This is the submitted version of the journal article:
Liang, Zhifu; Yang, Dawei; Tang, Pengyi; [et al.]. «Atomically dispersed Fe
in a C2N based catalyst as a sulfur host for efficient lithium-sulfur batteries».
Advanced Energy Materials, Vol. 11, issue 5 (Feb. 2021), art. 2003507. DOI
10.1002/aenm.202003507


This version is available at https://ddd.uab.cat/record/271942

under the terms of the             license
 Atomically dispersed Fe in a C2N based Catalyst as Sulfur Host for Efficient Lithium-
                                        Sulfur Batteries

Zhifu Liang+, Dawei Yang+, Pengyi Tang+, Chaoqi Zhang, Jordi Jacas Biendicho, Yi Zhang,
Jordi Llorca, Xiang Wang, Junshan Li, Marc Heggen, Jeremy David, Rafal E. Dunin-Borkowski,
Yingtang Zhou*, Joan Ramon Morante, Andreu Cabot*, Jordi Arbiol*



Z.F Liang, Dr. J. David, Prof. J. Arbiol

Catalan Institute of Nanoscience and Nanotechnology (ICN2),CSIC and BIST

Campus UAB, Bellaterra, 08193 Barcelona, Catalonia, Spain

Email: arbiol@icrea.cat



Z.F Liang, D. W. Yang, C. Q. Zhang, X. Wang, Dr. J. J. Biendicho, Prof. J. R. Morante, Prof. A.
Cabot

Catalonia Institute for Energy Research - IREC

Sant Adrià de Besòs, Barcelona, 08930, Catalonia, Spain

Email: acabot@irec.cat



Dr. P. Y Tang, Dr. M Heggen, Prof. R E. Dunin-Borkowski

Ernst Ruska-Centre for Microscopy and Spectroscopy with Electrons and Peter Grünberg
Institute Forschungszentrum Jülich GmbH 52425 Jülich, Germany



Prof. Y. T. Zhou

Key Laboratory of Health Risk Factors for Seafood and Environment of Zhejiang Province

Institute of Innovation & Application

Zhejiang Ocean University, Zhoushan, Zhejiang Province, 316022, China

Email: zhouyingtang@zjou.edu.cn




                                               1
Prof. J. Llorca
Institute of Energy Technologies, Department of Chemical Engineering and Barcelona
Research Center in Multiscale Science and Engineering
Universitat Politècnica de Catalunya, EEBE, 08019 Barcelona, Catalonia Spain.


Prof. Yi Zhang

School of Energy Sciences and Engineering, Nanjing Tech University, Nanjing 211816, Jiangsu
Province, China



Prof. A. Cabot, Prof. J. Arbiol

ICREA

Pg. Lluís Companys 23, 08010 Barcelona, Catalonia, Spain



Dr. Junshan Li

Institute of Fundamental and Frontier Sciences, University of Electronic Science and
Technology of China, 610054 Chengdu, P. R. China



D. W. Yang, C. Q. Zhang, X. Wang, Prof. J. R. Morante
Department of Electronic and Biomedical Engineering
Universitat de Barcelona, 08028 Barcelona, Catalonia, Spain


+
    These authors contribute equally to this work

* Corresponding authors




                                                2
Abstract

Lithium-sulfur batteries (LSBs) are considered as one of the most promising next generation
energy storage systems due to their high energy density and low materials cost. However, there
are still some challenges for the commercialization of LSBs, such as the sluggish redox reaction
kinetics and the shuttle effect of lithium polysulfides (LiPS). Here we report on a covalent
organic framework, C2N, loaded with atomically dispersed iron as effective sulfur host in LSB
cathodes. X-ray absorption fine spectroscopy and density functional theory calculations prove
the atomically dispersed Fe/C2N catalysts to promote the reversible electrochemical conversion
reaction and to immobilize LiPS to enhance the LSBs performance. As a result, Fe/C2N-based
cathodes demonstrate significantly improved rate performance and long-term cycling stability.
Fe/C2N-based cathodes display initial capacities up to 1540 mAh g−1 at 0.1 C and 678.7 mAh
g−1 at 5 C, while retaining 496.5 mAh g−1 after 2600 cycles at 3 C with a decay rate as low as
0.013% per cycle. Even at a high sulfur loading of 3 mg cm−2, Fe/C2N-based cathodes deliver
remarkable specific capacity retention of 587 mAh g−1 after 500 cycles at 1 C. Overall the
present work provides a rational structural design strategy to inspire the development of high
performance cathodes based on atomically dispersed catalysts for the next generation LSBs.




Keywords: C2N-based covalent organic frameworks (COFs), atomically dispersed iron,
electrocatalytic polysulfide conversion, lithium-sulfur batteries




                                               3
Introduction

Lithium-sulfur batteries (LSBs) are considered one of the main candidate technologies for next
generation energy storage systems. Main advantages of LSBs are their high theoretical capacity
(~1675 mAh g−1) and high energy density (~2600 Wh kg−1), which is about five times higher
than that of lithium-ion batteries.[1-4] Besides, sulfur, the active cathode material, is highly
abundant in the earth crust and it has a low cost and no major environmental, health, and safety
issues.[5] While LSBs are extremely appealing, some serious drawbacks still impede their
practical application. These drawbacks include the low electrical conductivity of sulfur and
lithium sulfides, the sluggish Li-S reaction kinetics, the large volume changes during
lithiation/delithiation and the shuttle effect of soluble lithium polysulfide (LiPS) Li2Sx (3<x≤8).
The latter has associated an irreversible loss of active material at the cathode and the corrosion
of the lithium metal anode, overall resulting in a rapid capacity fading and a poor coulombic
efficiency that greatly reduce the rate performance, cycling stability and lifespan of LSBs.[6][7]

An effective strategy to increase electrical conductivity and accommodate volume changes is
the use of high surface area and high porosity carbon-based materials as sulfur hosts,[15] e.g.
mesocarbon,[8] graphene,[9] porous carbon,[10] carbon nanotubes,[11] conducting polymers such
as polyaniline[12] or covalent organic frameworks (COFs)[13][14]. These porous materials also
limit LiPS diffusion by hampering their transport, but they just weakly physically interact with
LiPS, which is insufficient to inhibit the shuttle effect. To overcome this limitation, heteroatoms
with stronger electronegativity, e.g. O, N, S, P, have been introduced in carbon-based materials
to improve their interaction with LiPS.[16][17][18] While several of these heteroatom-doped
carbon-based sulfur hosts have demonstrated improved electrochemistry performance, this
strategy is far from its optimization. The main difficulty toward optimizing these complex
materials is the intricate characterization of the heteroatom dispersion, which translates in a
very challenging control of the materials parameters during synthesis and an unreliable
characterization of the electrochemical mechanisms and the structure-performance
relationships.


                                                 4
Recently, COFs with a periodic porous structure, large surface area, and controllable chemical
composition and functionality have been demonstrated as excellent sulfur hosts in LSBs.[19][20]
On the other hand, single-atom catalysts (SACs) based on atomically dispersed metal atoms
have demonstrated outstanding catalytic performances in several reactions, including Li-S
redox reactions.[21-25] Besides maximizing the metal dispersion, the main advantage of SACs is
their high surface energy, which allows decreasing the energy barrier of several catalytic
processes.

C2N, a 2D graphene-like layered COF was synthesized for the first time in 2015 through a wet
chemistry method by Beak and co-workers.[43] Since then, several C2N-based materials have
been developed for their application in electrocatalysis and batteries, e.g Ru/C2N for hydrogen
evolution,[26][27] Fe/C2N for oxygen reduction,[28] and C2N for lithium-ion batteries.[29] In C2N,
pores are surrounded by six pyridine nitrogen atoms, which can be either coordinated with a
metal atom or used themselves as traps for polysulfides and lithium ions, as predicted
theoretically.[30] Using DFT calculations, Lin and co-workers further predicted that metal atoms
embedded in C2N as SACs could hinder the shuttle effect and accelerate the electrochemical
conversion between sulfur and Li2S.[31] Nevertheless, the extremely challenging preparation of
C2N-based catalysts containing an atomic metal distribution has so far prevented to
experimentally corroborate these expectations.

Here, we report on the synthesis of atomically and uniformly dispersed iron on C2N frameworks
(Fe/C2N). These new catalysts present several advantages as sulfur host in LSBs. First, C2N
COF shows a high polarity and excellent electrical conductivity due to the abundance of
pyrazine nitrogen and its planar two dimensional (2D) π-conjugated properties. Second, C2N is
a highly porous and high surface area framework, allowing the efficient transport of lithium
ions and the effective absorption of polysulfide. Third, two iron atoms can be trapped in each
hole, coordinating to the neighbour nitrogen atoms, to act as active sites for the conversion
reaction of polysulfide during the charging and discharging processes. For all these reasons, the
produced Fe/C2N based catalysts were tested as sulfur cathode host materials in LSBs.



                                                 5
Results and Discussion

Fe/C2N composites were synthesized as illustrated in Figure 1a. First, C2N was prepared via a
polycondensation reaction and a subsequent annealing process.[32][33] Next, Fe/C2N was
obtained by a pyrolysis treatment of a mixture of C2N and iron (III) nitrate. Fe/C2N displayed
a granule-type morphology, as observed by scanning electron microscopy (SEM) (Figure S1a).
High resolution transmission electron microscopy (HRTEM) analysis showed no iron
nanoparticles attached or near the Fe/C2N structure (Figure S1b). Energy-dispersive X-ray
spectroscopy (EDS) elemental mapping (Figure S1c) showed the Fe, C, N and O signals to be
uniformly distributed through the nanostructures. It is worth noting that the presence of oxygen
was mainly due to the solution used for the TEM sample preparation and the fact that the C2N
structure present high polarity holes with high affinity for trapping oxygen and moisture. As
shown in Figure 1b and the zoomed detail in Figure 1c, isolated Fe atoms (labeled in yellow)
and double iron atoms (labeled in red) were clearly observed using high angle annular dark field
(HAADF) aberration corrected (AC) scanning transmission electron microscopy (STEM). On
the other hand, extensive SEM and fast Fourier transform (FFT) or power spectra of the
HRTEM images of the Fe/C2N structure indicated C2N to present a low crystallinity or to be
amorphous. To confirm this low crystallinity and discard any electron beam damage during
HRTEM characterization, Fe/C2N was further analyzed by powder X-ray diffraction (XRD,
Figure S2). XRD confirmed the C2N to present a low crystallinity, with a main broad and weak
diffraction peak at about 26.5°. This peak, common for graphene-like materials, corresponded
to a 0.33 nm d-spacing of the (002) crystal plane of the C2N layered structure. It is important to
highlight that no peak corresponding to an iron-based lattice structure was observed. The latter
experimental evidence confirmed the absence of Fe-related nanoparticles or clusters, thus
pointing toward an atomic dispersion of Fe. which is consistent with SEM-EDX, HRTEM and
HAADF AC STEM results.




                                                6
Figure 1. (a) Schematic illustration of the synthesis route of Fe/C2N composite (blue = nitrogen,
grey = carbon, red = oxygen, green = chlorine, orange = iron). (b) HAADF-STEM images of a
Fe/C2N catalyst showing the presence of atomically dispersed iron species: double iron clusters
are circled in red and single iron atoms are circled in yellow.

Thermogravimetric analysis (TGA) showed the weight of Fe/C2N samples to decrease to a 0.9 %
when heating them to 700 ℃ under air (Figure S3a). The remaining 0.9% mass was associated
to Fe2O3, which is consistent with a 0.67 wt% iron content in the initial Fe/C2N.

X-ray photoelectron spectroscopy (XPS) confirmed the Fe concentration to be around 1 wt%
(Figure S4a). Additionally, XPS was used to analyze the sample chemical structure. The high-
resolution N 1s XPS spectrum was fitted with three bands at 406.5 eV, 402.5 eV and 399.6 eV,
which correspond to oxidized nitrogen (406.5 eV and 402.5 eV)[35][51] and pyrazine nitrogen
(399.6 eV)[43] (Figure S4b). The C1s XPS spectrum was resolved into four bands (Figure S4c),
associated to C=C (284.3 eV), C-N (285.1 eV), C=O (287.3 eV) and the C-heteroatom (288.9
eV).[26] The high resolution Fe 2p XPS spectrum (Figure S4d) was fitted with 4 bands
corresponding to two iron oxidation states. The main two bands were located at 724 eV (2p1/2)

                                                7
and 710 eV (2p3/2) and corresponded to a Fe2+ chemical state. The second doublet at 712.7 eV
(2p1/2) and 726.5 eV (2p3/2) was assigned to a Fe3+ chemical environment.[35]

To further reveal the chemical structure of Fe/C2N and particularly the valence state of iron, X-
ray absorption near-edge structure (XANES) analyses were carried out using an Fe foil and
Fe2O3 as references. As shown in Figure 2a, the edge structure of Fe/C2N in the XANES spectra
is much closer to that of Fe2O3 than to Fe, meaning that the valence state of Fe in Fe/C2N is
higher than that of the metallic state, consistently with XPS results. The XANES spectrum of
Fe/C2N also displayed a small peak at 7112.9 eV, which is similar to that in iron
phthalocyanine.[34][45] This peak indicated the presence of a Fe-N bond in Fe/C2N catalysts.[35]
Fourier transform was applied to the extended X-ray absorption fine structure (FT-EXAFS)
spectra shown in Figure 2b and further used to investigate the coordination structure of Fe/C2N.
For the Fe foil, the main peak at 2.2 Å stands for the Fe-Fe bond, while for the Fe2O3, the peaks
at 1.5 Å and 2.6 Å are associated to Fe-O and Fe-Fe bonds, respectively.[36][46] The FT-EXAFS
spectrum of Fe/C2N displayed a peak at 1.76 Å, in between that of Fe-Fe and Fe-O bonds. This
peak was attributed to a Fe-N bond, demonstrating the coordination of iron with nitrogen in
Fe/C2N. To obtain a higher detail of the Fe coordination within Fe/C2N, the EXAFS spectra at
the Fe K-edge was fitted (Figure 2c, Table S1). Fitting results showed that the coordination
number of Fe-N is 3, and the average coordination number of Fe-Fe is 1.5, which discarded the
presence of iron and iron oxide nanoparticles, consistently with SEM-EDX, HRTEM, XRD and
XPS results. Notice that for atomically iron dispersed in the sample, the coordination number
of Fe-Fe bond in our EXAFS fitting results is 1.5, which is consistent with HAADF AC-STEM
results, further confirming the successful iron atomic dispersion.[37]

A wavelet transform (WT) analysis of the K3-weight EXAFS signal was carried out to further
characterize the atom back scattering (Figure 2d and Figure S5). The reference Fe foil
exhibited a WT maximum at 8.0 Å−1, associated to Fe-Fe. The reference Fe2O3 exhibited two
WT maxima, at 8.0 Å−1 and 4.0 Å−1, corresponding to Fe-Fe and Fe-O, respectively. Instead,
Fe/C2N displayed a single WT maximum located at 5.9 Å−1, that was assigned to Fe-N.



                                                8
Figure 2. (a) XANES spectra of a Fe/C2N sample and the reference Fe foil and Fe2O3. (b) k3-
weighted FT-EXAFS spectra corresponding to the Fe K-edge. (c) EXAFS fitting curves in R
space for the Fe/C2N sample. (d) Wavelet transform plot for Fe/C2N.

To investigate the performance of Fe/C2N as sulfur host in LSB cathodes, Fe/C2N composites
were loaded with ca. 70 wt% of elemental sulfur using the melt-impregnation method. In the
following, the Fe/C2N sample loaded with elemental sulfur will be named as S@Fe/C2N. The
XRD pattern obtained on the S@Fe/C2N sample (Figure S2) displayed the peak corresponding
to the orthorhombic sulfur phase, confirming the presence of sulfur. TGA quantified the final
percentage of sulfur in the S@Fe/C2N composite to be 69 wt% (Figure S6a). EDX analysis
showed the sulfur to be uniformly distributed on the surface of Fe/C2N (Figure S6b). The
Brunauer–Emmett–Teller (BET) specific surface area of Fe/C2N was 223.4 m2 g−1, and it
decreased with the sulfur loading to the 13.1 m2 g−1 measured for S@Fe/C2N. In parallel, the
overall pore volume decreased from 0.10 to 0.04 cm3 g−1 after sulfur loading. These results

                                              9
further confirmed sulfur to be infiltrated in the material cavities and pores (Figure S7). [23]

We evaluated the ability of Fe/C2N to adsorb LiPS by immersing the same amount (20 mg) of
Super P, C2N and Fe/C2N into a 10 mM solution of Li2S4. After 24 h, the solutions containing
C2N and Fe/C2N were completely transparent, while the blank solution and the solution
containing Super P showed a dark orange colour (Figure 3a). These results demonstrate the
excellent ability of C2N-based materials for LiPS adsorption. The high-resolution Fe 2p and N
1s XPS spectra obtained on the Fe/C2N before and after Li2S4 adsorption are displayed in
Figure 3b-c. After Li2S4 adsorption, the Fe 2p and N 1s peaks showed a significant shift to
higher binding energies, which denoted a strong chemical interaction between Li2S4 and Fe/C2N

To further investigate the strong interaction between LiPS and Fe/C2N, density functional
theory (DFT) calculations were conducted. For comparison, the interaction between LiPS and
C2N was also calculated. Based on the XPS and XAFS results, the Fe-Fe double atom on C2N
was used as model for the DFT calculations and C2N as reference (Figure S8). Figure 3d shows
the optimized adsorption configuration with Li2S4. Figure S9 exhibits the optimized adsorption
configuration with LiPS species (Li2S, Li2S2, Li2S4, Li2S6, Li2S8 and S8) on C2N and Fe/C2N.
The corresponding binding energies are displayed in Figure 3e. DFT calculations showed the
absolute binding energies for Fe/C2N with LiPS species to be higher than those for C2N,
indicating that Fe/C2N has a stronger ability to absorb soluble LiPS. These results are consistent
with the absorption test of Li2S4, suggesting that Fe/C2N could be effective to suppress the
“shuttle effect” of LiPS.




                                                10
Figure 3 (a) Adsorption test: Digital photograph of the Li2S4 solution before and after (24 h)
the addition of Super P, C2N and Fe/C2N, as marked in each flask cap. (b) High resolution XPS
spectra of Fe 2p from Fe/C2N before and after the Li2S4 adsorption test. (c) High resolution of
XPS spectra of N 1s from Fe/C2N before and after the Li2S4 adsorption test. (d) Adsorption
configurations for Li2S4 on Fe/C2N. (e) Binding energies between LiPS (Li2S, Li2S2, Li2S4,
Li2S6, Li2S8 and S8) and C2N ot Fe/C2N as calculated by DFT.

To prove the electrocatalytic activity of Fe/C2N for polysulfide conversion, CV tests of
symmetric cells with identical working and counter electrodes were conducted in 0.5 M Li2S6.
Electrodes were prepared using a slurry-casting process. Fe/C2N-based electrodes displayed
two cathodic and two anodic symmetric peaks at ±0.13 and ±0.08 V (Figure 4a), which are
associated with the electrochemical oxidation and reduction of Li2S6.[38][39] More specifically,
the cathodic peak at – 0.13 V and the anodic peak at 0.13 V are related to the reaction:

  2−
S6 + 10e− + 12Li +  6Li 2S

and the cathodic peak at 0.08 V and the anodic peak at – 0.08 V to the reaction:



                                               11
      2−
4S6         8e− + 3S8

On the other hand, redox peaks were barely observed in the CV profiles of symmetrical cells
based on C2N. C2N-based cells were also characterized by much lower current densities than
those based on Fe/C2N.

With the Fe/C2N-based electrode, redox peaks were clearly defined even when the scanning
rate was increased from 10 mV s−1 to 100 mV s−1 (Figure S10a). These results pointed out the
important role played by the atomically dispersed Fe on the catalytic reaction of polysulfides.
Besides, the excellent overlapping of the CV curves obtained from symmetric cells based on
Fe/C2N electrodes indicated excellent process reversibility and cell stability (Figure S10b).

Electrochemical impedance spectroscopy (EIS) analysis (Figure S10c) showed the charge
transfer resistance of the Fe/C2N-based electrode to be much lower than that of C2N. Here it is
important to take into account that the activation process is also related to the decreasing of the
charge-transfer resistance after cycling. Therefore, our results provide strong evidence that the
Fe/C2N based electrodes have a better reaction response than that of the C2N electrodes between
polysulfides and Li2S2/Li2S during charging and discharging process.

CV curves of Li-S coin cells based on S@Fe/C2N and S@C2N cathodes are shown in Figure
4b. These CV curves were found to almost overlap during the first cycles, indicating good
reversibility of the sulfur redox reaction (Figure S11). S@Fe/C2N-based cells displayed a peak
at 2.32 V (IB) during the cathodic scan attributed to the transformation reactions:

S8 + 2e − + 2Li + → Li 2S8


3Li 2S8 + 2e − + 2Li + → 4Li 2S6


2Li 2S6 + 2e − + 2Li + → 3Li 2S4 .


The second reduction peak in the cathodic scan at 2.04 V (IC) corresponds to the reactions:[40][41]

Li 2S4 + 2e − + 2Li + → 2Li 2S2

                                                12
Li 2S2 + 2e − + 2Li + → 2Li 2S


For the S@C2N-based cells, cathodic peaks appeared broader and shifted to lower potentials,
2.3 V (IB) and 2.0 V (IC).

During the anodic scan, S@Fe/C2N cells displayed two overlapping peaks at 2.35-2.40 V (IA),
that are attributed to the oxidation reaction from Li2S2/L2S to S8. Broader peaks, shifted around
50 mV to higher potentials (2.40-2.45 V) were obtained for the S@C2N cell. Overall,
S@Fe/C2N cathodes displayed cathodic peaks at a more positive potential and anodic peaks at
more negative potentials than S@C2N, suggesting improved kinetics for the polysulfide
transformation reaction (Figure 4c). The catalytic activity of Fe/C2N electrode was quantified
through the onset potential at a current density of 10 μA cm−2 beyond the baseline current
(Figure S12). As displayed in Figure 4c, S@Fe/C2N cathodes were characterized by higher
onset potentials for the reduction peaks and lower onset potentials for the oxidation peaks,
further demonstrating the important catalytic role played by the atomically dispersed iron in
S@Fe/C2N cathodes.

CV measurements at various scanning rates, from 0.1 mV s−1 to 0.4 mV s−1, were conducted to
explore the reaction kinetics. As shown in Figure 4d and Figure S13a, an increase of the scan
rate did not modify the shape of the redox peaks, indicating good electrochemical stability. The
linear relationship between the potential of the reduction and oxidation peak maxima and the
square root of the scan rate pointed toward a diffusion-limited process (Figure S13b). Thus,
the Li+ ion diffusion coefficients (DLi+) were calculated through the Randles-Sevcik equation:
[42][44]



I P = (2.69 10 5 )n1.5 SDLi + CLi + 0.5
                             0.5




where IP represents the peak current, n is the electron transferred number, S is the geometric
area of the electrode, DLi+ represents the lithium ion diffusion coefficient, CLi+ is the
concentration of lithium ions and v is the potential scanning rate. n, S and CLi+ are constant in
this equation, so DLi+ can be determined from the slope of IP vs v0.5. Slopes of the cathodic and

                                               13
anodic peaks were significantly higher for S@Fe/C2N (Figure S13c-e) than for S@C2N
electrodes, implying higher DLi+ values for the former (Figure S13f). This result suggested a
faster transport of lithium ions and a related more efficient conversion of polysulfides in
S@Fe/C2N than in S@C2N electrodes.

To further evaluate the catalytic effect of the electrode materials on the reversible reaction
between polysulfide and Li2S, potentiostatic nucleation and dissolution experiments were
carried out. As shown in Figure 4d, the deposition of Li2S on Fe/C2N electrodes was
considerably faster and at a larger current density under 2.05 V than on C2N. Based on Faraday’s
law, the Fe/C2N electrode exhibited a precipitation capacity of 252.9 mAh g−1, well above that
of the C2N electrode (180.8 mAh g−1). Similarly, the potentiostatic Li2S dissolution experiment
(Figure 4e) showed the Fe/C2N electrode to be characterized by a much higher current density
and dissolution capacity (525.2 mAh g−1) than C2N (406.8 mAh g−1). These results further
proved that the atomically dispersed iron in C2N enhanced the deposition and kinetic dissolution
of Li2S.

Overall, the dispersion of iron atoms not only provided strong polysulfide interaction sites
through the enhancement of the C2N polarity but also acted as excellent active sites for the
reversible transformation of Li2S.




                                               14
Figure 4 (a) CV curves of symmetrical cells with 0.1 mV s−1 scan rate. (b) CV profiles of Li-S
cells with S@Fe/C2N and S@C2N cathodes. (c) Peak potential and onset potential of
asymmetrical Li-S cells based on the CV curves. (d) CV profiles of the S@Fe/C2N electrode
with scan rates from 0.1 mV s−1 to 0.4 mV s−1. (e) Potentiostatic discharge profiles at 2.04 V on
Fe/C2N and C2N electrodes with Li2S8 catholyte. (f) Potentiostatic charge profiles at 2.32 V to
evaluate the dissolution kinetics of Li2S.

DFT calculations were conducted to reveal in detail the redox kinetics of the LIPS conversion.
Figure 5a and Figure S14 exhibit the initial state, transition state and final state of Li2S
decomposition on Fe/C2N and C2N. The calculated energy barrier for Li2S decomposition on
Fe/C2N and C2N surface was 0.62 eV and 1.52 eV, respectively (Figures 5b). These results
demonstrate that Fe/C2N can greatly reduce the Li2S decomposition energy barrier and enhance
the redox reversibility between Li2S and LiPSs. Next, the Gibbs free energies were calculated
for the S reduction pathways of both S@C2N and S@Fe/C2N cathodes. The optimized
configuration of the intermediates and their Gibbs free energy profiles are exhibited in Figure
5c. The largest increase of Gibbs free energy was obtained for the conversion from Li2S2 to Li2S
species, suggesting this step as the rate-limiting for the total discharge process.[22] The free
energy increase was lower for Fe/C2N (0.89 eV) than for C2N (0.92 eV), which suggested that

                                               15
the reduction of S is more thermodynamically favourable on Fe/C2N than on C2N substrate.




Figure 5 (a) The optimized adsorption configuration of Li2S decomposition on Fe/C2N. (b)
Energy barrier profiles of Li2S cluster decomposition on C2N and Fe/C2N along with different
reaction coordinates. (c) Energy profiles of the reduction of Fe/C2N and C2N substrate
respectively.

In Figure 6a the galvanostatic charge-discharge curves for S@Fe/C2N and S@C2N electrodes
at a current rate of 0.1 C are displayed. Two clear discharge and one charge plateaus are
observed in both cases. The first discharge plateau, at around 2.3 V, is attributed to the reduction
                                      2−        2−
of sulfur to soluble LiPS ( S8 → S6        → S4 ). The second discharge plateau, at about 2.1 V,
                                                                           2−
corresponds to the conversion of soluble LiPS into lithium sulfide ( S4         → Li 2S2 → Li 2S ).[50]

Defining Q1 and Q2 as the capacity of the first discharge and the second discharge plateaus,
respectively, the ratio between Q2 and Q1 (Q2/Q1) can be considered as a measure of the
catalytic activity of the electrode material. The higher the capacity ratio value, the better the
catalytic ability.[47,48] Figure 6b shows the capacity ratio for the S@Fe/C2N electrode to be

                                                     16
quite large (Q2/Q1=2.86), well above the ratio measured for the S@C2N electrode
(Q2/Q1=2.35). This result is consistent with the superior catalytic activity towards polysulfides
redox reaction of the Fe-loaded electrode.

As shown in Figure 6b, the polarization potential, i.e. the difference between the oxidation
potential and the second reduction potential, of the S@Fe/C2N electrode (∆E=142.3 mV) was
significantly lower than that of the S@C2N electrode (∆E= 180.9 mV). The lower overpotential
for the phase conversion between soluble LiPS and insoluble Li2S2/Li2S during the charge
(Figure S15a) and discharge (Figure S15b) processes, further confirmed the improved redox
kinetics of S@Fe/C2N.

Figure 6c and Figure S15c exhibit the galvanostatic charge-discharge voltage profiles at
various current densities ranging from 0.1 C to 5 C. The two discharge plateaus and the charge
plateau were clearly observed even at the highest charge/discharge rates. Figure 6d shows the
specific capacities at different discharge rates of the two electrode types tested. S@Fe/C2N
cathodes were characterized with average discharge capacities of 1480, 1250, 1085, 955, 856.4,
774 and 683 mAh g−1 at current rates from 0.1 C to 5 C, well above the capacities obtained for
S@C2N cathodes (Figure 6d). Besides, when switching back the current rate from 5 C to 0.2
C, fairly similar average capacities were recovered, 1172 mAh g−1 for the S@Fe/C2N electrode, ,
which pointed toward an excellent stability.

The energy efficiency of the energy storage device was calculated using the following formula:

E =  UIdt . As shown in Figure 6e, the S@Fe/C2N electrode was characterized by higher

energy efficiencies, up to 93 % at 0.1C, than S@C2N, which is consistent with the lower
polarization potential of the former, in turn associated with the exceptional catalytic properties
of Fe/C2N.

The cycling performance of S@Fe/C2N and S@C2N electrodes at 1C are shown in Figure 6f.
S@Fe/C2N electrodes displayed not only two-fold higher capacities than S@C2N but also
improved stability. S@Fe/C2N electrodes maintained a discharge capacity of 716.5 mAh g−1
after 900 cycles, with a coulombic efficiency of 99.7%. This value corresponds to a capacity

                                                17
retention of 75.14%, i.e. an average capacity reduction rate of 0.0276% per cycle. Figure 6g
displays the cycling performance of the S@Fe/C2N electrode at 3C current rate. At this high
current rate, the initial discharge capacity was 764.1 mAh g−1. After 2600 cycles, the capacity
was still 496.5 mAh g−1, which corresponds to a 0.013% average capacity attenuation per cycle.

Figure S16 displays results from EIS analysis before the first discharge and after 200 cycles at
1 C. Comparing the S@C2N with the S@Fe/C2N electrode, the latter showed a much lower Rct,
indicating improved electrode kinetics.

To explore the potential practical application of Fe/C2N-based cathodes, electrodes with higher
sulfur loading were produced and tested. Galvanostatic charging and discharging profiles at
various current rates from a S@Fe/C2N electrode with 3.0 mg cm−2 sulfur loading are shown in
Figure S17a. One charging plateau and two discharging plateaus were clearly observed even
at current rates up to 5 C. The average specific capacities were 972 mAh g−1 at 0.1 C and 524
mAh g−1 at 5 C, corresponding to areal capacities of 2.8 mAh cm−2 and 1.6 mAh cm−2,
respectively. Figure S17b shows the rate performances of the S@Fe/C2N electrode with a high
sulfur load, which displayed average discharge capacities of 1189.5, 966, 839.5, 734.7, 680.6,
601.3 and 525.4 mAh g−1 at current rates from 0.1 C to 5 C. Besides, the long-term cycling test
of this electrode displayed the initial capacity of 760 mAh g−1 to decrease at a rate of 0.046%
per cycle to 584.6 mAh g−1 after 500 cycles, which corresponds to a 76.92% capacity retention
(Figure 6h). During this process, a coulombic efficiency above 99.3% was maintained.




                                               18
Figure 6. Electrochemical performance of Fe/C2N and C2N-based electrodes. (a) Galvanostatic
charge-discharge profiles of S@Fe/C2N and S@C2N electrodes with a 0.1 C current rate. (b)
Values of ∆E and Q2/Q1 resulted from the charge/discharge curves. (c) Galvanostatic charge-
discharge profiles of S@Fe/C2N at various rates. (d) Rate capabilities at current rates from 0.1
C to 5 C. (e) Energy efficiency at various current rates. (f) Capacity retention of S@Fe/C2N and
S@C2N electrodes at 1 C. (g) Capacity retention of S@Fe/C2N electrode at 3 C. (h) Capacity
retention at 1 C with 3.0 mg cm−2 sulfur loading.


                                               19
Overall, the above results demonstrate that Fe/C2N-based cathodes show an excellent
electrochemical performance associated to the following properties: 1) The abundant presence
of pyrazine nitrogen and pores in the C2N COFs nanostructure, which can immobilize LiPS and
thus minimize the shuttle effect; 2) The high electrical conductivity of Fe/C2N which maximizes
the sulfur utilization; and 3) The presence of atomically dispersed iron, which is coordinated to
nitrogen and work as efficient active sites to promote the polysulfide conversion reaction
kinetics.


Conclusion

In summary, a novel catalyst based on atomically dispersed iron in a 2D covalent organic
framework (Fe/C2N) has been shown as an efficient sulfur host material for LSBs. Combination
of XPS, XANES, EXAFS, atomic resolution HAADF AC STEM and DFT calculations
confirmed that the synthesized Fe/C2N presents atomically dispersed iron. Iron atoms, which
are in some cases found to group in pairs, are coordinated to nitrogen in C2N COFs pores. These
atomically dispersed Fe atoms not only improve the material polarity to immobilize the soluble
polysulfides, but also act as catalytic active sites to promote the reaction kinetics between sulfur,
polysulfide and Li2S. As a result, S@Fe/C2N-baed cathodes exhibit excellent electrochemical
performance, showing a high capacity and a remarkable rate performance at different current
densities, while keeping outstanding cycling stability. This work demonstrates that atomically
dispersed metal atoms within the unique C2N structure can help to achieve excellent
electrochemical performance. The present strategy can be utilized to design active materials not
only for LSBs, but also for other energy conversion and storage systems.


Experimental Section

Materials: Ethylenediamine (99%), 1,3-dioxolane (DOL, 99.5%), lithium nitrate (99.98%),
lithium sulfide (99.9%), and chloroanilic acid (98%) were purchased from Alfa Aesar. Sulfuric
acid (95%~98%), N-methyl-pyrrolidone (NMP, 99.99%), diethyl ether (99.9%), iron (III)
nitrate nonahydrate (>98%), and tetraethylene glycol dimethyl ether (99%) were purchased
from Sigma-Aldrich. Lithium bis(trifluoromethanesulfonyl) imide (LiTFSI) (99%) was
                                                 20
purchased from Acros Organics and dimethoxymethane (DME, 99%) is from Honeywell. All
chemicals were used without further purification.

Synthesis of hexaaminobenzene (HAB):[32] First, chloroanilic acid was placed into a 15 mL
glass vial which was put in a 0 ℃ ice bath, under vigorous stirring. Next, 5.64 mL
ethylenediamine and a few drops of concentrated sulfuric acid were added. Then the ice bath
was removed and the obtained solution was warmed up to room temperature. Afterward, the
solution was transferred to a 15 mL Teflon autoclave, the autoclave was kept at 80 ℃ for 12 h
to complete the amination reactions. The solution was cooled to room temperature. The
obtained mixture was vacuum filtrated using a polytetrafluoroethylene (PTFE) membrane (0.47
µm), rinsed with diethyl ether and degassed ethanol three times, and freeze-dried overnight.
The identity of the final product was confirmed by 1H NMR . 1H NMR (400 MHz, D2O) δ 3.39
(s). (Figure S18).

Synthesis of C2N:[33] In a typical reaction experiment, an equal mol ratio of hexaaminobenzene
and chloroanilic acid, and degassed NMP were put into a three-necked round bottom flask under
argon gas placed in an ice bath. Under vigorous stirring, a few drops of concentrated sulfuric
acid were added. Then the ice bath was removed and the mixture was warmed up to room
temperature. The resulting solution was heated to 175 ℃ for 12 h. After cooling to room
temperature, the mixture was vacuum filtrated, and washed with ethanol and water for three
times and freeze-dried for 24 h. Finally, the obtained black solid was annealed at 450 ℃ for 3
h under argon gas with a ramp rate of 5 ℃/min.

Synthesis of Fe/C2N: C2N (200 mg) and iron(III) nitrate nonahydrate (404 mg) were added to
15 mL ethanol. The resulting solution was stirred for 48 h at room temperature, and then vacuum
filtrated and dried at 60 ℃ in an oven. The obtained precursor was annealed at 700 ℃ for 3 h.
After that, the black solid was placed into 10 mL 3M HCl solution and stirred for 2 days to
remove iron nanoparticles. The material was centrifuged, washed with water and ethanol for
several times, and finally vacuum dried at 120 ℃ for 12 h.

Preparation of S@C2N and S@Fe/C2N composites: Following a typical melt-diffusion

                                              21
procedure, Fe/C2N (or C2N) was mixed with sublimated sulfur in a mass ration of 1:3 through
uniform milling, then transferred to a clean vial, placed in a Teflon autoclave, sealed under
argon gas and heated for 12 h at 155 ℃. To remove the redundant sulfur outside of the Fe/C2N,
the powder was immersed in a 10 mL CS2 and ethanol solution (1:4, volume ratio) for 10 min
twice.

Li2S4 adsorption tests: Sulfur and Li2S with a molar ratio of 3:1 were mixed with appropriate
amounts of DME and DOL (volume ratio of 1:1) under vigorous stirring overnight, until a
homogeneous dark brown solution was formed. To test the adsorption ability for polysulfide,
20 mg active materials (Fe/C2N, C2N and Super P) were covered by a 3.0 mL 10 × 10−3 M Li2S4
solution in a glass vial, kept shaking and ageing overnight.

Materials characterization: Crystal structures were characterized using powder XRD in a
Bruker AXS D8 Advance X-ray diffractometer. (Cu-Kα radiation, λ = 1.5106 Å, 40 kV and 40
mA; Bruker, Germany). TGA was performed under air or nitrogen gas at a heating rate of
10 ℃/min using a Thermogravimetric Analyzer Q200. SEM images were obtained in a ZEISS
Auriga Field emission scanning electron microscopy (FE-SEM) operating at 20 kV. HRTEM
studies were conducted in a FEI Tecnai F20 microscope at an operating voltage of 200 keV.
HAADF-STEM images and elemental mapping were obtained in a spherical aberration-
corrected transmission electron microscope FEI Titan G2 80-200 ChemiSTEM with four EDX
detectors and operated at 200 keV. X-ray photoelectron spectroscopy (XPS) data was obtained
by using 150 W and a Phoibos 150 MCD-9 detector. The XANES and EXAFS were performed
at the Beijing Synchrotron Radiation Facility (BSRF). The nitrogen adsorption-desorption
isotherms were recorded in a Tristar II 3020 Micromeritics system at 77 K. The specific surface
area and the pore size distribution were calculated by Brunauer–Emmett–Teller (BET) and
Barrett-Joyner-Halenda (BJH) methods.

Electrochemical measurements: To prepare the cathodes, active materials (S@C2N,
S@Fe/C2N), Super P and PVDF binders were mixed with a mass ratio of 8:1:1 dispersing in N-
methyl-2-pyrrolidone (NMP, 99.99%, Sigma-Aldrich). The obtained homogeneous slurry was
cast on an aluminium foil current collector and then vacuum dried at 60 °C overnight. The sulfur
                                               22
was loaded in small plates (with a diameter of 12 mm) and was stamped in a coated aluminium
film with about 1 mg cm−2. For the high sulfur loading, we used a coating of 3.0 mg·cm−2. The
coin cells were assembled in an argon-filled glove box for the electrochemical tests. The anode
used was a Li foil and Celgard 2400 membranes were used as separators. 1.0 M lithium
bis(trifluoromethanesulfony) imide (LiTFSI) and 0.2 M LiNO3 dissolved in the mixed solvent
of DOL and DME (1:1 v/v) were used as the electrolyte. Each coin cell contained about 20 μL
of the electrolyte. To allow the electrolyte to penetrate the electrode sufficiently, all coin cells
were aged for several hours before testing. Galvanostatic charge/discharge (GCD)
measurements were conducted between 1.7 and 2.8 V (vs. Li+/Li) in a Neware BTS4008 battery
cycler. A battery tester BCS-810 from BioLogic was used to perform the cyclic voltammetry
(CV) tests with different scan rates, ranging from 0.1 to 0.4 mV s−1. Electrochemical impedance
spectroscopy (EIS) tests were conducted in the frequency range from100 kHz to 0.01 Hz.

Symmetric cell assembly and tests: Electrode for the symmetric cell were prepared using the
same method as that for the lithium-sulfur battery. The working and counter electrodes utilized
two pieces of the same electrode (with an average loading of ~ 0.5 mg cm−2). Each coin cell
contained 40 μL of electrolyte (0.5 M Li2S6 and 1 M LiTFSI in DOL/DME (v/v = 1/1)). CV
tests were carried out at a scan rate of 10 mV·s−1 and EIS measurements were performed in a
frequency range from 100 kHz to 0.01Hz.

Li2S2 nucleation and dissolution tests: Standard 2032 coin cells were used to analyze the
nucleation and dissolution of the Li2S. Equal amounts of Fe/C2N and C2N catalysts were
dispersed uniformly in ethanol. Then, the catalysts were loaded on carbon paper to work as
cathodes. Li foil worked as anode. The catholyte consisted of 20 μL of 0.25 M Li2S8 and 1.0 M
LiTFSI in tetraethylene glycol dimethyl ether solution. In the case of the anolyte, it consisted
of 20 μL of 1.0 M LiTFSI solution without Li2S8 in the same solvent as the catholyte. To
transform the polysulfide Li2Sx (x=6, 8) to Li2S4 we used the following procedure. The cells
were kept at 2.05 V until the current dropped to 0.01 mA. Fresh coin cells were assembled to
perform the dissolution of Li2S, which were discharged at 0.10 mA to 1.80 V firstly, following
galvanostatically discharge at 0.01 mA to 1.80 V for reducing S species into solid Li2S,

                                                23
completely. Subsequently, the cells were potentiostatically charged at 2.40 V for transforming
the Li2S into polysulfide until the charge current was lower than 0.01 mA.[49]


Supporting Information

Supporting Information is available online.


Acknowledgements

ICN2 is supported by the Severo Ochoa program from Spanish MINECO (Grant No. SEV-
2017-0706) and is funded by the CERCA Programme /Generalitat de Catalunya. Part of the
present work has been performed in the framework of Universitat Autònoma de Barcelona
Materials Science PhD program. JD has received funding from the European Union’s Horizon
2020 research and innovation programme under the Marie Sklodowka-Curie grant agreement
No 665919 (P-SPHERE) co-funded by Severo Ochoa Programme. Z. Liang acknowledges
funding from MINECO SO-FPT PhD grant (SEV-2013-0295-17-1). This project has received
funding from the European Union’s Horizon 2020 research and innovation programme under
grant agreement No 823717-ESTEEM3. The present work is supported by the European
Regional Development Funds and by the Spanish MINECO through the projects SEHTOP,
ENE2016-77798-C4-3-R, and ENE2017-85087-C3. D. Yang, X. Wang and C. Zhang thank the
China Scholarship Council for the scholarship support. Dr. P. Tang acknowledges the Humboldt
Research Fellowship. Dr. J. Li obtained International Postdoctoral Exchange Fellowship
Program (Talent-Introduction program) in 2019 and is grateful for the project (2019M663468)
funded by the China Postdoctoral Science Foundation. Authors acknowledge funding from
Generalitat de Catalunya 2017SGR327 and 2017SGR1246. J. Llorca is a Serra Húnter Fellow
and is grateful to MICINN/FEDER RTI2018-093996-B-C31, GC 2017 SGR 128 and to ICREA
Academia program.


Conflict of interest

The authors declare no conflict of interest.


                                               24
References

[1]    H.-J. Peng, J.-Q. Huang, X.-B. Cheng, Q. Zhang, Adv. Energy Mater. 2017, 7.

[2]    A. Manthiram, S. H. Chung, C. Zu, Adv. Mater. 2015, 27, 1980-2006.

[3]    P. G. Bruce, S. A. Freunberger, L. J. Hardwick, J. M. Tarascon, Nat. Mater. 2011, 11,
       19-29.

[4]    L. Lin, F. Pei, J. Peng, A. Fu, J. Cui, X. Fang, N. Zheng, Nano Energy 2018, 54, 50-58.

[5]    J. Song, M. J. Choo, H. Noh, J. K. Park, H. T. Kim, ChemSusChem 2014, 7, 3341-3346.

[6]    Q. Pang, X. Liang, C. Y. Kwok, L. F. Nazar, Nat. Energy 2016, 1.

[7]    M. Hagen, D. Hanselmann, K. Ahlbrecht, R. Maça, D. Gerber, J. Tübke, Adv. Energy
       Mater. 2015, 5.

[8]    B. Zhang, X. Qin, G. R. Li, X. P. Gao, Energy Environ. Sci. 2010, 3.

[9]    J.-Z. Wang, L. Lu, M. Choucair, J. A. Stride, X. Xu, H.-K. Liu, J. Power Sources 2011,
       196, 7030-7034.

[10]   G. Li, W. Lei, D. Luo, Y.-P. Deng, D. Wang, Z. Chen, Adv. Energy Mater. 2018, 8.

[11]   Y. S. Su, A. Manthiram, Chem. Commun. 2012, 48, 8817-8819.

[12]   P. Wei, M. Q. Fan, H. C. Chen, X. R. Yang, H. M. Wu, J. Chen, T. Li, L. W. Zeng, C. M.
       Li, Q. J. Ju, D. Chen, G. L. Tian, C. J. Lv, Renew. Energy 2016, 86, 148-153.

[13]   H. Liao, H. Wang, H. Ding, X. Meng, H. Xu, B. Wang, X. Ai, C. Wang, J. Mater. Chem.
       A 2016, 4, 7416-7421.

[14]   T. Sun, J. Xie, W. Guo, D. S. Li, Q. Zhang, Adv. Energy Mater. 2020, 10.

[15]   Y. Yang, G. Zheng, Y. Cui, Chem. Soc. Rev. 2013, 42, 3018-3032.

[16]   X. Zhou, J. Tang, J. Yang, J. Xie, B. Huang, J. Mater. Chem. A 2013, 1.

[17]   X. G. Sun, X. Wang, R. T. Mayes, S. Dai, ChemSusChem 2012, 5, 2079-2085.

[18]   F. Sun, J. Wang, H. Chen, W. Li, W. Qiao, D. Long, L. Ling, ACS Appl. Mater. Interfaces
       2013, 5, 5630-5638.

[19]   Y. Cao, C. Liu, M. Wang, H. Yang, S. Liu, H. Wang, Z. Yang, F. Pan, Z. Jiang, J. Sun,
       Energy Storage Mater. 2020, 29, 207-215.

[20]   X. Chen, Y. Xu, F. H. Du, Y. Wang, Small Methods 2019, 3.

[21]   J. Wang, L. Jia, J. Zhong, Q. Xiao, C. Wang, K. Zang, H. Liu, H. Zheng, J. Luo, J. Yang,
                                              25
       H. Fan, W. Duan, Y. Wu, H. Lin, Y. Zhang, Energy Storage Mater. 2019, 18, 246-252.

[22]   Z. Du, X. Chen, W. Hu, C. Chuang, S. Xie, A. Hu, W. Yan, X. Kong, X. Wu, H. Ji, L. J.
       Wan, J. Am. Chem. Soc. 2019, 141, 3977-3985.

[23]   J. Xie, B. Q. Li, H. J. Peng, Y. W. Song, M. Zhao, X. Chen, Q. Zhang, J. Q. Huang, Adv.
       Mater. 2019, 31, e1903813.

[24]   L. Zhang, D. Liu, Z. Muhammad, F. Wan, W. Xie, Y. Wang, L. Song, Z. Niu, J. Chen,
       Adv. Mater. 2019, 31, e1903955.

[25]   Y. Li, J. Wu, B. Zhang, W. Wang, G. Zhang, Z. W. Seh, N. Zhang, J. Sun, L. Huang, J.
       Jiang, J. Zhou, Y. Sun, Energy Stor. Mater. 2020, 30, 250-259.

[26]   J. Mahmood, F. Li, S. M. Jung, M. S. Okyay, I. Ahmad, S. J. Kim, N. Park, H. Y. Jeong,
       J. B. Baek, Nat. Nanotechnol. 2017, 12, 441-446.

[27]   J. Mahmood, S.-M. Jung, S.-J. Kim, J. Park, J.-W. Yoo, J.-B. Baek, Chemistry of
       Materials. 2015, 27, 4860-4864.

[28]   J. Mahmood, F. Li, C. Kim, H.-J. Choi, O. Gwon, S.-M. Jung, J.-M. Seo, S.-J. Cho, Y.-
       W. Ju, H. Y. Jeong, G. Kim, J.-B. Baek, Nano Energy 2018, 44, 304-310.

[29]   J. Xu, J. Mahmood, Y. Dou, S. Dou, F. Li, L. Dai, J. B. Baek, Adv. Mater. 2017, 29.

[30]   J. Wu, L.-W. Wang, J. Mater. Chem. A 2018, 6, 2984-2994.

[31]   H. Lin, R. Jin, A. Wang, S. Zhu, H. Li, Ceramics Int. 2019, 45, 17996-18002.

[32]   S. S. Shinde, C. H. Lee, J.-Y. Jung, N. K. Wagh, S.-H. Kim, D.-H. Kim, C. Lin, S. U.
       Lee, J.-H. Lee, Energy Environ. Sci. 2019, 12, 727-738.

[33]   S. S. Shinde, C. H. Lee, J. Y. Yu, D. H. Kim, S. U. Lee, J. H. Lee, ACS Nano 2018, 12,
       596-608.

[34]   N. Ramaswamy, U. Tylus, Q. Jia, S. Mukerjee, J. Am. Chem. Soc. 2013, 135, 15443-
       15449.

[35]   F. Xiao, G.-L. Xu, C.-J. Sun, M. Xu, W. Wen, Q. Wang, M. Gu, S. Zhu, Y. Li, Z. Wei,
       X. Pan, J. Wang, K. Amine, M. Shao, Nano Energy 2019, 61, 60-68.

[36]   A. Zitolo, V. Goellner, V. Armel, M. T. Sougrati, T. Mineva, L. Stievano, E. Fonda, F.
       Jaouen, Nat. Mater. 2015, 14, 937-942.

[37]   W. Ye, S. Chen, Y. Lin, L. Yang, S. Chen, X. Zheng, Z. Qi, C. Wang, R. Long, M. Chen,
       J. Zhu, P. Gao, L. Song, J. Jiang, Y. Xiong, Chem 2019, 5, 2865-2878.

[38]   H. Lin, L. Yang, X. Jiang, G. Li, T. Zhang, Q. Yao, G. W. Zheng, J. Y. Lee, Energy
       Environ. Sci. 2017, 10, 1476-1486.
                                             26
[39]   H. Lin, S. Zhang, T. Zhang, H. Ye, Q. Yao, G. W. Zheng, J. Y. Lee, Adv. Energy Mater.
       2019, 9.

[40]   S. S. Zhang, J. Power Sources 2013, 231, 153-162.

[41]   Y. Li, J. Fan, J. Zhang, J. Yang, R. Yuan, J. Chang, M. Zheng, Q. Dong, ACS Nano 2017,
       11, 11417-11424.

[42]   X. Zhu, W. Zhao, Y. Song, Q. Li, F. Ding, J. Sun, L. Zhang, Z. Liu, Adv. Energy Mater.
       2018, 8.

[43]   Mahmood, J. Lee, E. K. Jung, M. Shin, D. Jeon, I. Y. Jung, S. M. Choi, H. J.

       Seo, J. M. Bae, S. Y. Sohn, S. D. Park, N. Oh, J. H. Shin, H. J. Baek, J. B.

       Nat. Commun. 2015, 6, 6486.

[44]   G. Zhou, H. Tian, Y. Jin, X. Tao, B. Liu, R . Zhang, Z. W. Seh, D. Zhuo, Y.

       Liu, Sun, J. Zhao, J. C .Zu, D. S. Wu, Q. Zhang, Y. Cui, Proc. Natl. Acad. Sci.

       2017, 114, 840.

[45]   J. F. Kimer, W. Dow, W. R. Scheidt. Inorg. Chem. 1976, 15, 1685-1690.

[46]   Y. Chen, Ji, S. Wang, Y. Dong, J. Chen, W. Li, Z. Shen, R. Zheng, L. Zhuang,

       Z. Wang, Y D. Li, Angew. Chem. Int. Ed. 2017, 129, 7041-7045.

[47]   D. Su, M. Cortie, H. Fan, G. Wang, Adv. Mater. 2017, 29.

[48]   C.-Y. Fan, Y.-P. Zheng, X.-H. Zhang, Y.-H. Shi, S.-Y. Liu, H.-C. Wang, X.-L. Wu, H.-Z.
       Sun, J.-P. Zhang, Adv. Energy Mater. 2018, 8.

[49]   C. Zhang, J. J. Biendicho, T. Zhang, R. Du, J. Li, X. Yang, J. Arbiol, Y. Zhou, J. R.
       Morante, A. Cabot, Adv. Funct. Mater. 2019, 29.

[50]   H. Yuan, H.-J. Peng, B.-Q. Li, J. Xie, L. Kong, M. Zhao, X. Chen, J.-Q. Huang, Q.
       Zhang, Adv. Energy Mater. 2019, 9.

[51] W. Cheng, Lu. Weng, Y. Li, A. Lau, C K. Chan, C. Chan, Aerosol Science and Technology,
     2013, 47, 1118–1124.




                                               27