Nanopore enriched hollow carbon nitride nanospheres with extremely high visible-light photocatalytic activity in the degradation of aqueous contaminants of emerging concern

Article abstract of DOI:10.1039/c8cy02073f

Construction of highly efficient hollow nanosphere photocatalytic systems has been strongly attracting the attention of researchers. In the present work, nanopore enriched hollow carbon nitride nanospheres (HCNNSs) with a smaller particle size (200 nm) and a thinner shell thickness (40 nm) are successfully fabricated by a silica-nanocasting strategy. Such unique structures possess many advantages such as large BET surface area (122 m² g^-1), high light-harvesting ability, fast charge separation and transfer efficiency, plentiful exposed active sites and enhanced oxidation ability of photogenerated holes (h+VB). Therefore, HCNNSs in smaller sizes (HCNNS-200) exhibit extremely excellent visible-light photocatalytic efficiency towards the degradation of contaminants of emerging concern, e.g. levofloxacin (LEVO), in comparison with bulk g-C₃N₄ and HCNNSs in larger sizes (HCNNS-500). And it takes less than 10 min to finish the degradation of LEVO. The experimental results including those from indirect chemical probing, electron spin resonance, ion chromatography and high performance liquid chromatography-mass spectrometry confirm that h+VB and O₂^- are the active species that are responsible for the mineralization of LEVO to NO₃^-, F^-, H₂O and CO₂ under visible-light irradiation. Additionally, the degradation pathway of LEVO in the HCNNS-200 photocatalytic system is also proposed. It is expected that HCNNS-200 can be used as a promising photocatalyst for environmental remediation.

Full text of DOI:10.1039/c8cy02073f

Catalysis
Science &
Technology
View Article Online
View Journal
PAPER
Nanopore enriched hollow carbon nitride
nanospheres with extremely high visible-light
photocatalytic activity in the degradation of
aqueous contaminants of emerging concern†
Cite this: DOI: 10.1039/c8cy02073f
Yuxin Yang, ‡ An Hu,‡ Xinyue Wang, Jiaqi Meng, Yihang Guo,
*
Mingxin Huo and Suiyi Zhu
Construction of highly efficient hollow nanosphere photocatalytic systems has been strongly attracting the
attention of researchers. In the present work, nanopore enriched hollow carbon nitride nanospheres
(HCNNSs) with a smaller particle size (200 nm) and a thinner shell thickness (40 nm) are successfully fabri-
cated by a silica-nanocasting strategy. Such unique structures possess many advantages such as large BET
surface area (122 m2 g−1), high light-harvesting ability, fast charge separation and transfer efficiency, plenti-
ful exposed active sites and enhanced oxidation ability of photogenerated holes (h+VB). Therefore, HCNNSs
in smaller sizes (HCNNS-200) exhibit extremely excellent visible-light photocatalytic efficiency towards the
degradation of contaminants of emerging concern, e.g. levofloxacin (LEVO), in comparison with bulk
g-C3N4 and HCNNSs in larger sizes (HCNNS-500). And it takes less than 10 min to finish the degradation
of LEVO. The experimental results including those from indirect chemical probing, electron spin resonance,
ion chromatography and high performance liquid chromatography-mass spectrometry confirm that hV+B
Received 7th October 2018,
Accepted 22nd November 2018
and ˙O2 are the active species that are responsible for the mineralization of LEVO to NO3−, F−, H2O and
−
CO2 under visible-light irradiation. Additionally, the degradation pathway of LEVO in the HCNNS-200
photocatalytic system is also proposed. It is expected that HCNNS-200 can be used as a promising photo-
catalyst for environmental remediation.
DOI: 10.1039/c8cy02073f
rsc.li/catalysis
Inspired by the thylakoid membrane where plant photo-
synthesis occurs, hollow nanostructured materials, especially
1. Introduction
Graphitic carbon nitride (g-C3N4) has recently emerged as a
promising metal-free photocatalyst for water splitting, CO2 re-
duction, selective oxidation of alcohols and water
purification.1–6 Two-dimensional (2D) conjugated planes and
tri-s-triazine repeating units through Van der Waals interac-
tions provide g-C3N4 with unique advantages such as good
chemical stability as well as an appealing electronic structure
with a medium-bandgap (2.7 eV).7 However, the limited light
harvesting ability and fast recombination probability of the
photogenerated charge carriers seriously restrict the applica-
tion of g-C3N4 in the photocatalytic field.8 Therefore, fabrica-
tion of robust g-C3N4 material with excellent photocatalytic ac-
tivity is still a challenge.
hollow g-C3N4 nanospheres, have been constructed via mor-
phological engineering and have attracted much attention be-
cause of their unique structural, physical and chemical
properties:9–19 i) the hollow structures can enhance the light
harvesting efficiency owing to multiple light reflection events
within their void space; additionally, hollow structures can fa-
cilitate mass transfer via shortening the transfer distance as
compared with bulk catalysts; ii) the large surface areas of
both exterior and interior surfaces can promote surface-
dependent reactions by providing plentiful active sites.
Recently, much effort has been paid to fabricate hollow
g-C3N4-based nanospheres via a nanostructured silica tem-
plate, such as, hollow carbon nitride nanospheres
(HCNSs),13,20 post-annealing reinforced HCNSs,21 Co3O4/
HCNS/Pt,22 CdS/HCNS23 and MoS2/HCNS.24 Constructing
three-dimensional hollow nanospheres of g-C3N4 from two-
dimensional conjugated packed g-C3N4 planes not only sig-
nificantly enhances light harvesting but also facilitates the
mass transfer of the reactants and products as well as the mi-
gration of the photogenerated charge carriers. Therefore, all
School of Environment, Northeast Normal University, Changchun 130024, P.R.
China. E-mail: zhusy812@nenu.edu.cn; Tel: +86 431 89165610
† Electronic supplementary information (ESI) available. See DOI: 10.1039/
c8cy02073f
‡ Equally contributing authors.
This journal is © The Royal Society of Chemistry 2018
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
these hollow g-C3N4-based nanospheres achieve good H2 or/
and O2 evolution efficiency. In addition, by coupling the ap-
propriate cocatalysts on the exterior or interior surfaces, the
H2 and O2 evolution sites are spatially separated, and overall
photocatalytic water splitting can be achieved.22 Unfortu-
nately, such perfect hollow carbon nitride nanospheres are
rarely studied in the photodecomposition of organic pollut-
ants, which is seriously restricted by the lower oxidation abil-
ity of photogenerated holes, leading to slow photocatalytic ef-
ficiency and poor mineralization ability. In this context, it is
highly desirable to develop hollow carbon nitride nano-
spheres with high photocatalytic and mineralization effi-
ciency for pollutant removal.
then added dropwise to the above solution with vigorous stir-
ring for 10 min to create a thin mesoporous silica shell
around the dense silica core. The mixed solution was then
kept at ambient temperature for 3 h without stirring to pro-
mote the cohydrolysis and condensation of the TEOS and
C18TMOS on the nonporous silica cores. The resulting white
precipitate was recovered by filtration, dried at 80 °C for 12
h, and further calcined at 550 °C in air for 6 h with a heating
rate of 5 °C min−1. The as-prepared monodisperse silica tem-
plates were neutralized with 1 M HCl and then dried at 60 °C
for 12 h. The final silica templates were denoted as Tem-
plate-x, where x represented the particle size of the silica tem-
plate. Here, x = 200 and 500, respectively.
In the present work, nanopore enriched hollow carbon ni-
tride nanospheres (HCNNSs) with a smaller particle size (200
nm) and a thinner shell thickness (40 nm) are successfully
fabricated by a silica-nanocasting strategy. Owing to their
unique hollow nanostructure, the as-prepared HCNNSs pos-
sess enhanced light-harvesting efficiency, fast charge separa-
tion property and increased surface accessibility. What is
more, the smaller size of the HCNNSs also enhance the quan-
tum size effect, leading to increased oxidizing ability of
photogenerated holes in the HCNNSs. Therefore, HCNNSs ex-
hibit excellent visible-light photocatalytic activity in the
photodegradation of aqueous organic pollutants, as well as
deep oxidation of the organic pollutants into H2O, CO2 and
inorganic anions.
The morphologies, phase and chemical structure, optical
absorption and photoelectrochemical properties of the
HCNNSs are well characterized, and the visible-light photo-
catalytic activity of the HCNNSs is evaluated by the degrada-
tion of contaminants of emerging concern including levofl-
oxacin (LEVO) and trimethoprim (TMP) which are frequently
detected at low concentration levels in natural water and can
cause adverse effect in biota.25 Meanwhile, the active species
generated during the process of photodegradation are investi-
gated by electron spin resonance spectroscopy and free radi-
cal and hole scavenging experiments. On the basis of the
above results, the mechanism and pathway of the visible-
light photocatalytic degradation of LEVO over the HCNNSs
are revealed, which is one of the most important issues for
hollow carbon nitride nanosphere photocatalysts but is sel-
dom studied up till now.
2.1.2. HCNNS. A mixture of the Template-x powder (1 g)
and an aqueous cyanamide solution (50%, 1.5 mL) was soni-
cated for 2 h in a flask connected to a vacuum line, followed
by stirring at 60 °C for 14 h and then drying at 80 °C for 24
h. The resulting white powder was transferred to an alumina
crucible and calcined at 550 °C under flowing N2 for 2 h with
a heating rate of 5 °C min−1. Afterwards, the obtained yellow
g-C3N4/SiO2 hybrids were treated with NH4HF2 (4 M, 10 mL)
for 12 h to remove the silica template. The final yellow pow-
der was obtained by drying at 80 °C overnight. The product is
denoted as HCNNS-x with x = 200 and 500, respectively.
2.1.3. Bulk g-C3N4. Bulk g-C3N4 was synthesized as the ref-
erence sample by calcination of cyanamide at 550 °C under
flowing N2 atmosphere for 2 h with a heating rate of 5 °C
min−1.
2.2. Catalyst characterization
Transmission electron microscopy (TEM) images were
recorded on
a JEM-2100F high resolution transmission
electron microscope at an accelerating voltage of 200 kV. Ni-
trogen porosimetry measurements were performed on a
Micromeritics ASAP 2020 M surface area and porosity ana-
lyzer. Powder X-ray diffraction (XRD) patterns were obtained
on a Japan Rigaku D/max 2000 X-ray diffractometer (Cu Kα, λ
= 1.5418 Å). Fourier transform infrared (FT-IR) spectra were
recorded on a Nicolet Magna 560 IR spectrophotometer.
X-ray photoelectron spectroscopy (XPS) was performed on a
VG-ADES 400 instrument with a Mg K-ADES source. UV-vis
diffuse reflectance spectra (UV-vis/DRS) were recorded on a
Cary 500 UV-vis-NIR spectrometer. Photoelectrochemical
measurements were carried out on Chenhua CHI 660E
electrochemical station. Photoluminescence (PL) spectra were
2. Experimental
2.1. Catalyst preparation
obtained with
a Photon Technology International (PTI)
TimeMasterTM 400 with an excitation wavelength of 337 nm.
DMPO (5,5-dimethyl-1-pyrroline-N-oxide) spin-trapping EPR
spectra were recorded on a JES-FA200 electron spin reso-
nance spectrometer.
2.1.1. Silica templates. The monodisperse core–shell struc-
tured silica templates were synthesized through a modified
Stöber method.13 Typically, an aqueous ammonia solution
(7%, 16 or 30 mL) was added to ethanol (150 mL), and then
tetraethoxysilane (TEOS, 3.4 or 5.1 mL) was added dropwise
to the above basic ethanol solution with vigorous stirring at
room temperature for 1 h to yield uniform nonporous colloi-
dal silica spheres. A mixture of TEOS (5 or 7.5 mL) and
n-octadecyltrimethoxysilane (C18TMOS, 2.26 or 3.39 mL) was
2.3. Photocatalytic performance evaluation
The adsorption and photocatalytic activity of the as-prepared
photocatalysts were evaluated by the degradation of LEVO
and TMP under visible-light irradiation (400–680 nm) using a
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2018

View Article Online
Catalysis Science & Technology
Paper
300 W Xe lamp. For all adsorption and photocatalytic tests,
50 mL of the aqueous target pollutant solution with an initial
concentration of 20 mg L−1 was used, and the catalyst
amount was 50 mg. All tests were performed in a self-made
quartz photoreactor fitted with a circulation water system to
maintain a constant temperature. Prior to irradiation, the ad-
sorption–desorption equilibrium between the target com-
pound and the catalyst was monitored by the determination
of the concentrations of the target pollutant for a certain
time interval. During the photocatalytic tests, a fixed amount
of the reaction solution was taken out at given time intervals,
followed by centrifugation to remove the photocatalyst
completely. The concentrations of LEVO and TMP were deter-
mined using an Agilent 1200 high performance liquid chro-
matograph (HPLC): C18 column, UV detector. The concentra-
tions of the produced organic acids during the
photodegradation process were determined using a Dionex
DX-300 ion chromatograph (IC), while the concentrations of
inorganic anions were determined using a Metrohm 881
Compact IC. The intermediates produced during the photo-
degradation process were identified using an Agilent 1200
HPLC-micro TOFESI-TOF-MS high-resolution electrospray
ionization time-of-flight mass spectrometer equipped with
HPLC (HPLC-MS).
sis and condensation of TEOS in a mixture of ethanol and
ammonia aqueous solution; moreover, the particle size of the
silica cores can be adjusted by changing the initial concentra-
tions of TEOS and NH4OH in the preparation system. The
higher the TEOS and NH4OH concentrations, the larger the
particle size of the silica cores. Here, two silica cores are pre-
pared at different initial TEOS and NH4OH adding amounts,
which directly influence the particle sizes of the silica tem-
plates. After further cohydrolysis and condensation of the
TEOS and C18TMOS around the silica core followed by calci-
nation, core–shell structured silica templates with a thin po-
rous silica shell around the silica core are therefore fabri-
cated because of the removal of octadecyl groups from the
silica/carbon framework. In the next step, the as-prepared sil-
ica templates are used to cast the HCNNSs. Before adding the
g-C3N4 precursor of cyanamide (CY), the silica templates are
post-acidified with dilute HCl to enhance the surface reactiv-
ity of the silica to the NH2-rich CY molecules via acid–base
interaction. Subsequently, a combined technology of sonica-
tion and vacuum is applied to extract the air trapped in the
silica matrix as well as to pump the aqueous CY solution
(50%) sufficiently into the pore channels. Compared with
molten state CY as the precursor, the aqueous CY solution
possesses better liquidity so that the CY molecules can access
the pore channels of the silica templates more easily under
milder reaction conditions. After calcination of the dried CY/
SiO2 hybrids at 550 °C followed by the removal of silica tem-
plate via NH4HF2 treatment, HCNNS-200 (particle size of 200
nm) and HCNNS-500 (particle size of 500 nm) are fabricated.
3.1.1. Morphological characteristics and textural proper-
ties. The morphological characteristics of the as-prepared two
silica templates and the corresponding HCNNSs are revealed
by TEM observation (Fig. 1). Both silica templates exhibit a
monodispersed spherical nanostructure with a condensed
core and a porous shell (Fig. 1a–d). The diameter and shell
thickness of the smaller template (Template-200) is ca. 200
nm and ca. 40 nm, while the diameter and shell thickness of
the larger one (Template-500) is ca. 500 nm and ca. 130 nm
(data presented in Table 1). Removing the silica templates
yields highly uniform hollow carbon nitride nanospheres
(HCNNS-200 and HCNNS-500) that inherit most of their geo-
metrical features from the parent silica templates (Fig. 1e–h).
The estimated shell thickness of HCNNS-200 is 40 nm, which
is strongly related to the thickness of the porous silica shell
over the condensed core of the template. However, the shell
thickness of HCNNS-500 is only ca. 100 nm, smaller than the
shell thickness of its silica templates. The result implies that
50% aqueous CY is hard to penetrate sufficiently into the
pore channels of the thick shell of the silica template due to
the long diffusion path and air trapped in the deep hole of
Template-500. It should be noted that the bright-dark pat-
terns existing in the shell of HCNNS-200 or HCNNS-500 sug-
gest well-kept porous structures after removal of the silica
templates.
2.4. Photoelectrochemical measurements
All electrochemical measurements were performed on a con-
ventional
three-electrode
electrochemical
workstation
equipped with Ag/AgCl (saturated KCl) and a Pt wire as the
reference electrode and the counter electrode, respectively.
FTO coated with the as-prepared photocatalyst film was used
as the working electrode for the electrochemical impedance
spectroscopy (EIS) Nyquist analysis, photocurrent test and
Mott–Schottky plot determination.
3. Results and discussion
3.1. Preparation and characterization of HCNNSs
The preparation of HCNNSs includes two main steps. As il-
lustrated in Scheme 1, a modified Stöber method is applied
to produce monodisperse core–shell structured silica tem-
plates with a dense core covered by a nanopore enriched
shell. Colloidal silica cores are firstly obtained by the hydroly-
The textural properties of various g-C3N4 samples are stud-
ied by nitrogen gas porosimetry measurements. As shown in
Scheme 1 Illustration of the route of the preparation of the HCNNSs.
This journal is © The Royal Society of Chemistry 2018
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
Fig. 2 Nitrogen gas adsorption–desorption isotherms (a), nitrogen gas
adsorption isotherm at P/P0 < 0.1 (inset of a), and corresponding BJH
pore-size distribution curves (b) of g-C3N4 and the HCNNSs.
sures (P/P0) lower than 0.1 (type I isotherm) and a type IV iso-
therm with a H3 hysteresis loop at P/P0 = 0.45–1.0 (Fig. 2a),
indicating their interesting micro/mesoporosity.26–28 Com-
pared with HCNNS-500, HCNNS-200 with a smaller particle
size and thinner shell exhibits a larger hysteresis loop, indi-
cating that more abundant mesopores are created in the shell
of the hollow structure. This excellent hierarchical porosity
renders HCNNS-200 a larger BET surface area (122 m2 g−1)
than HCNNS-500 (116 m2 g−1).
The interesting textural properties of the hollow carbon ni-
tride nanospheres are also reflected in their corresponding
BJH pore size distribution curves (Fig. 2b). HCNNS-200 ex-
hibits a sharp BJH pore size distribution curve centering at
3.6 nm, suggesting its uniform mesopores within the polymer
shell. As for HCNNS-500, its mesopores mainly center at 3.4
nm.
Compared with HCNNS-200, the pore size of HCNNS-500
is less well-distributed; meanwhile, its pores are less abun-
dant. This is because the thicker shell of Template-500 re-
sults in the poor permeation of CY molecules. Uniform meso-
pores combined with the thin shell of HCNNS-200 may
increase the accessibility of the reactants to the active sites
by bridging the exterior and interior surfaces, which is
expected to improve the heterogeneous photocatalytic
activity.
Fig. 1 TEM images of Template-200 (a and b), Template-500 (c and
d), HCNNS-200 (e and f), and HCNNS-500 (g and h).
3.1.2. Composition and structural information. The crystal
structures of various g-C3N4 are characterized by XRD mea-
surements. As shown in Fig. 3a, bulk g-C3N4 shows the typical
graphitic stacking structure with the (100) diffraction peak at
Table 1 Particle size (d), shell thickness (δ), BET surface area (SBET) and
band gap (Eg) of various g-C3N4 samples
Catalyst
d (nm)
δ (nm)
SBET (m2 g−1
)
Eg (eV)
Bulk g-C3N4
HCNNS-200
HCNNS-500
—
200
500
—
40
100
22
122
116
2.74
2.89
2.85
Fig. 2a, bulk g-C3N4 prepared by direct thermal condensation
of cyanamide exhibits type IV isotherms with a very small H3
hysteresis loop, signifying its poor mesoporosity owing to its
bulk morphology (Fig. S1†). Correspondingly, bulk g-C3N4
possesses a very small BET surface area (22 m2 g−1, Table 1).
As for the HCNNSs prepared by the silica casting method,
their isotherms show combined characteristics of a rapidly
increasing nitrogen gas adsorbed amount at relative pres-
Fig. 3 XRD patterns (a) and FT-IR spectra (b) of g-C3N4 and the
HCNNSs.
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2018

View Article Online
Catalysis Science & Technology
Paper
13.1° and (002) peak at 27.5°, which represent the character-
istic interlayer stacking and in-plane repeating structural mo-
tifs of the conjugated aromatic systems, respectively.29,30 The
graphite-like structure can also be detected for the HCNNSs.
Nevertheless, compared with the XRD pattern of the bulk
g-C3N4, the (100) diffraction peaks of both HCNNSs become
somewhat weak, suggesting the decreased order degree of the
in-plane structural motif.31 The result is ascribed that the
narrowed porous channels within the template may confine
the polymerization degree of carbon nitride. In addition, the
(002) diffraction peak for both HCNNSs shifts from 27.5° (g-
C3N4) to 27.9°, suggesting that the average interlayer distance
of the HCNNSs is shortened,13,21 leading to the compact car-
bon nitride framework with the porous structures in it. Inter-
estingly, a new diffraction peak at 17.8° appears, which is
assigned to the (600) crystal plane of mesoporous g-C3N4.32,33
The result further indicates the mesoporous structure within
the shell of the HCNNSs.
FTIR further confirms the graphic structure of the
HCNNSs (Fig. 3b). The peaks at 1200–1600 cm−1 are typical
stretching vibration modes of CN heterocycles, while the
peak at 810 cm−1 originated from the bending vibration of
the triazine unit, illustrating the tri-s-triazine motifs of both
g-C3N4 and HCNNSs.27 Moreover, the broad absorption band
located in the range of 3000–3300 cm−1 is assigned to the
stretching vibration of the N–H and O–H groups.
N atom ratio closer to 4/3. It is speculated that HCNNSs may
possess a much better structured CN framework with lower
oxygen-content.
Fig. 4b shows the XPS of bulk g-C3N4 and HCNNSs in the
O 1s binding energy region. For g-C3N4, the peaks centering
at 531.2 and 533.1 eV are assigned to oxygen species from the
CO and C–O groups of g-C3N4.27 These two peaks can also
be detected in the O 1s spectra of the HCNNSs. However, the
peak intensity of the HCNNSs is much weaker than that of
the bulk g-C3N4, which is attributed to the fact that the con-
densation of cyanamide is much more thorough within the
mesopores of the silica template, further reducing the con-
tent of the CO and C–O groups.
The C 1s XPS spectra of the tested g-C3N4 samples display
three characteristic peaks centering at 284.6, 286.3 and 288.1
eV (Fig. 4c), corresponding to carbon species from the
adsorbed carbon, C–NH2 group and sp2 carbon atoms
bonded to nitrogen atoms in an aromatic ring (N–CN), re-
spectively.31 Compared with bulk g-C3N4, the peak intensity
of the C–NH2 group in the HCNNSs decreases, while the peak
intensity of N–CN increases. This result indicates that more
CN hybridized aromatic rings have been built during the
hard template assisted thermal condensation process, lead-
ing to the well-structured CN framework of carbon nitride
hollow spheres.
The N 1s XPS spectra of the tested g-C3N4 samples are
fitted into three peaks (Fig. 4d). For bulk g-C3N4, the peak at
398.2 eV is assigned to sp2 hybridized aromatic nitrogen
atoms bonded to carbon atoms (C–NC), while the peak at
398.9 eV relates to the tertiary nitrogen N–(C)3 groups linking
the structural motif (C6N7). The last peak at 400.5 eV corre-
sponds to the nitrogen atoms from the –NH2 group.27 As for
the HCNNSs, the peaks of the C–NC (398.6 eV), N–(C)3
(399.9 eV) and –NH2 (401.2 eV) groups slightly shift to higher
binding energy regions, which is in accordance with the hol-
low porous carbon nitride prepared from cyanamide.16 Due
to the restriction of mesopores in the SiO2 templates, the
thermal condensation of CY molecules undergoes a specific
situation and results in the better structured CN framework
of HCNNSs with less oxygen content than g-C3N4, which re-
duces the electron density around the nitrogen atom and
thereby increasing the binding energy.34
A high-resolution XPS surface probe technique is carried
out to investigate the chemical states of the as-prepared sam-
ples (Fig. 4). As shown in Fig. 4a, three elements, O, N and C,
are detected in both g-C3N4 and HCNNSs. Compared with
bulk g-C3N4, the peak intensity of O 1s in the HCNNSs de-
creases, indicating the lower O content in the HCNNSs, while
the intensity ratio of N 1s to C 1s increases, resulting in a C/
On the basis of the above discussion, hollow carbon ni-
tride nanospheres with a well-structured CN framework have
been fabricated. In the presence of multiple CN hybridized
aromatic rings with less oxygen-containing groups, the hol-
low carbon nitride nanospheres are expected to facilitate the
migration and separation of the photogenerated charge
carriers.
3.1.3. Optical absorption properties. The light absorption
property of the as-prepared photocatalysts is characterized by
UV-vis/DRS (Fig. 5). As shown in Fig. 5a, g-C3N4 exhibits a
semiconductor absorption in the region of 200 to 460 nm,
originating from the charge transfer from the valence band
(VB) populated by N 2p orbitals to the conduction band (CB)
formed by C 2p orbitals.35 Compared with the bulk g-C3N4,
Fig. 4 XPS survey (a) and high-resolution spectra in O 1s (b), C 1s (c)
and N 1s (d) binding energy regions of g-C3N4 and the HCNNSs.
This journal is © The Royal Society of Chemistry 2018
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
plying their n-type semiconductor structure. The estimated
flat band potentials are −1.09, −1.12 and −1.13 V vs. NHE for
g-C3N4, HCNNS-500 and HCNNS-200, respectively, indicating
the enhanced reduction capability of photogenerated
electrons (e−CB) generated on the HCNNSs. The relative posi-
tions of the VB for the samples are determined using the XPS
valence spectra. As shown in Fig. 5d, HCNNS-200 exhibits the
most positive location (1.76 V), which is 0.03 V and 0.11 V
higher than that of HCNNS-500 (1.73 V) and g-C3N4 (1.65 V),
respectively, indicating the strongest oxidizing ability for the
photogenerated holes (h+VB) generated on HCNNS-200. Addi-
tionally, such values of VB are consistent with the valence
bands calculated by combining the result of band gaps and
flat band potentials. Therefore, the band structure of g-C3N4
and the HCNNSs are well resolved as illustrated in Fig. 5e.
The most negative CB and positive VB positions of HCNNS-
200 is attributed to the strongest quantum size effect induced
by the smallest nanoparticle size of HCNNS-200, leading to
the largest bandgap energy of HCNNS-200.
3.1.4. Charge separation and transfer property. The photo-
generated charge carrier separation and transfer property is
evaluated by electrochemical impedance spectroscopy (EIS),
photoelectrochemistry tests and time-resolved fluorescence
spectroscopy. As shown in Fig. 6a, the Nyquist semicircles of
the HCNNSs show a smaller diameter compared with that of
the bulk g-C3N4; moreover, the Nyquist semicircle diameter
of HCNNS-200 is smaller than that of HCNNS-500. Therefore,
HCNNS-200 exhibits the fastest electronic conductivity and
the most efficient charge separation. This result is further
supplemented by the photocurrent measurements shown in
Fig. 6b. All samples show a stable and reproducible photocur-
rent response, and the photocurrent intensity follows the or-
der: HCNNS-200 > HCNNS-500 > bulk g-C3N4, which reflects
the fastest separation and transfer efficiency of the photo-
generated charge carriers within HCNNS-200. In addition, the
fluorescence decay spectra also illustrate the longer calcu-
lated average life-time of HCNNS-200 compared with HCNNS-
500 and g-C3N4 (Fig. S2†). For example, the calculated average
life-time values are 3.601 ns for HCNNS-200, 3.188 ns for
HCNNS-500 and 2.938 ns for the bulk g-C3N4.
Fig. 5 UV-vis/DRS (a), the corresponding band gap plots (b), Mott–
Schottky plots (c), VB XPS spectra (d) and band structures (e) of
g-C3N4 and the HCNNSs.
the HCNNSs show a slight blueshift of the absorption edge,
i.e. the absorption band edges are 440 and 450 nm for
HCNNS-200 and HCNNS-500, respectively, which might be
due to the quantum size effect.13,21 In addition, the enhanced
light absorption in the range of 440 to 800 nm for the
HCNNSs compared to the bulk g-C3N4 is believed to reflect
the ability of the hollow nanospheres to have multiple reflec-
tions within the structure, which increases the effective path
length for light absorption,21,36 making these materials prom-
ising candidates for photocatalytic applications. For two
hollow g-C3N4 nanospheres, HCNNS-200 with a smaller size
and thinner shell exhibits a slightly stronger light absorption
ability in the visible light region. The band gaps of the bulk
g-C3N4 and HCNNSs are calculated by the formula αhν = A
(hν − Eg)1/2, where α, ν, Eg and A are the absorption coeffi-
cient, the light frequency, the band gap and a constant, re-
spectively. The Eg values are therefore estimated to be 2.74 (g-
C3N4), 2.89 (HCNNS-200) and 2.85 eV (HCNNS-500), respec-
tively (Fig. 5b and Table 1).
Based on the above results, the fast charge separation and
transfer property of HCNNS-200 is due to its perfect small
and hollow spherical nanostructures with micro- and
Besides a suitable band gap, the suitable CB and VB edge
potentials of the semiconductor are also important for the
photocatalytic reaction. Herein, Mott–Schottky plots are ap-
plied to verify the flat-band of g-C3N4 and the HCNNSs, which
can be used to estimate the CB level.37 As shown in Fig. 5c,
all samples exhibit a positive slope in the linear region, im-
Fig. 6 EIS Nyquist plots (a) and photocurrent responses (b) of g-C3N4
and the HCNNSs electrodes.
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2018

View Article Online
Catalysis Science & Technology
Paper
mesopores in the shell. At first, the hollow nanostructure can
improve the light-harvesting efficiency via the multiple reflec-
tions within it, increasing the probability of electron transi-
tion and the population of photogenerated electrons and
holes. Secondly, the well-structured CN framework facilitates
the charge transfer to the surface without trapping at defect
sites. Finally, plentiful micro- and mesopores in the thin
shell can shorten the migration distance of the photo-
generated charge carriers to the surface of the photocatalyst.
Therefore, it is worth anticipating that HCNNS-200 will give
excellent photocatalytic activity.
of 99% LEVO over the HCNNSs, which is at least two times
faster than that of P25 TiO2 or bulk g-C3N4. In the cases of
the two HCNNSs, HCNNS-200 exhibits a much higher visible-
light photocatalytic activity than HCNNS-500. For example, af-
ter visible-light irradiation for 10 min, the degradation of
LEVO reaches 99% and 90%, respectively, over HCNNS-200
and HCNNS-500. The variations of the UV-vis absorption
spectra of LEVO under visible-light irradiation of HCNNS-200
show that the characteristic UV-vis absorption peaks at 227,
290 and 326 nm of LEVO reduced gradually with prolonga-
tion of the irradiation time (insert of Fig. 7a). The result is
consistent with those shown in Fig. 7a. HCNNS-200 still ex-
hibits the highest photocatalytic activity in the degradation
of TMP (Fig. 7b), and it takes less than 60 min to degrade
TMP completely.
3.2. Photocatalytic studies
3.2.1. Visible-light photocatalytic activity. The visible-light
photocatalytic activity of HCNNSs is firstly evaluated by the
degradation of levofloxacin (LEVO). LEVO is a well-known
fluoroquinolone (FQ) antibiotic that is extensively used in hu-
man and veterinary medicine, and it has strong recalcitrance
to microbial degradation and photolysis.38 For comparison,
bulk g-C3N4 and Degussa P25 TiO2 are also tested under the
same conditions. Before light irradiation, the adsorption be-
haviors of the tested photocatalysts are studied (Fig. 7a). The
adsorption–desorption equilibrium between LEVO molecules
and the photocatalyst was reached after stirring the suspen-
sion for 30 min, and the adsorption capacity of P25 TiO2,
bulk g-C3N4, HCNNS-500 and HCNN-200 towards LEVO mole-
cules is 6%, 8%, 18% and 19%, respectively. The stronger ad-
sorption ability of the HCNNSs than the bulk g-C3N4 is
closely related to their perfect hollow nanosphere structure
and large BET surface area. Considering that the photocata-
lytic reaction takes place on the surface of the catalyst, the
stronger adsorption ability of the photocatalyst toward the
substrate is beneficial to improving its photocatalytic activ-
ity.39 Under visible-light irradiation, the direct photolysis of
LEVO can be negligible, however, significant degradation of
LEVO occurs in the presence of both the photocatalyst and
visible-light irradiation. The photocatalytic activity of the
HCNNSs obviously outperforms P25 TiO2 and bulk g-C3N4.
Only 10–15 min was needed to finish the photodegradation
3.2.2. Mineralization of LEVO. Mineralization of organic
pollutants reflects the deep oxidation of the photocatalysts.
Here, the mineralization ability of the HCNNSs is evaluated
by monitoring the concentration changes of the yielded or-
ganic acids and inorganic anions during the HCNNS-200-
photocatalyzed LEVO degradation process. As shown in Fig.
S3,† both citric acid and acetic acid are found by IC analysis.
The concentration of the yield citric acid increases to 4 mg
L−1 after 4 h of visible-light irradiation and was kept at this
level for the next 12 h (Fig. 8a). While the concentration of
the yield acetic acid increases gradually to 16 mg L−1 with
visible-light irradiation for 16 h, suggesting the continuous
mineralization of LEVO. Additionally, F− and NO3 ions are
−
also detected by IC (Fig. 8b). The concentration of F− in-
creases rapidly to 0.75 mg L−1 after 4 h of irradiation, and
then slowly increases to 0.85 mg L−1 after the next 12 h,
finishing at 85% conversion of LEVO into the final product
−
F−. Differently, the concentration of NO3 increases gradually
to 4.3 mg L−1 with light irradiation for 16 h, which is only
−
43% of the expected concentration of NO3 . The above results
confirm that the as-prepared HCNNSs can efficiently mineral-
ize the pollutants under visible-light irradiation, working as
the robust photocatalysts to decompose organic pollutants in
water.
3.2.3. Regeneration and reusability. Taking HCNNS-200 as
the representative photocatalyst, the regeneration and reus-
ability of the as-prepared hollow g-C3N4 nanospheres are
Fig. 7 Adsorption property and photocatalytic activity of the HCNNSs
towards the degradation of LEVO (a) and TMP (b) under visible-light ir-
radiation. The UV-vis absorption spectra of LEVO (insert of a) solution
Fig. 8 Concentration changes of the yielded citric acid and acetic
acid (a), and F− and NO3 ions (b) in the HCNNS-200-photocatalyzed
−
as
a
function of irradiation time during the HCNNS-200-
photocatalyzed degradation process (catalyst amount 50 mg; c0 = 20
mg L−1; volume 50 mL; 400 nm < λ < 680 nm).
LEVO degradation system (catalyst amount 50 mg; c0 = 20 mg L−1
volume 50 mL; 400 nm < λ < 680 nm).
;
This journal is © The Royal Society of Chemistry 2018
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
investigated by selecting LEVO as the target compound. For
each recycling run, HCNNS-200 is collected by centrifugation,
and then it is washed with ethanol at 60 °C three times. As
shown in Fig. 9a, no apparent deactivation of the photo-
catalyst is observed after six consecutive runs. The XRD mea-
surements (Fig. 9b) and TEM observations (Fig. 9c and d)
confirm that the crystal structure and the morphology of the
used HCNNS-200 have scarcely changed after the photocata-
lytic reaction. These results suggest that the HCNNSs can
work as recoverable visible-light-driven photocatalysts for the
decomposition of aqueous organic pollutants.
−
Fig. 10 EPR spectra of DMPO–˙O2 in methanol dispersion and
DMPO–˙OH in aqueous dispersion with irradiation for 120 s in the
presence of HCNNS-200 (a), and influence of various scavengers on
the visible-light photocatalytic activity of the HCNNS-200 toward the
degradation of LEVO (b).
3.3. Mechanism consideration
In order to understand the mechanism underlying the photo-
catalytic degradation of aqueous pollutants over the HCNNSs,
the reactive species generated during the process of the
HCNNS-200-photocatalyzed aqueous LEVO degradation are
identified by ESR measurements as well as free radical and
hole trapping experiments.
−
duct. This result evidences the generation of ˙O2 radicals in
the HCNNS-200 photocatalytic system.
Free radical and hole trapping experiments obtain the
same results with those of the ESR measurements. As shown
in Fig. 10b, compared with the scavenger-free HCNNS-200
photocatalytic system, the degradation rate of LEVO is decel-
erated obviously in the presence of 1 mmol L−1 of EDTA-2Na
(hole scavenger), while the degradation of LEVO is almost
−
In order to detect the formation of ˙O2 and ˙OH radicals
in the HCNNS-200 photocatalytic system, 5,5-dimethyl-1-
pyrroline-N-oxide (DMPO) spin-trapping EPR measurements
are carried out. As shown in Fig. 10a, the signals correspond-
inhibited after the addition of
1
mmol L−1 of 1,4-
−
benzoquinone (BQ, scavenger of ˙O2 radicals). However, in
the presence of tert-butyl alcohol (t-BuOH, 1 mmol L−1, scav-
enger of ˙OH radicals), the degradation rate of LEVO is hardly
−
ing to the DMPO–˙O2 and DMPO–˙OH adducts are hardly ob-
served in the dark. After visible-light irradiation, the signal of
the DMPO–˙OH adduct is still not found in this system,
which suggests that the ˙OH radical is not the active species
for the photodegradation of organic pollutants. However, an
obvious quartet signal with an intensity ratio of 1 : 1 : 1 : 1 is
observed in this system under visible-light irradiation, which
inhibited. This result indicates that h+VB, especially ˙O2 , is the
−
main active species that is responsible for the oxidization of
LEVO completely under the visible-light irradiation.
The combination of the physicochemical properties,
photoelectrochemical properties and the identified active spe-
cies confirms that the enhanced visible-light photocatalytic
activity of nanopore enriched hollow g-C3N4 nanospheres
with a smaller particle size (HCNNS-200) originated from the
contribution of its perfect morphological characteristics. On
the one hand, the hollow nanostructure of HCNNS-200 can
improve the light-harvesting efficiency via multiple reflec-
tions within the nanostructure; meanwhile, the smaller size
and thinner shell of HCNNS-200 are beneficial for the light
penetrating and reflecting within their void space. Both fac-
tors give rise to higher electron transition probability and
thereby higher population of the photogenerated electrons
and holes. On the other hand, the rapid separation and
transfer of these photogenerated charge carriers within
HCNNS-200 is another important reason for its excellent
photocatalytic activity. The well-structured CN framework
with lower oxygen-content has been built during the synthe-
sis of HCNNS-200, which can promote the separation and
transfer of the photogenerated charge carriers owing to the
reduced defect sites. Additionally, the thinner shell of
HCNNS-200 with plentiful pore channels within it efficiently
shortens the migration distance of the photogenerated
charge carriers to the surface and thereby minimizes the re-
combination of e−CB and hV+B. Additionally, owing to the quan-
tum size effect, HCNNS-200 exhibits an enlarged bandgap
−
is assigned to the characteristic signal of the DMPO–˙O2 ad-
Fig. 9 Recycling experiments of the visible-light photocatalytic degra-
dation of LEVO over HCNNS-200 (a). XRD patterns of HCNNS-200 be-
fore and after the reaction (b). TEM images of HCNNS-200 before (c)
and after (d) the reaction.
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2018

View Article Online
Catalysis Science & Technology
Paper
framework and shortened photogenerated charge carrier mi-
gration distance, eC−B and hV+B rapidly separate from each
other and then transfer/migrate to the inner and external sur-
faces of the catalyst. Subsequently, e−CB reacts with the
−
dissolved dioxygen to produce ˙O2 radicals. Under the attack
−
of both the ˙O2 radicals and h+VB, LEVO is finally degraded to
NO3 , F−, H2O and CO2.
−
The proposed degradation pathway of LEVO over visible-
light irradiated HCNNS-200 can support the above mecha-
nism considered. As displayed in Fig. S4,† ESI-MS identifies
eleven intermediates generated during the process of the
photocatalytic degradation of LEVO; additionally, the IC tech-
nique finds two aliphatic acids (citric acid and acetic acid,
Scheme 2 Mechanism of the visible-light photocatalytic degradation
of aqueous LEVO over HCNNS-200.
with the improved oxidation ability of h+VB for deep oxidation
of organic pollutants. Finally, uniform pore channels com-
bined with thin shell as well as monodispersed spherical
nanostructures render HCNNS-200 with a large BET surface
area, which can facilitate the adsorption of the reactants and
increase the population of active sites. Consequently, the ac-
cessibility of the reactants to the active sites is improved sig-
nificantly, promoting the efficiency of photocatalytic
degradation.
On the basis of the above discussion, a reasonable mecha-
nism of the visible-light photocatalytic degradation of LEVO
over the nanopore enriched hollow carbon nitride nano-
spheres is revealed and illustrated in Scheme 2. Under
visible-light irradiation, an increased amount of photo-
generated charges carriers are generated in HCNNS-200 be-
cause of the multiple light reflections within the hollow
spherical nanostructures. Due to the well-structured CN
Fig. S3†) as well as inorganic anions (NO3 and Cl−, Fig. 8) in
−
the catalytic system. On the basis of these results, the path-
way of the visible-light photocatalytic degradation of aqueous
LEVO over HCNNS-200 is tentatively put forward (Scheme 3).
Under the attack of ˙O2 and h+VB, LEVO is firstly decomposed
−
to compound 2 via the cleavage of the piperazinyl ring. Sub-
sequently, compound
2
suffers from
a
series of
decarbonylations to produce compounds 3, 4 and 5. Further
photooxidation of compound 5 generates compound 6, and
then compound 6 suffers decarboxylation to produce com-
pound 7. After the leakage of the C–F bond from compound
7, compound 9 is yielded, accompanied by the release of a F−
anion into the reaction media. In addition, the leakage of the
C–F bond also occurs in compound 5, which gives rise to
compound 8 and a F− anion. Continuous photooxidation of
compound 8 leads to compound 9. Subsequently, simulta-
neous photooxidation and decarboxylation from compound 9
Scheme 3 Pathway of the visible-light photocatalytic degradation of aqueous LEVO over HCNNS-200.
This journal is © The Royal Society of Chemistry 2018
Catal. Sci. Technol.

View Article Online
Paper
Catalysis Science & Technology
results in compound 10. Further photooxidation and deami-
nation from compound 10 produces compound 11, and the
process is accompanied by the production of a NO3 anion.
5 L. F. Cui, J. L. Song, A. F. McGuire, S. F. Kang, X. Y. Fang,
J. J. Wang, C. C. Yin, X. Li, Y. G. Wang and B. X. Cui, ACS
Nano, 2018, 12, 5551.
−
Continuous photooxidation of compound 11 leads to the
opening of the aromatic ring, together with the formation of
citric acid and acetic acid. Under the continuous attack by
6 Z. Y. Fang, Y. Z. Hong, D. Li, B. F. Luo, B. D. Mao and W. D.
Shi, ACS Appl. Mater. Interfaces, 2018, 10, 20521.
7 Y. Yang, Y. Guo, F. Liu, X. Yuan, Y. Guo, S. Zhang, W. Guo
and M. Huo, Appl. Catal., B, 2013, 142–143, 828.
8 L. B. Jiang, X. Z. Yuan, Y. Pan, J. Liang, G. M. Zeng, Z. B. Wu
and H. Wang, Appl. Catal., B, 2017, 217, 388.
9 Y. X. Yang, L. Geng, Y. H. Guo and Y. N. Guo, J. Chem.
Technol. Biotechnol., 2016, 92, 1236.
˙O2 radicals and h+VB, citric acid and acetic acid are eventu-
−
ally decomposed into CO2 and H2O.
4. Conclusions
Nanopore enriched hollow carbon nitride nanospheres are
successfully synthesized by using silica templates composed
of a solid core and a polyporous shell. HCNNS-200 with a
small particle size of 200 nm and a thin shell thickness of
40 nm exhibits extremely high visible-light photocatalytic
ability toward the degradation of aqueous contaminants of
emerging concern. It takes less than 10 and 60 min for
HCNNS-200 to finish the decomposition of LEVO and TMP,
respectively, which outperforms bulk g-C3N4 and HCNNS-
500. Moreover, HCNNS-200 can be reused six times without
obvious activity loss. Such an excellent photocatalytic ability
of HCNNS-200 is attributed to its unique hollow nanostruc-
ture which endows HCNNS-200 with enhanced light-
harvesting ability, faster charge separation and transfer effi-
ciency, easier accessibility of the reactants to the active sites
and stronger oxidation ability of h+VB. The active species
10 L. J. Gao, Y. G. Li, J. B. Ren, S. F. Wang, R. N. Wang, G. S. Fu
and Y. Hu, Appl. Catal., B, 2017, 202, 127.
11 T. W. P. Seadira, G. Sadanandam, T. Ntho, C. M. Masuku
and M. S. Scurrell, Appl. Catal., B, 2018, 222, 133.
12 G. Prieto, H. Tüysüz, N. Duyckaerts, J. Knossalla, G.-H. Wang
and F. Schüth, Chem. Rev., 2016, 116, 14056.
13 J. H. Sun, J. S. Zhang, M. W. Zhang, M. Antonietti, X. Z. Fu
and X. C. Wang, Nat. Commun., 2012, 3, 1139.
14 D. D. Zheng, C. Y. Pang and X. C. Wang, Chem. Commun.,
2015, 51, 17467.
15 X. Zhong, L. Liu, Y. Jiang, X. D. Wang, L. Wang, G. L.
Zhuang, X. N. Li, D. H. Mei, J.-g. Wang and S. Su Dang,
ChemCatChem, 2015, 7, 1826.
16 H. X. Wu, D. Y. Chen, N. J. Li, Q. F. Xu, H. Li, J. H. He and
J. M. Lu, Nanoscale, 2016, 8, 12066.
17 Y.-S. Jun, Z. Lee Eun, X. C. Wang, H. Hong Won, D.
Stucky Galen and A. Thomas, Adv. Funct. Mater., 2013, 23,
3661.
18 Z. Meng, S. J. Li, H. J. Ying, X. Xu, X. L. Zhu and W.-Q. Han,
Adv. Mater. Interfaces, 2017, 4, 1601195.
19 X. J. Wang, J. Feng, Y. C. Bai, Q. Zhang and Y. D. Yin, Chem.
Rev., 2016, 116, 10983.
20 D. D. Zheng, C. Y. Pang, Y. X. Liu and X. C. Wang, Chem.
Commun., 2015, 51, 9706.
including ˙O2 radicals and h+VB produced in the HCNNS-200
−
system are responsible for the photooxidization and miner-
alization of LEVO and TMP. Additionally, a more systematic
photocatalytic degradation pathway of LEVO is proposed
based on IC and HPLC-MS analysis. It is expected that
HCNNS-200 can be used as a genuine heterogeneous visible-
light-driven photocatalyst that may find wide applications in
environmental remediation.
21 D. D. Zheng, C. J. Huang and X. C. Wang, Nanoscale,
2015, 7, 465.
22 D. D. Zheng, X.-N. Cao and X. C. Wang, Angew. Chem., Int.
Ed., 2016, 55, 11512.
Conflicts of interest
There are no conflicts to declare.
23 D. D. Zheng, G. G. Zhang and X. C. Wang, Appl. Catal., B,
2015, 179, 479.
Acknowledgements
24 D. D. Zheng, G. G. Zhang, Y. D. Hou and X. C. Wang, Appl.
Catal., A, 2016, 521, 2.
25 A. Molinari, E. Sarti, N. Marchetti and L. Pasti, Appl. Catal.,
B, 2017, 203, 9.
This work was supported by the Natural Science Fund Council
of China (51608102; 21573038) and Jilin province science and
technology development projects (20160520177JH).
26 Y. N. Sun, Q. Q. Zhang, P. P. Zhang, D. Y. Song and Y. H.
Guo, ACS Sustainable Chem. Eng., 2018, 6, 6771.
27 Y. X. Yang, L. Geng, Y. N. Guo, J. Q. Meng and Y. H. Guo,
Appl. Surf. Sci., 2017, 425, 535.
Notes and references
1 W.-J. Ong, L.-L. Tan, Y. H. Ng, S.-T. Yong and S.-P. Chai,
Chem. Rev., 2016, 116, 7159.
2 Z. X. Zhou, Y. Y. Zhang, Y. F. Shen, S. Q. Liu and Y. J.
Zhang, Chem. Soc. Rev., 2018, 47, 2298.
3 Y. Shiraishi, S. Shiota, Y. Kofuji, M. Hashimoto, K. Chishiro,
H. Hirakawa, S. Tanaka, S. Ichikawa and T. Hirai, ACS Appl.
Energy Mater., 2018, 1, 4169.
28 S. An, D. Y. Song, Y. N. Sun, Q. Q. Zhang, P. P. Zhang and
Y. H. Guo, ACS Sustainable Chem. Eng., 2018, 6, 3113.
29 P. Niu, L. Zhang, G. Liu and H.-M. Cheng, Adv. Funct.
Mater., 2012, 22, 4763.
30 S. Yang, Y. Gong, J. Zhang, L. Zhan, L. Ma, Z. Fang, R.
Vajtai, X. Wang and P. M. Ajayan, Adv. Mater., 2013, 25,
2452.
4 C. Cometto, R. Kuriki, L. J. Chen, K. Maeda, T.-C. Lau, O.
Ishitani and M. Robert, J. Am. Chem. Soc., 2018, 140, 7437.
Catal. Sci. Technol.
This journal is © The Royal Society of Chemistry 2018

View Article Online
Catalysis Science & Technology
Paper
31 Y. X. Wang, H. Wang, F. Y. Chen, F. Cao, X. H. Zhao, S. G.
Meng and Y. J. Cui, Appl. Catal., B, 2017, 206, 417.
36 X. C. Wang, K. Maeda, X. F. Chen, K. Takanabe, K. Domen,
Y. D. Hou, X. Z. Fu and M. Antonietti, J. Am. Chem. Soc.,
2009, 131, 1680.
37 S. Chu, Y. Wang, Y. Guo, J. Y. Feng, C. C. Wang, W. J. Luo,
X. X. Fan and Z. G. Zou, ACS Catal., 2013, 3, 912.
38 X.-J. Wen, C.-G. Niu, H. Guo, L. Zhang, C. Liang and G.-M.
Zeng, J. Catal., 2018, 358, 211.
32 J. Zhang, F. Guo and X. Wang, Adv. Funct. Mater., 2013, 23, 3008.
33 S. Y. Tan, Z. P. Xing, J. Q. Zhang, Z. Z. Li, X. Y. Wu, J. Y. Cui,
J. Y. Kuang, Q. Zhu and W. Zhou, J. Catal., 2018, 357, 90.
34 X. L. Yang, F. F. Qian, G. J. Zou, M. L. Li, J. R. Lu, Y. M. Li
and M. T. Bao, Appl. Catal., B, 2016, 193, 22.
35 X. C. Wang, K. Maeda, A. Thomas, K. Takanabe, G. Xin, J. M.
Carlsson, K. Domen and M. Antonietti, Nat. Mater., 2008, 8, 76.
39 Q. Liu, Y. R. Guo, Z. H. Chen, Z. G. Zhang and X. M. Fang,
Appl. Catal., B, 2016, 183, 231.
This journal is © The Royal Society of Chemistry 2018
Catal. Sci. Technol.