Enhanced nonradical catalytic oxidation by encapsulating cobalt into nitrogen doped graphene: highlight on interfacial interactions

Article abstract of DOI:10.1039/d0ta10662c

Supported metal catalysts are widely used for heterogeneous catalytic processes (e.g., Fenton-like reaction), but the mechanisms of interfacial processes are still ambiguous. Herein, unique nanocarbon based catalysts with Co nanoparticles encapsulated in

Full text of DOI:10.1039/d0ta10662c

Journal of
Materials Chemistry A
View Article Online
View Journal | View Issue
PAPER
Enhanced nonradical catalytic oxidation by
encapsulating cobalt into nitrogen doped
graphene: highlight on interfacial interactions†
Cite this: J. Mater. Chem. A, 2021, 9,
7198
a
Xiaoyong Yu,‡a Lijing Wang,‡b Xin Wang,
Hongzhi Liu,a Ziyuan Wang,a
Yixuan Huang,a Guoqiang Shan,a Weichao Wang
and Lingyan Zhu
b
a
*
*
Supported metal catalysts are widely used for heterogeneous catalytic processes (e.g., Fenton-like
reaction), but the mechanisms of interfacial processes are still ambiguous. Herein, unique nanocarbon
based catalysts with Co nanoparticles encapsulated in nitrogen (N)-doped graphene (Co@NG) were
prepared by calcination of Co based metal–organic frameworks (MOFs), which showed excellent
catalytic performance for peroxymonosulfate (PMS) activation. Comprehensive characterization revealed
that there were strong interfacial interactions between Co and the NG layer due to the presence of
special nitrogen species, especially graphitic-N. Density functional theory calculations suggested that the
strong interfacial interactions provided optimal active sites with low adsorption energy (ꢀ1.99 eV) for
PMS accumulation, which enabled the generation of highly oxidizing NG–PMS* intermediates as
evidenced by in situ Raman microscopy. Electrochemical analyses revealed that the interfacial
interactions facilitated surface-to-surface electronic communication across atomic interface-bonding
(N–Co). Consequently, phenol was quickly degraded by the NG–PMS* via direct oxidation by an anodic-
like nonradical process, and reasonable graphitic-N (G-N) content and pore size are important for this
process. Phenol at 1 mM was completely removed within only 12 min by Co@NG-900 (which was
prepared at 900 ꢁC), with an apparent rate constant 20 times higher than that of pure NG-900. This
work sheds new insights on the critical role of interfacial interactions in nonradical PMS activation.
Received 2nd November 2020
Accepted 11th February 2021
DOI: 10.1039/d0ta10662c
rsc.li/materials-a
radical scavenging by nontarget compounds in complex water
matrices.5 The mechanism for nonradical persulfate activation
1. Introduction
by carbocatalysts is frequently ascribed to singlet oxygenation or
mediated electron-transfer.5–7 However, the relative importance
of these potential nonradical pathways is debatable.
Carbocatalysts including graphene,8 nanodiamonds,9
carbon nanotubes,2 carbon nitride,3 and biochar10 have been
investigated for nonradical persulfate activation, but their
performances are not satisfactory. In order to improve the
activation efficiency of persulfate, multiple methods were
proposed by adjusting surface functional groups,3,11 modulating
defects,12 and adjusting the electron state of C atoms.6
In the last decade, particular attention has been paid to
supported carbon-based material mediated synthetic
approaches to prepare highly effective catalysts.13,14 Recent
investigations revealed the advantages in preparing highly
efficient heterogeneous catalytic materials from MOF precur-
sors by means of thermal decomposition for energy storage and
conversion,15 and hydrogenation.16 Construction of strong
interaction interfaces by encapsulating metal nanoparticles
with supporters has been proven to be a promising strategy to
greatly enhance the catalytic performance attributed to the
novel characteristics originating from the physical and chem-
ical interactions between the metal species and carbon
In recent decades, the discharge of large amounts of organic
pollutants via industrial wastewater and municipal sewage
leads to their severe contamination in natural water systems,
thus causing unexpected adverse effects on the health of
ecosystem and humans. Persulfate-based advanced oxidation
processes (PS-AOPs) have attracted extensive attention for
removal of phenolic compounds because of the merits of high
oxidation capacity, rapid reaction kinetics, and facile opera-
tion.1–3 Unlike most metal oxide-catalyzed persulfate activation
that produces a variety of radicals,4 carbocatalysts trigger non-
radical activation of persulfate, which is less susceptible to
aMOE Key Laboratory of Pollution Processes and Environmental Criteria, Tianjin Key
Laboratory of Environmental Remediation and Pollution Control, College of
Environmental Science and Engineering, Nankai University, No. 38 Tongyan Road,
Jinnan District, Tianjin, 300350, China. E-mail: Zhuly@nankai.edu.cn
bDepartment of Electronics, National Institute for Advanced Materials, Renewable
Energy Conversion and Storage Center, Tianjin Key Laboratory of Photo-Electronic
Thin Film Device and Technology, Nankai University, Tianjin 300071, China.
E-mail: weichaowang@nankai.edu.cn
† Electronic supplementary information (ESI) available. See DOI:
10.1039/d0ta10662c
‡ Yu and Wang contributed equally to this work.
7198 | J. Mater. Chem. A, 2021, 9, 7198–7207
This journal is © The Royal Society of Chemistry 2021

View Article Online
Paper
Journal of Materials Chemistry A
materials.17–20 Ito and Ohto encapsulated NiMo alloys in gra- All solutions were prepared in 18.2 MU cm ultra-pure water
phene to realize both high catalytic activity and high chemical produced by a RephiLe Biocel water system (Shanghai, China).
stability. The interface between NiMo and graphene signi-
cantly reduced the adsorption energy of intermediates, thereby
remarkably enhancing the oxygen evolution reaction (OER)
activity.20 Zeolitic imidazolate frameworks (ZIFs), as a member
of the MOF family, always serve as excellent precursors for
nanocarbon catalysts allowing abundant mesopores21 and
active nitrogen species.17,22 Calcination of ZIFs can generate
a core–shell structure with transition metal nanomaterials
2.2 Preparation of ZIF-based catalysts
The ZIF-67 precursor was synthesized according to a previous
report with few modications.24 Solution A was prepared by
dissolving 2 mmol of cobaltous nitrate hexahydrate in 50 mL of
methanol under agitated stirring for nearly 5 min. Solution B
was prepared by dissolving 8 mmol of 2-MeIM in 50 mL of
methanol. Solution B was shaken for 5 min and poured in
solution A with continuing stirring. The color of solution A
changed from pink to dark purple immediately. Aer stirring
for 10 min, the reaction solution was allowed to settle for 24 h.
The resulting precipitates were collected by centrifugation,
thoroughly washed with methanol several times and nally
which were tightly encapsulated by
a carbon shell or
network.22,23 Such capsule-like morphology coupled with
nitrogen doping displayed signicant interfacial interactions
between embedded metallic nanoparticles and the coated
carbon shells.25
In light of these results, special capsule-like catalysts
(Co@NG) with strong interfacial interactions were prepared by
adjusting the calcination process of ZIF-67 for nonradical per-
ꢁ
dried in a vacuum at 60 C.
In a typical process for the synthesis of Co@NG composites
(Co@NGs), 100 mg of ZIF-67 was put in a ceramic boat, heated
to 500 ꢁC at a ramp rate of 5 ꢁC minꢀ1 and maintained for 1 h
in a tube furnace. The furnace temperature was increased to
a predetermined temperature (700, 800, 900,ꢀ11000 and
1100 ꢁC respectively) at a ramp rate of 2 ꢁC min and that
temperature was maintained for 2 h to produce the desired
Co@NG-T materials, where T is the pyrolysis temperature.
The furnace was allowed to cool down to room temperature at
a ramp rate of 5 ꢁC minꢀ1. During the whole process, the
furnace was in an atmosphere of Ar. The as prepared Co@NGs
were put in 0.1 M H2SO4 for 2 h to remove the surface Co
species. To understand the critical roles of Co nanoparticles
encapsulated in Co@NGs,ꢁ the obtained Co@NG-900 was
treated in aqua regia at 80 C for 4 h to remove Co from the
composite, and the product was labeled NG-900. CoxO was
prepared by calcining Co@NG-900 in a muffle furnace for 1 h
oxymonosulfate
(PMS)
activation.
Such
capsule-like
morphology coupled with nitrogen doping might equip the
catalysts with signicant interfacial interactions between
embedded metallic nanoparticles and the coated carbon shells.
The nonradical peroxymonosulfate (PMS) catalytic efficiencies
of the as prepared catalysts were evaluated using phenol as
a model molecule. The physico-chemical properties of the
catalysts, including morphology, specic surface area, electro-
chemical properties, binding affinities and so on, were well
characterized using a series of characterization techniques.
Density functional theory (DFT) simulations were applied to
calculate the interfacial interactions between the Co nano-
particles and capsulated graphene. Various electrochemical
analysis methods were performed to study the nonradical
oxidation process of phenol in the Co@NG/PMS system. To the
best of our knowledge, this study is among the very rst studies
to in-depth explore the mechanism of metal–carbon interfacial
interactions for peroxide activation in AOP reaction systems.
The outcomes provide a new insight into heterogeneous catalyst
design for improving peroxide activation.
ꢁ
at 350 C.
2.3 Characterization techniques
The crystal structures and morphologies of the as-prepared
catalysts were examined using a powder X-ray diffractometer
(XRD, ESCALAB 250Xi) equipped with Cu Ka radiation (l ¼
0.15406 nm), eld-emission scanning electron microscope
(FESEM, G2F20 S-Twin), and high-resolution transmission
2. Experimental section
2.1 Chemicals and materials
PMS (KHSO5$0.5KHSO4$0.5K2SO4), 2-methylimidazole (2- electron microscope (HR-TEM, JEM-2100F). The elemental
MeIM), tert-butanol (TBA), p-phthalic acid (PTA), 5,5-dimethyl- compositions were analyzed by continuum source atomic
1-pyrroline N-oxide (DMPO), phenol, cobalt nitrate hexahy- absorption spectrometry (CS-AAS, Analytikjena contrAA®700).
drate, zinc nitrate hexahydrate, boric acid, sodium tetraborate The specic surface areas were measured using the Brunauer–
decahydrate, sodium thiosulfate, methanol and cobalt (Co NPs, Emmett–Teller (BET) method on a Micromeritics ASAP 2460
99.9%, 30 nm) were purchased from Aladdin Co., Ltd., instrument at liquid-nitrogen temperature. The pore size
Shanghai, China. 2,2,6,6-Tetramethyl-4-piperidinol (TMP) was distribution (PSD) curves were recorded by nonlocal density
purchased from Sinopharm Chemical Reagent Co., Ltd., China. functional theory (DFT) measurements. The X-ray photoelec-
2-Hydroxyterephthalic acid (2-OH PTA) was purchased from tron spectroscopy (XPS) spectra tted by the Thermo Advantage
Tokyo Chemical Industry Co., Ltd., China. Methanol, acetoni- soware using Shirley-type background were recorded on
trile, formic acid, acetic acid and phosphoric acid of HPLC a Thermo Scientic ESCALAB 250Xi photoelectron spectrometer
grade were bought from Aladdin Co., Ltd., Shanghai, China. equipped with a monochromated Al Ka source. The thermog-
Other reagents used were of analytical-reagent grade and ob- ravimetric (TG) studies were conducted on a Mettler Toledo
tained from Tianjin Chemical Reagent Co., Ltd. All of the TGA/DSC1 thermo-analyzer with a heating rate of 2 K minꢀ1 in
chemicals were used as received without further purication. Ar ow. The ESR spectra were collected on a JES-FA200
This journal is © The Royal Society of Chemistry 2021
J. Mater. Chem. A, 2021, 9, 7198–7207 | 7199

View Article Online
Journal of Materials Chemistry A
Paper
spectrometer with a center eld at 328.9 mT and a sweep width ultraperformance liquid chromatography system (UPLC)
of 10.0 mT at room temperature.
coupled with a Waters Xevo TQ-S tandem mass spectrometer
(MS/MS) equipped with an electrospray interface (ESI). The
details of the chromatographic and mass conditions of UPLC-
MS in the negative ESI mode are described in Text S2.†
The reaction rate was evaluated using a pseudo rst order
kinetic model (eqn (1))
2.4 Electrochemical characterization
The electrochemical studies were performed in 100 mM phos-
phate buffer solution (pH 7). A standard three-electrode
conguration was applied using Ag/AgCl/KCl as the reference
electrode, a platinum wire as the counter electrode (if needed),
and an activator-coated glassy carbon electrode as the working
electrode. Electrochemical impedance spectroscopy (EIS) was
conducted using a potentiostat (PGSTAT 302N, Metrohm,
Switzerland) to provide the redox information. The EIS was
measured at open circuit potential over a frequency range from
100 kHz to 0.01 Hz with a sinusoidal excitation of 10 mV.
Nyquist plots were tted according to the equivalent circuit
reported previously.25,26 The other electrochemical analysis
tests, including linear sweep voltammetry (LSV), chro-
noamperometry, and chronopotentiometry, were carried out on
an electrochemical workstation (CHI 760E, China). The details
of the operation and the preparation of the activator-coated
glassy carbon electrode are shown in Text S1.†
ln(C0/Ct) ¼ kt
(1)
where C0 is the initial phenol concentration, Ct is the concen-
tration at a certain time t during the degradation process, and k
is the reaction rate constant.
2.7 Computational methods
The spin-polarized density functional theory (DFT) calculations
were performed by using the VASP code30 within the local
density approximation (LDA)31 for all calculations. The cutoff
energy for the plane wave basis was set to be 500 eV. Grids of 5 ꢂ
5 ꢂ 1 and 15 ꢂ 15 ꢂ 1 Monkhorst–Pack k-point grids were
sampled for geometry optimization and electronic property
calculations, respectively. The atoms were relaxed with conver-
gence criteria for energy and force set to be 10ꢀ4 eV and 0.02 eV
2.5 Detection of reactive oxygen species (ROS)
˚
A
ꢀ1, respectively. The charge transfer was analyzed with the
1
In order to verify the generation of O2 in the reaction system,
Bader code.32 To avoid the image interaction along the
TMP was chosen as a spin-trapping reagent, which is oxidized to
2,2,6,6-tetramethyl-4-piperidinol-N-oxyl radical (TMPN) by
1O2.27 DMPO was selected as the trapping agent for the detec-
tion of cOH, SO4cꢀ. PTA was used as a probe molecule to detect
˚
perpendicular direction, a vacuum layer of 15 A was taken. The
supercells of 3 ꢂ 3 were taken to simulate the structures of Co
(111), graphene and their composite. Four atom layers were
used to simulate the surface of Co (111) and the bottom two
layers were xed.
The adsorption energy (Ead) of PMS on different systems is
dened as
cOH production to give 2-OH PTA with a rate constant of 3.3 ꢂ
109 Mꢀ1 sꢀ1
.
28
2.6 Catalytic activity tests
The catalytic performance was assessed in a 200 mL reactor
containing 1 mM phenol with an initial pH of 7.2. Sodium
borate (50 mM) rather than phosphate was used as a buffer
reagent considering that the latter may activate PMS under
neutral conditions.29 The reaction temperature was maintained
Ead ¼ Ecatal+PMS ꢀ Ecatal ꢀ EPMS
where Ecatal+PMS, Ecatal and EPMS represent the total energy of the
relaxed catalyst with PMS, the energy of the catalyst and the
energy of PMS, respectively.
ꢁ
at 25 C using an oscillator (180 rpm). In a typical experiment,
5 mg of the catalyst was added in 100 mL of 1 mM phenol
solution, and the reaction was initiated by adding a certain
amount of PMS aqueous solution. At predetermined time
intervals, 1 mL of the reaction solution was withdrawn and
3. Results and discussion
3.1 Structure and characterization of the as-prepared
catalysts
immediately ltered into a high-performance liquid chroma- The powder X-ray diffraction (XRD) patterns of the as-
tography (HPLC) vial containing 0.1 mL of sodium thiosulfate synthesized ZIF-67 material (Fig. S1†) matched well with the
solution (0.1 mM) to terminate the catalytic reaction. To identify standard patterns of pure ZIF-67.24 The MOF material was of
the phenol degradation products, 50 mL of sample was typical rhombic dodecahedral shape with a particle size of 500–
collected and concentrated to 2 mL by solid-phase extraction 700 nm, which was evidenced by the FE-SEM images (Fig. S2†).
(SPE) following the previously described procedure (Text S2†).
The obtained ZIF-67 was thermally transferred to Co@NGs in an
The concentration of phenol in the reaction solution was atmosphere of argon (Scheme 1). The TG-DSC curves in Fig. S3†
analyzed by HPLC (Agilent, 1260-Innity) at a detection wave- illustrate that decomposition of ZIF-67 mainly occurred at ca.
length of 273 nm with a mixed solution of methanol and 500–600 ꢁC. Thus, ZIF-67 was calcined at four selected
0.1 vol% acetic acid (50 : 50, v/v) as the mobile phase. To temperatures (i.e., T ¼ 700, 800, 900, 1000, and 1100 ꢁC) to
disclose the catalysis mechanisms, several other organic obtain the Co@NGs. The SEM images (Fig. S4†) indicate that
compounds (PTA, 2-OH PTA) were also tested (listed in Table the Co@NGs had a relatively smaller size and rougher surface
S1,† with detailed HPLC analytical conditions). The trans- than ZIF-67, which might be due to gradual decomposition and
formation products of phenol were identied using a Waters carbonization of the framework, and NC-900 maintained the
7200 | J. Mater. Chem. A, 2021, 9, 7198–7207
This journal is © The Royal Society of Chemistry 2021

View Article Online
Paper
Journal of Materials Chemistry A
ꢁ
higher than 900 C. A broad and weak second-order (2D) band
was observed at approximately 2700 cmꢀ1 in Co@NG-900, 1000
and 1100 (Fig. S6a†), and the shapes of the D and 2D bands
(especially the lack of a typical graphite shoulder) are charac-
teristic of few-layered graphene,33–35 which was conrmed by the
TEM images in Fig. 1. Other peaks, such as Eg (478 cmꢀ1) and
F2g (517 and 618 cmꢀ1), correspond to the Co element.36 The
HR-TEM image indicates that the Co nanoparticles were
covered by graphene shells, and well dispersed without distinct
aggregation in the Co@NGs pyrolyzed at a temperature as high
as 900 ꢁC (Table S2†), which could be attributed to the well
separated carbon atoms in the 2-MeIM linker of ZIF-67. The two
Scheme 1 Preparation route and model of the N-doped graphene
encapsulated Co nanoparticles (Co@NG) with well-controlled
morphology, and the proposed overall Fenton-like reaction mecha-
nism on Co@NG catalysts.
˚
fringe spacings of 2.05 and 1.79 A in Co@NG-900 (Fig. 1f)
correspond to the (111) and (200) planes of Co nanoparticles,
respectively.22 Energy dispersive spectroscopy (EDS) elemental
mapping images were taken to investigate the elemental
distribution of the sample (Co@NG-900), as displayed in
Fig. S8.† The elemental mapping images demonstrated the
homogeneous distribution of N and C atoms within the entire
nanoarchitecture, while the distribution of N atoms matches
well with that of Co atoms.
The XRD patterns of the Co@NGs (Fig. 2a) exhibited ve
diffraction peaks at around 44.2, 51.6, 76.0, 92.4, and 97.8ꢁ,
respectively, characteristic of metallic Co (JCPDS no. 15-0806).
The peak at ca. 25.8ꢁ in the high temperature materials corre-
dodecahedron morphology very well (Fig. S5a†). In the Raman
spectra (Fig. S6a†), the G band at ꢃ1350 cmꢀ1 is a characteristic
feature of graphitic layers, while the D (1590 cmꢀ1) and D0
(1840 cmꢀ1) bands correspond to the disordered carbon or
defective graphitic structures.33 A high ID/IG band intensity ratio
(0.63–1.01) (Fig. S6b†) indicates the generation of a large
amount of defects,33,34 implying that the N atoms were
successfully doped in the graphitic layer given that the original
concentration of nitrogen was relatively high (Fig. S7†).35 The ID/
IG (Fig. S6b†) of Co@NG-900 (0.75) was higher than that of
Co@NG-1000 (0.68) and Co@NG-1100 (0.63), suggesting that
more N atoms were present in Co@NG-900. In agreement with
this, the elemental analysis (Fig. S7†) indicated that the N
content declined distinctly as the calcination temperature was
˚
sponded to the interlayer spacing of 3.42 A, arising from the
Fig. 1 HR-TEM images of (a) Co@NG-700, (b) Co@NG-800, (c)
Co@NG-900, (d) Co@NG-1000, and (e) Co@NG-1100; (f) an enlarged Fig. 2 (a) Powder XRD patterns of NG-900 and Co@NG materials; (b)
image of a Co nanoparticle capsulated in the graphene layers.
N2 adsorption and desorption isotherms of Co@NG materials.
This journal is © The Royal Society of Chemistry 2021
J. Mater. Chem. A, 2021, 9, 7198–7207 | 7201

View Article Online
Journal of Materials Chemistry A
Paper
turbostratic ordering of the carbon and nitrogen atoms in the the C–C bond (379 kJ molꢀ1),40 and the cleavage of C–N bonds
graphite layers.37 Fig. 2b shows that all of the N2 adsorption– led to partial loss of nitrogen (Fig. S10†).38 The N2 and G-N were
desorption isotherms were classied as type IV sorption proved to be effective in improving the adsorption capacity and
isotherms. A typical type H4 hysteresis loop at the relative catalytic activity in PMS activation.41 As shown in Fig. 3a, c and
pressures (P/P0) from 0.45 to 1.0 appeared when the pyrolysis Table S3,† the nitrogen species experienced distinct changes
temperature was higher than 900 ꢁC, indicating the presence of during pyrolysis at high temperatures. The percentages of N1
a large amount of mesopores. The DFT pore-size distribution and O-N decreased, leading to predominant N–Co and N2 in the
indicated there were a number of micropores smaller than 2 nm high temperature materials. It is worth noting that Co@NG-900
in the Co@NGs (Fig. S9†), which presumably originated from and Co@NG-1000 had a relatively high content of G-N, which
the intrinsic pore structure of ZIF-67. When the T was 1000 and was more thermostable than N1. However, the N–Co content in
1100 ꢁC, more mesopores in the range of 10–30 nm were the Co@NG-1000 declined obviously compared to Co@NG-900,
produced. This suggested that Co@NGs with a hierarchically which might be attributed to the serious aggregation of Co
porous structure and large surface area were produced at high nanoparticles, loss of nitrogen and destruction of graphene
calcination temperatures, which are benecial for efficient shells. All these would lead to a weakened interaction between
permeation of the reactants to access the abundant active the graphene shells and the encapsulated Co nanoparticles.
catalytic sites.
Co@NG-900 possessed an appropriate nitrogen content and
The high-resolution XPS spectra of the Co@NGs indicated a relatively high ratio of G-N and N–Co, equipping it with rela-
that the content of oxygen decreased with an increase in the tively higher electron mobility and more active sites, which are
pyrolysis temperature (Fig. S10a†), suggesting that the oxygen- benecial for interfacial interactions between Co and the NG
containing functional groups on the carbon surface decom- shells.42
posed at high pyrolysis temperatures.38 Doping nitrogen is
The high resolution spectrum of Co 2p3/2 in Co@NGs was
considered as an efficient way to improve the electronic and deconvoluted into three peaks at 778.4, 780.2 and 782.6 eV,
catalytic properties of graphene.5 Therefore, it is imperative to corresponding to Co0, CoCxNy and CoNx, respectively (Fig. 3b,
understand the types and contents of nitrogen in the Co@NG d and Table S4†).43 The peak corresponding to the Co–Co bond
materials. Five peaks with binding energies at 398.4, 399.2, shied to a higher energy compared to the pristine Co nano-
400.5, 401.2 and 403.1–403.5 eV were deconvoluted, which can particles (778.1 eV).44 This implied there was a strong interac-
be assigned to pyridine-N (N1), N–Co, pyrrole-N (N2), graphitic- tion between the conned Co nanoparticles and N doped
N (G-N) and oxidized-N (O-N), respectively (Fig. 3a).24,39 The C–N graphene shells. Previous studies demonstrated that the strong
bond is more vulnerable during the high-temperature calcina- interaction between encapsulated metal nanoparticles and
tion due to its lower binding energy (305 kJ molꢀ1) than that of carbon shells could greatly facilitate electron transfer via the
atomic interface-bonding channels, which helped to reduce the
local work function and the adsorption quantity.44,45
3.2 Catalytic activities of the Co@NGs
The catalytic performance of the Co@NGs was evaluated for
phenol removal by activating PMS. As shown in Fig. S11,† less
than 1.5% of phenol was degraded by PMS alone. A control test
revealed that about 4.5% of phenol was adsorbed by Co@NG-
900 within 12 min. The phenol removal efficiency was
distinctly enhanced as Co@NGs (Co@NG-700, 800, 900, 1000
and 1100) were applied, which was 48.5, 57.5, 100, 93.2, and
87.5% in 12 min, respectively. The phenol removal kinetics
tted the rst-order reaction model very well (R2 > 0.96), and the
apparent rate constant (k) and the surface area normalized rate
constants (k0) were calculated (Table S2†). As shown in Fig. 4a,
when the temperature rose from 700 to 900 ꢁC, the k increased
from 0.045 to 0.397 minꢀ1 with an increase in the content of G-
N (Fig. 3). The k increased sharply from Co@NG-800 to Co@NG-
900 due to the mass production of mesopores (Fig. S9 and Table
S2†), which is benecial for phenol and/or PMS to approach the
catalytic sites. A slightly decline of k was observed when the
temperature further increased from 900 to 1100 ꢁC, which
might be owing to the reduced specic surface area (Table S2†)
and severe damage of the encapsulated structure. The k value of
Co@NG-900 was 0.397 minꢀ1 (R2 ¼ 0.988), which was approxi-
Fig. 3 High-resolution XPS N 1s (a), and Co 2p (b) spectra; and the
contents of different N (c) and Co (d) species of Co@NG materials.
7202 | J. Mater. Chem. A, 2021, 9, 7198–7207
mately 20.9 times of NG-900 (k ¼ 0.019), although the surface
This journal is © The Royal Society of Chemistry 2021

View Article Online
Paper
Journal of Materials Chemistry A
a heterogeneous catalyst. The performance of the Co@NG-900
nanoparticles was tested for 5 consecutive cycles using the
same reaction parameters. The results in Fig. S14† show that,
even aer 5 consecutive cycles, the catalyst exhibited a high
catalytic activity as indicated by ꢃ85% degradation of phenol
within 12 min.
3.3 Mechanisms for the highly efficient catalytic effect
To conrm the primary ROS in the PMS/Co@NG-900 system,
ESR with spin-trapping reagents of TMP and DMPO was applied
to detect the generation of 1O2 and free radicals accordingly. As
illustrated in Fig. S15,† a characteristic 1 : 1 : 1 triplet signal
with a g-value of 2.0056 appeared in the presence of Co@NG-
900, which was indexed to 2,2,6,6-tetramethylpiperidine-N-oxyl
(TMPN).27 The 1O2 evolution during phenol degradation is
illustrated in Fig. S16.† In the presence of phenol, the 1O2
formation rate had no correlation with the removal rate of
phenol, implying the minor role of 1O2 in the reaction. As shown
in Fig. 4c, the characteristic 1 : 2 : 2 : 1 signals of DMPO–cOH
adducts (with aN ¼ aH ¼ 14.9 G)47 were detectable in the
presence of Co@NG-900, while the sulfate radical was not
detected. To certify the presence of cOH, PTA was used as
a probe molecule to detect cOH production (Fig. 4d). 2-OH PTA
was not detected in the entire reaction period when Co@NG-900
was added. However, obvious signals of 2-OH PTA were detected
when CoxO was used as the catalyst. These suggest that the
detected signals of DMPO–cOH might be due to the false posi-
tive detection of cOH in the Co@NG-900/PMS system.28
The false positive detection involves a direct electron transfer
reaction of the spin trap (R) followed by a nucleophilic attack by
water (reaction (2)).
Fig. 4 (a) First-order kinetic model of phenol removal with Co@NG
materials within 12 min; (b) rate constant of Co@NG materials
compared with Co NPs and NG-900; (c) ESR spectra with DMPO
(DMPOc–OH, green ; DMPO-X, black
); (d) bulk oxidation of
0.1 mM PTA and the production of 2-OH PTA with Co@NG-900 and
CoxO. Reaction conditions: [phenol] ¼ 1.0 mM, [PMS] ¼ 3.0 mM,
catalyst ¼ 0.05 g Lꢀ1, T ¼ 298 K, initial solution pH ¼ 7.2.
area of NG-900 (344.3 m2 gꢀ1) was 2.5 times of Co@NG-900
(Table S2†). In addition, the turnover frequency (TOF) of
Co@NG-900, which was calculated by dividing the k value by the
catalyst concentration, was as high as 7.94 minꢀ1 for phenol
degradation. This TOF was remarkably higher than that re-
ported for other catalysts for non-radical and radical PMS acti-
vation, which was in the range of 0.12–2.27 (Table S5†) and
0.16–3.10 (Table S6†), respectively. The nanoparticles of pure
CoxO and Co, which did not contain nitrogen doped graphene
layers, only achieved a phenol removal efficiency of ꢃ43 and
65% (k ¼ 0.056 and 0.107 minꢀ1) in 12 min, respectively
(Fig. S12†). Attributed to the proper nitrogen species (especially
G-N and CoCxNy) distribution, Co@NG-900 displayed the
strongest interaction between the encapsulated metal nano-
particles and carbon shells, which helped to reduce the local
work function. In addition, the much abundant mesopores
enabled it to have higher adsorption capacity, and to facilitate
the diffusion of pollutants to the active sites. These collectively
contributed to its best catalytic performance. These results
suggested that the strong interaction between the conned Co
species and N-doped graphene shells played a vital role in
improving the PMS activation.
The effect of oxidants (PMS, PDS, and H2O2) on the catalytic
activity of Co@900-NG was also investigated. Co@900-NG
exhibited a higher catalytic performance with PMS (100%
removal of phenol) than with PDS activation (34% removal)
within 12 min (Fig. S13†). This was attributed to the difficulty in
breaking O–O bonds in a symmetrical structure like PDS
molecules.46 H2O2 cannot be activated by Co@900-NG, may be
due to the stronger bond energy of O–O in H2O2 than that in
persulfate and a poor affinity toward the catalyst. The reus-
ability is important for practical applications of
ꢀeꢀ
þOHꢀ
DMPO
DMPOꢄþ
DMPO ꢀ OH
(2)
ƒƒ!
ƒƒƒƒ!
Radical quenching experiments were conducted to identify
the active species responsible for phenol removal. As shown in
Fig. S17,† the addition of methanol (MeOH, radical scavenger
for cOH and SO4cꢀ 48,49) and tert-butyl alcohol (TBA, radical
scavenger for cOH48) displayed marginal impacts on phenol
removal, conrming that the free radicals did not play vital roles
in phenol oxidation in the Co@NG-900/PMS systems.
The electrochemical properties of the Co@NGs were evalu-
ated to probe the charge transfer capacity and redox potential in
PMS or/and phenol solution. Two semicircles of Bode plots were
distinct for Co@NGs (Fig. S18a†), implying that two chemical
processes might be involved.50,51 The Nyquist plots contained
two-time constants for Co@NGs (Fig. S18b†), which was
consistent with the Bode plots. The equivalent circuit of the EIS
data is shown in Fig. S19,† in which the resistance, R (Rct and
Rr), and constant phase element (pseudo-capacitance), CPE,
could be tted to all the data.25,26 The Rct was much smaller than
the reaction resistance Rr, suggesting that the PMS activation
was dominated by Rr (Table S7†). The Rr of Co@NG-900 (0.91
kU) was the smallest, while that of Co@NG-700 (16.4 kU) was
the largest among all the Co@NGs. However, the pure Co
nanoparticles displayed a single semicircle (Fig. 5a), implying
This journal is © The Royal Society of Chemistry 2021
J. Mater. Chem. A, 2021, 9, 7198–7207 | 7203

View Article Online
Journal of Materials Chemistry A
Paper
deduced that the oxidation potential of phenol on NGs mate-
rials was around 0.57 V.2,12 However, the current density grad-
ually declined as the potential surpassed 0.82 V, which might be
due to the fact that the mass transfer velocity of phenol mole-
cules in the solution was much slower than the oxidation rate
on the anode because of the fast consumption of phenol. The
corresponding Tafel plots were constructed according to the
LSV curves, and the Tafel slope was determined by tting LSV
data to the Tafel (Text S1†). The Tafel slope, which is determined
by the rate-limiting step in electrocatalysis, is an inherent
property of catalysts.52 The Tafel slopes of Co@NG-900 and NG-
900 were calculated to be 266 and 188 mV decꢀ1, respectively
(inset in Fig. 5c), suggesting that Co@NG-900 manifested
a faster reaction rate than NG-900 with the elevated applied
potentials, which might benet from the strong interfacial
interaction between Co and the NG shell in Co@NG-900.
In situ measurements of open circuit potential of surface
HSO5ꢀ* on Co@NG-900 and NG-900 were conducted to reveal
the formation of NG–PMS*. Upon addition of PMS in the
solutions, the open circuit potential in both Co@NG-900 and
NG-900 systems increased immediately (Fig. 5d), corroborating
the formation of NG–PMS* complexes. A larger change was
observed for the potential of Co@NG-900/PMS (0.64 V) than for
NG-900/PMS (0.37 V), fortifying that the intrinsic redox poten-
tial of NG–PMS* on Co@NG-900 was higher than that on NG-
900, which might be because that encapsulation of Co in the
catalysts enhanced the interaction between PMS and NG.
However, as phenol was added, the potential of the Co@NG-900
and NG-900 dropped immediately to around +0.60 V, and then
declined slowly. To further explore the relationship between the
potential of NG–PMS* and phenol oxidation, different amounts
of phenol were added. As can be seen in Fig. 5e, when the
phenol concentration was lower than PMS, the potential drop-
ped distinctly with the increase of phenol, but it remained
relatively constant (ꢃ0.60 V) if the amount of phenol further
increased. This might be due to the continuous consumption of
the surface-active complexes until the potential dropped below
the phenol's oxidation potential (0.57 V).
To understand the charge transfer in the oxidation process
of phenol, amperometric i–t curves were recorded to monitor
charge migration in the reaction systems (Fig. 5f), and the
working electrode was biased to the applied potential +0.6 V.
When PMS was injected, a negative current emerged promptly,
whereas the addition of phenol generated a positive current.
These observations indicated that the electrons transferred
from phenol to PMS in the presence of Co@NG-900 or NG-900.12
The higher DI observed for Co@NG-900 (79.7 mA) than NG-900
(16.2 mA) might be attributed to the excellent catalytic activity
of Co@NG-900. The degradation products of phenol identied
in the Co@NG-900/PMS and CoxO/PMS systems are listed in
Table S8.† In the CoxO/PMS system, the identied products
included benzoquinone, hydroquinone and some small ring
opening products (such as lactic acid and maleic acid), while
the oligomeric intermediates like di- or terphenyls (m/z ¼ 185
and 277) were observed in the Co@NG-900/PMS system.
According to the anodic-like oxidation mechanism, phenoxyl
radicals in various resonance forms were produced via one-
Fig. 5 (a) Nyquist plot of Co@NG-900 and Co NPs; (b) in situ Raman
spectra partial and whole views in the Co@NG-900/PMS and NG-900/
PMS systems; (c) LSV and Tafel slope (inset of c) on the Co@NG-900
and NG-900 electrodes in phenol solutions; the changes of open-
circuit potential curves of NG–PMS complexes on Co@NG-900 and
NG-900 electrodes (d) and in different phenol proportions (e); (f)
amperometric i–t curve measurements upon the addition of PMS and
phenol with Co@NG-900 and NG-900 electrodes.
that completely different mechanisms were involved in PMS
activation compared to Co@NGs.
In situ Raman analyses (Fig. 5b) were conducted to detect the
surface chemistry evolution during PMS activation for phenol
degradation. Besides the characteristic peaks of HSO5ꢀ for PMS
alone at 884 and 1063 cmꢀ1, a new peak (HSO5ꢀ*) emerged at
around 796 cmꢀ1 upon the addition of Co@NG-900 or NG-900
into the PMS solution.2 This new peak might be ascribed to
the bending vibrations of the prolonged peroxo O–O bond in the
metastable NG–PMS* intermediate. Moreover, the peak inten-
sity of HSO5ꢀ was weaker and that of HSO5ꢀ* disappeared in the
presence of phenol, suggesting that HSO5ꢀ* was consumed by
phenol during the oxidation process.
To understand the mechanisms, linear sweep voltammetry
(LSV) was performed, and the results are shown in Fig. 5c. The
LSV curve could be divided into three parts. As the electrode
potential (4) was in the range of 0.40–0.57 V, phenol could not
be oxidized, and the j uctuated around 1.5 mA cmꢀ2, sug-
gesting that the electrode potential was lower than the oxidation
potential of phenol. As the 4 increased and was in the range of
0.57–0.82 V, phenol started to be oxidized on the anode to
generate a sharp and increasing current, and the surged current
density of Co@NG-900 was greater than that of NG-900. It was
7204 | J. Mater. Chem. A, 2021, 9, 7198–7207
This journal is © The Royal Society of Chemistry 2021

View Article Online
Paper
Journal of Materials Chemistry A
Table 1 PMS adsorption onto different catalysis models. Q is the
charge transfer from catalysis to PMS. QCo is the charge transfer from
Co to graphene pristine or doped
electron oxidation of phenol in the Co@NG-900/PMS system,
and they were further oxidized through C–O and C–C coupling
between phenol radicals to generate oligomeric intermediates.53
According to the above analyses, the mechanism of non-
radical oxidation was elucidated via an anodic-like oxidation
process by Co@NGs. Due to the strong interfacial interaction
between Co and the NG shells, the intrinsic redox potential of
NG–PMS* was greatly improved. When the equilibrium poten-
tial of the complexes exceeded the oxidation potential of
phenol, phenol was decomposed via an anodic-like oxidation
over Co@NG-900, and the higher oxidation potential greatly
promoted the phenol oxidation.
˚
dO–O (A)
Ead (eV)
Q (e)
QCo (e)
Free PMS
G
Graphitic NG
Pyridinic NG
Pyrrolic NG
Co
1.31
1.39
1.44
1.39
1.39
1.44
1.44
1.45
1.43
1.43
—
—
—
—
—
—
—
—
1.78
1.71
1.95
2.15
ꢀ0.92
ꢀ1.99
ꢀ1.11
ꢀ1.14
ꢀ3.16
ꢀ1.69
ꢀ2.05
ꢀ1.55
ꢀ1.59
0.52
0.86
0.57
0.55
0.73
0.8
0.87
0.78
0.78
Co@G
Co@graphitic NG
Co@pyridinic NG
Co@pyrrolic NG
3.4 DFT calculations
Co@NG composites (Fig. 6b and c). The O–O bond in PMS tends
to be adsorbed above the C atoms, and the vertical distance
To clarify the mechanisms of the signicant increase in the
activity of the Co@NG system, DFT calculations were also per-
formed. Co (111) with single-layer graphene was used to simu-
late the structure of Co@NG according to the HR-TEM images
(Fig. 1). The structures of NG and the Co@NG are shown in
Fig. S20 and S21.† The lattice parameters of graphene and Co
˚
between PMS and the surface is less than 2.5 A (Fig. S25†).
To investigate the activating ability of the different catalysts,
the length of the O–O bond (dO–O) of PMS, adsorption energy
(Ead) and charge transfer (Q) from the catalysts to PMS are
calculated and listed in Table 1. For the single-layer graphene,
doping with nitrogen introduced extra electrons which serve as
Lewis basic sites to facilitate PMS adsorption on the NG surface,
which facilitated electron transfer during the process of phenol
oxidation. The graphitic NG impressively enhances the PMS
adsorption with more prolonged O–O bonds than the pristine
G. Analogously, the construction of Co encapsulated in
graphitic NG (Co@graphitic NG) leads to the longest dO–O and
strongest adsorption, implying that Co@graphitic NG
predominates PMS activation in the composite systems, which
was consistent with the degradation kinetics (Fig. 4a), XPS
analysis (Fig. 3c) and electrochemical analysis (Fig. 5c, d and f).
To further unravel the mechanisms of the enhanced activity, the
charge density differences and work function (WF) of the
composite catalysts were calculated, which are important for
the activities of catalysts44,55 (Fig. 6d–f). The increased C pz
charge on the surface of graphene owing to the electronegativity
difference between C and Co atoms improved the interfacial
interactions between the encapsulated Co and NG shells. Due to
the signicant decrease in the Fermi level by graphene encap-
sulation and N-doping, the work function of the Co@NG
composites all decreased to <4 eV (Fig. 6f), which is signicantly
lower than that of Co (111) (5.41 eV) or single-layer graphene
(4.49 eV) (Fig. S26†). Considering that Co might be coated by
several layers of graphene, the Co (111) surface covered with
double-layer graphene was also mimicked. Fig. S27† shows that
there is also charge distributed on the surface layer of graphene
and the WF is 4.10 eV. Therefore, it is the extra charges
distributed on the surface of the catalyst and the decreased
work function that may improve the PMS activating ability and
the redox ability of formed HSO5ꢀ*.
˚
˚
(111) are 2.447 A and 3.425 A, respectively. The interlayer
˚
spacing of Co (111) is 1.98 A, which is similar to the measured
˚
result 2.05 A. The 1 ꢂ 1 unit cell was used to test the adsorption
of graphene on the surface of Co (111). The hcp conguration
where the center of the carbon ring is above the surface Co
atoms was found to be the most stable adsorption conformation
with the distance of 2.07 A between them.54 To obtain the PMS
˚
adsorption conguration, the lateral and vertical adsorption
structures of PMS (Fig. S22†) were considered. The higher
lateral adsorption energy of PMS than the vertical one suggested
that PMS is apt to be laterally adsorbed on the surface of
Co@NG (Table S9, Fig. S23 and S24†). However, PMS was
adsorbed on Co nanoparticles vertically, and was directly
decomposed into O and HSO4 (Fig. 6a), certifying that PMS was
activated in different ways by pure Co nanoparticles and the
Fig. 6 Side views of PMS adsorbed on (a) Co (111), (b) graphitic NG, and
(c) Co@graphitic NG respectively. The blue, gray, violet, red, yellow and
light pink represent Co, C, N, O, S and H atoms, respectively. (d) Charge
density difference (rtotal ꢀ rsubstrate ꢀ rPMS) between Co@graphitic NG
and PMS (isosurface contour is 0.01 e A˚ꢀ3). The khaki and green denote
the electron accumulation and electron depletion, respectively. (e)
Work function of Co@NG-g. (f) Work function of composite systems.
NG-g, NG-pd and NG-pl represent graphitic, pyridinic and pyrrolic
nitrogen doped graphene.
4. Conclusions
In this work, unique composite catalysts of graphene-
encapsulated Co nanoparticles with extremely high catalytic
This journal is © The Royal Society of Chemistry 2021
J. Mater. Chem. A, 2021, 9, 7198–7207 | 7205

View Article Online
Journal of Materials Chemistry A
Paper
activity for remediation of organic pollutants were prepared via
facile thermal decomposition of MOFs in an Ar atmosphere. A
Co/N-doped graphene interface was generated, and the strong
interactions between the N-doped graphene shells and conned
Co nanoparticles improved the interfacial electron shuttling via
CoCxNy atomic interface-bonding channels. DFT calculations
indicated that the conned metal nanoparticles in graphene
shells not only reduced the local work function for PMS
adsorption, but greatly improved the potential of surface-active
complexes on the graphene shells, which greatly improved the
redox ability of HSO5ꢀ*, and the appropriate N species doping
5 X. Duan, H. Sun and S. Wang, Acc. Chem. Res., 2018, 51, 678–
687.
6 Y. Gao, Z. Chen, Y. Zhu, T. Li and C. Hu, Environ. Sci.
Technol., 2020, 54, 1232–1241.
7 E.-T. Yun, J. H. Lee, J. Kim, H.-D. Park and J. Lee, Environ. Sci.
Technol., 2018, 52, 7032–7042.
8 P. Sun, H. Liu, M. Feng, Z. Zhai, Y. Fang, X. Zhang and
V. K. Sharma, Appl. Catal., B, 2020, 272, 119005.
9 H. Lee, H.-i. Kim, S. Weon, W. Choi, Y. S. Hwang, J. Seo,
C. Lee and J.-H. Kim, Environ. Sci. Technol., 2016, 50,
10134–10142.
facilitates this process. Thus the interfacial interaction was 10 S. Zhu, X. Huang, F. Ma, L. Wang, X. Duan and S. Wang,
greatly boosted and the surface nonradical oxidation perfor- Environ. Sci. Technol., 2018, 52, 8649–8658.
mance was remarkably enhanced. As a result, 1 mM phenol 11 W. Ren, L. Xiong, G. Nie, H. Zhang, X. Duan and S. Wang,
could be completely decomposed in only 12 min with the
Environ. Sci. Technol., 2020, 54, 1267–1275.
reaction rate constant k ¼ 0.397 minꢀ1. Our study provides 12 P. Shao, S. Yu, X. Duan, L. Yang, H. Shi, L. Ding, J. Tian,
novel insights into the active interfacial interactions for
peroxide activation and enhancing the non-radical oxidation
L. Yang, X. Luo and S. Wang, Environ. Sci. Technol., 2020,
54, 8464–8472.
process, which provides a facile strategy for development of 13 H.-J. Niu, L. Zhang, J.-J. Feng, Q.-L. Zhang, H. Huang and
other “interface engineering catalysts” in future.
A.-J. Wang, J. Colloid Interface Sci., 2019, 552, 744–751.
14 Z. Wei, Y. Chen, J. Wang, D. Su, M. Tang, S. Mao and
Y. Wang, ACS Catal., 2016, 6, 5816–5822.
Author contributions
15 H. Yang and X. Wang, Adv. Mater., 2019, 31, 1800743.
L. Y. Zhu supervised the project. X. Y. Yu and L. Y. Zhu 16 Q. Shen, X. Li, R. Li and Y. Wu, ACS Sustainable Chem. Eng.,
conceived the project and designed the experiments. X. Y. Yu 2020, 8, 17608–17621.
synthesized the catalysts and performed most of the reactions 17 H. Yang, S. J. Bradley, A. Chan, G. I. N. Waterhouse, T. Nann,
and characterization. With the help of X. Wang carried out
electrochemical characterization. With the help of G. Q. Shan
P. E. Kruger and S. G. Telfer, J. Am. Chem. Soc., 2016, 138,
11872–11881.
carried out ROS analysis. L. J. Wang who is supervised by W. C. 18 S.-J. Kim, J. Mahmood, C. Kim, G.-F. Han, S.-W. Kim,
Wang performed the DFT calculation. X. Y. Yu, L. J. Wang and L.
Y. Zhu prepared the manuscript.
S.-M. Jung, G. Zhu, J. J. De Yoreo, G. Kim and J.-B. Baek, J.
Am. Chem. Soc., 2018, 140, 1737–1742.
19 F. Zhang, F. Jiao, X. Pan, K. Gao, J. Xiao, S. Zhang and X. Bao,
ACS Catal., 2015, 5, 1381–1385.
20 S. Jeong, K. Hu, T. Ohto, Y. Nagata, H. Masuda, J.-i. Fujita
and Y. Ito, ACS Catal., 2020, 10, 792–799.
Conflicts of interest
The authors declare no conict of interest.
21 A. Aijaz, N. Fujiwara and Q. Xu, J. Am. Chem. Soc., 2014, 136,
6790–6793.
22 W. Zang, A. Sumboja, Y. Ma, H. Zhang, Y. Wu, S. Wu, H. Wu,
Z. Liu, C. Guan, J. Wang and S. J. Pennycook, ACS Catal.,
2018, 8, 8961–8969.
Acknowledgements
The authors gratefully acknowledge the nancial support of the
National Natural Science Foundation of China (41991313,
21737003, and 21975136), the Ministry of Science and Tech- 23 G. Lu, S. Z. Li, Z. Guo, O. K. Farha, B. G. Hauser, X. Y. Qi,
nology of China (2018YFC1801003), the Tianjin Municipal
Science and Technology Commission (17JCYBJC23200), the
Yangtze River Scholar Program, the 111 Program, Ministry of
Education, China (T2017002) and the Fundamental Research
Funds for the Central Universities.
Y. Wang, X. Wang, S. Y. Han, X. G. Liu, J. S. DuChene,
H. Zhang, Q. C. Zhang, X. D. Chen, J. Ma, S. C. J. Loo,
W. D. Wei, Y. H. Yang, J. T. Hupp and F. W. Huo, Nat.
Chem., 2012, 4, 310–316.
24 H. Chen, K. Shen, Q. Mao, J. Chen and Y. Li, ACS Catal.,
2018, 8, 1417–1426.
25 B. M. Klahr, S. Gimenez, F. Fabregatsantiago, T. W. Hamann
and J. Bisquert, J. Am. Chem. Soc., 2012, 134, 4294–4302.
References
1 J. Lee, U. von Gunten and J.-H. Kim, Environ. Sci. Technol., 26 A. Hankin, F. E. Bedoyalora, C. K. Ong, J. C. Alexander,
2020, 54, 3064–3081.
2 W. Ren, G. Nie, P. Zhou, H. Zhang, X. Duan and S. Wang,
Environ. Sci. Technol., 2020, 54, 6438–6447.
F. Petter and G. H. Kelsall, Energy Environ. Sci., 2017, 10,
346–360.
27 J. Moan and E. Wold, Nature, 1979, 279, 450–451.
3 J. Miao, W. Geng, P. J. J. Alvarez and M. Long, Environ. Sci. 28 Y. Jing and B. P. Chaplin, Environ. Sci. Technol., 2017, 51,
Technol., 2020, 54, 8473–8481. 2355–2365.
4 W. D. Oh, Z. L. Dong and T. T. Lim, Appl. Catal., B, 2016, 194, 29 X. Lou, L. Wu, Y. Guo, C. Chen, Z. Wang, D. Xiao, C. Fang,
169–201.
J. Liu, J. Zhao and S. Lu, Chemosphere, 2014, 117, 582–585.
7206 | J. Mater. Chem. A, 2021, 9, 7198–7207
This journal is © The Royal Society of Chemistry 2021

View Article Online
Paper
Journal of Materials Chemistry A
¨
30 G. Kresse and J. Furthmuller, Phys. Rev. B: Condens. Matter 43 Z. Liang, X. Fan, H. Lei, J. Qi, Y. Li, J. Gao, M. Huo, H. Yuan,
Mater. Phys., 1996, 54, 11169–11186.
31 D. M. Ceperley and B. J. Alder, Phys. Rev. Lett., 1980, 45, 566–
569.
32 W. Tang, E. Sanville and G. Henkelman, J. Phys.: Condens.
Matter, 2009, 21, 084204.
W. Zhang, H. Lin, H. Zheng and R. Cao, Angew. Chem., Int.
Ed., 2018, 57, 13187–13191.
44 J. Meng, C. Niu, L. Xu, J. Li, X. Liu, X. Wang, Y. Wu, X. Xu,
W. Chen, Q. Li, Z. Zhu, D. Zhao and L. Mai, J. Am. Chem.
Soc., 2017, 139, 8212–8221.
33 F. Zheng, Y. Yang and Q. Chen, Nat. Commun., 2014, 5, 5261. 45 Y. Yao, H. Chen, J. Qin, G. Wu, C. Lian, J. Zhang and S. Wang,
34 A. C. Ferrari and D. M. Basko, Nat. Nanotechnol., 2013, 8,
235–246.
Water Res., 2016, 101, 281–291.
46 J. Wang and S. Wang, Chem. Eng. J., 2018, 334, 1502–1517.
35 P. Jiang, J. Chen, C. Wang, K. Yang, S. Gong, S. Liu, Z. Lin, 47 M. Wiedau-Pazos, J. J. Goto, S. Rabizadeh, E. B. Gralla,
M. Li, G. Xia, Y. Yang, J. Su and Q. Chen, Adv. Mater.,
2018, 30, 1705324.
J. A. Roe, M. K. Lee, J. S. Valentine and D. E. Bredesen,
Science, 1996, 271, 515.
36 K. R. Yoon, K. Shin, J. Park, S.-H. Cho, C. Kim, J.-W. Jung, 48 G. V. Buxton, C. L. Greenstock, W. P. Helman and A. B. Ross,
J. Y. Cheong, H. R. Byon, H. M. Lee and I.-D. Kim, ACS
Nano, 2018, 12, 128–139.
37 A. Vinu, K. Ariga, T. Mori, T. Nakanishi, S. Hishita,
J. Phys. Chem. Ref. Data, 1988, 17, 513–886.
49 P. Neta, R. E. Huie and A. B. Ross, J. Phys. Chem. Ref. Data,
1988, 17, 1027–1284.
D. Golberg and Y. Bando, Adv. Mater., 2005, 17, 1648–1652. 50 S. Nong, W. Dong, J. Yin, B. Dong, Y. Lu, X. Yuan, X. Wang,
38 P. Hu, H. Su, Z. Chen, C. Yu, Q. Li, B. Zhou, P. J. J. Alvarez
K. Bu, M. Chen and S. Jiang, J. Am. Chem. Soc., 2018, 140,
5719–5727.
and M. Long, Environ. Sci. Technol., 2017, 51, 11288–11296.
39 C. Tang, B. Wang, H.-F. Wang and Q. Zhang, Adv. Mater., 51 S. Wang, H. Xu, W. Li, A. Dolocan and A. Manthiram, J. Am.
2017, 29, 1703185. Chem. Soc., 2018, 140, 250–257.
40 J. Liu, S. Webster and D. L. Carroll, J. Phys. Chem. B, 2005, 52 A. A. Gewirth, J. A. Varnell and A. M. DiAscro, Chem. Rev.,
109, 15769–15774. 2018, 118, 2313–2339.
41 X. Li, X. Huang, S. Xi, S. Miao, J. Ding, W. Cai, S. Liu, X. Yang, 53 C. Chen, Z. Wu, S. Zheng, L. Wang, X. Niu and J. Fang,
H. Yang, J. Gao, J. Wang, Y. Huang, T. Zhang and B. Liu, J.
Am. Chem. Soc., 2018, 140, 12469–12475.
42 C. Hu and L. Dai, Angew. Chem., Int. Ed., 2016, 55, 11736–
11758.
Environ. Sci. Technol., 2020, 54, 8455–8463.
´ ˘
54 J. C. W. Swart, E. van Steen, I. M. Ciobıca and R. A. van
Santen, Phys. Chem. Chem. Phys., 2009, 11, 803–807.
55 X. Duan, J. Kang, W. Tian, H. Zhang, S.-H. Ho, Y.-A. Zhu,
Z. Ao, H. Sun and S. Wang, Appl. Catal., B, 2019, 256, 117795.
This journal is © The Royal Society of Chemistry 2021
J. Mater. Chem. A, 2021, 9, 7198–7207 | 7207