Facet effect of Pd cocatalyst on photocatalytic CO2 reduction over g-C3N4

Article abstract of DOI:10.1016/j.jcat.2017.02.005

The separation and transfer of charge carriers, adsorption of CO₂ molecules, and desorption of product molecules are crucial factors that affect the CO₂ photoreduction process. Herein, we demonstrate the significant facet effect of Pd cocatalyst toward CO₂ photoreduction over graphitic carbon nitride (g-C₃N₄). The surface atomic structure of Pd cocatalyst can be precisely controlled by adjusting the amount of {1?1?1} and {1?0?0} facets to modulate the interfacial charge carrier transfer, CO₂ adsorption and CH₃OH desorption. It is shown that the tetrahedral Pd nanocrystals with exposed {1?1?1} facets function as a more efficient cocatalyst as compared to cubic Pd nanocrystals with exposed {1?0?0} facets, which is reflected by enhancing the CO₂ photoreduction over graphitic carbon nitride. The origin of such remarkable shape-induced effect is explained on the basis of experimental studies of charge transfer dynamics and the atomic-scale DFT modeling of CO₂ adsorption and CH₃OH desorption.

Full text of DOI:10.1016/j.jcat.2017.02.005

Journal of Catalysis 349 (2017) 208–217
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Facet effect of Pd cocatalyst on photocatalytic CO reduction over g-C N
2
3 4
a
a
a
b
a,c,
⇑
Shaowen Cao , Yao Li , Bicheng Zhu , Mietek Jaroniec , Jiaguo Yu
a
State Key Laboratory of Advanced Technology for Materials Synthesis and Processing, Wuhan University of Technology, Wuhan 430070, PR China
Department of Chemistry and Biochemistry, Kent State University, Kent, OH 44242, USA
Department of Physics, Faculty of Science, King Abdulaziz University, Jeddah 21589, Saudi Arabia
b
c
a r t i c l e i n f o
a b s t r a c t
Article history:
The separation and transfer of charge carriers, adsorption of CO2 molecules, and desorption of product
molecules are crucial factors that affect the CO photoreduction process. Herein, we demonstrate the sig-
nificant facet effect of Pd cocatalyst toward CO photoreduction over graphitic carbon nitride (g-C ).
The surface atomic structure of Pd cocatalyst can be precisely controlled by adjusting the amount of
111} and {100} facets to modulate the interfacial charge carrier transfer, CO adsorption and CH OH
Received 24 December 2016
Revised 4 February 2017
Accepted 6 February 2017
2
2
3 4
N
{
2
3
desorption. It is shown that the tetrahedral Pd nanocrystals with exposed {111} facets function as a more
efficient cocatalyst as compared to cubic Pd nanocrystals with exposed {100} facets, which is reflected by
Keywords:
CO photoreduction
2
Pd cocatalyst
Surface structure
Charge transfer
enhancing the CO
induced effect is explained on the basis of experimental studies of charge transfer dynamics and the
atomic-scale DFT modeling of CO adsorption and CH OH desorption.
Ó 2017 Elsevier Inc. All rights reserved.
2
photoreduction over graphitic carbon nitride. The origin of such remarkable shape-
2
3
1
. Introduction
energy and reducing the overpotential [18–20]. However, these
metal nanoparticles are normally deposited on the surface of pho-
tocatalysts by in-situ photodeposition or impregnation methods,
leading to the formation of an irregularly spherical shape. Hence
it is challenging to precisely probe the effects of surface atomic
structure of metal cocatalysts on the photocatalytic activity.
Importantly, the surface atomic structure of metal cocatalysts is
expected to have a significant impact on the molecular absorption
and charge separation in photocatalytic reactions [21–23]. There-
fore, it is indispensable to investigate the shape effects of metal
cocatalysts by rational methods for designing a highly efficient
photocatalytic system.
The rapid consumption of fossil fuels has led to a significant
2
increase of emitted carbon dioxide (CO ) into the atmosphere, con-
sequently causing not only the global energy shortage but also a
serious greenhouse effect [1–3]. To balance the global carbon cycle
and slow down the energy crisis, the conversion of CO
tainable hydrocarbon fuels is a potentially promising strategy [4–
]. However, the high thermodynamic stability of the CO molecule
2
into sus-
6
2
requires a high input of energy and efficient catalysts to break the
C@O bond during the conversion process [7]. Therefore, the photo-
catalytic reduction of CO into hydrocarbon fuels with utilization
2
of renewable solar energy has emerged as an effective means of
low energy cost [8–10].
3 4
Graphitic carbon nitride (g-C N ) is a promising photocatalyst
with high stability, appropriate electronic structure and excellent
light absorption ability [24,25]. The conduction band position of
Metal nanoparticles are frequently used as efficient catalysts in
many important reactions [11–13]. The surface facets of metal
nanoparticles strongly affect the catalytic activity due to the
diverse arrangement of surface atoms, which usually serve as pre-
dominant active sites [14–17]. In photocatalysis, metal nanoparti-
cles act as cocatalysts to enhance the performance of
g-C
CO
attracted a lot of attention in the field of photocatalytic CO
3
N
4
is much more negative than the reduction potentials of
to various hydrocarbon fuels, thus in recent years g-C has
reduc-
2
3 4
N
2
tion [26,27]. Very recently, density functional theory (DFT) calcula-
tions have revealed that a strong interaction could occur between
Pd atoms and the lone-pair electrons of the neighbouring pyridinic
photocatalysts for reactions such as H
2 2
evolution and CO reduc-
tion, by facilitating the charge separation, lowering the activation
nitrogen atoms of g-C
as the model photocatalyst and demonstrate a remarkable
shape-induced effect of Pd cocatalyst on g-C photocatalyst
towards photocatalytic CO reduction. Cubic and tetrahedral Pd
nanoparticles with different exposed facets can be precisely syn-
thesized and ex-situ deposited onto the surface of g-C via elec-
3 4
N [28]. Hence, in this work, we employ g-
3 4
C N
3 4
N
⇑
Corresponding author at: State Key Laboratory of Advanced Technology for
Materials Synthesis and Processing, Wuhan University of Technology, Wuhan
2
4
30070, PR China. Fax: +86 27 87879468.
3 4
N
E-mail address: jiaguoyu@yahoo.com (J. Yu).
http://dx.doi.org/10.1016/j.jcat.2017.02.005
021-9517/Ó 2017 Elsevier Inc. All rights reserved.
0

S. Cao et al. / Journal of Catalysis 349 (2017) 208–217
209
trostatic assembly. The photocatalytic CO
2
reduction performance
is found to be highly sensitive to the surface atomic structure of
Pd nanoparticles due to the different charge transfer ability,
adsorption energy of CO
OH molecules.
2 3
molecules and desorption energy of CH -
2
. Experimental and computational methods
2.1. Materials
Urea (98.0%), Na
pyrrolidone (PVP, MW ꢀ 1,300,000),
2
PdCl
4
(99.99%), Pd(acac)
-ascorbic acid (AA, 99.0%),
KBr (99.0%), diglycol (DEG, 99.0%), tetraethylene glycol (TTEG,
0%) and ethylene glycol (EG, 99.0%) were all purchased from
2
(99.0%), poly vinyl
L
9
Sigma–Aldrich and used as received without further purification.
Deionized (DI) water was used for all experiments.
2.2. Preparation of bulk g-C
3 4
N
3 4
Fig. 1. Schematic illustration of the process used for the synthesis of Pd/g-C N
hybrids.
g-C was prepared by thermal treatment of urea [29]. Typi-
3 4
N
cally, 70 g of urea powder was put into an alumina crucible with
a cover, and heated in a muffle furnace at 500 °C for 2 h with a
1.5 mL of Pd nanoparticles-containing ethanol solution were dis-
persed in 200 mL of DI water by ultrasonication for 3 h to obtain
a suspension. After that, the pH of the mixed suspension was
adjusted to 2.5 (the zeta potentials of CN and Pd nanoparticles sus-
pensions at pH = 2.5 are 30.4 mV and ꢁ0.232 mV, respectively)
with HCl solution (0.5 M) under vigorous stirring. The suspension
was kept under stirring for 1 h and sonicated for 2 h. The obtained
precipitates were washed with DI water for 3 times and dried in an
oven at 80 °C overnight. The obtained different shapes of palladium
heating rate of 5 °C/min. The resultant g-C
3
N
4
yellow powder
(
CN) was collected for use without further treatment.
2.3. Preparation of different shapes of Pd nanoparticles
The Pd nanocubes (NCs) and nanotetrahedrons (NTs) were pre-
pared by using the slightly modified methods reported previously
30,31]. For the synthesis of Pd NCs [30], 105 mg of PVP, 60 mg of
[
(
3 4
Pd NCs and NTs) coupled with g-C N photocatalysts were
AA and 500 mg of KBr were dissolved in 8.0 mL of deionized water
at ambient temperature. The mixture was placed in a home-made
denoted as CACN and TACN, respectively. In order to eliminate
the influence of the synthesis process on the photocatalytic perfor-
mance, the same treatment process was applied to prepare CN
without Pd nanoparticles, and the obtained product was denoted
as CN1.
5
1
0 mL capped glass vial and pretreated in an oil bath at 80 °C for
0 min under vigorous stirring. Then, the reaction was maintained
at 80 °C for 3 h after rapid injection of 3.0 mL of Pd-containing
aqueous solution (57 mg of Na PdCl ).
2
4
Pd NTs were synthesized using a seed-mediated approach
30,31]. For the synthesis of cuboctahedral Pd seeds, 2.0 mL of EG
[
2.5. Characterization
containing 30 mg of PVP was introduced into a 20 mL glass vial
and preheated in an oil bath at 160 °C for 10 min under vigorous
stirring. Then, the reaction was maintained for 3 h after rapid injec-
The X-ray diffraction (XRD) measurements were carried out on
an X-ray diffractometer (Rigaku, Japan) with Cu K
a radiation at a
tion of 1.0 mL of Pd-containing EG solution (15.5 mg of Na
2
4
PdCl ).
ꢁ1
scan rate (2h) of 0.05° s , and the accelerating voltage and applied
current were 40 kV and 80 mA, respectively. Transmission electron
microscopy (TEM) and high-resolution TEM (HRTEM) images were
recorded by using a JEOL JEM-2100F electron microscope, working
at an accelerating voltage of 200 kV. X-ray photoelectron spec-
troscopy (XPS) measurements were performed on an ultra-high-
The product was quenched by ice-water bath for 15 min after the
vial was removed from the oil bath. The seeds were washed three
times with acetone, n-hexane and ethanol and then re-dispersed in
2
.5 mL of EG. Afterward, 3.0 mL of TTEG containing 10 mg of PVP
and 0.1 mL of EG containing 0.2 mg of cuboctahedral Pd seeds were
placed in a home-made 50 mL glass vial with a reflux condenser.
Then, the reaction temperature was maintained at 140 °C for 1 h
after rapid injection of 0.5 mL of Pd-containing TTEG solution
vacuum VG ESCALAB 210 electron spectrometer with an Al K
a
source. All the binding energies were referenced to the C1s peaks
at 284.8 eV of the surface adventitious carbon. UV–visible diffuse
reflectance spectra (DRS) were obtained by using a UV–visible
spectrometer (UV-2600, SHIMADZU, Japan) in the wavelength
(
2.1 mg of Pd(acac)
2
).
Both the products were quenched by maintaining in an ice-
water bath for 30 min, and washed several times with acetone,
n-hexane and DI water to remove the solvent, excess of PVP (PVP
was employed as a dual-function agent to reduce the Pd precursor
with its hydroxyl end groups as well as to stabilize the Pd
nanocrystals) [32] and other ions by centrifugation. The products
were collected and stored in pure ethanol, keeping the same con-
range of 200–800 nm, with BaSO
concentrations of Pd in the prepared Pd/g-C
4
as the reference standard. The
structures were
3
N
4
determined by the inductively coupled plasma atomic emission
spectrometry (ICP-AES) (PerkinElmer, Optima 4300DV, American)
after dissolving the particles with aqua regia. The Fourier trans-
form infrared spectra (FTIR) of the samples were recorded on an
IRAffinity-1 FTIR spectrometer in a KBr pellet, scanning from 400
ꢁ1
centration (2 mg mL ) for further usage and characterization.
ꢁ1
to 4000 cm at room temperature. The Brunauer–Emmett–Teller
(BET) specific surface areas (SBET) were measured by a multipoint
BET method using adsorption data obtained on a Micromeritics
ASAP 3020 nitrogen adsorption apparatus (USA). Pore size distribu-
tions were obtained by the Barrett–Joyner–Halenda (BJH) method
using nitrogen desorption data. The pore volume was calculated
2
.4. Preparation of Pd/g-C hybrid photocatalysts
3 4
N
As shown in Fig. 1, the Pd/g-C
synthesized via electrostatic assembly of Pd nanoparticles onto
the surface of g-C . Typically, 300 mg of CN powder and
3 4
N hybrid photocatalysts were
3
N
4

2
10
S. Cao et al. / Journal of Catalysis 349 (2017) 208–217
by using the nitrogen adsorption volume at the relative pressure
P/P ) of 0.994. Time-resolved photoluminescence (TRPL) decay
mization of the models by using the CASTEP package, the adsorp-
tion energies (Eads) of these gas molecules were calculated using
the following equation:
(
0
measurements were carried out using a FLS920 fluorescence life-
time spectrophotometer (Edinburgh Instruments, UK), using a
diode laser (340.2 nm) as the excitation source. The in-situ FT-IR
spectra were obtained on an in-situ diffuse reflectance infrared
Fourier transform spectroscopy (Nicolet iS50, TMO, US). The Pd dis-
persion (Dm, %) was measured via CO pulse chemisorption on a
BELCAT-B-293 Catalyst Analyzer (Bel Japan, Japan) and calculated
using the CO chemisorption data as follows:
Eads ¼ Eðgas molecule=PdÞ ꢁ ½EðPdÞ þ Eðgas moleculeÞꢄ
ð2Þ
where E(gas molecule/Pd) and E(Pd) are the energies of Pd with and
without gas molecule (CO , CH OH or CH ), respectively, E(gas
molecule) is the energy of a free gas molecule (CO , CH OH or
CH ). From this definition, a negative value of Eads indicates that
2
3
4
2
3
4
the adsorption process is exothermic, and a more negative value
corresponds to a more readily and preferred adsorption scenario.
In addition, the CO2 molecule exhibits a slight deformation after
geometry optimization of the models. Here, the deformation energy
of CO2 (Edef) was calculated using the following equation:
V
chem=22414 ꢂ SF ꢂ MW
m ꢂ p=100
Dm ¼
ð1Þ
3
where Vchem is the amount of adsorption (cm ), SF is the stoichiom-
ꢁ
1
etry factor, MW is the atomic weight of Pd (g mol ), m is the
weight (g) of the sample and p is the weight percentage of the sup-
ported metal Pd content (wt%).
Edef ¼ Eðdeformed CO
2
Þ ꢁ EðCO
2
Þ
ð3Þ
where E(CO ) is the energy of a free CO molecule and E(deformed
2
2
CO
2
) is the energy of separated CO
2
from Pd with the deformed geo-
metric structure remaining unchanged. From this definition, a neg-
ative value of the Edef represents that the energy of CO decreases
after being adsorbed onto Pd, and a more negative value reflects a
more stable CO molecule. It should be noted that strong adsorption
of CO is beneficial for photocatalytic reaction, while a stable CO
2
.6. Photocatalytic CO
2
reduction
2
Photocatalytic reduction of CO
2
was carried out in a 200 mL
home-made two-neck Pyrex top-irradiation reactor with a special
groove on one neck at ambient temperature and atmospheric pres-
sure. The two openings of the reactor were sealed with two silicone
rubber septum to form a closed system. A 300 W Xe arc lamp
2
2
2
molecule is not favorable for photoreduction because of the diffi-
culty of breaking the C@O bond. Generally, more negative adsorp-
tion energy and less negative deformation energy of CO
desirable for the achievement of excellent photocatalytic activity
for CO reduction.
2
are
(
Changzhou Siyu Science Co. Ltd., China) was used as a light source
and positioned ꢃ10 cm above the photocatalytic reactor. In a typ-
ical photocatalytic experiment, 50 mg of the catalyst was sus-
pended into 6 mL of DI water in a glass reactor via ultrasonic
dispersion for 10 min. Then the reactor was placed in a constant
temperature oven, preheated to 80 °C over 2–3 h to evaporate
water. The catalyst was deposited onto the bottom of the reactor
forming a thin and smooth film. Prior to the light irradiation, the
reactor was sealed and purged by blowing nitrogen for 30 min to
2
2.8. Photoelectrochemical measurements
Photocurrent measurements were measured using an electro-
chemical analyzer (CHI 660D electrochemical workstation, Chen-
hua Instrument, Shanghai, China) in a standard three-electrode
configuration. The working electrodes were prepared with as-
prepared samples, and a Pt wire and Ag/AgCl (saturating KCl) were
used as the counter electrode and reference electrode, respectively.
A low-power LED (3 W, 420 nm) (Shenzhen LAMPLIC Science Co,
ensure the anaerobic conditions of the reaction system. CO
O vapor were produced by the reaction of 0.084 g of NaHCO
introduced into the groove before sealing) and 0.3 mL of 2 M
SO solution, which was injected into the groove by a syringe
2
and
H
(
H
2
3
2
4
before light irradiation. After 1 h of irradiation, the gas product
extracted from the reactor was analyzed using a gas chro-
matograph (GC-2014C, Shimadzu) equipped with a flame ionized
detector (FID). High-purity nitrogen was used as carrier gas in
the GC-2014C. The products generated in the experiment were cal-
ibrated with a standard gas mixture and its retention time was
2 4
Ltd, China) was utilized as the light source. A 0.5 M Na SO aque-
ous solution was used as the electrolyte. The working electrodes
were prepared as follows: 0.05 g of the as-obtained photocatalyst
was ground together with a small amount of silver-epoxy adhesive
and 2 mL of ethanol to form a slurry. The slurry was then coated
ꢁ
2
onto a pre-prepared 2 ꢂ 1.2 cm
2
F-doped SnO -coated glass
ꢁ2
determined. Each photocatalytic CO
was performed in duplicates.
2
reduction cycle in this study
(FTO glass) substrate with an active area of about 1 cm by the
doctor-blade method, using adhesive tape as the space, and then
the film was dried in air. The photoresponse of the samples after
light on and off were measured at 0.5 V. All working electrodes
studied had a similar film thickness.
2.7. Computational methods
Density functional theory (DFT) calculations were conducted by
the CASTEP package based on the plane-wave-pseudo-potential
approach. The Perdew–Burke–Ernzerhof (PBE) of the generalized
gradient approximation (GGA) was used as the exchange-
correlation function. The interaction between valence electrons
and the ionic core was described by the ultrasoft pseudopotential.
A plane wave cut-off energy of 350 eV and Monkhorst–Pack k-
points of 3 ꢂ 3 ꢂ 1 were used for the accurate geometry optimiza-
tions and electronic structure calculations. The vacuum space at z
axis was set to 15 Å to avoid the interactions between layers. The
convergence tolerance of energy, maximum force and maximum
3. Results and discussion
3.1. Structure analysis
The XRD analysis was carried out to investigate the phase struc-
tures of the samples, as shown in Fig. 2. The patterns of all samples
feature two distinct diffraction lines at 13.1° and 27.4°, which
3 4
match well with the reported data of the layered g-C N [33,34].
The strongest XRD line of (002) at around 27.4° (d = 0.325 nm) is
characteristic for inter-planar stacking of conjugated aromatic sys-
tems. Another weak line of (100) at 13.1° (d = 0.675 nm) can be
attributed to the in-planar structural packing motif [33,35]. In
ꢁ5
displacement were set to 1.0 ꢂ 10 eV/atom, 0.03 eV/Å and
0
.001 Å, respectively.
Models for different gas molecules (CO
adsorbed on Pd sites were constructed by placing a single gas
molecule on the Pd{100} or Pd{111} facets. After geometry opti-
2
, CH
3
OH or CH
4
)
comparison with CN (pristine g-C
without Pd modification), both CACN (g-C
Pd nanocrystals) and TACN (g-C with deposited tetrahedral Pd
3
N
4
) and CN1 (treated g-C
3 4
N
3 4
N
with deposited cubic
3
N
4

S. Cao et al. / Journal of Catalysis 349 (2017) 208–217
211
the existence of slit-shaped pores from aggregates of flake-like par-
ticles [35,42], which is consistent with the typical microstructure
of the g-C N . The pore size distributions (inset in Fig. 5) are broad
3 4
and include mesopores in the entire range and small macropores
with a peak centered at ꢃ45 nm. The BET surface areas and pore
volumes for all samples are summarized in Table 1. The CN sample
shows the largest BET surface area and pore volume among the
four samples studied, suggesting that the pore structure of g-
3
C N
4
has been slightly influenced during the synthetic process of
Pd/g-C hybrids. This is because the technique of electrostatic
assembly may slightly disturb the surface structure of pristine g-
and a small number of Pd nanoparticles can fill the pores of
. The similar BET surface areas and pore distributions of
both Pd/g-C hybrids imply that the effect of these parameters
on their photocatalytic performance is analogous.
3 4
N
3 4
C N
g-C N
3 4
3 4
N
3.2. UV–visible and XPS spectra
Fig. 2. XRD patterns of CN, CN1, CACN and TACN.
UV–visible (DRS) was used to investigate the optical properties
of the samples. As can be seen from Fig. 6, the intrinsic absorption
edges of CN and CN1 are all around 450 nm, corresponding to the
band gap of ꢃ2.75 eV [43]. Both Pd/g-C N hybrids show compara-
nanocrystals) show characteristic lines of face-centered cubic Pd
(
3 4
JCPDS No. 05-0681). Note that the two distinct lines of g-C N vis-
3
4
ible on the patterns of all samples exhibit only a negligible shift,
suggesting that the synthetic process has no effect on the phase
structure of g-C N .
3 4
3 4
ble light absorption without altering the absorption edge of g-C N ,
indicating that the electrostatic assembly of Pd cocatalyst did not
affect the electronic structure of g-C N . In comparison with CN
3
4
Fig. 3a and b shows the TEM and HRTEM images of Pd nano-
cubes (NCs) and Pd nanotetrahedrons (NTs) at the same magnifica-
tion level. The TEM images indicate that Pd NCs and Pd NTs exhibit
the well-defined cubic and tetrahedral shapes with a purity of
ꢃ90% and ꢃ80%, respectively. The HRTEM images reveal the
and CN1, the samples of CACN and TACN both exhibit an enhanced
absorption in the broad visible region originated from the loaded
Pd nanoparticles, which can concentrate photoenergy to generate
a localized photothermal effect to boost the catalytic reactions
[44].
The XPS analysis was conducted to determine the Pd states in
the samples of CACN and TACN. As shown in Fig. 7, two character-
istic peaks of Pd 3d5/2 and Pd 3d3/2 emerging at binding energies of
334.6 and 339.8 eV, respectively, are in good agreement with the
metallic Pd [32,45], which can serve as excellent active sites for
{
100} facets of Pd NCs with the corresponding spacing of lattice
fringes of 0.195 nm. However, the {111} facets of Pd NTs feature
the corresponding spacing of lattice fringes of 0.225 nm. The size
distribution histograms (Fig. 3c and d) suggest that the Pd NCs
and NTs possess similar average sizes of 11.2 and 11.4 nm, respec-
tively. Fig. 3e and f shows the TEM images of Pd/g-C
N
3 4
hybrids
photocatalytic CO reduction [46].
2
with different shapes of Pd nanoparticles. As can be seen from this
figure, the shapes of Pd nanoparticles are well preserved and the Pd
3.3. Photocatalytic CO reduction
2
nanoparticles are well distributed on the surface of g-C
ing the superiority of electrostatic assembly approach for prepar-
ing the shape-controlled Pd/g-C hybrid photocatalysts. The
3 4
N , reveal-
Control experiment indicates that no noticeable products were
N
3 4
observed without introducing CO into the reactor, when using g-
2
real loading contents of Pd nanoparticles measured by ICP-AES
are 0.41 and 0.39 wt% for CACN and TACN samples, respectively.
Moreover, the values of Pd dispersion of CACN (3.0%) and TACN
C N as the photocatalyst and keeping other conditions the same.
3
4
Fig. 8a shows a comparison of the photocatalytic CO -reduction
2
performance for the as-prepared samples. As can be seen from this
(
3.2%) are very close (Table 1), indicating the similar amount of
figure, the products consist mainly of CH OH and a small amount
3
active surface atoms of the two types of Pd nanoparticles in the
hybrid photocatalysts.
of CH ; thus, the deposition of Pd cocatalyst mainly enhances the
4
CH OH production. In detail, the pristine CN shows very weak
3
Fig. 4 presents the FTIR spectra of CN, CN1, CACN and TACN
samples. There is no noticeable difference in the spectra of the four
samples, confirming that the technique of electrostatic assembly
activity, while acid treatment of CN1 indeed promotes the photo-
catalytic CO reduction. Importantly, both Pd/g-C N hybrids show
2
3
4
much higher activity towards CO2 photoreduction than CN1. In
did not affect the pristine structure of g-C
band between 3000 and 3500 cm
3
N
4
. In detail, the broad
is related to the adsorbed
O molecules and NAH vibration of the uncondensed amine
groups [36]. A number of obvious absorption bands in the region
particular, CACN exhibits lower photocatalytic activity, with a
ꢁ1
ꢁ1 ꢁ1
CH OH-production rate of 2.23
3
l
mol h
g . It is noteworthy that
ꢁ1
ꢁ1
H
2
the CH OH-production rate for TACN (3.17
l
mol h
g ) was 1.42
3
times higher than that of the CACN hybrid photocatalyst, revealing
a strong shape-dependent CO2 photoreduction over Pd cocatalyst
deposited on CN.
ꢁ1
of 1200–1650 cm can be ascribed to the typical stretching modes
of CN heterocycles [37,38]. The characteristic absorption bands at
ꢁ1
1
318 and 1241 cm correspond to the stretching vibration of con-
The in-situ FTIR spectra of TACN were analyzed to understand
nected units of CAN(AC)AC (full condensation) or CANHAC (par-
tial condensation) [39]. The absorption band at 887 cm belongs
the conversion process of photocatalytic CO reduction. As shown
2
ꢁ1
in Fig. 8b, the exposure of the sample surface to the mixture of
1
to the deformation mode of NAH [40]. The sharp absorption band
CO2 and water vapor (red curve) resulted in the formation of
ꢁ1
at 810 cm is attributed to the typical breathing mode of triazine
units [41].
primary carbonate species with water molecule, i.e. carboxylate
ꢁ
ꢁ1
ꢁ1
(CO , 1687 and 1242 cm ) [47], H O (1639 cm ) [48], bicarbonate
2
2
2
The N adsorption–desorption isotherms and the corresponding
pore-size distribution curves (inset) of the four samples are dis-
played in Fig. 5. All of the samples show type IV isotherms with
H3 hysteresis loops, indicating the presence of mesopores and
1
For interpretation of color in Fig. 8, the reader is referred to the web version of
this article.

2
12
S. Cao et al. / Journal of Catalysis 349 (2017) 208–217
Fig. 3. TEM, HRTEM images of (a) Pd NCs and (b) Pd NTs; size distribution histograms of (c) Pd NCs and (d) Pd NTs; TEM images of (e) CACN and (f) TACN.
Table 1
Physical properties of the samples.
BET (m2
g
ꢁ1
)
V
pore (cm3
g
ꢁ1
)
dpore (nm)
Sample
Colour
Pd content (wt%)
Pd dispersion (%)
S
CN
CN1
CACN
TACN
Yellow
Yellow
Gray
N/A
N/A
0.41
0.39
N/A
N/A
3.0
35
29
25
24
0.17
0.13
0.11
0.11
20
18
18
18
Gray
3.2
Fig. 5. Nitrogen adsorption-desorption isotherms and the pore size distribution
curves (inset) obtained for CN, CN1, CACN and TACN.
Fig. 4. FTIR spectra of CN, CN1, CACN and TACN.
ꢁ1
3
.4. Photocatalytic mechanism
ꢁ
(
(
(
HCO
3
, 1652, 1436 and 1206 cm ) [47,49,50], bidentate carbonate
First, to investigate the effect of the trace-amount of adsorbed
2
ꢁ
ꢁ1
b-CO
3
, 1589 and 1355 cm ) [47,48,51], monodentate carbonate
stabilizers/surfactants on the Pd surface towards the CO
duction performance, we have prepared Pd/g-C hybrid photo-
catalysts with a clean surface of Pd nanoparticles and evaluated
the corresponding CO photoreduction activity. The approach is
to remove the stabilizers/surfactants from the Pd surface by UV/
Ozone cleaning treatment prior to depositing onto g-C . UV/
Ozone oxidation has been demonstrated as an effective way for
generating nearly atomically clean surfaces without altering the
geometric feature of metal nanoparticles, which has been applied
in the removal of various capping agents from the surface of Pt,
Au, and Pd nanoparticles [60–63]. The technique uses 185- and
257-nm UV light, where the former generates ozone by interacting
2
photore-
2
ꢁ
ꢁ1
m-CO
3
, 1540, 1516, 1464 and 1311 cm ) [49,52,53], and polyden-
3 4
N
2
ꢁ
ꢁ1
tate carbonate (p-CO
curve), new absorption bands characteristic for formate (1669 and
3
, 1406 cm ) [54]. Upon light irradiation (blue
2
ꢁ1
ꢁ1
1
2
557 cm ) [55–57], dCH (HCHO, 1485 cm ) [58] and methoxyl
ꢁ1
groups (CH
3
OH, 1156 cm ) [55,59] show up. The in-situ FTIR results
on the surface of
hybrids undergoes a multi-step process involving inter-
mediate products such as HCOOH and HCHO, and finally they are
converted into CH and CH OH. Due to the nonpolar nature of
molecular structure and its low affinity toward the photocatalyst
surface, the CH molecule cannot be detected in spectra.
3 4
N
indicate that the photocatalytic conversion of CO
Pd/g-C
2
3 4
N
4
3
4

S. Cao et al. / Journal of Catalysis 349 (2017) 208–217
213
Fig. 6. UVꢁvisible DRS of CN, CN1, CACN and TACN.
Fig. 7. High-resolution XPS spectra of Pd 3d region of CACN and TACN.
with molecular oxygen. The synergistic action of ozone and
high-energy UV light can effectively break the interaction between
stabilizers/surfactants and the metal surface, and oxidize organic
contaminants into carbon dioxide and water. Fig. 9 indicates that
after UV/ozone cleaning treatment of Pd nanoparticles, the
as-obtained CACN exhibits
a
CH
3
OH-production rate of
ꢁ
1
ꢁ1
ꢁ1 ꢁ1
2
.13
lmol h
g
and a CH
4
-production rate of 0.58
lmol h
g ,
Fig. 8. (a) CH
in-situ FTIR spectra over TACN sample.
4
and CH
3
OH production over CN, CN1, CACN and TACN samples; (b)
while the as-obtained TACN shows a CH
3
OH-production rate of
ꢁ1
ꢁ1
ꢁ1 ꢁ1
3
.03
l
mol h
g
4
and a CH -production rate of 0.61
lmol h
g .
The results reveal that the removal of the surface-adsorbed
stabilizers/surfactants indeed improves the total CO conversion
efficiency (in the case of CH production) of the hybrid photocata-
2
4
lysts. Note that after UV/ozone cleaning treatment of Pd surface,
the TACN photocatalyst still exhibits higher photocatalytic activity
than that of as-obtained CACN photocatalyst, confirming the
remarkable shape-dependent CO
lyst deposited on CN. Moreover, the ratio of CH
and TACN is 3.7 and 5.0, respectively. The results indicate that the
2
photoreduction over Pd cocata-
3
OH/CH for CACN
4
Pd{111} surface shows a higher selectivity towards CH
production.
3
OH
Schottky barriers can usually be formed at the interface of a
semiconductor photocatalyst and metal cocatalyst, which results
in improving separation of the photoexcited charge carriers by
the accumulation of electrons and holes on the surfaces of metal
and semiconductor, respectively [64]. As illustrated in Fig. 10, the
Pd nanoparticles deposited on the surface of g-C
3 4
N can capture
photoexcited electrons from the conduction band of g-C
through such metal–semiconductor heterojunction. The accumu-
3
N
4
Fig. 9. CH4 and CH OH production over CACN and TACN samples with UV/ozone
cleaning treatment of Pd nanoparticles.
3
lated electrons on the Pd nanoparticles subsequently react with
+
the surface-adsorbed CO
while H
valence band of g-C
trated by Eqs. (4)–(9) according to the in-situ FTIR study.
2
and H to produce CH
3
OH and CH
4
gas,
and H in the
. The possible reaction pathway is illus-
! hþ þ eꢁ þ g-C
ð4Þ
ð5Þ
+
g-C
3
N
4
þ h
Pd þ CO
m
3 4
N
2
O molecules consume holes to release O
2
3
N
4
þ 2Hþ þ 2eꢁ ! HCOOH þ Pd
2

2
14
S. Cao et al. / Journal of Catalysis 349 (2017) 208–217
Fig. 10. Schematic illustration of the mechanism for enhanced CO
2
reduction in Pd-
hybrids.
Pd þ HCOOH þ 2Hþ þ 2eꢁ ! HCHO þ H
O þ Pd
ð6Þ
ð7Þ
ð8Þ
ð9Þ
2
Pd þ HCHO þ 2Hþ þ 2eꢁ ! CH
OH þ Pd
þ H O þ Pd
3
Pd þ CH
3
OH þ 2Hþ þ 2eꢁ ! CH
4
2
g-C
3
N
4
þ 2H
2
O þ 4hþ ! O
2
þ 4Hþ þ g-C
3 4
N
To gain an in-depth understanding of the above-mentioned
mechanism for enhanced photocatalytic reduction of CO , the
2
time-resolved photoluminescence (TRPL) analysis was carried out
to investigate the lifetime of photogenerated charge carriers in
CN, CN1, CACN and TACN samples, as shown in Fig. 11a. A tri-
exponential fitting was performed for the luminescence decay
curves by using the following equation [65,66]:
ꢀ
tꢁ
ꢀ
ꢁ
ꢀ
ꢁ
ꢁ
ꢁt
ꢁt
yðtÞ ¼ y þ B
1
exp
þ B
2
exp
þ B
3
exp
ð10Þ
0
s
1
s
2
s
3
where y
exponential factors, and
stants: the radiative process (
energy transfer process ( ) [65,66]. The mean lifetimes (
0
is the baseline correction (y-offset), B
and
), non-radiative process (
1
, B
2
and B
3
are pre-
s
1
,
s
2
s
3
are the three decay time con-
) and
s
1
s
2
Fig. 11. Photoelectrochemical measurement of samples. (a) TRPL decay curves, (b)
transient photocurrent responses and (c) EIS Nyquist plots.
s
3
s
m
) are cal-
culated using the following equation [67]:
P
n
2
B
i
si
i¼1
s
m
¼ P
ð11Þ
n
confirming that the charge transfer from the conduction band of
B s
i i
i¼1
g-C
g-C
3
N
N
4
is more sensitive to the Pd{111} surface than that from
to the Pd{100} surface. Note that for the curves obtained
The decay time constants, the corresponding pre-exponential
3
4
factors, and the mean lifetime for all the samples are listed in
Table 2. Data listed in this table indicate that the fluorescence life-
time is much smaller after introduction of Pd nanoparticles, imply-
ing the remarkably suppressed charge recombination in the Pd/g-
3 4
for Pd/g-C N hybrids, after a rapid increase, the photocurrent
gradually reaches a stable value under a continuous light irradia-
tion. This is because the photogenerated electrons gradually trans-
3 4
fer from the conduction band of g-C N to the Pd cocatalyst. In
C
3
N
4
hybrids due to the electron transfer process from g-C
3
N
4
to
contrast, the accumulated electrons gradually release from the Pd
cocatalyst when switching off the light, resulting in a gradual
decay of photocurrent to zero [69,70].
Pd cocatalysts. Particularly, the fluorescence lifetime of TACN is
shorter than that of CACN, indicating more efficient charge transfer
from g-C
Pd{100} surface.
3
N
4
to the Pd{111} surface than that from g-C
3
N
4
to the
The electrochemical impedance spectra (EIS) of the as-prepared
samples were also examined (see Fig. 11c). As can be seen from
Photoelectrochemical measurements were also performed to
obtain the transient photocurrent responses to further investigate
the transfer and separation of photogenerated charge carriers. Nor-
mally, a high photocurrent value suggests more efficient transfer
and separation of electrons and holes [68]. Fig. 11b shows the I–t
curves for all the samples with several on-off cycles at a bias poten-
3 4
this figure the Nyquist plots obtained for both Pd/g-C N hybrids
(CACN and TACN) exhibit a smaller arc radius than those of CN
and CN1, because the Pd nanoparticles can reduce the charge
transfer resistance and increase the electronic conductivity, which
leads to a more efficient charge separation [71]. It is noteworthy
that the TACN sample possesses a smaller radius of the Nyquist
plots as compared to that of CACN, confirming that the Pd{111}
surface is more preferable than the Pd{100} surface to capture
tial of 0.5 V. Obviously, both the Pd/g-C
photocurrent intensity than pure g-C
3
N
4
hybrids exhibit higher
3
4
N due to the excellent
electron-sink effect of Pd nanoparticles that can greatly promote
the separation of photogenerated charge carriers. Interestingly, a
higher photocurrent intensity corresponds to the TACN sample,
3 4
photogenerated electrons from the conduction band of g-C N .
In light of the combined TRPL analysis, the transient photocur-
rent response and EIS, one can conclude that the deposition of Pd

S. Cao et al. / Journal of Catalysis 349 (2017) 208–217
215
Table 2
The kinetics of emission decay parameters of CN, CN1, CACN and TACN samples.
Sample
s
1
(ns) (rel. %)
s
2
(ns) (rel. %)
s
3
(ns) (rel.%)
m
s (ns)
CN
CN1
CACN
TACN
1.30 (60.75)
1.26 (61.58)
0.98 (62.40)
0.74 (57.58)
5.32 (35.51)
5.72 (35.47)
4.19 (31.40)
3.49 (35.98)
26.41 (3.74)
25.76 (2.95)
15.49 (6.20)
12.31 (6.44)
10.13
9.03
7.27
5.84
2
Fig. 12. Optimized geometry structures for CO adsorption on Pd{100} facets and Pd{111} facets at (a) bridge position, (b) top1 position and (c) top2 position.
nanoparticles onto the surface of g-C
charge transfer and separation, leading to the remarkably
enhanced photocatalytic CO reduction. More importantly, the Pd
111} surface is more efficient than the Pd{100} surface to func-
tion as the electron sink to capture photogenerated electrons from
the conduction band of g-C , further enhancing the photocat-
alytic performance of Pd/g-C hybrids for CO reduction. This
3
N
4
greatly improves the
exothermic process of CO
the Pd{111} surface is more sensitive towards CO
toward the Pd{100} surface, which is beneficial for photocatalytic
CO reduction. Meanwhile, it is concluded that the most preferen-
tial adsorption position of CO molecules on the Pd{100} and Pd
2
adsorption. These results reveal that
2
adsorption than
2
{
2
2
3
N
4
{111} surfaces are the bridge and top1 positions, respectively.
Therefore, the following calculations, including the deformation
3
N
4
2
is reasonable because the Pd{111} surface possesses a larger work
function as compared to that of the Pd{100} surface [72,73].
2 3 4
energy (Edef) of CO and adsorption energies of CH OH and CH ,
are carried out at the bridge position on the Pd{100} surface and
top1 position on the Pd{111} surface. Consequently, the calculated
It is known that the CO
tocatalytic CO reduction [74]. Hence we calculated the adsorption
energies of CO molecules on different Pd surfaces. Fig. 12 shows
the optimized geometry structures for CO adsorption on the Pd
100} and Pd{111} surfaces at three different positions. It turns
out that the adsorption energy at the bridge position on the Pd
111} surface (Eads = ꢁ0.0083 eV) is more negative than that on
can
2
adsorption is a critical step for the pho-
2
E
def of CO
ꢁ0.008 eV, respectively, suggesting that the CO
adsorbed on the Pd{111} surface has higher energy and weaker
stability. Therefore, the CO molecule can be activated and partic-
2
on the Pd{100} and Pd{111} surfaces are ꢁ0.017 eV and
2
2
molecule
2
{
2
ipate in subsequent photoreduction reaction more readily on the
Pd{111} surface.
{
the Pd{100} surface (Eads = ꢁ0.0076 eV), indicating that CO
2
On the other hand, the product desorption from the catalyst
be more strongly adsorbed onto the Pd{111} surface rather than
on the Pd{100} surface. Whereas, for top1 and top2 positions,
surface is of great importance for the photocatalytic CO
too. As such, we have calculated the adsorption energies of CH
and CH molecules on the Pd{100} surface (bridge position) and
the Pd{111} surface (top1 position). As shown in Fig. 13, the CH
2
reduction
3
OH
the Pd{100} surface is not able to adsorb CO
itive Eads (0.133 and 0.181 eV, respectively). In contrast, the Eads
values for CO at the top1 and top2 positions on the Pd{111} sur-
faces are ꢁ0.021 and ꢁ0.012 eV, respectively, suggesting a viable
2
because of the pos-
4
3
-
2
OH adsorption energies on the Pd{100} surface (bridge position)
and Pd{111} surface (top1 position) are ꢁ0.118 and ꢁ0.014 eV,

2
16
S. Cao et al. / Journal of Catalysis 349 (2017) 208–217
Fig. 13. Optimized geometry structures for (a) CH
3
OH adsorption and (b) CH
4
adsorption on Pd{100} facets (bridge position) and Pd{111} facets (top1 position).
respectively, indicating that the desorption of CH
from the Pd{111} surface is easier than that from the Pd{100} sur-
face. Meanwhile, the CH adsorption energies on the Pd{100} sur-
3
OH molecule
[4] F. Lei, W. Liu, Y. Sun, J. Xu, K. Liu, L. Liang, T. Yao, B. Pan, S. Wei, Y. Xie, Nat.
Commun. 7 (2016) 12697.
[
5] S. Gao, Y. Lin, X.C. Jiao, Y.F. Sun, Q.Q. Luo, W.H. Zhang, D.Q. Li, J.L. Yang, Y. Xie,
Nature 529 (2016) 68–71.
4
face (bridge position) and the Pd{111} surface (top1 position) are
[6] B.A. Rosen, A. Salehi-Khojin, M.R. Thorson, W. Zhu, D.T. Whipple, P.J.A. Kenis, R.
I. Masel, Science 334 (2011) 643–644.
very close to each other (ꢁ0.011 and ꢁ0.012 eV, respectively), sug-
[
7] T.R. Chen, F.S. Wu, H.P. Lee, K.H. Chen, J. Am. Chem. Soc. 138 (2016) 3643–
646.
[8] T. Inoue, A. Fujishima, S. Konishi, K. Honda, Nature 277 (1979) 637–638.
4
gesting the equal probability for desorption of CH molecule from
3
the Pd{111} and Pd{100} surfaces. These calculations strongly
indicate that the Pd{111} surface is more efficient than the Pd
[
9] H. Mistry, A.S. Varela, C.S. Bonifacio, I. Zegkinoglou, I. Sinev, Y.W. Choi, K.
Kisslinger, E.A. Stach, J.C. Yang, P. Strasser, B.R. Cuenya, Nat. Commun. 7 (2016)
{
100} surface for CO
desorption, and thus, for catalytic reduction of CO
also can be deduced that the higher selectivity of Pd{111} surface
towards CH OH production should be due to the fact that the Pd
111} surface has better ability for CH OH desorption, but equal
ability for CH desorption, as compared to Pd{100} surfaces.
2
adsorption and activation, as well as CH
3
OH
12123.
2
. Moreover, it
[10] A. Corma, H. Garcia, J. Catal. 308 (2013) 168–175.
11] X. Liu, D. Gregurec, J. Irigoyen, A. Martinez, S. Moya, R. Ciganda, P. Hermange, J.
Ruiz, D. Astruc, Nat. Commun. 7 (2016) 13152.
12] Y.N. Xia, Y.J. Xiong, B. Lim, S.E. Skrabalak, Angew. Chem. Int. Ed. 48 (2009) 60–
103.
13] G.A. Filonenko, W.L. Vrijburg, E.J.M. Hensen, E.A. Pidko, J. Catal. 343 (2016) 97–
[
[
[
3
{
3
4
105.
[
14] T. Van Cleve, E. Gibara, S. Linic, ChemCatChem 8 (2016) 256–261.
[
15] T. Bian, H. Zhang, Y. Jiang, C. Jin, J. Wu, H. Yang, D. Yang, Nano Lett. 15 (2015)
4
. Conclusions
7808–7815.
[
16] S. Kumar-Krishnan, E. Prokhorov, O.A. de Fuentes, M. Ramirez, N.
Bogdanchikova, I.C. Sanchez, J.D. Mota-Morales, G. Luna-Barcenas, J. Mater.
Chem. A 3 (2015) 15869–15875.
In summary, the shape-dependent photocatalytic CO
2
reduction
over Pd nanoparticles supported on g-C photocatalyst is well
3
N
4
[
[
17] M. Jin, H. Zhang, Z. Xie, Y. Xia, Energy Environ. Sci. 5 (2012) 6352–6357.
18] J. Ran, J. Zhang, J. Yu, M. Jaroniec, S.Z. Qiao, Chem. Soc. Rev. 43 (2014) 7787–
7812.
demonstrated. It is found that during this reduction reaction, tetra-
hedral Pd nanoparticles with exposed {111} facets are more effi-
cient to serve as the cocatalyst than cubic Pd nanoparticles with
exposed {100} facets. Experimental results and theoretical calcula-
[19] J. Yang, D. Wang, H. Han, C. Li, Acc. Chem. Res. 46 (2013) 1900–1909.
[20] Y. Choi, H.-I. Kim, G.-H. Moon, S. Jo, W. Choi, ACS Catal. 6 (2016) 821–828.
[21] S. Cao, F. Tao, Y. Tang, Y. Li, J. Yu, Chem. Soc. Rev. 45 (2016) 4747–4765.
tions reveal that the structure-sensitive photocatalytic CO
tion activity is mainly attributed to the more preferable electron-
sink effect, CO adsorption capability and CH OH desorption capa-
2
reduc-
[22] X. Li, G. Li, W. Zang, L. Wang, X. Zhang, Catal. Sci. Technol. 4 (2014) 3290–3297.
[
[
23] S. Cao, J. Jiang, B. Zhu, J. Yu, Phys. Chem. Chem. Phys. 18 (2016) 19457–19463.
24] P. Qiu, H. Chen, C. Xu, N. Zhou, F. Jiang, X. Wang, Y. Fu, J. Mater. Chem. A 3
2
3
(
2015) 24237–24244.
bility of the Pd{111} surface, as compared to those observed on the
Pd{100} surface, due to their different surface atomic structures.
This work provides a new insight for the structure tuning of metal
[25] X. Wang, K. Maeda, A. Thomas, K. Takanabe, G. Xin, J.M. Carlsson, K. Domen, M.
Antonietti, Nat. Mater. 8 (2009) 76–80.
[
26] M. Li, L. Zhang, X. Fan, M. Wu, M. Wang, R. Cheng, L. Zhang, H. Yao, J. Shi, Appl.
Catal. B-Environ. 201 (2017) 629–635.
cocatalysts to enhance the photocatalytic CO
performance.
2
reduction
[
27] H. Zhou, P. Li, J. Liu, Z. Chen, L. Liu, D. Dontsova, R. Yan, T. Fan, D. Zhang, J. Ye,
Nano Energy 25 (2016) 128–135.
[
[
28] G. Gao, Y. Jiao, E.R. Waclawik, A. Du, J. Am. Chem. Soc. 138 (2016) 6292–6297.
29] F. Dong, L. Wu, Y. Sun, M. Fu, Z. Wu, S.C. Lee, J. Mater. Chem. 21 (2011) 15171–
15174.
Acknowledgments
[
[
[
[
30] S.-I. Choi, J.A. Herron, J. Scaranto, H. Huang, Y. Wang, X. Xia, T. Lv, J. Park, H.-C.
Peng, M. Mavrikakis, Y. Xia, ChemCatChem 7 (2015) 2077–2084.
31] Y. Wang, S. Xie, J. Liu, J. Park, C.Z. Huang, Y. Xia, Nano Lett. 13 (2013) 2276–
2281.
The authors acknowledge the financial support from the 973
Program (2013CB632402), NSFC (51472191, 21407115,
1272199, 51320105001 and 21433007), the Natural Science
32] S. Bai, X. Wang, C. Hu, M. Xie, J. Jiang, Y. Xiong, Chem. Commun. 50 (2014)
5
6094–6097.
Foundation of Hubei Province of China (2015CFA001), the Special
Financial Grant from the China Postdoctoral Science Foundation
33] Y. Zhang, J. Liu, G. Wu, W. Chen, Nanoscale 4 (2012) 5300–5303.
[34] Q. Tay, X. Wang, X. Zhao, J. Hong, Q. Zhang, R. Xu, Z. Chen, J. Catal. 342 (2016)
5–62.
5
(
2015T80843), the Fundamental Research Funds for the Central
[
[
35] S. Martha, A. Nashim, K.M. Parida, J. Mater. Chem. A 1 (2013) 7816–7824.
36] J. Hong, X. Xia, Y. Wang, R. Xu, J. Mater. Chem. 22 (2012) 15006–15012.
Universities (WUT: 2015-III-034) and the Innovative Research
Funds of SKLWUT (2017-ZD-4).
[37] S.C. Yan, Z.S. Li, Z.G. Zou, Langmuir 25 (2009) 10397–10401.
[
38] V.W. Lau, I. Moudrakovski, T. Botari, S. Weinberger, M.B. Mesch, V. Duppel, J.
Senker, V. Blum, B.V. Lotsch, Nat. Commun. 7 (2016) 12165.
References
[39] J. Liu, T. Zhang, Z. Wang, G. Dawson, W. Chen, J. Mater. Chem. 21 (2011)
4398–14401.
1
[
[
40] J. Ma, Q. Yang, Y. Wen, W. Liu, Appl. Catal. B-Environ. 201 (2017) 232–240.
41] X. Zhang, X. Xie, H. Wang, J. Zhang, B. Pan, Y. Xie, J. Am. Chem. Soc. 135 (2013)
[
[
[
1] C.W. Li, J. Ciston, M.W. Kanan, Nature 508 (2014) 504–508.
2] Y. Tachibana, L. Vayssieres, J.R. Durrant, Nat. Photonics 6 (2012) 511–518.
3] T. Avanesian, G.S. Gusmão, P. Christopher, J. Catal. 343 (2016) 86–96.
18–21.

S. Cao et al. / Journal of Catalysis 349 (2017) 208–217
217
[
42] J. Mao, T. Peng, X. Zhang, K. Li, L. Ye, L. Zan, Catal. Sci. Technol. 3 (2013) 1253–
260.
[58] G. Busca, J. Lamotte, J.-C. Lavalley, V. Lorenzelli, J. Am. Chem. Soc. 109 (1989)
5197–5202.
1
[
43] G. Liu, G. Zhao, W. Zhou, Y. Liu, H. Pang, H. Zhang, D. Hao, X. Meng, P. Li, T.
Kako, J. Ye, Adv. Funct. Mater. 26 (2016) 6822–6829.
44] Z. Lou, Q. Gu, Y. Liao, S. Yu, C. Xue, Appl. Catal. B-Environ. 184 (2016) 258–263.
45] T. Borkowski, J. Dobosz, W. Tylus, A.M. Trzeciak, J. Catal. 319 (2014) 87–94.
46] J. Xu, X. Su, X. Liu, X. Pan, G. Pei, Y. Huang, X. Wang, T. Zhang, H. Geng, Appl.
Catal. A 514 (2016) 51–59.
[59] F. Boccuzzi, A. Chiorino, M. Manzoli, J. Power Sources 118 (2003) 304–310.
[60] J.R. Vig, J. Vac. Sci. Technol. A 3 (1985) 1027–1034.
[61] C. Aliaga, J.Y. Park, Y. Yamada, H.S. Lee, C.-K. Tsung, P. Yang, G.A. Somorjai, J.
Phys. Chem. C 113 (2009) 6150–6155.
[62] S. Pang, Y. Kurosawa, T. Kondo, T. Kawai, Chem. Lett. 34 (2005) 544–545.
[63] M. Crespo-Quesada, J.-M. Andanson, A. Yarulin, B. Lim, Y. Xia, L. Kiwi-Minsker,
Langmuir 27 (2011) 7909–7916.
[
[
[
[
47] L. Liu, Y. Jiang, H. Zhao, J. Chen, J. Cheng, K. Yang, Y. Li, ACS Catal. 6 (2016)
1
097–1108.
[64] X. Gong, W.Y. Teoh, J. Catal. 332 (2015) 101–111.
[65] X. Guo, W. Song, C. Chen, W. Di, W. Qin, Phys. Chem. Chem. Phys. 15 (2013)
14681–14688.
[
[
48] L. Liu, C. Zhao, Y. Li, J. Phys. Chem. C 116 (2012) 7904–7912.
49] J. Mao, L. Ye, K. Li, X. Zhang, J. Liu, T. Peng, L. Zan, Appl. Catal. B-Environ. 144
(
2014) 855–862.
50] L. Ye, J. Mao, T. Peng, L. Zan, Y. Zhang, Phys. Chem. Chem. Phys. 16 (2014)
5675–15680.
51] K. Li, T. Peng, Z. Ying, S. Song, J. Zhang, Appl. Catal. B-Environ. 180 (2016) 130–
38.
[66] P. Niu, L. Zhang, G. Liu, H.-M. Cheng, Adv. Funct. Mater. 22 (2012) 4763–4770.
[67] K. Das, S.N. Sharma, M. Kumar, S.K. De, J. Phys. Chem. C 113 (2009) 14783–
14792.
[68] Z.A. Huang, Q. Sun, K. Lv, Z. Zhang, M. Li, B. Li, Appl. Catal. B-Environ. 164
(2015) 420–427.
[
[
1
1
[
[
[
[
52] L. Liu, C. Zhao, J. Xu, Y. Li, Appl. Catal. B-Environ. 179 (2015) 489–499.
53] H. Zhao, L. Liu, J.M. Andino, Y. Li, J. Mater. Chem. A 1 (2013) 8209–8216.
54] S. Collins, M. Baltanas, A. Bonivardi, J. Catal. 226 (2004) 410–421.
55] M.-T. Chen, C.-F. Lien, L.-F. Liao, J.-L. Lin, J. Phys. Chem. B 107 (2003) 3837–
[69] Q. Xiang, J. Yu, M. Jaroniec, J. Phys. Chem. C 115 (2011) 7355–7363.
[70] K. Wang, Q. Li, B. Liu, B. Cheng, W. Ho, J. Yu, Appl. Catal. B-Environ. 176–177
(2015) 44–52.
[71] J. Jiang, J. Yu, S. Cao, J. Colloid Interface Sci. 461 (2016) 56–63.
[72] N.E. Singh-Miller, N. Marzari, Phys. Rev. B 80 (2009).
[73] J. Hulse, J. Kuppers, K. Wandelt, G. Ertl, Appl. Surf. Sci. 6 (1980) 453–463.
[74] D. Maiti, Y.A. Daza, M.M. Yung, J.N. Kuhn, V.R. Bhethanabotla, J. Mater. Chem. A
4 (2016) 5137–5148.
3
843.
56] L.-F. Liao, Wen-Chun Wu, C.-Y. Chen, J.-L. Lin, J. Phys. Chem. B 105 (2001)
678–7685.
57] W. Wu, K. Bhattacharyya, K. Gray, E. Weitz, J. Phys. Chem. C 117 (2013) 20643–
0655.
[
[
7
2