalysts must laterally diffuse to the
unirradiated area of photocatalysts, C paper, and then to the
C electrode (anode) connected to the external circuit. For
the photocathode, the electron flow from the C paper to the
unirradiated area of LDH1 and then laterally to the irradiated
serpentine area of LDH1 is needed. For photofuel cell-2, the
electrons separated under irradiation easily diffuse into the
thin WO3 photocatalyst vertically to the C paper,36 and from
the C paper vertically to the thin LDH1 (Fig. 2A, C).
Acknowledgements
The authors are grateful for the financial support from the
Grant-in-Aid for Scientific Research C (proposal no. 22550117)
from Monbukagakusho and the Feasibility Study Stage of
A-STEP (proposal no. AS231Z01459C and AS251Z00906L) from
the Japan Science and Technology Agency. The X-ray absorption
experiments were performed with the approvals of the Photon
Factory Proposal Review Committee (no. 2011G033 and
2009G552) and the grant of the Priority Program for Disaster-
Affected Quantum Beam Facilities (2011A1977, SPring-8 & KEK).
Notes and references
1 M. Mikkelsen, M. Jørgensen and F. C. Krebs, Energy Environ.
Sci., 2010, 3, 43.
Of the two photofuel cells analyzed in this study, the maxi-
mum photocurrent was observed (Fig. 5b) in photofuel cell-2
consisting of WO3 and the LDH1 photoelectrode on Cu foil pre-
pared without Nafion dispersion solution (Fig. 2B). The current
was quantitatively in accord with dominant H2 formation
(Fig. 6b). Nafion dispersion solution used to mount LDH1 on
Cu foil had a negative effect in covering the active photoreduc-
tion sites (Table 1c, d). Furthermore, X-ray diffraction studies
of the LDH1 sample indicate that the layered structure of
LDH1 is not destroyed during the photoreduction test for 5 h.
In spite of this, LDHs as hydroxide-based materials needing to
be in close contact with Nafion dispersion solution/film would
be unstable in the longer-term application of photofuel cell-1.
2 J. H. Alstrum-Acevedo, M. K. Brennaman and T. J. Meyer,
Inorg. Chem., 2005, 44, 6802.
3 D. Voet and J. G. Voet, Biochemistry, Wiley, New York, 2nd
edn, 1995, ch. 22, pp. 626–661.
4 K. S. Joya, Y. F. Joya, K. Ocakoglu and R. van de Krol, Angew.
Chem., Int. Ed., 2013, 52, 10426.
5 K. J. Young, L. A. Martini, R. L. Milot, R. C. Snoeberger III,
V. S. Vatista, C. A. Schmuttenmaer, R. H. Crabtree and
G. W. Brudvig, Coord. Chem. Rev., 2012, 256, 2503.
6 J. Sun, D. K. Zhong and D. R. Gamelin, Energy Environ. Sci.,
2010, 3, 1252.
7 K. Sayama, K. Mukasa, R. Abe, Y. Abe and H. Arakawa,
Chem. Commun., 2001, 2416.
8 K. Fujihara, T. Ohno and M. Matsumura, J. Chem. Soc.,
Faraday Trans., 1998, 94, 3705.
9 C. D. Windle and R. N. Perutz, Coord. Chem. Rev., 2012,
256, 2562.
Conclusions and future prospects
A reverse photofuel cell to form O2 and methanol was
designed using WO3 as the photooxidation catalyst of water
10 Y. Izumi, Coord. Chem. Rev., 2013, 257, 171.
1650 | Catal. Sci. Technol., 2014, 4, 1644–1651
This journal is © The Royal Society of Chemistry 2014

View Article Online
Catalysis Science & Technology
Paper
11 A. D. Handoko, K. Li and J. Tang, Curr. Opin. Chem. Eng.,
2013, 2, 200.
12 L. Alibabaei, H. Luo, R. L. House, P. G. Hoertz, R. Lopez and
T. J. Meyer, J. Mater. Chem. A, 2013, 1, 4133.
13 M. Morikawa, N. Ahmed, Y. Ogura and Y. Izumi, Appl.
Catal., B, 2012, 117/118, 317.
25 M. G. Walter, E. L. Warren, J. R. McKone, S. W. Boettcher,
Q. Mi, E. A. Santori and N. S. Lewis, Chem. Rev., 2010,
110, 6446.
26 A. Kudo and Y. Miseki, Chem. Soc. Rev., 2009, 38, 253.
27 Y. Ogura and Y. Izumi, Japanese Patent, 2012, 223765;
Y. Ogura and Y. Izumi, Japanese Patent, 2012, 254796;
Y. Ogura, Y. Fujishima and Y. Izumi, Japanese Patent, 2013,
211956; Y. Ogura, S. Okamoto, T. Itoi, Y. Fujishima,
Y. Yoshida and Y. Izumi, Chem. Commun., 2014, 50, 3067.
28 R. Yadav and P. S. Fedkiw, J. Electrochem. Soc., 2012, 159, B340.
29 F. Quentel, G. Passard and F. Gloaguen, Chem. - Eur. J.,
2012, 18, 13473.
30 S. Kaneco, H. Kurimoto, Y. Shimizu, K. Ohta and T. Mizuno,
Energy, 1999, 24, 21.
31 N. M. Dimitrijevic, B. K. Vijayan, O. G. Poluektov, T. Rajh,
K. A. Gray, H. He and P. Zapal, J. Am. Chem. Soc., 2011, 133, 3964.
32 N. Zhang, S. Ouyang, T. Kako and J. Ye, Chem. Commun.,
2012, 48, 1269.
14 J. K. Hurst, Science, 2010, 328, 315.
15 H. Gray, Nat. Chem., 2009, 1, 7.
16 N. S. Lewis and D. G. Nocera, Proc. Natl. Acad. Sci. U. S. A.,
2006, 103, 15729.
17 H. Lv, Y. V. Geletii, C. Zhao, J. W. Vickers, G. Zhu, Z. Luo,
J. Song, T. Lian, D. G. Musaev and C. L. Hill, Chem. Soc.
Rev., 2012, 41, 7572.
18 N. Serpone, P. Maruthanuthu, P. Pichat, E. Pelizzetti and
H. Hidaka, J. Photochem. Photobiol., A, 1995, 85, 247.
19 F. Li and X. Duan, Struct. Bonding, 119, 193.
20 K. Teramura, S. Iguchi, Y. Mizuno, T. Shishido and
T. Tanaka, Angew. Chem., Int. Ed., 2012, 51, 8008.
21 M. Morikawa, N. Ahmed, Y. Yoshida and Y. Izumi, Appl.
Catal., B, 2014, 144, 561.
33 S. N. Habisreutinger, L. Schmidt-Mende and J. K. Stolarczyk,
Angew. Chem., Int. Ed., 2013, 52, 7372.
22 N. Ahmed, M. Morikawa and Y. Izumi, Catal. Today, 2012,
185, 263.
34 Y. Yoshida, Y. Mitani, T. Itoi and Y. Izumi, J. Catal., 2012,
287, 190.
23 N. Ahmed, Y. Shibata, T. Taniguchi and Y. Izumi, J. Catal.,
2011, 279, 123.
35 C. Wang, R. L. Thompson, J. Baltrus and C. Matranga,
J. Phys. Chem. Lett., 2010, 1, 48.
24 C. Genevese, C. Ampelli, S. Parathoner and G. Centi,
J. Energy Chem., 2013, 22, 202.
36 Y. T. Liang, B. K. Vijauyan, K. A. Gray and M. C. Hersam,
Nano Lett., 2011, 11, 2865.
This journal is © The Royal Society of Chemistry 2014
Catal. Sci. Technol., 2014, 4, 1644–1651 | 1651