B?N Pairs Enriched Defective Carbon Nanosheets for Ammonia Synthesis with High Efficiency

Article abstract of DOI:10.1002/smll.201805029

Electrochemical synthesis has garnered attention as a promising alternative to the traditional Haber–Bosch process to enable the generation of ammonia (NH₃) under ambient conditions. Current electrocatalysts for the nitrogen reduction reaction (NRR) to produce NH₃ are comprised of noble metals or transitional metals. Here, an efficient metal-free catalyst (BCN) is demonstrated without precious component and can be easily fabricated by pyrolysis of organic precursor. Both theoretical calculations and experiments confirm that the doped B?N pairs are the active triggers and the edge carbon atoms near to B?N pairs are the active sites toward the NRR. This doping strategy can provide sufficient active sites while retarding the competing hydrogen evolution reaction (HER) process; thus, NRR with high NH₃ formation rate (7.75 μg h^?1 mg_cat.^?1) and excellent Faradaic efficiency (13.79%) are achieved at ?0.3 V versus reversible hydrogen electrode (RHE), exceeding the performance of most of the metallic catalysts.

Full text of DOI:10.1002/smll.201805029

FULL PAPER
Ammonia Synthesis
www.small-journal.com
BN Pairs Enriched Defective Carbon Nanosheets
for Ammonia Synthesis with High Efficiency
Chen Chen, Dafeng Yan, Yu Wang, Yangyang Zhou, Yuqin Zou,* Yafei Li,*
and Shuangyin Wang*
be explored for nitrogen reduction reac-
tion (NRR) under ambient conditions.
Extensive studies have been conducted
for NRR, e.g., biomimetic catalysis, photo
Electrochemical synthesis has garnered attention as a promising alternative
to the traditional Haber–Bosch process to enable the generation of
ammonia (NH ) under ambient conditions. Current electrocatalysts for the
3
[
17,18]
catalysis, and photo(electro)catalysis.
nitrogen reduction reaction (NRR) to produce NH are comprised of noble
3
Even though NH synthesis under milder
3
metals or transitional metals. Here, an efficient metal-free catalyst (BCN)
is demonstrated without precious component and can be easily fabricated
by pyrolysis of organic precursor. Both theoretical calculations and
conditions was achieved, but it still
suffers from drawbacks such as necessary
sacrificed reagents, expensive electrolytes,
[
17,19,20]
experiments confirm that the doped BN pairs are the active triggers and
the edge carbon atoms near to BN pairs are the active sites toward the NRR.
This doping strategy can provide sufficient active sites while retarding the
competing hydrogen evolution reaction (HER) process; thus, NRR with high
and low NH3 formation rate.
On
this basis, the electrochemical NH3 syn-
thesis, operating in cost-effective aqueous
electrolytes and utilizing protons directly
from water splitting emerged as a prom-
NH formation rate (7.75 µg h− mgcat. ) and excellent Faradaic efficiency
1
−1
[3–16]
3
ising process for NRR.
Electrocatalysts
(
13.79%) are achieved at −0.3 V versus reversible hydrogen electrode (RHE),
studies so far are associated with noble
[
3,4,7,10,14]
metals,
materials,
transitional metal-based
and conducting polymers.
exceeding the performance of most of the metallic catalysts.
[
6,9]
[21]
The current progress of electrochemical
synthesis of NH3 was summarized in
Figure 1, despite deeper understanding toward reaction mecha-
nisms and obtaining relatively higher performance, problems
1
. Introduction
[
3,4,7,10,14]
[6]
Ammonia (NH ) is a valuable chemical with rapidly growing
demand along with the ever-increasing population around the
world, and further development of human society would be
hindered due to the insufficient NH -derived fertilizers.
to now, the industrial-scale NH synthesis has been dominated
like high cost,
efficiency still exist.
low NH yield, and nonideal Faradaic
3
3
[
5,6]
The conversion of N2 to NH3 is a complex process with
sluggish kinetics due to the inertness of N2 and hindrance
by the competing low over-potential hydrogen evolution
[
1,2]
Up
3
3
[
3–15]
by traditional Haber–Bosch method operating under high tem-
reaction (HER).
Ideal catalysts should provide sufficient
perature (350–550 °C) and high pressure (150–350 atm), which
active sites for NRR, especially facilitating the protonation
[
3–16]
consumed about 1% of the annual energy worldwide.
of *N2 (* indicates an adsorption site) to *N H, which was
2
[
3,4,16]
Moreover, as one of the feedstock, high purity hydrogen was
confirmed to be the rate-determining step for most catalysts.
obtained from fossil fuels, which is accompanied by large
Meanwhile,theHERperformanceofcatalystsshouldbesuppressed
amount of greenhouse gases emission.[
4–15]
For energy saving
to achieve NH synthesis more efficiently. Thus, cost-efficient
3
and sustainable development, alternative approaches should
catalysts with sufficient active sites for NRR and higher
overpotential for HER need to be developed.
Dr. C. Chen, D. Yan, Y. Zhou, Prof. Y. Zou, Prof. S. Wang
State Key Laboratory of Chem/Bio-Sensing and Chemometrics
College of Chemistry and Chemical Engineering
Hunan University
2
. Results and Discussion
2
.1. Theoretical Calculations and Electrocatalytic Activities
Changsha, Hunan 410082, P. R. China
E-mail: yuqin_zou@hnu.edu.cn; shuangyinwang@hnu.edu.cn
Dr. Y. Wang, Prof. Y. Li
College of Chemistry and Materials Science
Nanjing Normal University
Nanjing, Jiangsu 210023, P. R. China
E-mail: liyafei@njnu.edu.cn
Theoretical analyses by means of density functional theory
(DFT) calculations were carried out for preliminary under-
standing of the catalytic ability. The NRR initiates with the N2
activation proceeding with a proton-coupled electron transfer
(
associative path) to form a key intermediate, adsorbed N H
2
The ORCID identification number(s) for the author(s) of this article
can be found under https://doi.org/10.1002/smll.201805029.
[
15]
species (*N H). We mainly focus on the associative mecha-
2
nism because it is well established that the dissociative route
is energetically quite unfavorable for carbon-based metal-free
DOI: 10.1002/smll.201805029
Small 2019, 1805029
1805029 (1 of 9)
© 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.small-journal.com
active transition metals.[23,24] This initial *N H formation step
2
is generally the rate-determining step in the overall pathway
[
25]
owing to the strong N N triple bond. Therefore, the catalytic
activity of each specified configuration can be preliminarily
assessed using the free energy change in the protonation of N2
to *N H (ΔG*N H). A lower ΔG*N H indicates stronger N H
2
2
2
2
bonding and higher activity.
As a typical 2D carbon material, graphene was theoretically
analyzed first (illustrated in Figure 2). The results indicated
that pure graphene has less activation toward NRR due to a
high value of ΔG*N H (2.01 eV for C-1) and corresponding
2
experiment tests were conducted. Just as what we expected, low
−
1
−1
NH formation rates (<1 µg h mgcat. ) and nonideal Faradaic
3
efficiencies were obtained for pure graphene in the whole
potential range (−0.2 V to −0.8 V vs RHE). Given heteroatom
doping of graphene could improve performance by adjusting
the electronic structure and regulating the active sites,[
26–28]
we
prepared boron-doped graphene and nitrogen-doped graphene
by thermal annealing of precursor containing graphene oxide
and corresponding heteroatom precursor. Nevertheless, the the-
oretical calculations indicated that the B or N mono-doping is
Figure 1. Recent progress of electrochemical NH3 synthesis under
ambient conditions.
electrocatalysts due to the prohibitively high energy barrier for
dissociation.[ Actually, under ambient chemical conditions
almost ineffective on improving the ΔG*N H (2.24 and 1.99 eV
for B-1 and N-1, respectively) compared with the undoped gra-
phene, which was confirmed by the poor NRR performance
2
22]
the dissociation of N molecule is very difficult even on some
2
Figure 2. a) Schematic of the computational models. Gray (black), pink, red, blue, and green balls represent C, B, O, N, and H atoms, respectively. b) The
calculated ΔG*N H for different configurations, except C-3, B-3, and BN-6, which cannot bind to N H species with chemical adsorption. c) Density of
2
2
states for the active site (the edge C atom) of C-1, B-1, N-1, and BN-1 systems. The red dashed line represents Fermi level. The energy level with the
highest states (Ep) of C-1, B-1, N-1, and BN-1 are highlighted in black, pink, blue, and green dashed lines, respectively. d) The calculated free energy
diagram for N reduction over BN-1 system. The inset is the geometrical structure of BN-1, of which the active site is highlighted by the dashed red cycle.
2
Small 2019, 1805029
1805029 (2 of 9)
© 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.small-journal.com
experimentally. Except for single doping, the strategy of
co-doping has been extensively utilized to acquire efficient
catalysts. Previously, researchers have systematically studied
the influence of B,N co-doping states on the oxygen reduction
reaction (ORR) performance and conclusions were drawn that
the doping states of B and N have significant influence on the
performance and carbon materials with separated B and N
doping show obvious enhancement of ORR than bonded or
sites (BN-2) with relatively lower ΔG*N H comparing to mono-
2
doping samples. It indicates that the BN pairs are responsible
for the high NRR performance.
The ammonia concentration was quantified by the indophenols
blue method and the ammonia-selective electrode method to
ensure the accuracy of the ammonia yield. The measurement
details are afforded in the Supporting Information. The results
indicated that similar ammonia formation rates were obtained
by these two methods at each applied potential, thus making the
presented ammonia formation rates reliable.
[
28]
undoped cases, that is the BN bond pairs are inert for ORR.
However, in this work, theoretical calculation results showed
that significant improvement for N H adsorption can be real-
ized in B,N co-doping cases, especially in BN-1 (1.72 eV), the
bonded mode with B-N pair.
As the activity of catalysts are essentially governed by their
2
[
30]
electronic structures,
we further performed the density of
states (DOS) analysis at the active center of *N H to gain some
2
Herein, two different synthetic routes were applied to obtain
graphene with distinct doping states. Graphene with bonded B
and N (BNG-B) was prepared by thermal treatment of graphene
oxide and boric acid under ammonia atmosphere. Because
stable BN bonds can be favorably formed by reaction of boric
acid and ammonia and BN pairs tend to be doped into gra-
phene. On the contrary, the separated one (BNG-S) was obtained
by annealing graphene oxide and boric acid under argon
atmosphere to dope B followed by treatment under ammonia
atmosphere to achieve N doping, and the post N doping only
achieved in the defect sites, and had less interaction with the
pre-formed B-doping configuration.[ The NRR performances
of BNG-S and BNG-B were listed in Figure 3. An obviously
deep understanding. The hybridization between the electronic
states of *N H and active atom would trigger the splitting
2
of hybridized energy levels into antibonding states and the
bonding states. In this case, the higher location of the highest
peak of active atom DOS (Ep) would correspondingly lead to
a higher location of antibonding states with a lower occupy,
resulting in a stronger interaction between *N H and active
2
atom, and vice versa. Therefore, the position of Ep is a good
descriptor of the binding strength of *N H. Figure 2c shows
2
the projected p-orbital DOS of the active carbon of the four
typical doping models. The Ep position of C-1 (−8.22 eV) is far
from the Fermi level (0 eV) and subjected to slight change after
the sole B-1(-7.69 eV) or N-1 (−8.11 eV) doping, resulting in a
28]
−
1
−1
improved ammonia formation rate value of 4.87 µg h mgcat.
weak *N H adsorption for C-1, B-1, and N-1. In sharp contrast,
2
at −0.2 V vs RHE was achieved for BNG-B. It is equally
important to retard the side reaction for high efficient catalysis.
As shown in Figure S2 in the Supporting Information, the
B,N co-doping cases have exhibited poor activity toward HER,
which is good for NRR. Specially, the adsorption free energy
of *H species (ΔG*H) of BN-1 is calculated to 0.65 eV, which
is far away from the optimal value (0 eV), confirming its poor
HER performance. Note here that the potential HER active
site of BN-1 is the inner C atom near the NB pair, as the H2
formation step is quite energetically unfavorable on the edge
C site.[ Due to high activation toward NRR and high overpo-
tential to HER, the Faradaic efficiency of BNG-B was 9.77% at
the Ep position of B,N-1(−2.51 eV) is much closer to the Fermi
level than those of C-1, B-1, and N-1, which well explains the
stronger *N H binding of B,N-1. The above results indicate that
2
singular B doping or N doping has barely sense for NRR per-
formance optimization of graphene, but for the co-doping case,
the BN pairs could serve as active triggers for NRR.
Meanwhile, no N H was detected for all samples, indicating
2
4
[
3,4]
high selectivity of the catalysts and a distal pathway of NRR.
Based on this result, we examined the free energy changes
along the possible N reduction pathway over BN-1 to clarify
2
29]
the whole reduction mechanisms. As shown in Figure 2d, the
initial conversion of N to *N H is the rate-determining step,
2
2
−
0.2 V vs RHE, exceeding most of the electrocatalysts under
and the rest of electrochemical steps are considerably mod-
erate. Specially, for the rate-determining step, the BN-1 exhibit
physical adsorption to the inert N2 molecule, indicating that
ambient conditions. But for BNG-S, only a slight improvement
of performance was achieved which resulted from the active
Figure 3. Ammonia formation rates (the average value on 2 h reaction) and corresponding Faradaic efficiencies of G, BG, NG, BNG-S, BNG-B, and
BCN.
Small 2019, 1805029
1805029 (3 of 9)
© 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.small-journal.com
the co-transfer of electron and proton is very important for N2
gas, three characteristic peaks appeared and the positions
activation to form *N H on BN-1. The similar phenomenon can
were well matched with the standard curve of 14NH Cl. Sub-
2
4
1
5
15
be also found in the CO adsorption on carbon-catalysts during
sequently, electrochemical test was conducted with
N
N
2
CO2 reduction. Compared with the conversion of *N H to
as feeding gas and only 15N was detected in the product. The
above results confirmed the conversion of nitrogen to ammonia
by NRR, and the doped nitrogen could not form ammonia due
to the fact that the species of nitrogen elements in the catalyst
was only 14N.
2
*
NHNH via an alternating pathway, the distal hydrogenation to
form *NNH is more energetically favorable, indicating that the
2
associative reduction of N on B,N co-doped graphene prefers
2
[32]
to proceed through the distal pathway. Moreover, the formation
of *NHNH , an important intermediate in N H route, was
2
2
4
significantly overwhelmed by the step of *NNH →*N+NH .
2
3
This explained the limiting detection of N H in the conclusion
2.2. Chemical Bonding Structure and Morphology
Characterization
2
4
of electrocatalysis, as observed experimentally.
Since nitrogen element was introduced to the catalysts, the
nitrogen source of the ammonia should be verified. According
To determine the chemical structure and composition of BNG-S
and BNG-B, electron energy loss spectroscopy (EELS) measure-
to the published literature,[
5,31]
the nitrogen species could be
distinguished by 1H NMR spectra due to the chemical shift
of triplet coupling of 14N and doublet coupling of N. First,
Ar was used as feeding gas to conduct the NRR test for 4 h at
ments were carried out and the K-edge absorption for B, C,
15
[33,34]
and N were located at 191, 284, and 401 eV respectively,
as
seen in Figure 4a. There appears a pair of peak at each absorp-
tion location which resulted from the transitions of 1s-π* and
14
−
0.3 V vs RHE and the result indicated that neither N nor
N was detected by NMR (illustrated in Figure S10, Supporting
1
5
2
1s-σ*, indicating sp hybridization states of B, C, and N atoms
Information). However, when 14N N was used as feeding
14
[34]
both in BNG-S and BNG-B. With quantification, the atom
Figure 4. a) Electron energy loss spectroscopy (EELS) spectra of BNG-S and BNG-B. b) XPS Survey spectra of BNG-S and BNG-B. c) B1s and N1s
spectra of BNG-S and BNG-B. The B1-B4 correspond to BC , BN, BC O, and BCO . N1-N4 are assigned to NB, pyridinic N, pyrrolic N, and quater-
3
2
2
nary N. d) EPR spectra of BNG-B and BCN. e) Survey XPS spectrum of BCN. f) High-resolution spectrum of N1s of BCN. g) High-resolution spectrum of
B1s of BCN. h) C K-edge XANES spectra of G, BG, NG, BNG-S, BNG-B, and BCN. i) N K-edge XANES spectra of G, BG, NG, BNG-S, BNG-B, and BCN.
Small 2019, 1805029
1805029 (4 of 9)
© 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.small-journal.com
percentages of B, C, and N elements in BNG-S were 2.63%,
nanosheets was successfully fabricated and no regular lattice
fringe can be seen in Figure 5b, indicating that this carbon-
based material obtained by pyrolysis of organic precursor was in
8
7.72%, and 9.65%, respectively. However, much higher atom
percentages of B (20.33%) and N (48.88%) were obtained in
BNG-B; the increase of doping amounts may be attributed to
the formation of h-BN during synthetic process.[26] The pres-
ence of B, C, and N were further confirmed by survey X-ray
photoelectron spectroscopy (XPS) spectra. For both BNG-S and
BNG-B, four characteristic peaks at 190.6, 284.6, 397.6, and
[
27]
an amorphous form. The uniform distributions of B, C, and
N in BCN, as shown in Figure 5c, suggest that these elements
were well dispersed at the steps of precursor synthesis and
the post treatment. The bond structure of BCN was analyzed
by XPS and summarized in Figure 4e; similar to BNG-B, quite
high doping levels of B and N were obtained and the B1s
and N1s spectra were both dominated by BN bonds. So the
as-prepared BCN was a carbon-based catalyst with large amount
5
32.6 eV appeared, which belonged to B1s, C1s, N1s, and O1s
[26]
respectively, indicating successful synthesis of B,N co-doped
carbon catalysts. The contents of doping elements determined
by XPS results also showed that the substitutions of BN pairs
favored higher doping levels for B and N. The high-resolution
spectra were given in Figure 4c to clarify the bond configura-
tions of dopants. The B1s spectrum of BNG-S was dominated
[
27]
of BN substitutions. However, just as reported before, this
type of material would show a polycrystalline characterization
as well as BNG-B (Figure S4d, Supporting Information) due
to grain boundaries of h-BN domains. The existence of defect-
rich structure of BCN might be a rational explanation for this
phenomenon; thus electron paramagnetic resonance (EPR)
was used to evaluate the defect levels of BNG-B and BCN.
As shown in Figure 4d, a symmetric Lorentzian line with a g
value of 2.0057 was obtained for BNG-B and BCN. The EPR
intensity of BCN was overwhelming compared to BNG’s, indi-
by BC O at 190.9 eV but it was dominated by BN at 190.7 eV
2
[28]
for BNG-B. The pyridinic N was the main component in N1s
spectrum of BNG-S; however, BN made up 77.10% of BNG-B.
The results indicated that distinct samples with BN bonds
(
BNG-B) and without BN bonds (BNG-S) were successfully
synthesized, which was further proved by Fourier transform
infrared spectroscopy (FTIR). As presented in Figure S3 in the
Supporting Information, two extra peaks can be seen at 766 and
[
35]
cating the existence of unpaired electron and more defects.
After electrochemical tests in Ar-saturated electrolyte and in
−
1
1
375 cm for BNG-B, which was attributed to the in-plane
N -saturated one (Figure S5, Supporting Information), an
2
[
3,5]
stretching of BN and the out-of-plane stretching of BNB
bending modes of h-BN, respectively.[ The FTIR results also
confirmed the existence of BN bonds in BNG-B but not in
BNG-S.
obvious gap of current density appeared because of the NRR.
26]
Meanwhile, the high overpotential of HER for BCN (Figure S2,
Supporting Information) indicated that BCN was an unsuit-
able catalyst toward HER but a promising one toward NRR.
The NRR performance of BCN is presented in Figure 3 in
comparison with BNG-S and BNG-B. Due to sufficient active
For further improvement of NRR performance, we con-
structed a carbon-based catalyst (BCN) with BN substitutions
and more defects by annealing of an organic precursor. The
interaction of boric acid (a Lewis acid and B source) and urea
sites and abundant defects, a further increased NH forma-
3
−
1
−1
tion rate of 7.75 µg h mgcat. at −0.3 V vs RHE was achieved
and the corresponding Faradaic efficiency was 13.79%. The
ammonia formation rate of BCN could be comparable to the
reported value achieved by the metal-containing catalysts
and even higher than the noble-metal catalysts’. Due to the
(a Lewis base and N source) in aqueous solution facilitates the
formation of BN bonds and favors BN substitutions. The
morphology and element distribution of BCN were shown in
Figure 5. Figure 5a presented that a catalyst which made up of
[
5–7]
[
4]
Figure 5. a) SEM images of BCN. b) High-resolution TEM image and the selected area electron diffraction (SAED) of BCN. c) Element mappings of
boron, carbon, and nitrogen.
Small 2019, 1805029
1805029 (5 of 9)
© 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.small-journal.com
lower applied potential and the strategy of providing suffi-
cient active sites while retarding competitive HER, the BCN
has achieved the most efficient ammonia synthesis than the
as BNG-x-y, where x to y is the mass ratio of graphene oxide
to boric acid. As presented in Table S2 and Figure S12 in the
Supporting Information, we can obtain that the contents of B
and N increased along with the increase of boric acid dosage,
and the NRR performance was always enhanced except a
decrease happened when the mass ratio exceeded 1 to 2. Lower
ratios led to little amounts of doping and insufficient active
sites, but excessive doping may result in a decline of electronic
conductivity not favorable for electrocatalysts. Meanwhile,
there is no need for excessive doping because of the limited
edge sites for graphene and the internal doping of BN (BN-6)
was considered inactive by theoretical calculations that BN-6
[8,13]
reported nitrogen-doped carbon materials.
Same as other
catalysts,[
3–7]
despite slight enhancement of ammonia forma-
tion along with the more negative applied potential, the HER
was greatly enhanced, leading to rapid decrease for Faradaic
efficiency. Figure 4h is the C K-edge X-ray absorption near edge
structure (XANES) spectra of various catalysts. The G feature
2
centered at 284.8 eV is assigned to the graphitic-like sp bonding
structure and the C2 feature is ascribed to the CC bond
within the plane. The intensities of G and C2 (290.9 eV) were
significantly reduced for BCN with the highest percentages of
heteroatom doping which indicated an amorphous structure
of BCN and confirmed that the doping atoms were stacking
along the out-of-plane of carbon and embedded into the carbon
substrate. In addition, the increase of C1 (≈287.8 eV) pre-
sented that the surface and the edge sites of BCN were rich of
cannot activate N2 to form chemisorbed *N H (Figure 2a,b).
2
With the aid of sufficient active sites, BNG-1-2 presented
superior ammonia formation rate and Faradaic efficiency and,
therefore, following studies examining the effect of thermal
treat temperature were carried out with this mass ratio. Con-
sequently, a series of BNG-T (T is the temperature value) were
successfully synthesized by adjusting the thermal treatment
temperature while keeping other conditions unchanged. The
results indicated that lower temperatures leaded to little doping
into graphene but an excessively high temperature would result
in a decline of the BN percentage (Figure S14, Supporting
Information). Thus BNG-900 possessed a relatively higher NRR
performance due to suitable chemical structure.
To confirm the importance of chemical structure on the
performance, NRR tests of BCN-10-1, BCN-10-3, BCN-10-5,
BCN-1-1, and BCN-1-2 were carried out. The NRR perfor-
mances and the BN proportions are shown in Figures S15 and
S16 in the Supporting Information, respectively. Along with the
increase of the BN proportions, the NRR performance of the
catalysts exhibited an improvement trend which was attributed
to the increase of the active sites. However, a decline of NRR
performance appeared for BCN-1-2 while the fraction of BN
pairs was increased, resulting from the nonactive internal
doping of BN and the poor electronic conductivity. Thus,
carbon with an appropriate fraction of BN pairs would achieve
the efficient ammonia synthesis.
[36]
CB/CN and some other functional groups. The N K-edge
XANES spectra of various catalysts were shown in Figure 4i.
The spectrum of NG showed a typical configuration of nitrogen
doped graphene as reported in literature. Peak N1 (397.6 eV)
is attributed to the pyridinic nitrogen and Peak N4 (406 eV)
is associated with the transitions from the N 1s core level to
C-N σ* states.[ A similar spectrum and the same located
peaks were acquired for BNG-S. However, the peaks at 401.0,
37]
4
07.4, and 414.0 eV both in the spectra of BNG-B and BCN are
[36]
attributed to the formation of h-BN, indicating the successful
synthesis of catalysts with substitutions of BN pairs.
Considering that the substrate of NRR test was a type of
carbon material, its performance should be excluded. The NRR
test for the substrate was conducted at −0.3 V vs RHE, as shown
in Figure S6 in the Supporting Information; the NH forma-
3
tion with the aid of bare carbon paper was far less than the rate
obtained by catalysis with BCN and Faradaic efficiency was also
quite low. To eliminate the effect of possible residual ammonia
during synthetic process on the results, the source of ammonia
was traced by chemical test in electrolyte with saturated Ar
or N . As shown in Figure S7 in the Supporting Information,
Since the stability of ammonia production is an important
parameter during electrochemical tests, subsequent NRR test
was conducted for BCN at −0.3 V vs RHE for 4 h and the
ammonia concentration in the electrolytes was measured every
half hour. The relationship between the reaction time and the
ammonia yield is listed in Figure S19 in the Supporting Infor-
mation and good linear relationship was acquired. Meanwhile,
the stable current densities in Figures S20 and S21 in the
Supporting Information also indicate the stable catalytic activity
of BCN-1-1 toward NRR.
2
only in N -saturated electrolyte the BCN would achieve a high
2
ammonia formation rate and it was negligible in Ar-saturated
electrolyte at −0.4 V vs RHE. The NRR performances of BCN
in Ar-saturated electrolytes at various potentials are illustrated
in Figure S8 in the Supporting Information. And only trace
amount of ammonia was detected for each applied potential in
Ar-saturated electrolyte, indicating almost no nitrogen-leaching
from BCN with applied potentials and no ammonia was gen-
erated without feeding of gaseous nitrogen, as confirmed by
the 1 N isotope labeling experiment (Figure S10, Supporting
5
To gain more information about the states of the catalysts, the
Raman spectra of BCN with different ratios in the component
of the organic precursor were acquired and shown in Figure
S22 in the Supporting Information. The spectrum of BCN-10-1
2
Information).
The compositions of precursors and the thermal treatment
temperatures have great influence on the doping amounts of
elements and bond configurations, resulting in performance
differences; thus systematic studies were conducted for BNG-B
and BCN. Precursors for BNG-B were prepared by homoge-
neous mixing of graphene oxide with constant mass and boric
acid with various amounts, and consequent thermal treat-
ments were performed at 900 °C for two hours under ammonia
atmosphere with the same gas flow rate. Samples were denoted
−
1
centered at 1337 and 1583 cm which could be assigned to
disorder carbon and graphite carbon respectively and the inten-
sity ratio of disorder carbon to graphite carbon (I /I ) is 1.16,
D
G
indicating a kind of material with defective structure was suc-
cessfully fabricated. However, the location of disorder carbon
shifted to higher wave number along with the increasing
dosage of boric acid and the value of I /I was also improving.
D
G
Small 2019, 1805029
1805029 (6 of 9)
© 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.small-journal.com
[
38]
According to the literature,
the Raman spectrum of h-BN
which are tightly associated with NRR performance. By rational
optimization, a metal-free BCN with sufficient active sites and
abundant defects demonstrated excellent NRR performance
that was better than most of catalysts under the same ambient
conditions. The determination of active sites and the develop-
ment of an efficient metal-free NRR catalyst provide original
insight into catalysts design toward NRR.
−1
peaked at 1369 cm , thus the blue shift and the increasing
intensity ratio could be attributed to the increasing amount of
h-BN. The specific surface areas of catalysts are listed in Figure S23
in the Supporting Information presenting a downward trend
with the increase of boric acid dosage. The BCN-10-1 pos-
sesses the highest value of 117.80 but the lowest active activity
toward NRR; meanwhile, BCN-1-1 with a lower specific surface
2
−1
area of 59.01 m g exhibited the highest ammonia formation
rate which could be attributed to the difference of the chemical
structure between BCN-1-1 and BCN-10-1, and the high specific 4. Experimental Section
surface area of BCN-10-1 may provide the possibility of efficient
Fabrication of Electrocatalysts: Graphene oxide (GO) was synthesized
hydrogen evolution reaction. Based on the above results, we
can concluded that the BCN-1-1 possess the highest intrinsic
active activity toward NRR. The electrochemical impedance
spectroscopes have been conducted at −0.3 V vs RHE and the
applied frequency range is from 0.01 to 10000 Hz. As shown
in Figure S25 in the Supporting Information, high impedances
were acquired for all catalysts due to the less active activity
by the Hummers method and was reduced to form graphene (G) via
annealing at 900 °C for 2 h. To prepare boron doped graphene (BG), the
mixture of GO (0.2 g) and boric acid (0.4 g) was dispersed in aqueous
solution followed by freeze drying and the thermal treatment was
conducted under Argon atmosphere with the same conditions as the
preparation of G. NG was synthesized by the annealing of GO under
ammonia atmosphere. BNG-S was fabricated by thermal treatment of BG
under NH atmosphere at 900 °C for another 2 h. BNG-B was prepared by
3
[
39]
toward the dominated hydrogen evolution reaction.
With
one-step annealing of mixture containing GO (0.2 g) and boric acid (0.4 g)
under ammonia. With this mass ratio in the precursor, BNG-700, BNG-
the increasing amount of dopants, the impedance showed an
enlarged trend which resulted from the formation of h-BN with
less active activity toward HER, and the decline of electronic
conductivity may influence. The test conditions and the capaci-
tances of BCN with different ratios are presented in Figure S26
in the Supporting Information. Compared to other catalysts,
800, BNG-900, and BNG-1000 were prepared with various temperatures
of 700, 800, 900, and 1000 °C, respectively. BNG-x-y (x to y was the mass
ratio of GO to boric acid) was prepared by thermal treatment at 900 °C for
2
h with different ratios. BCN was fabricated by annealing of an organic
precursor consisting of PEG-2000 (0.5 g), boric acid (0.5 g), and urea (5 g)
under Ar atmosphere at 900 °C for 2 h. BCN-700, BCN-800, BCN-900, and
BCN-1000 were prepared with the annealing temperatures of 700, 800,
−
2
the BCN-1-1 possessed a lower capacitance of 2.28 mF cm
9
00, and 1000 °C, respectively. BCN-x-y (x to y was the mass ratio of PEG-
and the highest active activity toward NRR. It was also con-
firmed that the BCN-1-1 possessed the highest intrinsic active
activity toward NRR based on the linear relationship between
the capacitance and the active surface area.[
2000 to boric acid and the dosage of urea was constant) was prepared by
thermal treatment at 900 °C for 2 h with various ratios. Hexagonal boron
nitride (h-BN) with few-layered nanosheets was prepared by the thermal
treatment of commercial h-BN, followed by the controlled gas exfoliation.
The above B-doped samples were washed in hot water (≈80°C) with
constant stirring to remove the boric oxides.
40]
To further confirm the importance of electronic conduc-
tivity and chemical environment to NRR performance, electro-
chemical tests on h-BN nanosheets were also carried out. Even
though h-BN possesses abundant BN bonds but very little
Materials Characterizations: The morphology was analyzed by field
emission scanning electron microscopy (FESEM, Zeiss, *IGMAHD).
High-resolution transmission electron microscopy (TEM) images and
the selected area electron diffractions (SAED) were acquired with a
Zeiss LIBRA 200FE transmission electron microscope equipped with
EELS. The synchrotron X-ray spectroscopies were performed at the
National Synchrotron Radiation Research Center (NSRRC), Taiwan. The
XANES at C K-edge and N K-edge were collected with total electron
yield mode at BL20A. The X-ray diffraction (XRD) measurements used
a Rigaku D/MAX 2500 diffractometer with Cu Kα radiation. The EPR
measurements were carried out on a JEOL JES-FA200 spectrometer. The
FTIR spectra were obtained on an IR Affinity-1 spectrometer with KBr
as the diluents. The XPS analysis was conducted on with an ESCALAB
amount of NH was generated within the whole potential range
3
(illustrated in Figure S27, Supporting Information). It should
be noted that the substitutions of BN pairs into carbon sub-
strate would act as active triggers and the actually active sites
toward NRR are the edge carbon atoms near to BN pairs.
Without the participation of carbon atoms, the h-BN exhibited
the poor performance toward NRR due to the inappropriate
chemical environment. Since the theoretical calculations
confirmed that the internal doping of BN pairs (BN-6, seen
in Figure 2) was inert for NRR, the h-BN which dominated by
internal BN pairs would possess poor performance rationally.
Moreover, the poor electronic conductivity of h-BN may have
possible negative effect on the catalytic activity.
2
50 Xi X-ray photoelectron spectrometer using Mg as the excitation
source. The absorbance data of spectrophotometer were obtained
by a UV1902PC ultraviolet and visible (UV–vis) spectrophotometer.
The 1H NMR spectra were measured on an AscendTM 400 Nuclear
Magnetic Resonance Spectrometer. The Raman spectra of catalysts
were acquired on a RENISHAW inVia Raman Microscope. The specific
surface areas of the catalysts were measured on a JW-BK200C Specific
Surface Area and Aperture Analyzer. An ammonia ion concentration
meter (Bante931-NH3) and an ammonia ion selective electrode
(NH3-US) were used to detect the concentration of ammonia ion
by the ammonia-selective electrode method. The electrochemical
impedance spectroscopies (EIS) were conducted on a Metrohm Autolab.
Electrochemical Test: The electrochemical experiments were carried
out on a CHI1000C multichannel electrochemical station using a
three-electrode configuration with a gas diffusion layer (GDL) working
electrode, Pt mesh counter electrode, and Ag/AgCl reference electrode
(saturated KCl electrolyte), respectively. All potentials were measured
3
. Conclusion
In summary, we constructed a robust and cost-efficient
NRR catalyst by simple pyrolysis of organic precursor. Both
theoretical calculations and practical experiment confirmed that
the incorporation of B-N pairs in carbon matrix would be the
active triggers and the edge carbon atoms near to the B-N
pairs are the active sites toward NRR. The doping levels and
bond states were regulated by changing synthetic conditions,
Small 2019, 1805029
1805029 (7 of 9)
© 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.small-journal.com
against a Ag/AgCl reference electrode (saturated KCl electrolyte) and
converted to the RHE reference scale using E (vs RHE) = E (vs Ag/AgCl)
Keywords
+
0.0591 × pH + 0.197 V. Electrolysis was performance under room
active sites, defects, electrocatalysts, metal-free, nitrogen reduction
reaction
temperature in a two-compartment electrochemical cell with the above-
mentioned membrane as the separator, and the electrolyte volume for
each compartment is 90 mL. Before NRR tests, the electrolyte of cathode
Received: November 28, 2018
Revised: December 31, 2018
Published online:
−
1
was purged with high purity nitrogen (99.999%, 50 mL min ) for 30 min
and then the flow rate of nitrogen was adjusted to 10 mL min and
maintained stable during the constant potential test for 2 h.
−
1
Working Electrode Fabrication: 10 mg catalyst was dispersed in a mixed
solution of 950 µL of absolute ethyl alcohol and 50 µL of 5 wt% Nafion
solution (DuPont) by ultrasonication to form a homogeneous ink.
[
[
1] R. F. Service, Science 2014, 345, 610.
2] S. Licht, B. Cui, B. Wang, F. F. Li, J. Lau, S. Liu, Science 2014, 345,
637.
3] M. M. Shi, D. Bao, B. R. Wulan, Y. H. Li, Y. F. Zhang, J. M. Yan,
Q. Jiang, Adv. Mater. 2017, 29, 1606550.
4] D. Bao, Q. Zhang, F. L. Meng, H. X. Zhong, M. M. Shi, Y. Zhang,
J. M. Yan, Q. Jiang, X. B. Zhang, Adv. Mater. 2017, 29, 1604799.
Thereafter, 100 µL of ink was loaded onto each piece of carbon paper
2
(
1 × 1 cm ), thus the mass loading of catalyst on each working electrode
−
2
is 1 mg cm .
[
[
Quantification of Ammonia: To ensure the accuracy of the ammonia
formation rates in the manuscript, the ammonia concentrations in the
electrolyte were measured by the indophenols blue method and the
ammonia-selective electrode method.
Quantification of Hydrazine: The concentration of hydrazine was
quantified using the method of Watt and Chrisp. The color reagent
was prepared by mixing of para-(dimethylamino) benzaldehyde
[5] G. F. Chen, X. Cao, S. Wu, X. Zeng, L. X. Ding, M. Zhu, H. Wang,
J. Am. Chem. Soc. 2017,139, 9771.
[6] S. Chen, S. Perathoner, C. Ampelli, C. Mebrahtu, D. Su, G. Centi,
Angew. Chem. 2017, 129, 2743.
(
5.99 g), HCl (concentrated, 30 mL), and ethanol (300 mL), and then
saturated by nitrogen for reservation. The absorbance at 458 nm was
measured after the reaction of resultant electrolyte and color reagent
for 20 min.
[7] S. J. Li, D. Bao, M. M. Shi, B. R. Wulan, J. M. Yan, Q. Jiang, Adv.
Mater. 2017, 29, 1700001.
[
8] Y. Liu, Y. Su, X. Quan, X. Fan, S. Chen, H. Yu, H. Zhao, Y. Zhang,
Computational Equation: The Faradaic efficiency is the ratio of the
J. Zhao, ACS Catal. 2018, 8, 1186.
[9] C. Lv, C. Yan, G. Chen, Y. Ding, J. Sun, Y. Zhou, G. Yu, Angew.
Chem., Int. Ed. 2018, 57, 1.
10] H. K. Lee, C. S. L. Koh, Y. H. Lee, C. Liu, I. Y. Phang, X. Han,
C.-K. Tsung, X. Y. Ling, Science Adv. 2018, 4, eaar3208.
11] J.-L. Ma, D. Bao, M.-M. Shi, J.-M. Yan, X.-B. Zhang, Chem 2017, 2,
525.
number of electrons transferred for the generation of NH to the total
3
amount of electricity that flows through the circuit. Assuming three
electrons were needed to form one NH molecule, the Faradaic efficiency
3
[
[
[
(FE) can be calculated as follows
3
×F ×C ×V
FE =
(1)
1
7 ×Q
12] J. Zhao, Z. Chen, J. Am. Chem. Soc. 2017, 139, 12480.
The amount of NH was calculated by the equation
[13] S. Mukherjee, D. A. Cullen, S. Karakalos, K. Liu, H. Zhang, S. Zhao,
H. Xu, K. L. More, G. Wang, G. Wu, Nano Energy 2018, 48, 217.
[14] R. Lan, J. T. Irvine, S. Tao, Sci. Rep. 2013, 3, 1145.
3
m = C ×V
(2)
[
[
[
15] C. J. Van der Ham, M. T. Koper, D. G. Hetterscheid, Chem. Soc. Rev.
014, 43, 5183.
16] C. Guo, J. Ran, A. Vasileff, S. Z. Qiao, Energy Environ. Sci. 2018, 11,
5.
17] H. Li, J. Shang, Z. Ai, L. Zhang, J. Am. Chem. Soc. 2015, 137, 6393.
where F is the Faraday constant, C is the concentration of generated
ammonia, and V is the volume of the electrolyte. The formation rate of
ammonia was averaged by the time, the presented ammonia formation
rate is the average value within the time frame of the tests.
2
4
1
5N2 Isotope Labeling Experiments: An isotopic labeling experiment
15
15
[18] S. Wang, X. Hai, X. Ding, K. Chang, Y. Xiang, X. Meng, Z. Yang,
H. Chen, J. Ye, Adv. Mater. 2017, 29, 1701774.
utilized N-15 enriched gas (98% N N) as the isotopic N2 source
15
to verify the nitrogen source in ammonia. The flow rate of the N gas
2
during the experiment was adjusted to 5 sccm according to the literature.
After NRR for 4 h, the acquired nitrogen species in the electrolyte
was identified by 1H (nuclear magnetic resonance, 400 MHz) NMR
spectrometer.
[19] T. M. Pappenfus, K. M. Lee, L. M. Thoma, C. R. Dukart, ECS Trans.
2009, 16, 89.
[20] V. Kordali, G. Kyriacou, C. Lambrou, Chem. Commun. 2000, 0, 1673.
[21] F. Köleli, T. Röpke, Appl. Catal., B 2006, 62, 306.
[
22] Y. Jiao, Y. Zheng, M. Jaroniec, S. Z. Qiao, J. Am. Chem. Soc. 2014,
36, 4394.
1
[
[
23] G. Ertl, Catal. Rev. 1980, 21, 201.
24] S. Dahl, A. Logadottir, R. C. Egeberg, J. H. Larsen, I. Chorkendorff,
E. Törnqvist, J. K. Nørskov, Phys. Rev. Lett. 1999, 83, 1814.
Supporting Information
Supporting Information is available from the Wiley Online Library or
from the author.
[25] J. H. Montoya, C. Tsai, A. Vojvodic, J. K. Nørskov, ChemSusChem
015, 8, 2180.
2
[
26] S. Wang, L. Zhang, Z. Xia, A. Roy, D. W. Chang, J. B. Baek, L. Dai,
Angew. Chem., Int. Ed. 2012, 51, 4209.
Acknowledgements
[
27] F. Guo, P. Yang, Z. Pan, X. N. Cao, Z. Xie, X. Wang, Angew. Chem.,
Int. Ed. 2017, 56, 8231.
C.C., D.Y., and Y.W. contributed equally to this work. The authors
acknowledge support from the National Natural Science Foundation of
China (Grant No. 21573066, 21825201, and 21873050).
[28] Y. Zhao, L. Yang, S. Chen, X. Wang, Y. Ma, Q. Wu, Y. Jiang, W. Qian,
Z. Hu, J. Am. Chem. Soc. 2013, 135, 1201.
[
[
29] Y. Jiao, Y. Zheng, K. Davey, S.-Z. Qiao, Nat. Energy 2016, 1, 16130.
30] C. Tsai, K. Chan, F. Abild-Pedersen, J. K. Nørskov, Phys. Chem.
Chem. Phys. 2014, 16, 13156.
Conflict of Interest
[
31] Y. Luo, G. F. Chen, L. Ding, X. Chen, L. X. Ding, H. Wang, Joule
2018, 3, 1.
The authors declare no conflict of interest.
Small 2019, 1805029
1805029 (8 of 9)
© 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.small-journal.com
[
[
32] G. F. Chen, S. Ren, L. Zhang, H. Cheng, Y. Luo, K. Zhu, L. X. Ding,
H. Wang, Small Methods 2018, 1800337.
33] C. Huang, C. Chen, M. Zhang, L. Lin, X. Ye, S. Antonietti, M. Lin,
X. Wang, Nat. Commun. 2015, 6, 7698.
34] E. Iyyamperumal, S. Wang, L. Dai, ACS Nano 2012, 6, 5259.
35] M. Zhang, Z. Luo, M. Zhou, C. Huang, X. Wang, Sci. China Mater.
[37] L.-S. Zhang, X.-Q. Liang, W.-G. Song, Z.-Y. Wu, Phys. Chem. Chem.
Phys. 2010, 12, 12055.
[38] S. M. Kim, A. Hsu, M. H. Park, S. H. Chae, S. J. Yun, J. S. Lee,
D.-H. Cho, W. Fang, C. Lee, T. Palacios, Nat. Commun. 2015, 6,
8662.
[
[
[39] L. Pospíšil, M. Hromadová, M. Gál, J. Bulíčková, R. Sokolová,
2
015, 58, 867.
N. Fanelli, Electrochim. Acta 2008, 53, 7445.
[
36] B.-Y. Wang, H. Wang, L.-Y. Chen, H.-C. Hsueh, X. Li, J. Guo, Y. Luo,
[40] L. Tao, C.-Y. Lin, S. Dou, S. Feng, D. Chen, D. Liu, J. Huo, Z. Xia,
S. Wang, Nano Energy 2017, 41, 417.
J.-W. Chiou, W.-H. Wang, P.-H. Wang, Carbon 2016, 107, 857.
Small 2019, 1805029
1805029 (9 of 9)
© 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim