Photocatalytic water splitting and hydrogenation of CO2 in a novel twin photoreactor with IO3-/I- shuttle redox mediator

Article abstract of DOI:10.1016/j.apcata.2015.08.027

One of the major drawbacks in photoreduction of carbon dioxide (CO₂) into hydrocarbons is the high thermodynamic barrier involved, resulting in low photoreduction quantum efficiency (PQE). A novel twin photoreactor system has been developed to enhance the PQE of CO₂ reduction as a solution to this limitation. The twin photoreactor divides the oxygen (O₂)-generating photocatalyst and the dual-function photocatalyst having both CO₂ reduction and hydrogen (H₂) production capabilities in two compartments by a membrane thus preventing the undesired reverse reaction. Herein, the dual-function photocatalyst can hydrogenate CO₂ via the produced H₂, allowing the overall reaction to be more thermodynamically favorable. The charge balance was accomplished by the aid of IO₃^-/I^- redox mediator in the twin photoreactor for shuttling electrons. Two visible-light photocatalysts, Pt/WO₃ (O₂-generating photocatalyst) and GaN:ZnO-Ni/NiO (both CO₂-reducing and H₂-generating photocatalysts) were used. We found out that through the application of the twin photoreactor system under artificial sunlight (AM1.5G 300 W Xenon lamp), the PQE was enhanced more than 4-folds in comparison with that in the single photoreactor, an increase from 0.015% to 0.070%. A possible enhancing mechanism for twin photoreactor, in compare with the conventional single photoreactor, is also proposed based on the knowledge of species present during the photocatalytic reaction.

Full text of DOI:10.1016/j.apcata.2015.08.027

G Model
APCATA-15521; No. of Pages9
ARTICLE IN PRESS
Applied Catalysis A: General xxx (2015) xxx–xxx
Contents lists available at ScienceDirect
Applied Catalysis A: General
journal homepage: www.elsevier.com/locate/apcata
Photocatalytic water splitting and hydrogenation of CO2 in a novel
−
twin photoreactor with IO3 /I− shuttle redox mediator
Sheng-Hung Yua, Cheng-Wei Chiua, Yi-Ting Wua, Chi-Hung Liaoa, Van-Huy Nguyena,b
,
Jeffrey C.S. Wua,∗
a Department of Chemical Engineering, National Taiwan University, Taipei 10617, Taiwan
b Department of Chemical Engineering, National Taiwan University of Science and Technology, Taipei City 106, Taiwan
a r t i c l e i n f o
a b s t r a c t
Article history:
One of the major drawbacks in photoreduction of carbon dioxide (CO2) into hydrocarbons is the high
thermodynamic barrier involved, resulting in low photoreduction quantum efficiency (PQE). A novel
twin photoreactor system has been developed to enhance the PQE of CO2 reduction as a solution to this
limitation. The twin photoreactor divides the oxygen (O2)-generating photocatalyst and the dual-function
photocatalyst having both CO2 reduction and hydrogen (H2) production capabilities in two compartments
by a membrane thus preventing the undesired reverse reaction. Herein, the dual-function photocatalyst
can hydrogenate CO2 via the produced H2, allowing the overall reaction to be more thermodynamically
favorable. The charge balance was accomplished by the aid of IO3−/I− redox mediator in the twin photore-
actor for shuttling electrons. Two visible-light photocatalysts, Pt/WO3 (O2-generating photocatalyst) and
GaN:ZnO–Ni/NiO (both CO2-reducing and H2-generating photocatalysts) were used. We found out that
through the application of the twin photoreactor system under artificial sunlight (AM1.5G 300 W Xenon
lamp), the PQE was enhanced more than 4-folds in comparison with that in the single photoreactor, an
increase from 0.015% to 0.070%. A possible enhancing mechanism for twin photoreactor, in compare with
the conventional single photoreactor, is also proposed based on the knowledge of species present during
the photocatalytic reaction.
Received 17 June 2015
Received in revised form 20 August 2015
Accepted 22 August 2015
Available online xxx
Keywords:
Twin photoreactor
Photoreduction
CO2
Hydrogenation
Water splitting
© 2015 Elsevier B.V. All rights reserved.
1. Introduction
technology has been developed, that is, the use of photocatalyst in
the reduction of CO2 into renewable hydrocarbons such as methane
Global warming and energy crisis have become the most con-
cerned global challenges in the 21st century. Since the industrial
revolution, the emission of carbon dioxide (CO2) into the atmo-
sphere has risen rapidly throughout the years, along with that of
other greenhouse gases, is cited as one of the major causes of global
warming [1]. On the other hand, the usage of conventional fossil
fuels has gradually led to resource depletion and environmental
pollution [2]. Recently, a popular idea towards artificial photosyn-
thesis was proposed which considered CO2 as a potential building
block for the development of alternative energy [3]. If this strategy
were succeeded, sustainable economic and environmental devel-
opment may be realized.
(CH4), methyl formate, methanol (CH3OH), etc. [3,5,6]. It is worth
mentioning that the presence of hydrogen (H2) can significantly
improve the yield of hydrocarbons [7–10]. This is due to the fact that
hydrogenation by H2 is more favorable than by H2O in the reduction
of CO2. However, adding fossil-H2 directly for CO2 hydrogena-
tion is not a practical way because more CO2 would be produced
by reforming of fossil fuels. Looking forward with a more practi-
cal route, to develop a solar-H2 is necessary for providing the H2
needed in CO2 hydrogenation [11].
One of the most promising ideas for solar H2 is to apply Z-scheme
water splitting driven by IO3−/I− shuttle redox mediator [12],
then perform CO2 hydrogenation in the photoreactor system. Our
group recently reported that a twin photoreactor based on the dual
photocatalytic system, which consists of H2- and O2-generating
photocatalysts divided by a membrane. The main advantage of Z-
scheme is that H2 and O2 can be separately produced [13]. As shown
in Scheme 1, in the right-hand side, proton is reduced to hydrogen
by photo-excited electrons and CO2 can be hydrogenated to hydro-
In 1979, Inoue et al. had firstly conducted several CO2 experi-
ments in a slurry batch reactor by using WO3, TiO2, ZnO, CdS, GaP,
and SiC photocatalysts [4]. After years of research, a revolutionized
∗
Corresponding author. Fax: +886 2 23623040.
E-mail address: cswu@ntu.edu.tw (J.C.S. Wu).
carbons at the same time, while I− is oxidized to IO3 by holes. In
−
http://dx.doi.org/10.1016/j.apcata.2015.08.027
0926-860X/© 2015 Elsevier B.V. All rights reserved.
Please cite this article in press as: S.-H. Yu, et al., Appl. Catal. A: Gen. (2015), http://dx.doi.org/10.1016/j.apcata.2015.08.027

G Model
APCATA-15521; No. of Pages9
ARTICLE IN PRESS
2
S.-H. Yu et al. / Applied Catalysis A: General xxx (2015) xxx–xxx
Scheme 1. The diagram of combined H2O water splitting and CO2 photo-hydrogenation by a Z-scheme system.
−
the left-hand side, the photo-excited electrons reduce IO3 back
reaction from H2 and O2 into H2O [21,22]. In brief, it is reasonable
to develop GaN:ZnO–Ni/NiO photocatalyst for CO2 hydrogenation
through overall water splitting. First, H2O is split into H2 and O2 via
the water splitting. Then, the generated H2 is directly used to per-
form the CO2 photo-hydrogenation to design products. In regard
to the photocatalyst for O2-generating, WO3 is found as the most
attractive candidate with high and stable photocatalytic activity
[13]. To further improve the efficiency of photocatalyst, Pt-loaded
WO3, which facilitates the separation of photo-generated electrons
and holes by trapping electrons, is also considered.
This study proposes the combined H2O water splitting and CO2
hydrogenation in the twin photoreactor with IO3−/I− shuttle redox
mediator, which is a highly promising process for photocatalytic
reduction of CO2. Based on the knowledge of species present dur-
ing the photocatalytic reaction, a possible photo-activity enhancing
mechanism for the twin photoreactor is also proposed in compar-
ison with the conventional single photoreactor.
to I−, while water is oxidized to O2 by holes. The proton gener-
ated in the left-hand side is diffused to the right-hand side due
to concentration gradient. The redox mediators, I−/IO3−, are circu-
lated between two sides via the Neosepta membrane. As shown in
Eqs. ((1)–(4)), after the reduction and oxidation of the redox medi-
ators, IO3 and I− are counter-diffused toward the counter sides
−
due to the concentration gradients caused by the photoreaction.
Thus through this cycle, the mass and charge balances are main-
tained when the generations of O2 and hydrocarbons are carried
out simultaneously in each side.
1.1. Oxidation of I−
6H+ + 6e− → 3H2
(1)
(2)
I
− + 3H2O + 6h+ → IO3− + 6H+
−
1.2. Reduction of IO3
2. Experimental
3H2O + 6h+ → 1.5O2 + 6H+
IO3− + 6H+ + 6e− → I− + 3H2O
(3)
(4)
2.1. Preparation of photocatalysts
GaN:ZnO catalyst was prepared by heating a mixture of 2.16 g ␤-
Ga2O3 (Acros, 99.99%) and 1.88 g ZnO (J.T. Baker 99%) at 1125 K for
15 h under NH3 flow [20]. Prior to the nitridation process, ␤-Ga2O3
and ZnO were pulverized in a mortar for 30 min. After the nitri-
dation process, the resulting powder was pulverized for another
30 min. For the final step, GaN:ZnO was calcined under air flow at
873 K for 2 h. GaN:ZnO–Ni/NiO catalyst was prepared by incipient
wetness impregnation as described previously [23]. Firstly, 0.3 g
of GaN:ZnO powder was impregnated with 3–4 mL of de-ionized
water containing 2.5 wt.% of Ni precursor (Ni(NO3)2·6H2O, Nacalai
Tesque Inc.) over an evaporating dish. After that, the slurry was
placed in an oven at 353 K to ensure complete dryness. Lastly, the
resulting powder was reduced with H2 for 2 h at 573 K and then oxi-
dized by air for 1 h at 473 K to form a core-shell structure of Ni/NiO
cluster.
Pt/WO3 catalyst was prepared by a photo-deposition process.
Firstly, H2PtCl6 solution of desired concentration was mixed with
commercial WO3 (Aldrich 99.995%) powder to give a 0.5 wt.%
of Pt-loaded. The well-mixed solution was next irradiated by
UV lamp with an intensity of 5 W/cm2 for 1.5 h to perform the
photo-deposition process. The resulting powder was subsequently
washed by de-ionized water and centrifuged several times to
ensure that there was no residual Cl− remaining on the photocata-
Scheme 2 shows the schematics of twin photoreactor, which
consists of IO3−/I− as shuttle redox mediator. There are several
advantages in the twin photoreactor system. Firstly, CO2 could be
directly hydrogenated to hydrocarbons by the produced H2. Sec-
ondly, the separation of H2 and O2 prevents the backward reaction
of water splitting and the oxidation of the produced hydrocarbons
by O2 back into CO2. Thirdly, with the aid of IO3−/I− shuttle redox
mediator, more electron transfer can be achieved as compared to
the Fe3+/Fe2+ system [13]. Last but not least, the neutral condition of
I−/IO3−, with an anion-exchanged membrane, could dissolve more
CO2 into the reacting solution, leading to higher product yield as
compared to the acidic condition of Fe3+/Fe2+
.
Numerous investigators have reported that the photocatalytic
reduction of CO2 in the presence of H2 proceeds over many simple
oxides such as MgO [7], ZrO2, [14], ZnO [15], and Ga2O3 [16,17]. It is
interesting that (Ga1−xZnx)(N1−xOx), which is typically synthesized
by nitriding a mixture of Ga2O3 and ZnO powders, is also poten-
tially applicable in water splitting [18–20]. In order to enhance
efficiency of overall water splitting on (Ga1−xZnx)(N1−xOx) photo-
catalyst, loading with Ni/NiO has attracted considerable interest
[21]. When Ni particles are reoxidized at 473 K, its shell become
NiO can effectively prevent the loss of products for the reverse
Please cite this article in press as: S.-H. Yu, et al., Appl. Catal. A: Gen. (2015), http://dx.doi.org/10.1016/j.apcata.2015.08.027

G Model
APCATA-15521; No. of Pages9
ARTICLE IN PRESS
S.-H. Yu et al. / Applied Catalysis A: General xxx (2015) xxx–xxx
3
Scheme 2. The diagram of a novel twin photoreactor system.
Table 1
lyst. Lastly, the washed photocatalyst was dried in an oven at 353 K
to obtain Pt/WO3.
The diffusion of iodide and iodate ions through Neosepta membrane.
Ion species
Apparent diffusivity (cm2 s−1
)
Ion transfer rate
under 10 mM conc.
Difference
2.2. Characterization of photocatalysts
(␮mol h−1
)
The crystalline of photocatalysts was analyzed by Rigaku
M03XHF Ultima IV X-ray diffractometer (XRD) in the diffraction
angle (2ꢀ) between 10◦ and 80◦, using Cu-K␣ radiation as the
X-ray source. The X-ray (wavelength ꢁ = 1.5418 Å) tube equipped
with a copper target was operated at 40 mA and 40 kV. The UV–vis
absorption spectrum of photocatalysts was measured by the UV–vis
spectrometer (Varian Cary-100) in the wavelength range between
200 and 800 nm. Field emission scanning electron microscopy
(FE-SEM) was carried out on a JEOL JSM-7000 FE-SEM. The pho-
tocatalysts were sputtered with a thin layer of Pt film to prevent
surface charging. The morphology of Pt/WO3 photocatalyst was
observed by Philips FEI Tecnai F20 G2 high-resolution transmis-
sion electron microscope (HR-TEM) at an acceleration voltage of
200 kV. The specimen was prepared by ultrasonically dispersing the
sample into ethanol, depositing droplets of the suspensions onto a
copper mesh (200 meshes), and keeping 24 h in a vacuum dry oven
(DOV30). For surface analysis, X-ray photoelectron spectroscopy
(XPS) was conducted to identify the composition and chemical sta-
tus of photocatalysts by the Theta Probe XPS (Thermo Scientific)
with Mg target as the X-ray source of 1253.6 eV.
I−
IO3
9.0 × 10−7
7.4 × 10−7
24.77
20.37
−
2.4. CO2 photoreduction in a twin and a single batch
photoreactor system
For the CO2 photoreduction reaction in the twin photoreactor,
Pt/WO3 (0.3 g) and GaN:ZnO–Ni/NiO (0.3 g) photocatalysts were
placed in the counterpart compartments separated by the circular
Neosepta membrane (TAIYEN Co.). Then, de-ionized water (200 mL)
combined with NaI (15 mM) was added in each compartment, as
shown in Scheme 2. That the reason 15 mM of I− was selected
to carry out the subsequent CO2 reduction is mentioned in the
Supplementary material. Prior to the photocatalytic reaction, the
left and right compartments were purged with ultra-pure Ar gas
and ultra-pure CO2 for 30 min, respectively, to remove residual air
in both compartments and to introduce CO2 into the right com-
partment for the reaction. During the photo-reaction, the solution
in each compartment was stirred and irradiated with an artificial
sunlight (AM1.5G 300 W Xenon lamp, 100 mW cm−2) that has the
same power and spectral distribution of the sun at 48.5◦ zenith
angle.
2.3. Counter-diffusion of Neosepta membrane
The produced O2 and H2 were collected separately in each com-
partment every 1 h with an on-line sampling loop (1 mL) connected
to a thermal conductivity detector (China GC TCD 2000) for analysis.
A molecular sieve 5 A packed column of length 3.5 m was installed,
and ultra-pure Ar was used as carrier gas in the GC-TCD for high
sensitivity of H2. For the detection of hydrocarbons in the right-
hand compartment, liquid samples (1–10 ␮L) were withdrawn and
injected into the flame ionization detector (China GC FID 9800) after
filtering the photocatalyst. To further improve the CO detection
limit, a methanizer packed with Ni catalyst was connected to GC-
FID to convert CO into CH4 with H2 at 633 K. A Porapak Q column
of length 2 m was installed, and pure N2 was used as the carrier gas
for FID in the detection of hydrocarbons.
The Neosepta membrane (TAIYEN Company, Taiwan) was used
in this experiment as an anion-exchanged membrane for the
counter-diffusion of I− and IO3− ions in the twin photoreactor. The
membrane was pretreated by immersing it into the same concen-
tration of NaI solution under which the photocatalytic reaction was
performed. A concentration of 10 mM NaI or NaIO3 solution was
placed in the counterpart compartment of the twin photoreactor
separated by Neosepta membrane to evaluate the ion transfer rates
and the diffusivities of I− and IO3−. The concentrations of I− and
−
IO3 were measured in each compartment with ion chromatog-
raphy (ICS-1000) after several hours of diffusion, and the results
of ion transfer rates and diffusivities were summarized in Table 1.
The transfer rates of I−/IO3 redox mediators through the mem-
−
For comparison, the CO2 photoreduction reaction was also
performed in a single photoreactor. At first, GaN:ZnO–Ni/NiO
(0.3 g) photocatalyst was placed in a quartz photoreactor filled
in de-ionized water (180 mL) with 0.2 M NaOH. Then, the solu-
brane were found to be much higher than the reaction rate of the
CO2 reduction, suggesting that the reaction rate would not be con-
strained during the photocatalytic reaction.
Please cite this article in press as: S.-H. Yu, et al., Appl. Catal. A: Gen. (2015), http://dx.doi.org/10.1016/j.apcata.2015.08.027

G Model
APCATA-15521; No. of Pages9
ARTICLE IN PRESS
4
S.-H. Yu et al. / Applied Catalysis A: General xxx (2015) xxx–xxx
Fig. 2. UV–vis spectra of photocatalysts: (a) ZnO, Ga2O3, GaN:ZnO, and
GaN:ZnO–Ni/NiO; and (b) WO3, and Pt/WO3.
3. Results and discussion
3.1. Characterization of photocatalyst
Fig. 1 depicts the X-ray diffraction (XRD) patterns of all photo-
catalysts and their references. In details, Fig. 1a reveals patterns of
the prepared GaN:ZnO sample and its precursors of ␤-Ga2O3 and
ZnO. Clearly, the position of the d (1 0 0) diffraction peak of GaN:ZnO
shifted slightly to higher angles (2-theta), in comparison with that
of ZnO. Moreover, no peaks can be assigned to either Ga2O3 or ZnO,
indicating that complete nitridation of the precursors was achieved
by nitriding a mixture of ␤-Ga2O3 and ZnO at 1125 K for 15 h. This
phenomenon is consistent with that observed in previous stud-
ies [19,20]. There is no noticeable change in their XRD patterns
before and after loading the Ni/NiO as co-catalysts, (Fig. 1b), sug-
gesting that Ni/NiO co-catalysts did not change the structure of
bulk. However, the presence of Ni/NiO co-catalyst slightly reduces
the GaN:ZnO crystallites. All the peaks of GaN:ZnO have inten-
sity higher than that of GaN:ZnO–Ni/NiO. Additionally, we also
observed no change in the XRD pattern of GaN:ZnO–Ni/NiO after
8 h of reaction, indicating that this photocatalyst was essentially
stable in the reaction. Fig. 1c shows that the XRD pattern of WO3
which is similar to that of orthorhombic WO3 (JCPDS file 20-1324)
[13]. Obviously, the peaks of Pt are not detected, showing that Pt
loading is very small and highly dispersed on the WO3 surface.
Fig. 1. XRD patterns of photocatalysts: (a) GaN:ZnO in compared with starting
materials of ZnO and Ga2O3; (b) GaN:ZnO in compared with fresh and spent
GaN:ZnO–Ni/NiO; (c) commercial WO3 and Pt/WO3.
tion was purged with pure CO2 for 30 min before performing the
photocatalytic reaction under visible-light (300W Xenon lamp,
270 mW cm−2). The liquid samples (1–10 ␮L) and gas samples
(0.5–1 mL) were analyzed by GC-FID for the detection of hydro-
carbons in both aqueous and gas phases.
Please cite this article in press as: S.-H. Yu, et al., Appl. Catal. A: Gen. (2015), http://dx.doi.org/10.1016/j.apcata.2015.08.027

G Model
APCATA-15521; No. of Pages9
ARTICLE IN PRESS
S.-H. Yu et al. / Applied Catalysis A: General xxx (2015) xxx–xxx
5
Fig. 4. HR-TEM image of Pt/WO3 photocatalyst.
To evaluate the morphology of GaN:ZnO and WO3 photocata-
lysts, the FE-SEM analysis was performed, as shown in Fig. 3. The
particles of GaN:ZnO sample are uniform, and the average particle
size is around 400–600 nm. On the other hand, the mean parti-
cle size of WO3 is about 200–400 nm. The morphology of Pt/WO3
photocatalyst was further characterized by transmission electron
microscopy (TEM). Fig. 4 clearly shows that a number of tiny Pt
nano-particles are well dispersed on the WO3 surface. Further-
more, the Pt nano-particles have spherical or hemispherical shape;
their particle size is about 4–7 nm. The lattice fringe (inset, Fig. 4)
with d = 0.23 nm is corresponded to Pt (1 1 1) (JCPDS file 65-2868,
d = 0.227 nm). The TEM observation is in accordance with the XRD
results.
Further evidence of co-catalysts loaded on the prepared photo-
catalysts was observed through the XPS analysis. Fig. 5a exhibits
the Ni 2p spectrum of Ni/NiO-loaded GaN:ZnO. After curve fitting,
the spectrum of Ni 2p can be divided into two pairs of peaks. In
details, peaks at 858.0 and 874.6 eV are corresponding to Ni 2p3/2
and Ni 2p1/2, respectively, while peaks at 867.8 and 880.4 eV are
assigned to Ni satellite [25]. Fig. 5b reveals the Pt 4f spectrum of
Pt-loaded WO3. As expected, the principal peaks are attributed to
Pt0 at 70.9 eV (4f7/2) and 74.2 eV (4f5/2) [26]. It suggests that all Pt
ions are completely reduced by the photo-deposition process.
Fig. 3. FE-SEM images of photocatalysts: (a) GaN:ZnO and (b) WO3.
Regarding the light harvesting, the optimal photocatalyst is
expected to absorb light efficiently over the whole wavelength
range of the solar spectrum. Fig. 2 shows the UV–vis absorp-
tion spectra of all samples, such as Ga2O3, ZnO, GaN:ZnO,
GaN:ZnO–Ni/NiO, WO3 and Pt/WO3 photocatalysts. Obviously, the
absorption edge of GaN:ZnO after the nitridation process is at
longer wavelengths than those of starting materials Ga2O3 and
ZnO, which are only active in the ultraviolet region. As suggested
from the previous study [18], the visible light response of GaN:ZnO
comes from the repulsion between N2p and Zn3d orbitals. It is
noted that both two photocatalysts GaN:ZnO–Ni/NiO and Pt/WO3
are active in the visible-light region up to 500 and 480 nm, respec-
tively. With the presence of Ni/NiO, a slightly shift of the absorption
edge to longer wavelengths with a gradual increase in absorption
can be observed. On the other hand, the significant red-shift in
absorption edge over Pt/WO3 is attributed to the Pt loading. The
scattering of photons by crystal defects created upon metal-loaded
and the free carrier absorption of photons contribute to the shift of
the absorption edge and the enhancement of absorption [24].
3.2. Photocatalytic CO2 reduction
3.2.1. Blank test
Table 2 shows the result of blank experiments, which are con-
ducted before performing the photocatalytic reaction to prove that
the formation of the product (hydrocarbons, H2) actually comes
from photoreduction and not from CO2 contaminations or the pho-
tocatalyst itself. Argon was purged into the reacting solution to
bring out any impurity contained within. Note that a carbon source
is needed to be reduced into hydrocarbons. There are two possi-
ble sources: (1) the impurity from the reaction system, and (2) the
carbon residue on the photocatalyst. For the first blank experiment
(Entry 1, Table 2), CO2 was not purged into the reacting system,
resulting in no product generated even with the presence of the
photocatalyst and light irradiation. This proves that both the reac-
tion system and the photocatalyst contain a minimal contamination
Please cite this article in press as: S.-H. Yu, et al., Appl. Catal. A: Gen. (2015), http://dx.doi.org/10.1016/j.apcata.2015.08.027

G Model
APCATA-15521; No. of Pages9
ARTICLE IN PRESS
6
S.-H. Yu et al. / Applied Catalysis A: General xxx (2015) xxx–xxx
Fig. 5. XPS spectra: (a) Ni 2p of Ni/NiO loaded on GaN:ZnO and (b) Pt 4f of Pt/WO3 photocatalysts.
Table 2
The result of blank experiment for slurry batch reaction.
Entry
Experimental conditions
Products (␮mol gcat−1 h−1
)
Ar
CO2
Catalyst
Light source
1
2
3
O
O
O
X
O
O
O
X
O
O
O
X
BDL
BDL
BDL
Light source: 300 W Xenon lamp 270 mW cm−2
.
Abbreviations: X—absent in the photoreactor; O—present in the photoreactor; BDL—below detection limit of gas chromatography.
Table 3
ically, while both CO and CH4 yields increase initially in the first
2–3 h, and then decrease dramatically as reaction time goes on.
This can be understood by considering the back reaction of O2. Due
to the presence of O2 producing during the photocatalytic reac-
tion, CH4 can be easily oxidized to CO, resulting in decrease the
CH4 yield after 2–3 h in reaction (Fig. 6c). One should note that
CO might be converted into HCHO and CH3OH (Eqs. (9) and (10),
Table 3) to promote their product yields. Meanwhile, CO can be
more readily reacted with O2 to form CO2 (Eq. (8), Table 3). It is
no surprise that this exothermic oxidation, which exhibits negative
ꢂH◦ value (–283.2 kJ mol−1), is associated with highly negative ꢂG◦
value (–257.3 kJ mol−1) and, is highly thermodynamically favor-
able. Therefore, the CO yield dropped dramatically as reaction time
goes on (Fig. 6b). Based on the knowledge of species present during
the photo-reaction, a possible reaction pathway for single reac-
tion is proposed and illustrated in Scheme 3. We suggest that CO2
can be directly hydrogenated into several possible products, such
as CH3OH, and CH4. On the other hand, CO can be directly pro-
duced by catalytic oxidation of CH4. As a result, CO will react with
H2O and H2 to produce HCHO and further favor the CH3OH yield,
respectively [28]. However, CO is easily converted to CO2 by react-
ing with O2, resulting in the decrease CO yield after 3 h in reaction
(Eq. (10), Table 3). In short summary, GaN:ZnO–Ni/NiO photocat-
alyst successfully performs the production of H2 and reduction of
CO2 simultaneously with an efficient catalytic activity.
The Enthalpy (ꢂH◦) and Gibbs free energy (ꢂG◦) values changes in some reactions.
Eqs. Reactions
ꢂH◦ (kJ mol−1
)
ꢂG◦ (kJ mol−1
)
(5)
(6)
(7)
(8)
(9)
CO2 (g) + 3H2 (g) → CH3OH (l) + H2O (l) −137.8
−10.7
−132.4
−86.5
−257.3
27.3
CO2 (g) + 4H2 (g) → CH4 (g) + H2O (l)
CH4 (g) + ½ O2 (g) → CO (g) + 2H2 (g)
2CO (g) + O2 (l) → 2CO2 (g)
−259.9
−59.7
−283.2
−5.5
CO (g) + H2 (l) → HCHO (l)
(10) CO (g) + 2H2 (g) → CH3OH (l)
Aspen was used in calculating the ꢂH◦ and ꢂG◦ data [27].
−131.6
−29.9
that is too small to be detected by GC. Additionally, the photocat-
alyst itself contains nearly no carbon residue because its synthesis
process did not use any carbon-containing precursor. There are also
three essential elements in performing the photocatalytic reaction,
including (1) CO2, (2) photocatalyst and (3) light source. Clearly,
with any of these parts missing, no product could be detected.
3.2.2. Single photoreactor
Photocatalysts were first screened for their photocatalytic
CO2 reduction activity by using a single photoreactor under
visible-light (300 W Xenon lamp, 270 mW cm−2). We found that
GaN:ZnO–Ni/NiO was a dual-function photocatalyst with both
CO2-reducing and H2-generating capability. In details, CH3OH was
found as the main product. A total of 18.96, 2.08, 13.44, 1.87 and
0.43 ␮mol gcat−1 of H2, CH4, CH3OH, formaldehyde (HCHO) and CO,
respectively, were produced over 6 h of photo-reaction. Table 3
summaries some possible reaction paths with their ꢂH◦ and ꢂG◦
values involved in CO2 hydrogenation [27]. Most exothermic reac-
tions (ꢂH◦ < 0) are thermodynamically favorable with negative
ꢂG◦ values. The correlation between the amount of H2, CO, and CH4
produced and the reaction over GaN:ZnO–Ni/NiO photocatalyst
are presented in Fig. 6. The photocatalyst produces H2 monoton-
3.2.3. Twin photoreactor
Due to the weather conditions changed all the time (even during
a sunny day). It is impossible to have consistent results under real-
istic sunlight illumination. Therefore, photocatalytic CO2 reduction
in the twin photoreactor was performed under artificial sunlight,
which was provided by the 300-W xenon lamp integrating with
an AM1.5G filter. The light intensity in front of the twin photore-
Please cite this article in press as: S.-H. Yu, et al., Appl. Catal. A: Gen. (2015), http://dx.doi.org/10.1016/j.apcata.2015.08.027

G Model
APCATA-15521; No. of Pages9
ARTICLE IN PRESS
S.-H. Yu et al. / Applied Catalysis A: General xxx (2015) xxx–xxx
7
Fig. 6. The time-dependent yields of various products over GaN:ZnO–Ni/NiO photocatalyst under visible-light 300 W Xenon lamp in the single photoreactor: (a) H2 evolution,
(b) CO evolution, and (c) CH4 evolution.
Scheme 3. The proposed diagram and reaction pathways of combined water-splitting/photo-hydrogenation over GaN:ZnO–Ni/NiO photocatalyst under visible-light 300 W
Xenon lamp in the single photoreactor.
Fig. 7. The time-dependent yields of various products over GaN:ZnO–Ni/NiO photocatalyst under artificial sunlight AM1.5G 300 W Xenon lamp in the twin photoreactor: (a)
O2 evolution in left-hand compartment, (b) H2 evolution, MeOH formation and H2 theoretical evolution in right-hand compartment.
actor was approximately 100 mW cm−2, which was similar to the
power and spectral distribution of the realistic illumination of sun
at 48.5◦ zenith angle. In the twin photoreactor system, both water
splitting and CO2 reduction reactions occur simultaneously. The
advantage of using the twin photoreactor is that the produced O2
and H2–hydrocarbon products could be separated into two differ-
ent compartments that can prevent either the backward reaction
between H2 and O2 or the oxygenation of hydrocarbon products. As
a result, the generated H2 is effectively used to photo-hydrogenate
CO2 into design products under mild conditions. Fig. 7 shows the
time-dependent yields of various products over GaN:ZnO–Ni/NiO
photocatalyst in the twin photoreactor under AM1.5G 300 W Xenon
lamp. In the left-hand compartment, th−e1amount of O2 produced
after 8 h of reaction is 66.96 ␮mol gcat
(Fig. 7a). In the right-
Please cite this article in press as: S.-H. Yu, et al., Appl. Catal. A: Gen. (2015), http://dx.doi.org/10.1016/j.apcata.2015.08.027

G Model
APCATA-15521; No. of Pages9
ARTICLE IN PRESS
S.-H. Yu et al. / Applied Catalysis A: General xxx (2015) xxx–xxx
Table 4
8
hand compartment, the amount of CH3OH and H2 produced after
8 h of reaction is 59.7 and 1.6 ␮mol gcat−1, respectively (Fig. 7b).
Theoretically, the amount of H2 and O2 produced in pure water
splitting have a ratio of two-to-one [29]. Therefore, the theoret-
ical amount of H2 should be 134.08 ␮mol gcat−1. However, only
1.6 ␮mol gcat−1 of H2 was observed in the right-hand compartment
(Fig. 7b). It suggests that CH3OH is mainly formed via the direct
hydrogenation of CO2. It −is1 noted that a required amount of H2
to produce 59.7 ␮mol gcat of CH3OH by direct hydrogenation is
Performance comparison of the twin photoreactor and single photoreactor.
Conditions and
results
Single photoreactor
Twin photoreactor
Hydrogen
3.16
0.35
2.24
0.31
0.07
73.6
0.015
0.20
ND
(␮mol gcat−1 h−1
Methane
)
)
)
)
)
(␮mol gcat−1 h−1
Methanol
7.46
ND
(␮mol gcat−1 h−1
Formaldehyde
(␮mol gcat−1 h−1
CO
−1
179.1 ␮mol gcat which is greater than the calculated theoretical
amount of H2 (134.08 ␮mol gcat−1). Thus, some of the CH3OH is not
fully produced through the direct hydrogenation, but is also pro-
duced by the reduction of CO2 with H2O forming hydrocarbons [30].
Importantly, CH3OH was found to be the major product. The rea-
son is that the photo-hydrogenation of CO2 to form CH4 requires
8 photoelectrons, which is higher than that to form CH3OH (6
photoelectrons required). In our study, visible-light (300 W Xenon
lamp, 270 mW cm−2), which is used to conduct single reaction,
can produce more electron–hole pairs on the photocatalysts than
the artificial sunlight (AM1.5G 300 W Xenon lamp, 100 mW cm−2).
Hence, as we observed, CH3OH was the main product when exper-
iment was conducted under artificial sunlight. The result is in
good agreement with previous study [31]. The absence of CH4
pathway in current condition is also attributed to the missing of
HCHO.
ND
(␮mol gcat−1 h−1
Selectivity for CO2
photoreduction (%)
PQE (%)
99.1
0.070
Light sources
300 W Xenon lamp: 270 mW cm−2 AM1.5G 300 W
Xenon lamp:
100 mW cm−2
Note: ND—not detected; PQE—photoreduction quantum efficiency; the product
(␮mol gcat−1 h−1) is an average rate of batch reaction in 6 or 8 h.
dual-function visible-light photocatalyst, which provides the abil-
ity to reduce CO2 via the produced H2. The photocatalytic reaction
is enhanced because hydrogenation of CO2 is thermodynamically
favorable and is a spontaneous reaction. Last but not least, IO3−/I−
was incorporated in the twin photoreactor for the first time. Due to
the shuttle redox mediator, the energy barrier for water splitting
and reduction of CO2 was diminished, resulting in a higher PQE.
3.2.4. Comparison of efficiency in different photoreactors
The photoreduction efficiency is compared between different
photoreactors by a term called the photoreduction quantum effi-
ciency (PQE) which was evaluated in details previously [13].
Photoreduction quantum efficiency (%) = 100%
× (number of moles of photoelectrons required for reduction
4. Conclusions
× product formation rate)/incident photon rate
(11)
In summary, H2O water-splitting and CO2 photo-hydrogenation
are successfully explored and adopted by using the twin pho-
toreactor system. Our present work has demonstrated that
GaN:ZnO–Ni/NiO and Pt/WO3 photocatalysts can significantly
enhance the efficient photocatalytic water splitting and hydrogena-
tion. GaN:ZnO–Ni/NiO photocatalyst reduces CO2 via the produced
H2 while Pt/WO3 photocatalyst produces O2 simultaneously in
each compartment of twin photoreactor. Additionally, IO3−/I− is
incorporated in the twin photoreactor for the first time to bal-
ance charge. Applying twin photoreactor can improve significantly
PQE more than 4 folds from 0.015% to 0.070%, in compared with
single photoreactor, with the same photocatalyst. Moreover, the
selectivity for CO2 reduction in the twin photoreactor reaches to
99.1% which is higher than that in the single photoreactor (73.6%).
Despite the prominent results achieved, there is a need for further
improving the efficiency of photocatalytic CO2 hydrogenation. The
bottleneck is obviously the limited efficiency of the H2-generating.
Therefore, the goal toward solving global warming and energy
crisis will be one step closer if a high-efficient photocatalyst is
discovered in the future and applied to the twin photoreactor sys-
tem.
To consider an competition between the reduction of H2O to
H2 and CO2 to CO, HCHO, CH3OH and CH4; the selectivity for CO2
reduction based on a photoelectron basis is defined as following
[32].
Selectivity for CO2 reduction (%) = 100%
× (No. of photoelectrons required × CO2 reduction products)/
×(No. of photoelectrons required
× CO2 reduction products + 2 × hydrogen)
(12)
The product yields, selectivity for CO2 reduction and PQE of the
single photoreactor and the twin photoreactor are summarized
in Table 4. The result clearly demonstrates that twin photo-
photoreactor can significantly promote the photo-hydrogenation
of CO2, even the light intensity of single reactor (270 mW cm−2) is
higher than that of the twin reactor (100 mW cm−2). The selectivity
for CO2 reduction reaches to 99.1%. We believe that the bottleneck is
the limited efficiency of the H2-generating; therefore, how to pref-
erentiallyacceleratethe producingH2 may be a key point for further
improving the activity of CO2 photo-hydrogenation. Remarkably,
the PQE of the twin photoreactor is improved more than 4 folds
comparing to that of the single photoreactor (an enhancement from
0.015% to 0.070%). Such significant improvement is attributed to
three main reasons. Firstly, the twin photoreactor separates the O2
and H2 produced in each compartment, thus prevents the oxida-
tion reaction of O2 with the hydrocarbons generated. Furthermore,
the H2 is used immediately for the hydrogenation of CO2. Secondly,
GaN:ZnO–Ni/NiO applied in the twin photoreactor system acts as a
Acknowledgements
This work was financially supported by the Taiwan Ministry
of Science and Technology under the project NSC 101-2621-M-
002-012, and was also financially supported by the NSFC (Grant
U1305242). In addition, we appreciated the Taiyen Tongsiao Fac-
tory for providing Neosepta membrane.
Please cite this article in press as: S.-H. Yu, et al., Appl. Catal. A: Gen. (2015), http://dx.doi.org/10.1016/j.apcata.2015.08.027

G Model
APCATA-15521; No. of Pages9
ARTICLE IN PRESS
S.-H. Yu et al. / Applied Catalysis A: General xxx (2015) xxx–xxx
9
Appendix A. Supplementary data
[17] K. Teramura, H. Tsuneoka, T. Shishido, T. Tanaka, Chem. Phys. Lett. 467 (2008)
191–194.
[18] K. Maeda, K. Teramura, T. Takata, M. Hara, N. Saito, K. Toda, Y. Inoue, H.
Kobayashi, K. Domen, J. Phys. Chem. B 109 (2005) 20504–20510.
[19] K. Maeda, K. Domen, Chem. Mater. 22 (2010) 612–623.
[20] K. Maeda, T. Takata, M. Hara, N. Saito, Y. Inoue, H. Kobayashi, K. Domen, J. Am.
Chem. Soc. 127 (2005) 8286–8287.
Supplementary data associated with this article can be found,
in the online version, at http://dx.doi.org/10.1016/j.apcata.2015.08.
027.
[21] K. Maeda, K. Teramura, K. Domen, Catal. Surv. Asia 11 (2007) 145–157.
[22] T. Takata, K. Shinohara, A. Tanaka, M. Hara, J.N. Kondo, K. Domen, J.
Photochem. Photobiol. A 106 (1997) 45–49.
References
[23] C.-H. Liao, C.-W. Huang, J.C.S. Wu, Catalysts 2 (2012) 490–516.
[24] Y.-T. Wu, Y.-H. Yu, V.-H. Nguyen, K.-T. Lu, J.C.-S. Wu, L.-M. Chang, C.-W. Kuo, J.
Hazard. Mater. 262 (2013) 717–725.
[1] A. Creamer, B. Gao, Overview of Greenhouse Gases and Global Warming,
Carbon Dioxide Capture: An Effective Way to Combat Global Warming,
Springer International Publishing, 2015, pp. 1–15.
[25] M.A. Peck, M.A. Langell, Chem. Mater. 24 (2012) 4483–4490.
[26] C.D. Wagner, G.E. Muilenberg, Handbook of X-ray Photoelectron
Spectroscopy: a Reference Book of Standard Data for Use in X-ray
Photoelectron Spectroscopy, PerkinElmer Corp., Physical Electronics Division,
Eden Prairie, Minn, 1979.
[27] Xiaoding, J.A. Moulijn, Energy Fuels 10 (1996) 305–325.
[28] Y.-H. Cheng, V.-H. Nguyen, H.-Y. Chan, J.C.S. Wu, W.-H. Wang, Appl. Energy
147 (2015) 318–324.
[29] C.-C. Lo, C.-W. Huang, C.-H. Liao, J.C.S. Wu, Int. J. Hydrogen Energy 35 (2010)
1523–1529.
[30] H. Yamashita, Y. Fujii, Y. Ichihashi, S.G. Zhang, K. Ikeue, D.R. Park, K. Koyano, T.
Tatsumi, M. Anpo, Catal. Today 45 (1998) 221–227.
[2] N. Armaroli, V. Balzani, Angew. Chem. Int. Ed. 46 (2007) 52–66.
[3] C.S. Jeffrey, Developments and Innovation in Carbon Dioxide (CO2) Capture
and Storage Technology, in: M.M. Maroto-Valer (Ed.), Woodhead Publishing,
2010, pp. 463–501.
[4] T. Inoue, A. Fujishima, S. Konishi, K. Honda, Nature 277 (1979) 637–638.
[5] K. Li, X. An, K.H. Park, M. Khraisheh, J. Tang, Catal. Today 224 (2014) 3–12.
[6] A.D. Handoko, K. Li, J. Tang, Curr. Opin. Chem. Eng. 2 (2013) 200–206.
[7] Y. Kohno, H. Ishikawa, T. Tanaka, T. Funabiki, S. Yoshida, Phys. Chem. Chem.
Phys. 3 (2001) 1108–1113.
[8] Y. Kohno, T. Tanaka, T. Funabiki, S. Yoshida, Phys. Chem. Chem. Phys. 2 (2000)
2635–2639.
[9] Y. Kohno, T. Tanaka, T. Funabiki, S. Yoshida, Chem. Commun. (1997) 841–842.
[10] K. Teramura, T. Tanaka, H. Ishikawa, Y. Kohno, T. Funabiki, J. Phys. Chem. B
108 (2004) 346–354.
[31] P.-Y. Liou, S.-C. Chen, J.C.S. Wu, D. Liu, S. Mackintosh, M. Maroto-Valer, R.
Linforth, Energy Environ. Sci. 4 (2011) 1487–1494.
[32] Q. Zhai, S. Xie, W. Fan, Q. Zhang, Y. Wang, W. Deng, Y. Wang, Angew. Chem.
Int. Ed. 52 (2013) 5776–5779.
[11] H. Ahmad, S.K. Kamarudin, L.J. Minggu, M. Kassim, Renew. Sust. Energy Rev.
43 (2015) 599–610.
[12] R. Abe, K. Sayama, K. Domen, H. Arakawa, Chem. Phys. Lett. 344 (2001)
339–344.
[13] W.-H. Lee, C.-H. Liao, M.-F. Tsai, C.-W. Huang, J.C.S. Wu, Appl. Catal. B 132–133
(2013) 445–451.
[14] Y. Kohno, T. Tanaka, T. Funabiki, S. Yoshida, Phys. Chem. Chem. Phys. 2 (2000)
5302–5307.
[15] S. Ichikawa, Energy Convers. Manage. 36 (1995) 613–616.
[16] K. Teramura, S.-i. Okuoka, H. Tsuneoka, T. Shishido, T. Tanaka, Appl. Catal. B 96
(2010) 565–568.
Please cite this article in press as: S.-H. Yu, et al., Appl. Catal. A: Gen. (2015), http://dx.doi.org/10.1016/j.apcata.2015.08.027