A Directional Synthesis for Topological Defect in Carbon

Article abstract of DOI:10.1016/j.chempr.2020.05.010

Recently, defect electrocatalysis has become a research focus with significant advances. However, the control synthesis of target defects is still challenging to date, which is prerequisite for deeply understanding the intrinsic activity origin of metal-free catalysts. Herein, inspired by the theoretical demonstration, we report a general edge-engineering strategy to fulfill controlling definitive defect configurations in carbons by the direct removal of specific nitrogen (N) doping sites, representing as one-to-one conversion; e.g., graphitic-N to divacancy (C585), pyridinic-N to separate pentagon (S-C5), and pyrrolic-N to adjacent pentagons (A-C5). Electrochemical measurements reveal that A-C5 defects prefer oxygen reduction reaction (ORR) catalysis, whereas C585 defects are more favorable toward hydrogen evolution reaction (HER). This work provides insights into the design of high-performance carbon-based catalysts based on the principles of defect formation.Metal-free carbon catalysts, especially with rich topological defects, have exhibited extraordinary performance in various electrochemical reactions. To fundamentally understand the relation of activity with specific defect, control of defect type in carbon is indispensable and unfortunately still remains a great challenge to date. Here, we developed a general edge-engineering method to produce a class of definitive defect configurations converted from corresponding specific nitrogen (N) doping sites. Theoretical simulations provide the fundamental capability and the design principles. Experimentally, structural characterizations clearly elucidate the specific one-to-one conversion of confirmative N configuration to carbon defect; e.g., graphitic-N to divacancy (C585), pyridinic-N to separate pentagon (S-C5), and pyrrolic-N to adjacent pentagons (A-C5), respectively. For electrocatalysis, A-C5 defects show the highest intrinsic activity in oxygen reduction reaction, whereas C585 defects perform the best in hydrogen evolution reaction. This work provides the main guidance in design of carbon-based catalysts via control of target defect synthesis.A directional synthesis for specific topological defect in carbon has been realized by a combining theoretical and experimental avenue. The relationship between electrocatalytic activity and defect is thereby uncovered, in that adjacent pentagons show the best oxygen reduction reaction (ORR) activity, whereas divacancy defects prefer hydrogen evolution reaction (HER).

Full text of DOI:10.1016/j.chempr.2020.05.010

ll
Article
A Directional Synthesis for Topological Defect
in Carbon
Xin Wang, Yi Jia, Xin Mao, ...,
Kang Wang, Aijun Du,
Xiangdong Yao
aijun.du@qut.edu.au (A.D.)
x.yao@griffith.edu.au (X.Y.)
HIGHLIGHTS
Topological defect control in
carbon is theoretically designed
A class of specific carbon defect
configurations is experimentally
realized
Adjacent pentagons serve as the
best ORR active sites
Divacancy defect exhibits the
highest activity toward HER
A directional synthesis for specific topological defect in carbon has been realized
by a combining theoretical and experimental avenue. The relationship between
electrocatalytic activity and defect is thereby uncovered, in that adjacent
pentagons show the best oxygen reduction reaction (ORR) activity, whereas
divacancy defects prefer hydrogen evolution reaction (HER).
Wang et al., Chem 6, 1–15
August 6, 2020 ª 2020 Elsevier Inc.
https://doi.org/10.1016/j.chempr.2020.05.010

Please cite this article in press as: Wang et al., A Directional Synthesis for Topological Defect in Carbon, Chem (2020), https://doi.org/10.1016/
j.chempr.2020.05.010
ll
Article
A Directional Synthesis for Topological
Defect in Carbon
Xin Wang,1,8 Yi Jia,1,8 Xin Mao,2,8 Longzhou Zhang,1 Daobin Liu,3 Li Song,3 Xuecheng Yan,1 Jun Chen,4
1,9,
Dongjiang Yang,5 Jizhi Zhou,6 Kang Wang,7 Aijun Du,2, and Xiangdong Yao
*
*
SUMMARY
The Bigger Picture
Recently, defect electrocatalysis
has become a research focus with
significant advances. However,
the control synthesis of target
defects is still challenging to date,
which is prerequisite for deeply
understanding the intrinsic
Metal-free carbon catalysts, especially with rich topological defects,
have exhibited extraordinary performance in various electrochemi-
cal reactions. To fundamentally understand the relation of activity
with specific defect, control of defect type in carbon is indispensable
and unfortunately still remains a great challenge to date. Here, we
developed a general edge-engineering method to produce a class
of definitive defect configurations converted from corresponding
specific nitrogen (N) doping sites. Theoretical simulations provide
the fundamental capability and the design principles. Experimen-
tally, structural characterizations clearly elucidate the specific one-
to-one conversion of confirmative N configuration to carbon defect;
e.g., graphitic-N to divacancy (C585), pyridinic-N to separate
pentagon (S-C5), and pyrrolic-N to adjacent pentagons (A-C5),
respectively. For electrocatalysis, A-C5 defects show the highest
intrinsic activity in oxygen reduction reaction, whereas C585 defects
perform the best in hydrogen evolution reaction. This work provides
the main guidance in design of carbon-based catalysts via control of
target defect synthesis.
activity origin of metal-free
catalysts. Herein, inspired by the
theoretical demonstration, we
report a general edge-
engineering strategy to fulfill
controlling definitive defect
configurations in carbons by the
direct removal of specific nitrogen
(N) doping sites, representing as
one-to-one conversion; e.g.,
graphitic-N to divacancy (C585),
pyridinic-N to separate pentagon
(S-C5), and pyrrolic-N to adjacent
pentagons (A-C5).
INTRODUCTION
Carbon-based metal-free catalysts are widely regarded as promising candidates to
substitute costly noble-metal catalysts in green electrochemical conversion devices
since nitrogen (N)-doped carbon nanotubes were discovered to be active toward ox-
ygen reduction reaction (ORR) in 2009.1–5 It is traditionally accepted that N dopants
in carbon lattice are the active sites to help accelerate the whole catalytic process,
but controversial views still exist: Some reports clamed pyridinic-N acted as the cat-
alytic species in ORR,6,7 whereas others argued the activity origin was graphitic-N.8,9
Recently, a new concept toward defect mechanism of ORR is presented, which has
been boosting the research in a new direction of defect electrocatalysis.10,11 It
should be emphasized that Jia et al. confirmed that pentagon defects rather than
pyridinic-N were the main active sites for acidic ORR.12 This is the first case to con-
trollably synthesize vacancy defects in terms of type and density. However, the
developed method limits to produce pentagon defects from the removal of pyri-
dinic-N. Moreover, the catalyst is only for a proof-of-concept research and cannot
be used for practical electrocatalysis because the active surface occupies only a
very small portion of the catalyst. Accordingly, it is critical to develop a general
methodology for defect control synthesis.
Electrochemical measurements
reveal that A-C5 defects prefer
oxygen reduction reaction (ORR)
catalysis, whereas C585 defects
are more favorable toward
hydrogen evolution reaction
(HER). This work provides insights
into the design of high-
performance carbon-based
catalysts based on the principles
of defect formation.
To accomplish this goal, the directional synthesis of carbon defects is prerequisite,
while current methods for preparing defect-rich carbon materials, including plasma
Chem 6, 1–15, August 6, 2020 ª 2020 Elsevier Inc.
1

Please cite this article in press as: Wang et al., A Directional Synthesis for Topological Defect in Carbon, Chem (2020), https://doi.org/10.1016/
j.chempr.2020.05.010
ll
Article
technique,13,14 potassium hydroxide (KOH) etching,15 template-based carboniza-
tion,11,16 and edge reconstruction by N dopants subtraction10,17–19 are very difficult
to control the synthesis of certain type of defects and can only create ‘‘random’’ de-
fects in carbon matrix. Due to the previous investigations on the N removal to create
topological defect, the removal of pyridinic-N will form pentagons.12 It is hypothe-
sized that the removal of graphitic-N or pyrrolic-N could be likely to form different
type of defects under proper conditions; e.g., removal of a specific N to produce
a definite defect configuration. Consequently, the first challenge is how to precisely
control the synthesis of graphitic-N, pyridinic-N, and pyrrolic-N separately, at least,
in a majority of the certain type of N configuration. Based on the difference of doping
locations between graphitic-N and pyridinic-N, Guo et al. designed an ingenious
experiment to deliberately increase the graphitic-N or pyridinic-N content to a high-
ly selective percentage using highly oriented pyrolytic graphite (HOPG).20 This result
obviously inspires us that the control of synthesis for N configuration is possible, but
their research has its own disadvantages: (1) It is a proof-of-concept investigation
and only a very small portion of the HOPG (the side surface of the channels) is active
for ORR; (2) It cannot be used as a general strategy to control all the N configurations
including pyrrolic-N; and (3) it cannot be used to massively produce useful catalysts
for practical applications. Therefore, it is highly desirable to develop a general
approach to control the synthesis of specific N configuration. The method should
be simple, cheap, and easy to scale up. In this work, an additional benefit is able
to clarify the effect of each type of N configuration (e.g., pyridinic-N, graphitic-N,
or pyrrolic-N) on ORR and identify the real active sites, which is a long-standing
and critical issue of metal-free catalysis.
1School of Environment and Science,
Herein, we report the principles to control of the synthesis of specific N configuration
Queensland Micro- and Nanotechnology Centre,
(e.g., pyridinic-N, graphitic-N, or pyrrolic-N) and the corresponding topological
Griffith University, Nathan Campus, Brisbane,
defect configuration by the direct consequence of N removal. Using density func-
tional theory (DFT) calculations, it is recommended that through N-doping, the orig-
inal carbon structure is crucial to obtain the certain type of N configuration: Perfect
carbon network leads to graphitic-N, edge-rich hexagonal structure leads to pyri-
dinic-N, and edge-rich pentagon leads to pyrrolic-N. The removal of the above spe-
cific N results in the corresponding defect structures; e.g., graphitic-N versus C585,
pyridinic-N versus S-C5, and pyrrolic-N versus A-C5. Experimentally, we use zinc (Zn)
evaporation methodology to create edge-rich hexagonal and pentagon, respec-
tively by control Zn concentration. Consequently, it is realized to control the N
configuration in a very high majority, e.g., graphitic-N and pyridinic-N exceed
75% and pyrrolic-N is greater than 85%. The consequent ORR measurements sug-
gested that neither pyridinic- and pyrrolic-N nor graphitic-N is the active center.
On the contrast, the defect derived from the removal of specific N is proved to be
very active for electrochemical reactions; e.g., A-C5 defects (from the removal of
pyrrolic-N) performed the best in ORR, whereas 585 defects (from the removal of
graphitic-N) favored hydrogen evolution reaction (HER).
QLD 4111, Australia
2School of Chemistry, Physics and Mechanical
Engineering, Queensland University of
Technology, Brisbane, QLD 4000, Australia
3National Synchrotron Radiation Laboratory, CAS
Center for Excellence in Nanoscience, University
of Science and Technology of China, Hefei, Anhui
230029, China
4Intelligent Polymer Research Institute, ARC
Centre of Excellence for Electromaterials
Science, AIIM Facility, University of Wollongong,
Innovation Campus, Wollongong, NSW 2522,
Australia
5Collaborative Innovation Center for Marine
Biomass Fibers, Materials and Textiles of
Shandong Province, Institute of Marine Biobased
Materials, College of Environmental Science and
Engineering, Qingdao University, Qingdao
266071, China
6School of Environmental and Chemical
Engineering, Shanghai University, No. 99
Shangda Rd., Shanghai 200444, China
7Beijing Key Laboratory, for Science and
Application of Functional Molecular and
Crystalline Materials, Department of Chemistry,
University of Science and Technology Beijing,
Beijing 100083, China
RESULTS AND DISCUSSION
Formation Energy Calculations
DFT calculations were used to explore the possibility of defect control synthesis via
choosing different original carbon structures and their consequent conversion to
specific N-doping configurations (Figure 1). Starting from three types of carbon lat-
tices (perfect carbon matrix, edge-rich hexagons, and edge-rich pentagons), it is
proposed that the structural feature differences may lead to the three typical N con-
figurations (graphitic-N locates at the inner part with bonding to three adjacent
8These authors contributed equally
9Lead Contact
*Correspondence: aijun.du@qut.edu.au (A.D.),
x.yao@griffith.edu.au (X.Y.)
https://doi.org/10.1016/j.chempr.2020.05.010
2
Chem 6, 1–15, August 6, 2020

Please cite this article in press as: Wang et al., A Directional Synthesis for Topological Defect in Carbon, Chem (2020), https://doi.org/10.1016/
j.chempr.2020.05.010
ll
Article
Figure 1. Computational Simulation of Specific N-Doping and Removing Process in Different Carbon Models
(A) Schematic and formation energy calculation of transformation from edge-deficient carbon to GN dominated carbon and then to divacancy-rich
carbon.
(B) Schematic and formation energy calculation of transformation from edge-rich carbon to PDN dominated carbon and then to pentagon-rich carbon.
(C) Schematic and formation energy calculation of transformation from pentagon-edge-rich carbon to PON dominated carbon and then to special
carbon reconstruction.
carbon atoms; pyridinic-N dopes at the exposed sites of edge hexagons; and pyrro-
lic-N exists at the apex of edge pentagons).21 Formation energy for every step was
calculated to confirm the theoretical feasibility. As observed clearly, specific N
configuration is energetically favorable in the three well-designed carbon models:
0.76 eV for graphitic-N at basal plane (Figure 1A); ꢀ2.48 and ꢀ1.21 eV for pyri-
dinic-N at zigzag and armchair edge, respectively (Figure 1B); ꢀ1.81 and ꢀ2.14
eV for pyrrolic-N with type-1 and type-2 configuration, respectively (Figure 1C). It
is thus logical to conclude, through intentional control of accessible doping sites
for each type of N, carbon structures with specific N configuration can be realized.
Through consequent removal of such N dopants, topological carbon defects with
certain types are obtained. To be specific, removing one graphitic-N atom can result
in one single vacancy. However, single vacancies are not stable, tending to migrate
and merge into divacancies (C585) with higher stability.22,23 Based on this rule, cal-
culations have given an Ef of 7.35 eV for the conversion from graphitic-N to C585
(Figure 1A). This high value indicates that it may require more energy input, like
high heating temperature, to fulfill such transformation. In contrast, the removal of
pyridinic-N and pyrrolic-N are much easier to take place due to their much lower
Chem 6, 1–15, August 6, 2020
3

Please cite this article in press as: Wang et al., A Directional Synthesis for Topological Defect in Carbon, Chem (2020), https://doi.org/10.1016/
j.chempr.2020.05.010
ll
Article
formation energy, below 2 eV (Figures 1B and 1C). Removal of one pyridinic-N atom,
either locating at zigzag site or armchair site, can lead to one pentagon defect at the
edge (S-C5), yet there are two possible choices in the case of pyrrolic-N removal.
Type 1 induces the recovery of regular hexagonal carbon lattice and type 2 produces
adjacent pentagon defects (A-C5). The calculations in Figure 1 theoretically account
for the rational design of starting carbon structures so as to realize the control syn-
thesis of specific N configurations and the corresponding confirmative defect type.
Synthesis of Carbon Materials with Specific N Configuration
The incorporation of Zn in carbon precursor has been reported to increase the
porosity and exposed edge defects.24–26 Therefore, here, we used a Zn-induced
edge-engineering strategy to synthesize different carbon matrix that simulated
the aforementioned three pristine carbon models (Figure 2A). A modified hydrother-
mal route was first applied to fabricate two kinds of phenol-formaldehyde nanofibers
(PFNs).27 After high-temperature pyrolysis, the PFN without Zn element was trans-
formed into edge-deficient carbon nanofibers (CNFs) (representing perfect carbon
networks) and the other PFNs prepared by adding a small amount of Zn ions
(0.7 mmol) was converted to edge-rich porous CNFs (P-CNFs) (representing edge-
rich hexagonal structure). With a following N-doping process, graphitic-N-domi-
nated CNFs (GN-CNFs) and pyridinic-N-dominated CNFs (PDN-CNFs) could be ob-
tained, respectively. To get pyrrolic-N dominated samples, Zn content should be
largely raised, which is not feasible in the CNF synthetic system. Hence, another hy-
persaline polymerization method was employed to prepare phenol-formaldehyde
monolith (PFM) containing a large amount of Zn species.28 With it, highly porous car-
bon monolith (P-CM) (representing edge-rich pentagon structures) and pyrrolic-N-
dominated CM (PON-CM) could be synthesized. This is also a demonstration of
the generality of this methodology for various materials. Transmission electron mi-
croscope (TEM) and scanning electron microscope (SEM) were used to characterize
the morphologies of as-prepared original carbons and their N-doped counterparts
(Figures S1 and S2). All the CNF samples before and after N-doping treatment,
including CNFs, P-CNFs, GN-CNFs, and PDN-CNFs show the similar nanofiber
structure with diameter of around 60 nm. Differently, the CM samples (P-CM and
PON-CM) are composed of interconnected frazzles with a much more porous
feature. The actual N configurations and content before and after N-doping were
identified by high-resolution X-ray photoelectron spectroscopy (XPS) spectra (Fig-
ures 2B and 2C). The results are well-consistent with our theoretical predictions
that graphitic-N (ꢁ401.0 eV), pyridinic-N (ꢁ398.6 eV), and pyrrolic-N (ꢁ400.1 eV)
are separately occupying a dominant part of the whole N species in GN-CNF,
PDN-CNF, and PON-CM samples, with a relative content of 83.5, 76.4, and 85.4
atom %, respectively (Figure S3). Besides, the total N contents (atom %) are all signif-
icantly increased after doping (1.27 for CNFs to 3.77 for GN-CNFs; 1.30 for P-CNFs
to 4.52 for PDN-CNFs; 0 for P-CM to 2.05 for PON-CM). The normalized N K-edge X-
ray absorption near-edge structure (XANES) spectra also confirm the main N forms,
consistent with the XPS data. As shown in Figure 2D, the three peaks at 398.0, 400.1,
and 410.0 eV can be attributed to p* resonance peaks of pyridinic-N, pyrrolic-N, and
graphitic-N, respectively.29 Apparently, control of specific N configuration is basi-
cally realized through this new method, with dominated pyridinic-N, pyrrolic-N,
and graphitic-N dopants in GN-CNFs, PDN-CNFs, and PON-CM, respectively.
The ORR performance of all starting carbon materials and their corresponding N-
doped counterparts was evaluated in 0.1 M KOH. Linear sweep voltammetry (LSV)
curves in Figure 2E show that the ORR activity for P-CNFs and PDN-CNFs is similar
(halfwave potential: 0.67 V versus 0.65 V; current density at 0.7 V, the fastest kinetics
4
Chem 6, 1–15, August 6, 2020

Please cite this article in press as: Wang et al., A Directional Synthesis for Topological Defect in Carbon, Chem (2020), https://doi.org/10.1016/
j.chempr.2020.05.010
ll
Article
Figure 2. Synthetic Scheme, Structural Characterization, and ORR Performance of Carbon Materials with Controllable N-Doping Types
(A) Schematic illustration of fabricating three types of starting carbon materials.
(B) High-resolution N 1s XPS data of GN-CNF, PDN-CNF, and PON-CM.
(C) Content of graphitic-N ( grap-N), pyridinic-N (pyri-N), and pyrrolic-N (pyrr-N) in the starting carbon materials and their derived N-doped products
based on XPS results.
(D) N K-edge XANES of GN-CNF, PDN-CNF, and PON-CM.
(E–G) ORR performance of P-CNF and PDN-CNF (E), CNF and GN-CNF (F), and P-CM and PON-CM (G).
Chem 6, 1–15, August 6, 2020
5

Please cite this article in press as: Wang et al., A Directional Synthesis for Topological Defect in Carbon, Chem (2020), https://doi.org/10.1016/
j.chempr.2020.05.010
ll
Article
variation of the ORR: 1.74 versus 2.01 mA cmꢀ2). It should be noted that the content
of pyridinic-N in PDN-CNFs (3.45 atom %) is about 15-fold larger than the counter-
part in P-CNFs (0.24 atom %) with almost constant content of graphitic-N in both
samples (1.06ꢁ 1.07 atom %), suggesting pyridinic-N is not the key motivator to-
ward ORR electrocatalysis. This is consistent of our previous research.12 Then we
focus on CNFs and GN-CNFs, with significant change in graphitic-N content (1.05
to 3.15 atom %). Apparently, LSV curves in Figure 2F reveal that the performance
of GN-CNFs is close to that of its pristine CNFs (halfwave potential: 0.62 versus
0.63 V; current density at 0.65 V-the fastest kinetics variation of the ORR: 1.42 versus
1.60 mA cmꢀ2). This indicates graphitic-N is also not active for ORR. Notably, pyri-
dinic-N is not stable at temperature of more than 500ꢂC (confirmed below by Fig-
ure 3D), thus N-doping treatment (700ꢂC) should be accompanied by partial
removal of pyridinic-N in original CNFs (0.22 atom % for CNFs; 0.24 atom % for P-
CNFs), which may result in slightly increased limit diffusion current densities after
CNFs and P-CNFs transform into their N-doped counterparts (GN-CNFs and PDN-
CNFs) (Figures 2E and 2F). By converting P-CM to PON-CM, with the presence of
pyrrolic-N-doping (0 to 1.75 atom %), an interesting fact is shown that the LSV curve
shifts negatively (halfwave potential: 0.72 versus 0.7 V; current density at 0.72 V, the
fastest kinetics variation of the ORR: 2.66 versus 1.97 mA cmꢀ2) (Figure 2G). The
decreased ORR activity by converting P-CM to PON-CM suggests pyrrolic-N may
be not as efficient as pentagon defects in catalyzing ORR. It is thereby concluded
that aforementioned results unambiguously verify neither pyridinic and pyrrolic-N
nor graphitic-N is the active center for ORR.
Formation of Specific Carbon Defects from Removal of Specific N Dopants
To obtain defective carbon samples, the GN-CNF, PDN-CNF, and PON-CM were all
treated with the same strategy as reported in our published paper,10,30 which aimed
to remove the N dopants and induce the rearrangement of the remaining carbon lat-
tice. SEM and TEM images in Figure S4 disclose all the defective samples, referred as
D-CNF-1, D-CNF-2, and D-CM have analogous morphologies as their former N-
doped carbons. XPS survey spectra reveal that, after heat treatment, the N signals
in all three defective samples are all absent, implying the successful subtraction of
N element (Figure S5). This can also be verified by the high-resolution N1s spectra,
showing no detected N species in Figure 3A. Raman spectra give the disorder level
(ID/IG ratio) of the whole nine carbon samples we have made (Figures 3B and S6). For
the starting carbon materials, the ID/IG ratio follows the order of CNF < P-CNF < P-
CM, demonstrating the critical role of Zn in creating edge sites. After incorporation
of N heteroatoms, the D band intensity and ID/IG ratio are increased due to the
disruption of regular carbon lattice. By heat-conversion of the N-doped samples
to defective counterparts, ID/IG value can be further increased, strongly supporting
the fact that new reconstructed defects are produced after removal of N dopants.31
Doping and removing N are both dynamic procedures, it is thereby of great signif-
icance to figure out what is the exact gaseous form of the releasing N fragments dur-
ing the heating process, which has a strong correlation with the resulted carbon de-
fects. Thermogravimetry-mass spectrometry (TG-MS) coupling technique was used
to examine the annealing of as-made specifically N-doped products (GN-CNF, PDN-
CNF, and PON-CM) (Figures 3C–3E).32 Structurally, there are two possible releasing
types of N atom outside carbon lattice, either mono-N molecule (NH3) or N-carbon
compound (HCN: the most typical configuration with one N atom and one C atom).
Accordingly, we set two fragments for test, NH3 (m/z = 17) and HCN (m/z = 27).
Quite encouragingly, no HCN but NH3 signal can be observed for all three N-doped
samples during the heating process, which means N-doping atom, no matter what
6
Chem 6, 1–15, August 6, 2020

Please cite this article in press as: Wang et al., A Directional Synthesis for Topological Defect in Carbon, Chem (2020), https://doi.org/10.1016/
j.chempr.2020.05.010
ll
Article
Figure 3. Investigation of N-Subtracting Process and Structural Characterization of As-Made Defective Samples (D-CNF-1, D-CNF-2, and D-CM)
(A) High-resolution N 1s XPS spectra of D-CNF-1, D-CNF-2, and D-CM.
(B) Comparison of ID/IG ratio obtained from Raman spectra for all starting, N-doped, and defective products.
(C–E) TG-MS analysis of gases evolved from the thermal degradation of GN-CNF (C), PDN-CNF (D), and PON-CM (E). Insets: the corresponding AC
HAADF-STEM images of D-CNF-1 (C), D-CNF-2 (D), and D-CM (E) after fast Fourier transformation filtering.
(F) C K-edge XANES of CNF, GN-CNF, and D-CNF-1.
(G) C K-edge XANES of P-CNF, PDN-CNF, and D-CNF-2.
(H) C K-edge XANES of P-CM, PON-CM, and D-CM.
types, independently escapes from the carbon lattice (NH3) without simultaneously
pulling away surrounding carbon atoms (HCN). This verifies the rationality of theo-
retical design in Figure 1 on structural evolution by subtraction of only one N
atom. Furthermore, according to the TG-MS results, GN-CNF shows the best ther-
mal stability, requiring the highest temperature (>800ꢂC) to release NH3 as
compared with PDN-CNF (ꢁ500ꢂC) and PON-CM (ꢁ250ꢂC), coincident with the
predicted formation energy by DFT calculations. The aberration-corrected atomic-
resolution high-angle annular dark field scanning transmission electron microscopy
Chem 6, 1–15, August 6, 2020
7

Please cite this article in press as: Wang et al., A Directional Synthesis for Topological Defect in Carbon, Chem (2020), https://doi.org/10.1016/
j.chempr.2020.05.010
ll
Article
(AC HAADF-STEM) was further used to directly identify the defect structures (insets
in Figures 3C–3E). C585, S-C5, and A-C5 can be observed clearly in D-CNF-1, D-
CNF-2, and D-CM, respectively, agreeing well with the reconstructed carbon defect
models in Figure 1.
To exactly probe the variation of regularity for carbon structure in as-prepared three
batches of samples, the normalized C K-edge XANES spectra are presented in Fig-
ures 3F and 3G. The sharp peak at approximately 285.5 eV is ascribed to C 1s / p*
transition, which can reflect the integrity level of sp2-hybridized carbon configura-
tion.19,33 For graphitic-N batch in Figure 3F, the intensity of p* feature of GN-CNF
is significantly less than that of starting CNF. However, the situation is totally
different in the other two batches, without obvious change in p* intensity after
doping N atoms (Figures 3G and 3H). The phenomena should come from the struc-
tural difference of N type in these samples. In GN-CNF, the dominant graphitic-N
locates at the center of carbon structure, causing serious damage of graphitic sp2
bond. For comparison, pyridinic-N and pyrrolic-N always locate at the edge sites
with slight impact on the internal p* structure. After removal of N atoms in GN-
CNF, the p* peak intensity rises but still smaller than pristine CNF, which gives an
important experimental evidence for our assumption in Figure 1A. The removal of
graphitic-N will generate single vacancies and their instability will lead to further
migration and merger to form C585 defects. This process can cause partial recovery
of carbon lattice, resulting in the stronger intensity of the p* peak. For PDN-CNF and
PON-CM, the p* peak intensities are both reduced a little by subtracting edge N
dopants (pyridinic-N and pyrrolic-N), suggesting edge reconstruction may occur
to induce more non-hexagon defects that slightly damage the whole p* conjugation.
These results experimentally demonstrate the realization of such specifically ori-
ented N-to-defects transformation.
Electrochemical Performance
Currently, although carbon defects are generally acknowledged to serve as active
sites in ORR, it still remains a great challenge to distinguish which configuration pos-
sesses the highest intrinsic activity. Our successful control of defect types via confir-
mative N-doping provides ideal models to experimentally investigate the activity
difference between N dopants and topological carbon defects. Particularly, the in-
fluence of mass transportation for our ORR measurement can be excluded, because
it is found that Brunauer-Emmett-Teller (BET) surface areas (m2 gꢀ1) for these sam-
ples follow an order of original carbon (CNF: 599; P-CNF: 863; P-CM: 1,264) > N-
doped carbon (GN-CNF: 522; PDN-CNF: 754; PON-CM: 1,150) > defective counter-
part (D-CNF-1: 435; D-CNF-2: 647; D-CM: 910), and the pore size distributions al-
ways keep constant during the N-doping and removing induced structural evolution
(Figure S7; Table S1). These mean only activity per site (N dopant or carbon defect)
affects most to the overall ORR performance. LSV curves and inset tables in Figures
4A–4C apparently show the tailored carbon defects (C585, S-C5, and A-C5) can act
as highly efficient ORR active sites in terms of halfwave potential (0.65 V for PDN-
CNF versus 0.78 V for D-CNF-2; 0.63 V for GN-CNF versus 0.7 V for D-CNF-1;
0.7 V for PON-CM versus 0.81 V for P-CM).
It is well accepted that the activity of catalyst originates from two factors (active site
configurations and accessible active site numbers).34,35 Hence, to truly manifest the
activity difference for various carbon defect types, specific activity is a better indica-
tor for comparison. Accordingly, electrochemical active surface areas (ECSAs) for all
samples were evaluated by double-layer capacitance (Cdl) calculated from cyclic vol-
tammetry (CV) tests in 0.1 M KOH electrolyte (Figures S8–S10).36 Apparently, they all
8
Chem 6, 1–15, August 6, 2020

Please cite this article in press as: Wang et al., A Directional Synthesis for Topological Defect in Carbon, Chem (2020), https://doi.org/10.1016/
j.chempr.2020.05.010
ll
Article
Figure 4. Identification of the Most Active Defect Sites toward ORR and HER in Carbon Catalysts via Electrochemical Test and DFT Calculations
(A–C) ORR performance of PDN-CNF and D-CNF-2 (A), GN-CNF and D-CNF-1 (B), and PON-CM and D-CM (C) in 0.1 M KOH.
(D) HER LSV curves of three defective samples (D-CNF-1, D-CNF-2, and D-CM) in 0.5 M H2SO4.
(E and F) The schematic of all N dopant types (E) and defect types (F) for DFT calculations. Gray and blue balls represent C and N atom, respectively.
(G) Volcano plots of overpotential versus adsorption energy of OH*, indicating carbon defects, especially A-C5 model, as more active sites than N
species for ORR electrocatalysis.
(H) DFT calculated energy profiles of HER for five carbon defect models.
(I) Comparison of calculated overpotential for the different carbon defect models in ORR and HER.
show the similar variation trends as their BET surface areas and the incorporation of
massive Zn indeed leads to more active sites after high-temperature pyrolysis. The
kinetic current density (Jk) is correlated with the intrinsic activity of catalyst.37 There-
fore, the Jk is normalized by ECSAs to study the relationship between carbon defect
types and ORR activity. Potential dependence of Jk specific for as-made nine samples
after normalization is shown in Figure S11. Obviously, the curves for primitive carbon
materials (CNF, P-CNF, and P-CM) are similar to their specific N-doping counter-
parts (GN-CNF, PDN-CNF, and PON-CM), consistent with above results that all N
dopants (graphitic-N, pyridinic-N, and pyrrolic-N) are not active for ORR (Figures
S11A–S11C). Moreover, after removing N dopants to induce defective carbons
Chem 6, 1–15, August 6, 2020
9

Please cite this article in press as: Wang et al., A Directional Synthesis for Topological Defect in Carbon, Chem (2020), https://doi.org/10.1016/
j.chempr.2020.05.010
ll
Article
(D-CNF-1, D-CNF-2, and D-CM), the absolute values of Jk specific are significantly
increased, further confirming the much higher activity of carbon defects on ORR
than specific N-doping sites (Figures S11A–S11C). The comparison of three defec-
tive carbons (D-CNF-1, D-CNF-2, and D-CM) indicates the superiority of A-C5
configuration to S-C5 and C585 toward ORR in 0.1 M KOH, displaying the most pos-
itive curves and the highest Jk specific value, following the order of D-CM > D-CNF-2 >
D-CNF-1 at potentials of less than 0.8 V (Figure S11D). The ORR performances were
also investigated in acidic (0.1 M HClO4) and neutral (0.1 M PBS) media (Figure S12),
which showed similar results as in alkaline electrolyte that A-C5 was the best active
center among the three defect configurations. The kinetic parameters of ORR by the
three defective samples in acidic, neutral, and alkaline electrolytes were further ex-
tracted, also indicating the highest activity for A-C5 defects, with small Tafel slopes
(119 mV decꢀ1 in 0.1 M HClO4; 94 mV decꢀ1 in 0.1 M PBS; 74 mV decꢀ1 in 0.1 M
KOH) at low overpotential for D-CM sample (Figure S13).The electron transfer num-
ber (n) and the yield of peroxide species of D-CM sample, representing the best A-
C5 defects, were calculated based on rotation ring-disk electrode (RRDE) data (Fig-
ure S14). Electron transfer number is close to 4 and peroxide yield remains extremely
low within a wide potential range, demonstrating a 4-electron pathway for A-C5
defect toward catalyzing ORR.
As for HER, the situation is different. HER LSV curves in Figure 4D show that C585
configuration presents significantly higher HER activity than S-C5 and A-C5. Accord-
ing to the calculated Tafel slopes (131, 96, and 67 mV decꢀ1 for D-CNF-2, D-CM, and
D-CNF-1, respectively), C585 defects are further verified to be more favorable to-
ward HER than the other two defect types (S-C5 and A-C5) (Figure S15). Electro-
chemical impedance spectroscopy (EIS) was used to study the charge transfer resis-
tance of all the defective samples, showing D-CM (A-C5) had the rapidest charge
transfer ability in ORR (Figure S16A), whereas D-CNF-1 (C585) had the quickest
HER kinetics (Figure S16B). Working stability is critical for the practical application
of electrocatalysts. Hence, we performed aggressive long-term durability tests on
the best ORR catalysts (D-CM, Figure S17A) and HER catalysts (D-CNF-1, Fig-
ure S17B) by means of chronoamperometry. The i-t curves display negligible current
decay over 10 h, indicating good ORR robustness for D-CM and HER robustness for
D-CNF-1. We next used Raman spectroscopy to analyze the defect stability during
electrocatalysis, revealing the ID/IG value keep unchanged for D-CM in ORR (1.04)
and D-CNF-1 in HER (Figure S18). This unambiguously manifests the carbon defects
can withstand continuous long-term electrocatalytic reactions. Together, all the
above results illustrate that selective durable electrocatalysis may be realized by
rationally controlling the defect types in carbon: Pre-designed A-C5 defects are
the preferential active sites for ORR, and C585 defects can catalyze HER more
efficiently.
DFT Calculations
DFT calculations were carried out to get an in-depth understanding on the relation-
ship between defect types and ORR-HER activity. On the basis of characterization
data, ten N-doped and defective models are built and schematically illustrated in
Figures 4E and 4F, including GN in the bulk phase, PDN at zigzag (PDN-1) and
armchair (PDN-2) edges, PON at two different locations (PON-1 and PON-2), diva-
cancies in the bulk phase and on the edge site (C585-1, C585-2), pentagons on
zigzag and armchair edges (S-C5-1, S-C5-2), and adjacent pentagons (A-C5). The
corresponding free energy evolutions of ORR sub-steps on these active sites are
calculated in Figures S19 and S20. Obviously, carbon defects require lower overpo-
tential than N species, giving a theoretical explanation why N-doping sites are
10
Chem 6, 1–15, August 6, 2020

Please cite this article in press as: Wang et al., A Directional Synthesis for Topological Defect in Carbon, Chem (2020), https://doi.org/10.1016/
j.chempr.2020.05.010
ll
Article
inferior to topological defects in activating ORR (Figure 4G). Notably, among all the
defect models, A-C5 shows the largest onset-potential of 0.81 V for ORR (overpoten-
tial: 0.42 V), unraveling high-edge pentagon density with small distance between
them may have a positive synergistic effect on improving ORR electrocatalysis (Fig-
ure 4I). In addition, HER energy profiles for five defect models are also summarized in
Figure 4H, which uncover that C585 configuration, especially C585-2, has a lower
overpotential as compared with other defect types, suggesting C585 defect is the
most energetically favorable model for HER (Figure 4I).
DISCUSSION
In summary, our work offers a theoretical demonstration and experimental realiza-
tion wherein control of specific N-doping removal mode is a general method to syn-
thesize a class of definitive topological defect types in nanocarbons. More impor-
tantly, electrochemical measurements and DFT calculations attest that neither
pyridinic- and pyrrolic-N nor graphitic-N is an efficient active site for ORR, which is
a long-standing and debate issue of metal-free electrocatalysis. N removal derived
carbon defects are confirmed to be much more active in electrochemical reactions
and the reactivity is highly dependent on the defect type: A-C5 defects prefer
ORR catalysis while C585 defects are more favorable toward HER. These findings
not only clarify the activity origin of ORR for metal-free carbon materials but also pro-
vide the design principles for promising carbon-based catalysts via defect control
synthesis.
EXPERIMENTAL PROCEDURES
Resource Availability
Lead Contact
Further information and requests for resources and reagents should be directed to
and will be fulfilled by the Lead Contact, Xiangdong Yao (x.yao@griffith.edu.au).
Materials Availability
This study did not generate new unique reagents.
Data and Code Availability
The data supporting the findings of this study are available for the article and Sup-
plemental Information or from the corresponding authors upon request.
Synthesis of CNF and P-CNF
Two kinds of PFN were first fabricated with a modified hydrothermal strategy re-
ported previously.27 Typically, hexamethylenetetramine (0.28 g), phenol (0.075 g),
and pre-synthesized Te nanowires (10.7 mg) were uniformly dispersed in 70 mL of
deionized water (DIW). Then, the homogeneous dispersion was transferred into a
Teflon vessel with a total volume of 100 mL. After the autoclave was maintained in
an oven under 160ꢂC for 4 h and naturally cooled down to room temperature, the
first PFN could be obtained by centrifugation and washing with ethanol and DIW.
For the other Zn-PFN, we can prepare it by adding 0.154 g of Zn(COOCH3)2$2H2O
in the initial reactants without changing any other synthetic parameters. Through
further annealing at 1,150ꢂC for 2 h with a ramp rate of 5ꢂC under N atmosphere,
the PFN and Zn-PFN can be converted into CNF and P-CNF, respectively.
Synthesis of P-CM
To attain higher Zn content in carbon precursor, we used a reported hypersaline-as-
sisted polymerization method to prepare PFM.28 Typically, ZnCl2 (8 g) and phenol
(1.882 g) were dispersed in 3 mL of formaldehyde under vigorous magnetic stirring
Chem 6, 1–15, August 6, 2020
11

Please cite this article in press as: Wang et al., A Directional Synthesis for Topological Defect in Carbon, Chem (2020), https://doi.org/10.1016/
j.chempr.2020.05.010
ll
Article
until to form a white viscous sol. Then, the mixture was transferred into a 100 mL
Teflon-lined autoclave and kept at 160ꢂC for 8 h. After atmospheric drying at
100ꢂC, the monolith was annealed at 1,150ꢂC for 2 h under N atmosphere, which
yielded the final P-CM.
Synthesis of GN-CNF, PDN-CNF, and PON-CM
The starting carbon materials were first mixed with melamine at a mass ratio of 1:1
and then annealed at 700ꢂC for 2 h under N atmosphere to generate our target
N-doped samples (CNF/GN-CNF, P-CNF/PDN-CNF, and P-CM/PON-CM).
Synthesis of D-CNF-1, D-CNF-2, and D-CM
To remove the N dopants in as-prepared N-doped carbon materials, a thermal treat-
ment method developed by us was used. By annealing at 1,150ꢂC for 2 h with a ramp
rate of 5ꢂC under an atmosphere of N, the N dopants could be successfully sub-
tracted to yield corresponding carbon defects (GN-CNF/D-CNF-1, PDN-CNF/
D-CNF-1, and PON-CM/D-CM).
Characterizations
The morphology was characterized by SEM (JEOL JSM-7001F) and TEM (Tecnai 20
FEG). XPS were recorded from a Kratos Axis ULTRA X-ray photoelectron spectrom-
eter, and the binding energy of C1s line at 284.6 eV from adventitious carbon was
used as a reference. Raman data were collected on a Renishaw InVia spectrometer
with a model 100 Ramascope optical fiber instrument using the 514 nm laser excita-
tion. N adsorption and desorption isotherms were examined from a TriStar II 3020
automated surface area and pore size analyzer. The N and C-K-edge XANES spectra
were obtained at the MCD Endstation of the BL12B-a beamline at the National Syn-
chrotron Radiation Laboratory (NSRL) in Hefei, China. High-angle annular dark field
(HAADF) images are collected from a probe-corrected JEOL ARM200F with an ac-
celeration voltage of 80 kV.
Electrochemical Measurements
All the electrochemical tests were carried out using a conventional three-electrode
system at an electrochemical station (CHI 760E), with graphitic carbon rod as counter
electrode, glassy carbon electrode as working electrode, and an Ag-AgCl (saturated
KCl solution) electrode as reference electrode. All potentials were calibrated relative
to reversible hydrogen electrode (RHE) by the following calculations: E (versus
RHE) = E (versus Ag-AgCl) + 0.197 + 0.059 pH. The catalyst ink was prepared by
dispersion of 5 mg catalyst powders in 1 mL solution containing 700 mL DIW,
250 mL ethanol, and 50 mL Nafion (5 wt %). The loading amount of catalysts on work-
ing electrode was determined to be 0.3 mg cmꢀ2. The polarization curves were per-
formed at a continuous rotating speed of 1,600 rpm and a sweeping rate of 10 mV
s
ꢀ1. ECSAs was assessed by CV tests with different scanning rate of 10, 20, 30, 40,
and 50 mV sꢀ1 at the potential window of 0.987–1.087 V (versus RHE). For RRDE
test, the electron transfer number (n) and peroxide intermediate production per-
centage (H2O2%) were calculated by the following formula:
Id
n = 4 3
(a)
(b)
Id + Ir =N
Ir =N
H2O2% = 200 3
Id + Ir =N
where Id is the disk current, Ir is the ring current, and N is the current collection effi-
ciency of the Pt ring, which was determined to be 0.43.
12
Chem 6, 1–15, August 6, 2020

Please cite this article in press as: Wang et al., A Directional Synthesis for Topological Defect in Carbon, Chem (2020), https://doi.org/10.1016/
j.chempr.2020.05.010
ll
Article
EIS measurements, performed by applying AC voltage with 10 mV amplitude in a
frequency range from 100 KHz to 1 Hz, were recorded at 0.8 V versus RHE for
ORR in O2 saturated 0.1 M KOH and at ꢀ0.4 V versus RHE for HER in 0.5 M
H2SO4. The ORR and HER stability of samples were tested by the i-t chronoampero-
metric responses.
DFT Computational Details
DFT as implemented in the Vienna ab initio simulation package (VASP) code was em-
ployed to optimize geometry structures.38,39 The exchange-correlation interactions
were described by the generalized gradient approximation (GGA)40 in the form of
the Perdew-Burke-Ernzerhof functional (PBE).41 A cutoff energy of 500 eV for
plain-wave basis was adopted and the convergence threshold was 10ꢀ5 eV and
˚
0.005 eV/A for energy and force, respectively. The weak interaction was described
by DFT + D3 method using empirical correction in Grimme’s scheme.42 The vacuum
˚
space was set to be more than 20 A, which was enough to avoid the interaction be-
tween periodical images. A 4 3 4 graphene super cell with different atomic config-
urations was used for the calculations.
The formation energy is obtained by following equation:
DEf = Edefect + iεC + jεN ꢀ Egraphene
where Edefect and Egraphene are the total energy of graphene with and without a
defect, respectively. εC, and εN are the energy of one C atom in graphene, and
one N atom of N2 molecule in gas phase. i and j indicate the number of C and N
atoms loss.
The overall reaction scheme of ORR process can be described as follows;
O2ðgÞ + eꢀ + H+ /OOHꢃ
OOHꢃ + eꢀ + H+ /H2O ðlÞ + Oꢃ
Oꢃ + eꢀ + H+ /OHꢃ
(a)
(b)
(c)
OHꢃ + eꢀ + H+ /H2O ðlÞ + ꢃ
(d)
The reaction Gibbs free energy changes (DG) for each elementary steps were based
on the computational hydrogen electrode model proposed by Norskov et al.,43
which can be calculated by the following equation;
DG = DE + DZPE ꢀ TDS + DGU + DGpH
where DE is obtained directly from DFT calculations, DZPE is the correction of zero-
point energies (ZPE), T is the temperature of 298.15K, and DS is the correction of en-
tropy for products and reactants. DGU = ꢀ eU is the contribution of electrode poten-
tial to DG, and DGpH = ꢀ kBT ln 10 ꢃ pH. The overpotential for ORR is obtained by:
maxðDGa; DGb; DGc; DGdÞ
h =
+ 1:23
e
in which, DGa; DGb; DGc; and DGd are the reaction Gibbs free energy change of re-
actions (a–d), when U = 0 V.
SUPPLEMENTAL INFORMATION
Supplemental Information can be found online at https://doi.org/10.1016/j.chempr.
2020.05.010.
Chem 6, 1–15, August 6, 2020
13

Please cite this article in press as: Wang et al., A Directional Synthesis for Topological Defect in Carbon, Chem (2020), https://doi.org/10.1016/
j.chempr.2020.05.010
ll
Article
ACKNOWLEDGMENTS
The authors thank the financial support from Australia Research Council (ARC
DP200103043). Y.J. also thanks ARC Discovery Early Career Researcher Award
(ARC DE180101030). The help for TEM characterization by Qiang Sun and Prof Jin
Zou and for XPS by Linzhou Zhuang at the University of Queensland are much appre-
ciated. The authors would like to thank the Australian National Fabrication Facility
(ANFF) Materials Node, University of Wollongong (UOW) Electron Microscopy
Centre (EMC), and Canadian Light Sources for equipment access.
AUTHOR CONTRIBUTIONS
X. Yao conceived, designed, and supervised the project; X.W. and Y.J. conducted
the synthesis and electrocatalytic measurements; X.M. and A.D. performed the
DFT calculations; X.W., L.Z., and X. Yan performed the characterizations including
BET surface area, SEM, TEM, and Raman; J.C. performed ACTEM; D.L. and L.S. per-
formed the XANES experiments; J.Z. and K.W. performed the TG-MS tests; D.Y. pro-
vided some useful suggestions; X.W., Y.J., X. Yao, and A.D. wrote and improved the
manuscript; X.W., Y.J., and X.M. contributed equally to this work; and all authors dis-
cussed the results on the manuscript.
DECLARATION OF INTERESTS
The authors declare no competing interests.
Received: March 4, 2020
Revised: April 3, 2020
Accepted: May 14, 2020
Published: June 18, 2020
REFERENCES
1. Li, G., Blake, G.R., and Palstra, T.T.M. (2017).
Vacancies in functional materials for clean
energy storage and harvesting: the perfect
imperfection. Chem. Soc. Rev. 46, 1693–1706.
oxygen reduction reaction. J. Phys. Chem. Lett.
1, 2622–2627.
13. Liu, Z., Zhao, Z., Wang, Y., Dou, S., Yan, D., Liu,
D., Xia, Z., and Wang, S. (2017). In situ
exfoliated, edge-rich, oxygen-functionalized
graphene from carbon fibers for oxygen
electrocatalysis. Adv. Mater. 29, 1606207.
8. Wang, N., Lu, B., Li, L., Niu, W., Tang, Z., Kang,
X., and Chen, S. (2018). Graphitic nitrogen is
responsible for oxygen electroreduction on
Nitrogen-doped carbons in alkaline
electrolytes: insights from activity attenuation
studies and theoretical calculations. ACS Catal.
8, 6827–6836.
2. Hong, J., Jin, C., Yuan, J., and Zhang, Z. (2017).
Atomic defects in two-dimensional materials:
From single-atom spectroscopy to
functionalities in opto-/electronics,
nanomagnetism, and catalysis. Adv. Mater. 29,
1606434.
14. Dou, S., Tao, L., Wang, R., El Hankari, S., Chen,
R., and Wang, S. (2018). Plasma-assisted
synthesis and surface modification of electrode
materials for renewable energy. Adv. Mater. 30,
e1705850.
9. Lu, X., Wang, D., Ge, L., Xiao, L., Zhang, H., Liu,
L., Zhang, J., An, M., and Yang, P. (2018).
Enriched graphitic N in nitrogen-doped
graphene as a superior metal-free
electrocatalyst for the oxygen reduction
reaction. New J. Chem. 42, 19665–19670.
3. Yan, X., Jia, Y., and Yao, X. (2018). Defects on
carbons for electrocatalytic oxygen reduction.
Chem. Soc. Rev. 47, 7628–7658.
15. Zhu, J., Huang, Y., Mei, W., Zhao, C., Zhang, C.,
Zhang, J., Amiinu, I.S., and Mu, S. (2019). Effects
of intrinsic pentagon defects on
electrochemical reactivity of carbon
nanomaterials. Angew. Chem. Int. Ed. 58,
3859–3864.
4. Wang, Q., Lei, Y., Wang, D., and Li, Y. (2019).
Defect engineering in earth-abundant
electrocatalysts for CO 2 and N 2 reduction.
Energy Environ. Sci. 12, 1730–1750.
10. Jia, Y., Zhang, L., Du, A., Gao, G., Chen, J., Yan,
X., Brown, C.L., and Yao, X. (2016). Defect
graphene as a trifunctional catalyst for
electrochemical reactions. Adv. Mater.
Weinheim 28, 9532–9538.
16. Wang, H., Li, X.B., Gao, L., Wu, H.L., Yang, J.,
Cai, L., Ma, T.B., Tung, C.H., Wu, L.Z., and Yu,
G. (2018). Three-dimensional graphene
5. Gong, K., Du, F., Xia, Z., Durstock, M., and Dai,
L. (2009). Nitrogen-doped carbon nanotube
arrays with high electrocatalytic activity for
oxygen reduction. Science 323, 760–764.
networks with abundant sharp edge sites for
efficient electrocatalytic hydrogen evolution.
Angew. Chem. Int. Ed. 57, 192–197.
11. Tang, C., Wang, H.F., Chen, X., Li, B.Q., Hou,
T.Z., Zhang, B., Zhang, Q., Titirici, M.M., and
Wei, F. (2016). Topological defects in metal-
free nanocarbon for oxygen electrocatalysis.
Adv. Mater. Weinheim 28, 6845–6851.
6. Lai, L., Potts, J.R., Zhan, D., Wang, L., Poh, C.K.,
Tang, C., Gong, H., Shen, Z., Lin, J., and Ruoff,
R.S. (2012). Exploration of the active center
structure of nitrogen-doped graphene-based
catalysts for oxygen reduction reaction. Energy
Environ. Sci. 5, 7936–7942.
17. Wang, X., Zhuang, L., He, T., Jia, Y., Zhang, L.,
Yan, X., Gao, M., Du, A., Zhu, Z., Yao, X., and Yu,
S.-H. (2018). Grafting cobalt diselenide on
defective graphene for enhanced oxygen
evolution reaction. iScience 7, 145–153.
12. Jia, Y., Zhang, L., Zhuang, L., Liu, H., Yan, X.,
Wang, X., Liu, J., Wang, J., Zheng, Y., Xiao, Z.,
et al. (2019). Identification of active sites for
acidic oxygen reduction on carbon catalysts
with and without nitrogen doping. Nat. Catal.
2, 688–695.
7. Rao, C.V., Cabrera, C.R., and Ishikawa, Y.
(2010). In search of the active site in nitrogen-
doped carbon nanotube electrodes for the
18. Qu, Y., Chen, B., Li, Z., Duan, X., Wang, L., Lin,
Y., Yuan, T., Zhou, F., Hu, Y., Yang, Z., et al.
(2019). Thermal emitting strategy to synthesize
14
Chem 6, 1–15, August 6, 2020

Please cite this article in press as: Wang et al., A Directional Synthesis for Topological Defect in Carbon, Chem (2020), https://doi.org/10.1016/
j.chempr.2020.05.010
ll
Article
atomically dispersed Pt metal sites from bulk Pt
metal. J. Am. Chem. Soc. 141, 4505–4509.
templating synthesis, casting films, and their
superhydrophobic property. Adv. Funct.
Mater. 26, 5086–5092.
efficient mass transport for high-performance
proton exchange membrane fuel cells. Nat.
Catal. 2, 259–268.
19. Li, D., Jia, Y., Chang, G., Chen, J., Liu, H., Wang,
J., Hu, Y., Xia, Y., Yang, D., and Yao, X. (2018). A
defect-driven metal-free electrocatalyst for
oxygen reduction in acidic electrolyte. Chem 4,
2345–2356.
35. Chen, L., Liu, X., Zheng, L., Li, Y., Guo, X., Wan,
X., Liu, Q., Shang, J., and Shui, J. (2019).
Insights into the role of active site density in the
fuel cell performance of Co-N-C catalysts.
Appl. Catal. B 256, 117849.
28. Yu, Z.L., Li, G.C., Fechler, N., Yang, N., Ma, Z.Y.,
Wang, X., Antonietti, M., and Yu, S.H. (2016).
Polymerization under hypersaline conditions: a
robust route to phenolic polymer-derived
carbon aerogels. Angew. Chem. Int. Ed. 55,
14623–14627.
20. Guo, D., Shibuya, R., Akiba, C., Saji, S., Kondo,
T., and Nakamura, J. (2016). Active sites of
nitrogen-doped carbon materials for oxygen
reduction reaction clarified using model
catalysts. Science 351, 361–365.
36. Chen, D., Qiao, M., Lu, Y.R., Hao, L., Liu, D.,
Dong, C.L., Li, Y., and Wang, S. (2018).
Preferential cation vacancies in perovskite
hydroxide for the oxygen evolution reaction.
Angew. Chem. Int. Ed. 57, 8691–8696.
29. Zhao, Y., Wan, J., Yao, H., Zhang, L., Lin, K.,
Wang, L., Yang, N., Liu, D., Song, L., Zhu, J.,
et al. (2018). Few-layer graphdiyne doped with
sp-hybridized nitrogen atoms at acetylenic
sites for oxygen reduction electrocatalysis. Nat.
Chem. 10, 924–931.
21. Liu, X., and Dai, L. (2016). Carbon-based metal-
free catalysts. Nat. Rev. Mater. 1, 16064.
37. Li, F., Bu, Y., Han, G.F., Noh, H.J., Kim, S.J.,
Ahmad, I., Lu, Y., Zhang, P., Jeong, H.Y., Fu, Z.,
et al. (2019). Identifying the structure of Zn-N2
active sites and structural activation. Nat.
Commun. 10, 2623.
22. Banhart, F., Kotakoski, J., and Krasheninnikov,
A.V. (2011). Structural defects in graphene. ACS
Nano 5, 26–41.
30. Jia, Y., Zhang, L., Gao, G., Chen, H., Wang, B.,
Zhou, J., Soo, M.T., Hong, M., Yan, X., Qian, G.,
et al. (2017). A heterostructure coupling of
exfoliated Ni–Fe hydroxide nanosheet and
defective graphene as a bifunctional
23. Vicarelli, L., Heerema, S.J., Dekker, C., and
Zandbergen, H.W. (2015). Controlling defects
in graphene for optimizing the electrical
properties of graphene nanodevices. ACS
Nano 9, 3428–3435.
38. Kresse, G., and Furthmuller, J. (1996). Efficient
¨
iterative schemes for ab initio total-energy
calculations using a plane-wave basis set. Phys.
Rev. B Condens. Matter 54, 11169–11186.
electrocatalyst for overall water splitting. Adv.
Mater. 29, 1700017.
31. Daiyan, R., Tan, X., Chen, R., Saputera, W.H.,
Tahini, H.A., Lovell, E., Ng, Y.H., Smith, S.C.,
Dai, L., Lu, X., and Amal, R. (2018).
39. Kresse, G., and Furthmuller, J. (1996). Efficiency
¨
24. Zhang, W., Wu, Z.Y., Jiang, H.L., and Yu, S.H.
(2014). Nanowire-directed templating
synthesis of metal-organic framework
nanofibers and their derived porous doped
carbon nanofibers for enhanced
of ab-initio total energy calculations for metals
and semiconductors using a plane-wave basis
set. Comput. Mater. Sci. 6, 15–50.
Electroreduction of CO2 to CO on a
mesoporous carbon catalyst with progressively
removed nitrogen moieties. ACS Energy Lett.
3, 2292–2298.
40. Perdew, J.P., Burke, K., and Ernzerhof, M.
(1996). Generalized gradient approximation
made simple. Phys. Rev. Lett. 77, 3865–3868.
electrocatalysis. J. Am. Chem. Soc. 136, 14385–
14388.
32. Arrigo, R., Havecker, M., Schlogl, R., and Su,
¨
¨
D.S. (2008). Dynamic surface rearrangement
and thermal stability of nitrogen functional
groups on carbon nanotubes. Chem.
Commun. (Camb.), 4891–4893.
25. Zhao, X., Zou, X., Yan, X., Brown, C.L., Chen, Z.,
Zhu, G., and Yao, X. (2016). Defect-driven
oxygen reduction reaction (ORR) of carbon
without any element doping. Inorg. Chem.
Front. 3, 417–421.
41. Perdew, J.P., Ernzerhof, M., and Burke, K.
(1996). Rationale for mixing exact exchange
with density functional approximations.
J. Chem. Phys. 105, 9982–9985.
33. Chuang, C.H., Wang, Y.F., Shao, Y.C., Yeh,
Y.C., Wang, D.Y., Chen, C.W., Chiou, J.W., Ray,
S.C., Pong, W.F., Zhang, L., et al. (2014). The
effect of thermal reduction on the
42. Grimme, S. (2006). Semiempirical gga-type
density functional constructed with a long-
range dispersion correction. J. Comput. Chem.
27, 1787–1799.
26. Ye, L., Chai, G., and Wen, Z. (2017). Zn-MOF-74
derived N-doped mesoporous carbon as ph-
universal electrocatalyst for oxygen reduction
reaction. Adv. Funct. Mater. 27, 1606190.
photoluminescence and electronic structures
of graphene oxides. Sci. Rep. 4, 4525.
43. Nørskov, J.K., Rossmeisl, J., Logadottir, A.,
Lindqvist, L., Kitchin, J.R., Bligaard, T., and
Jo´ nsson, H. (2004). Origin of the overpotential
for oxygen reduction at a fuel-cell cathode.
J. Phys. Chem. B 108, 17886–17892.
27. Wang, X., Zhang, M.-Q., Kou, R., Lu, L.-L., Zhao,
Y., Xu, X.-W., Liu, G.-M., Zheng, Y.-M., and Yu,
S.-H. (2016). Unique necklace-like phenol
formaldehyde resin nanofibers: scalable
34. Wan, X., Liu, X., Li, Y., Yu, R., Zheng, L., Yan, W.,
Wang, H., Xu, M., and Shui, J. (2019). Fe–N–C
electrocatalyst with dense active sites and
Chem 6, 1–15, August 6, 2020
15