123456789
101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657
Crystal Structure Modification Enhanced FeNb11O29 Anodesfor Lithium-Ion BatteriesXiaoming Lou,[a] Chunfu Lin,*[a] Qiang Luo,[b, c] Jinbo Zhao,[c] Bin Wang,[d] Jianbao Li,[a]
Qian Shao,[e] Xingkui Guo,[e] Ning Wang,*[a, b] and Zhanhu Guo*[c]
The recently explored FeNb11O29 is an advanced anode material
for lithium-ion batteries, owing to its high specific capacity and
safety. However, it suffers from poor rate capability. To tackle
this issue, a crystal structure modification is employed. Defective
FeNb11O29 (FeNb11O27.9) is fabricated by using a one-step solid-
state reaction method in N2. FeNb11O27.9 has the same ortho-
rhombic shear ReO3 crystal structure (Amma space group) as
FeNb11O29, but a larger unit-cell volume and 3.8 % O2� vacancies
(vs. all O2� ions), which improve the Li+-ion diffusion coefficient
by a factor of 88.3 %. The contained Nb4 + ions with free 4d
electrons significantly increase the electronic conductivity by
three orders of magnitude. Consequently, FeNb11O27.9 shows
improved pseudocapacitive behavior and electrochemical prop-
erties. In comparison with FeNb11O29, FeNb11O27.9 exhibits a
higher reversible capacity of 270 mAh g�1 with a higher first-
cycle coulombic efficiency of 90.6 % at 0.1 C. At 10 C, FeNb11O27.9
still retains a high capacity of 145 mAh g�1 with low capacity
loss of 6.9 % after 200 cycles, in contrast to the values of
99 mAh g�1 and 11.1 % obtained for FeNb11O29.
1. Introduction
Owing to the advanced performances of higher energy output,
longer storage life, lower maintenance and lighter weight than
other secondary batteries, lithium-ion batteries (LIBs) are
considered as very promising power sources for electric
vehicles (EVs).[1] The profitable use of LIBs in EVs requires high
safety performance, energy density, power density and cyclic
stability for LIBs. Graphite is the most widely investigated and
used anode material in consumer electronics due to its low
cost, high capacity (theoretically 372 mAh g�1) and good cyclic
stability. However, it suffers from a limited rate capability due to
its insufficient Li+-ion diffusivity. Moreover, its low working
potential of ~0.1 V (vs. Li/Li+) causes the formation and growth
of lithium dendrites on its surface, resulting in a limited safety
performance.[2] The limited rate capability and safety perform-
ance hinder the applications of the graphite anode material in
EVs. In order to solve these two problems, Li4Ti5O12 has been
developed as an alternative anode material in recent years.[3] It
exhibits a stable spinel-type crystal structure with the Fd�3m
space group, in which two thirds of the cations reside at
octahedral sites and the others at tetrahedral sites. Due to its
high working potential of ~1.57 V, the risk of forming lithium
dendrites on its surface is removed, solving the safety problem.
In addition, its poor rate capability can be improved by various
approaches.[4] However, its intrinsically low theoretical capacity
of only 175 mAh g�1 cannot be increased by any approaches,
leading to its poor energy density. Therefore, it is essential to
further explore new anode materials with similar merits to Li4Ti5
O12 but with much higher capacities for EVs.
As a new intercalating anode material with a high specific
capacity, FeNb11O29 has emerged as an interesting anode
material for LIBs since it was first reported by Pinus et al. in
2014.[5] Based on the Nb4 +/Nb5 +, Nb3 +/Nb4 + and Fe2 +/Fe3 +
redox couples, the twenty-three electron transfer per FeNb11O29
formula unit corresponds to a high theoretical capacity of
400 mAh g�1. This value is 1.29 times higher than that of Li4Ti5
O12 (175 mAh g�1) and even surpasses that of graphite
(372 mAh g�1), thereby solving the problem of low capacity.
FeNb11O29 shows a similar high working potential (~1.6 V) to Li4
Ti5O12, which enables a similarly good safety performance. It
exhibits an orthorhombic shear ReO3 crystal structure (Amma
space group) consisting of corner-sharing and edge-sharing
MO6 (M = Fe, Nb) octahedra (Figure 1a).[5] Therefore, the shear
ReO3 crystal structure purely constructed by octahedra has a
greater degree of openness than the spinel structure by both
octahedra and tetrahedra, inferring a larger Li+-ion diffusion
coefficient of FeNb11O29 than Li4Ti5O12. However, the Li+-ion
[a] X. Lou, Prof. C. Lin, Prof. J. Li, Prof. N. WangState Key Laboratory of Marine Resource Utilization in South China SeaCollege of Materials and Chemical EngineeringHainan University, Haikou 570228, PR ChinaE-mail: [email protected]
[b] Dr. Q. Luo, Prof. N. WangState Key Laboratory of Electronic Thin Film and Integrated DevicesUniversity of Electronic Science and Technology of ChinaChengdu 610054, PR ChinaE-mail: [email protected]
[c] Dr. Q. Luo, J. Zhao, Prof. Z. GuoIntegrated Composites Laboratory (ICL)Department of Chemical and Biomolecular EngineeringUniversity of Tennessee, Knoxville, TN 37996 USAE-mail: [email protected]
[d] B. WangEngineered Multifunctional Composites (EMC) Nanotech. LLCKnoxville, TN 37934 USA
[e] Prof. Q. Shao, Prof. X. GuoCollege of Chemical and Environmental EngineeringShandong University of Science and TechnologyQingdao 266590, PR China
Supporting information for this article is available on the WWW underhttps://doi.org/10.1002/celc.201700816
3171ChemElectroChem 2017, 4, 3171 – 3180 � 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ArticlesDOI: 10.1002/celc.201700816
Wiley VCH Dienstag, 05.12.20171712 / 99586 [S. 3171/3180] 1
123456789
101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657
diffusion coefficient of FeNb11O29 is still insufficient. In addition,
FeNb11O29 suffers from a poor electronic conductivity resulting
from the chemical valences of Fe3 + and Nb5 + in FeNb11O29. The
empty 4d orbitals in the Nb5 + ions indicate that no free
electrons in the Nb5 + ions can participate in the electronic
conduction. For the Fe3 + ions, half of their 3d orbitals are filled
by electrons. According to the Hunds rule, the Fe3 + ions with
half-full 3d orbitals are rather stable, and it is rather difficult for
the Fe3 + ions to lose their 3d electrons. The poor electronic
conductivity together with the insufficient Li+-ion diffusion
coefficient in FeNb11O29 significantly limits its rate capability. In
fact, Pinus et al. reported that FeNb11O29 had a poor rate
capability with a low reversible capacity of 140 mA h g�1 at a
small current rate of 0.4 C, which was only 35 % of its
theoretical capacity.[5] Therefore, it is crucial to modify FeNb11
O29 in order to achieve its practical applications. Since the first
report, however, no subsequent studies on the FeNb11O29 anode
material have been reported so far.
Crystal-structure modification is regarded as a very effective
and facile route for improving the electrochemical properties of
intercalating electrode materials as this method can improve
their conductivities.[6] In this work, through a simple solid-state
reaction in N2, defective FeNb11O29 containing O2� vacancies
and Nb4 + ions has been successfully fabricated for the first
time. The defective FeNb11O29 has been comprehensively
investigated using various techniques and compared with the
pristine FeNb11O29. The experimental results reveal that the
defective FeNb11O29 has superior properties in terms of
increased electronic conductivity (6.45 � 10�6 S cm�1), enhanced
Li+-ion diffusion coefficient (1.61 � 10�11 cm2 s�1 at lithiation),
significant pseudocapacitive contribution (78.1 % at 1.1 mV s�1),
high reversible capacity (270 mAh g�1 at 0.1 C), safe working
potential (~1.65 V), advanced first-cycle Coulombic efficiency
(90.6 %), outstanding rate capability (145 mAh g�1 at 10 C) and
good cyclic stability (93.1 % capacity retention over 200 cycles),
thereby fulfilling the four requirements for the LIBs in EVs.
2. Results and Discussion
Figure 2 depicts the final observed, calculated and error X-ray
diffraction (XRD) patterns for the pristine and defective FeNb11
O29 materials, and their Rietveld-refined data are listed in
Table 1. The sharp diffraction peaks of both the samples
indicate their good crystallinity. All the XRD peaks for the
pristine FeNb11O29 can conform to a orthorhombic shear ReO3
crystal structure with the Amma space group (JCPDS 22-352),
and no information about impurity phases (such as Nb2O5 or
Fe2O3) can be found, indicating the formation of pure FeNb11
Figure 1. Schematic representations of crystal structures of a) FeNb11O29 andb) Ti2Nb10O29.
Figure 2. XRD patterns of FeNb11O29 and FeNb11O27.9.
3172ChemElectroChem 2017, 4, 3171 – 3180 www.chemelectrochem.org � 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Articles
Wiley VCH Dienstag, 05.12.20171712 / 99586 [S. 3172/3180] 1
123456789
101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657
O29. The detailed crystal structure of FeNb11O29 is sketched in
Figure 1a. As can be seen, the structural unit consists of 3 � 4 �
1 ReO3-type octahedral-blocks. These structural units share
corners and edges, guaranteeing the structural stability of
FeNb11O29 during lithiation and delithiation. The cations are
distributed in six different types of 8 f sites, one of which is
occupied by Fe3 + and Nb5 + ions with an atomic ratio of 1 : 1
and others are fully occupied by Nb5 + ions.[5]
It is noteworthy that FeNb11O29 and Ti2Nb10O29 (another
popular anode material) show similar M12O29-type shear ReO3
crystal structures (Figure 1a and 1b).[6b,c,7] Thus, it can be
expected that FeNb11O29 can deliver similarly good electro-
chemical properties to Ti2Nb10O29. However, two differences
exist in their crystal structures. First, Ti2Nb10O29 shows a
monoclinic framework, which has a lower symmetry than the
orthorhombic framework of FeNb11O29. Second, all the cations
in Ti2Nb10O29 (i. e., Ti4 + and Nb5 + ions with an atomic ratio of
1 : 5) are disorderly distributed at the octahedral sites.
For the defective FeNb11O29, its XRD peaks match well with
those of the pristine FeNb11O29, suggesting that the calcination
in N2 does not change the phase structure of FeNb11O29. Two
previous studies confirmed that similar reactions in non-
oxidizing atmospheres (Ar or N2) led to the formation of oxygen
vacancies and lower-valence cations.[6a,b] Therefore, FeNb11O29�x
reasonably represents the defective FeNb11O29 compound
calcined in N2, in which O2� vacancies and lower-valence cations
exist. The x value in FeNb11O29�x was Rietveld-refined to be 1.1
(Table 1). Consequently, the resultant FeNb11O27.9 contains 3.8 %
O2� vacancies (vs. all O2� ions). In addition, FeNb11O27.9 has a
larger unit-cell volume than FeNb11O29. This increase can be
attributed to the existence of the O2� vacancies and the lower-
valence cations with larger ion sizes.[8] The O2� vacancies and
the larger unit-cell volume can respectively render larger
number and size of Li+-ion transport paths in FeNb11O27.9,
benefiting its Li+-ion transport.
To further understand the crystal structures of FeNb11O29
and FeNb11O27.9, their high-resolution transmission electron
microscopy (HRTEM) images were analyzed. The two sets of
lattice fringes with the d-spacing of 0.251 and 0.212 nm
(Figure 3a) can respectively correspond to the (2 1 6) and (9 1
5) crystallographic planes of FeNb11O29, and those with 0.289
and 0.373 nm (Figure 3b) match with the (10 0 0) and (111)
planes of FeNb11O27.9. This finding is further supported by the
selected-area electron diffraction (SAED) patterns of FeNb11O29
(Figure 3c) and FeNb11O27.9 (Figure 3d), which are in good
agreement with orthorhombic shear ReO3 crystal structure with
the Amma space group. The very sharp electron diffraction dots
in the SAED patterns together with the clear lattice fringes in
the HRTEM images further confirm the highly crystalline nature
of the two samples.
To clarify the valences of Fe and Nb elements in FeNb11O29
and FeNb11O27.9, their X-ray photoelectron spectrometry (XPS)
spectra are recorded in Figure 4a and 4b. The binding energies
of FeNb11O29 at 711.6 eV (Fe-2p3/2), 725.1 eV (Fe-2p1/2), 207.4 eV
(Nb-3d5/2) and 210.2 eV (Nb-3d3/2) respectively indicate the
presence of Fe3 + and Nb5 + in FeNb11O29.[9] After the calcination
in N2, the peaks of Fe-2p3/2 and Fe-2p1/2 in FeNb11O27.9 negligibly
shift, while those of Nb-3d5/2 and Nb-3d3/2 respectively red-shift
to 206.9 and 209.6 eV, demonstrating that the Nb5 + ions rather
than the Fe3 + ions are reduced and that the resultant Nb4 + ions
are the aforementioned lower-valence cations.
Figure 5a through 5d display the field-emission scanning
electron microscopy (FESEM) images of FeNb11O29 and FeNb11
O27.9. Both samples are observed to exhibit similar particle
morphologies with a platelet particle shape and similar particle
sizes ranging from 0.2 to 20 mm. The Brunauer�Emmett�Teller
(BET) specific surface area of FeNb11O27.9 (0.181 m2 g�1) is almost
equal to that of FeNb11O29 (0.176 m2 g�1), further confirming
similar particle sizes of these two samples. These results
suggest that the calcination in N2 has little influences on the
particle morphology or size. As a result, changes in the particle
morphology and size are negligibly responsible for the differ-
ence of the following rate capabilities.
Table 1. Results of crystal analyses by Rietveld refinements in FeNb11O29 and FeNb11O27.9.
Sample a [A] b [A] c [A] V [A3] f1[a] f2
[b] Rwp[c] Rp
[d]c2 [e]
FeNb11O29 28.70490(49) 3.82569(7) 20.62376(42) 2264.822(60) 1(�) 1(�) 0.0764 0.0558 7.481FeNb11O27.9 28.69993(53) 3.82695(7) 20.62579(39) 2265.395(91) 0.459 1(�) 0.0671 0.0496 5.938
[a]: Occupancy of O2� in O4 sites. [b]: Occupancy of O2� in other sites. [c]: Weighted profile residual. [d]: Profile residual. [e]: Goodness of fit.
Figure 3. HRTEM images and SAED patterns of a,c) FeNb11O29 and b,d)FeNb11O27.9.
3173ChemElectroChem 2017, 4, 3171 – 3180 www.chemelectrochem.org � 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Articles
Wiley VCH Dienstag, 05.12.20171712 / 99586 [S. 3173/3180] 1
123456789
101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657
The tested electronic conductivity of FeNb11O29 is as low as
4.89 � 10�9 S cm�1. After the calcination in N2, the electronic
conductivity of FeNb11O27.9 was significantly increased by three
orders of magnitude, reaching a large value of 6.45 �
10�6 S cm�1. FeNb11O27.9 contains a considerable amount of Nb4 +
ions with free 4d electrons, which are capable of transporting
the FeNb11O27.9 crystals and thus significantly increase the
electronic conductivity. Previous reports showed that the
electronic conductivities of doped niobium oxides were lower
than 10�7 S cm�1.[10] Clearly, the crystal-structure modification
based on the creation of O2� vacancies and lower-valence
cations is more effective to increase the electronic conductivity
of niobium oxides than the doping methodology.
To understand the electrochemical mechanisms, CV tests
were conducted on the FeNb11O29/Li and FeNb11O27.9/Li cells
within a potential window of 3.0–0.8 V, and their corresponding
CV curves were recorded and compared in Figure 6. The first
four cycles were operated at a scan speed of 0.2 mV s�1, and
the last three cycles at 0.4, 0.7 and 1.1 mV s�1, respectively. At
0.2 mV s�1, the locations of the cathodic CV peaks in the first
cycle are different from those in the following three cycles
(Figure 6a, 6b and 6c) in both the cells. This phenomenon could
be derived from the irreversible lithiation during the first
cycle.[11] The corresponding peak shifts of the FeNb11O27.9/Li cell
are smaller, which probably stems from its smaller degree of
irreversible lithiation during the first cycle indicated by its
higher first-cycle Coulombic efficiency (mentioned below).
As shown in Figure 6a, the FeNb11O29/Li cell exhibits four
pairs of cathodic/anodic CV peaks. From the second cycle, a
pair centered at 1.05/1.30 V can be ascribed to the Nb3 +/Nb4 +
redox couple. Two pairs at 1.54/1.76 V and 2.01/2.03 V can be
assigned to the Nb4 +/Nb5 + redox couple.[12] Another pair at
2.41/2.44 V can be related to the Fe2 +/Fe3 + redox couple,[13]
indicating that the Fe2 +/Fe3 + redox couple is also active in 3.0–
0.8 V. In Pinus et al.’s study, the active Fe2 +/Fe3 + redox couple
was missed since the upper potential-limit in the CV tests was
only 2.0 V.[5] Therefore, not only the Nb3 +/Nb4 + and Nb4 +/Nb5 +
redox couples but also the Fe2 +/Fe3 + redox couple contribute
to the large capacity of the FeNb11O29/Li cell in the following
galvanostatic discharge�charge measurements.
The intermediate potential of two sharp peaks at 1.54/1.76 V
can correspond to the average working potential of FeNb11O29.
Thus, its working potential was calculated to be ~1.65 V, which is
slightly lower than that of Ti2Nb10O29 (~1.71 V)[6b] and slightly higher
than that of Li4Ti5O12 (~1.57 V).[4a] This reasonably high working
potential of FeNb11O29 avoids the lithium-dendrite formation and
the electrolyte decomposition, guaranteeing its high safety
performance.
For the FeNb11O27.9/Li cell, besides the same four pairs of
cathodic/anodic peaks at similar positions, no other peaks can
be found (Figure 6a). Thus, the crystal modification through the
calcination in N2 does not affect the redox behavior of FeNb11
O29 in 3.0–0.8 V. However, in comparison with the FeNb11O29/Li
cell, the FeNb11O27.9/Li cell exhibits larger peak intensities and
smaller polarization at all the four scan speeds (Figure 6a, 6b
and 6c). Hence, the FeNb11O27.9 sample possesses better electro-
chemical kinetics, which are favorable for achieving better rate
capability. In addition, the CV curves of the FeNb11O27.9/Li cell
after the first cycle almost completely overlap (Figure 6c), in
sharp contrast to that of the FeNb11O29/Li cell (Figure 6b).
Therefore, the crystal-structure modification also improves the
cyclic stability.
The peak current Ip of the sharp cathodic/anodic peaks is
proportional to the square root of the scan speed v0.5 (Fig-
ure 6d). Consequently, the apparent Li+-ion diffusion coeffi-
Figure 4. XPS spectra of a) Fe-2p and b) Nb-3d in FeNb11O29 and FeNb11O27.9.
Figure 5. FESEM images of a,c) FeNb11O29 and b,d) FeNb11O27.9.
3174ChemElectroChem 2017, 4, 3171 – 3180 www.chemelectrochem.org � 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Articles
Wiley VCH Dienstag, 05.12.20171712 / 99586 [S. 3174/3180] 1
123456789
101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657
cients D of FeNb11O29 and FeNb11O27.9 can be obtained by using
the Randles-Sevcik equation [Eq. (1)]:[14]
Ip ¼ 2:69� 105 � n1:5SCD0:5v0:5 ð1Þ
where n, S and C respectively refer to the charge-transfer
number, the contact surface between FeNb11O29/FeNb11O27.9
and electrolyte, and the Li+-ion concentration in solid. The
obtained Li+-ion diffusion coefficient of FeNb11O29 for lithiation
is 8.55 � 10�12 cm2 s�1, which is smaller than that for delithiation
(3.39 � 10�11 cm2 s�1), revealing that lithiation is the rate-limit
step in the whole electrochemical reaction of FeNb11O29.
Through the crystal-structure modification, the Li+-ion diffusion
coefficient of FeNb11O27.9 in the rate-limit step is increased by a
factor of 88.3 %, reaching a value of 1.61 � 10�11 cm2 s�1
(lithiation). This value is even five orders of magnitude greater
than that of the popular Li4Ti5O12 (1.81 � 10�16 cm2 s�1).[10a] The
shear ReO3 crystal structure of FeNb11O27.9 purely built by MO6
(M=Fe, Nb) octahedra has a greater degree of openness than
the spinel crystal structure of Li4Ti5O12 with LiO4 tetrahedra and
M’O6 (M’=Li, Ti) octahedra. Furthermore, the larger unit-cell
volume of FeNb11O27.9 after the crystal-structure modification
can result in wider Li+-ion transport paths and the 3.8 % O2�
vacancies can provide more Li+-ion transport paths. These
crystal-structure advantages of FeNb11O27.9 can significantly
facilitate the Li+-ion transport, leading to its large Li+-ion
diffusion coefficient.
To further investigate the Li+-ion storage mechanisms of
FeNb11O29 and FeNb11O27.9, their pseudocapacitive behaviors are
quantitatively analyzed. According to a previous study,[15] the
peak current Ip and the scan speed v in Figure 6b and 6c can
follow the following Equation (2):
log ðIpÞ ¼ alog ðvÞ þ log ðbÞ ð2Þ
where a and b are adjustable parameters. When a is equal to
0.5, the electrochemical system is controlled by the ionic
diffusion, while an a-value of 1 indicates that the electro-
chemical system is controlled by the pseudocapacitive effect.
Figure 7a displays the plots of log(Ip) vs. log(v) at the four
cathodic/anodic peak pairs, from which the a-values for the
cathodic/anodic reactions in the FeNb11O29 and FeNb11O27.9
samples are determined to be 0.71/0.90 and 0.80/0.94,
respectively, indicating that these electrochemical systems are
controlled by both the Li+-ion diffusion and the pseudocapaci-
tive effect and that FeNb11O27.9 shows a more obvious
pseudocapacitive behavior than FeNb11O29. The detailed pseu-
Figure 6. CV curves of FeNb11O29/Li and FeNb11O27.9/Li cells at a) 0.2 mV s�1 and b,c) various scan speeds. d) Relationship between peak current of cathodic/anodic reaction Ip and square root of scan speed v0.5.
3175ChemElectroChem 2017, 4, 3171 – 3180 www.chemelectrochem.org � 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Articles
Wiley VCH Dienstag, 05.12.20171712 / 99586 [S. 3175/3180] 1
123456789
101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657
docapacitive contributions at various scan speeds can be
calculated through the Equation (3):[15]
I ¼ cv0:5 þ dv ð3Þ
where c and d are adjustable parameters, and cv0.5 and dv
represent the diffusion-controlled and pseudocapacitive contri-
butions, respectively. The obtained pseudocapacitive contribu-
tions in the FeNb11O29 sample are 43.1, 53.5, 64.1 and 72.0 % at
0.2, 0.4, 0.7 and 1.1 mV s�1, respectively (Figure 7b), while those
in the FeNb11O27.9 sample are increased to 57.6, 65.9, 73.1 and
78.1 %. For clear understanding, their detailed pseudocapacitive
contributions are illustrated in their CV curves at 1.1 mV s�1
(Figure 7c and 7d). Therefore, it is confirmed that FeNb11O27.9
possesses more significant pseudocapacitive effects at all the
scan speeds, probably due to its improved electronic/ionic
conductivity, which can benefit its rate capability and specific
capacity.[16]
Figure 8a plots the discharge�charge curves of the FeNb11
O29/Li and FeNb11O27.9/Li cells at various current rates (C-rates)
between 0.1 and 10 C in 3.0–0.8 V. Although these two samples
show differences in the reversible capacities, their
discharge�charge curves are similar in shape, further verifying
that the crystal-structure modification in this work has no
obvious influences on the redox behavior of FeNb11O29. Each
discharge/charge curve at 0.1 C contains a long plateau at
~1.7 V and a short one at ~2.4 V, which correspond to the Nb4 +
/Nb5 + and Fe2 +/Fe3 + redox couples, respectively. These
discharge�charge curves match with the CV curves in Figure 6a.
In particular, the discharge/charge plateau at ~1.7 V matches
well with the sharp cathodic/anodic peak.
At 0.1 C, the first-cycle discharge capacity of the FeNb11O29
sample is 304 mAh g�1, while a slightly decreased capacity
298 mAh g�1 for the FeNb11O27.9 sample. The first-cycle dis-
charge capacity comes from the reduction of Fe3 + to Fe2 +, Nb5 +
to Nb4 + and Nb4 + to Nb3 +. Since some of the Nb5 + ions were
reduced to Nb4 + ions after the calcination in N2, the FeNb11O27.9
sample lost a bit first-cycle discharge capacity. However, the
FeNb11O27.9 sample shows a higher first-cycle Coulombic
efficiency (90.6 %) than the FeNb11O29 sample (87.5 %) probably
due to the improved electronic/ionic conductivity after the
crystal-structure modification. As a result, the FeNb11O27.9
sample delivers a higher first-cycle charge capacity
(270 mAh g�1) than the FeNb11O29 sample (266 mAh g�1). It is
noteworthy that this reversible capacity of the FeNb11O29/Li cell
is obviously higher than that Pinus et al. reported
Figure 7. a) Plots of log(Ip) vs. log(v) and b) percentages of pseudocapacitive contributions of FeNb11O29/Li and FeNb11O27.9/Li cells. CV curves with detailedpseudocapacitive contributions of c) FeNb11O29/Li and d) FeNb11O27.9/Li cells at 1.1 mV s�1.
3176ChemElectroChem 2017, 4, 3171 – 3180 www.chemelectrochem.org � 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Articles
Wiley VCH Dienstag, 05.12.20171712 / 99586 [S. 3176/3180] 1
123456789
101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657
(185 mAh g�1),[5] which may be ascribed to the better material
and cell fabrication processes in this work. The capacities of the
FeNb11O29 and FeNb11O27.9 samples at 0.1 C in this work are
comparable with those of the previously reported Ti2Nb10O29
materials.[6c,7a,c,d,e]
Figure 8a reveals the rate capabilities of the FeNb11O29/Li
and FeNb11O27.9/Li cells. With increasing the C-rate, the
capacities of these two samples monotonically decrease, but
the FeNb11O27.9 sample exhibits much less capacity loss than the
FeNb11O29 sample. When the FeNb11O29 sample provides a
capacity of 142 mAh g�1 at 5 C, the FeNb11O27.9 sample offers
32 mAh g�1 more, reaching 174 mAh g�1. At 10 C, the FeNb11
O27.9 sample still retains a high capacity of 145 mAh g�1, which
is 47.9 % higher than that of the FeNb11O29 sample (98 mAh g�1).
For clearer observation, the relative capacities of the two
samples as a function of the C-rate are revealed in Figure 8b, in
which the capacities determined at 0.1 C are taken as standard.
As can be seen, the relative capacity of the FeNb11O27.9 sample
at 10 C (53.7 %) is even higher than that of the FeNb11O29
sample at 5 C (53.3 %). Therefore, the FeNb11O27.9 sample has a
significantly better rate capability than the FeNb11O29 sample.
After the crystal-structure modification, the Li+-ion diffusion
coefficient and electronic conductivity of FeNb11O27.9 are
increased by a factor of 88.3 % and three orders of magnitude,
respectively, and more significant pseudocapacitive effects in
both slow and fast electrochemical reactions are achieved.
Together, these factors work to promote the rate capability. It is
noteworthy that the rate capabilities of the FeNb11O29 and
FeNb11O27.9 samples outperform those of the micron-sized Ti2
Nb10O29 materials.[6b,7b]
Figure 8c illustrates the cyclic stability of the FeNb11O29/Li
and FeNb11O27.9/Li cells at a large C-rate of 10 C. As can be seen,
the FeNb11O27.9 sample shows an advanced cyclic stability with
a largely retained capacity of 135 mAh g�1 after 200 cycles,
corresponding to a small capacity loss of 0.035 % per cycle. In
contrast, larger losses of 0.056 % and 0.235 % were obtained for
the FeNb11O29 sample and micron-sized Ti2Nb10O29 material,
respectively.[6b,7b] This comparison demonstrates that the crystal-
structure modification in this work can also improve the cyclic
stability of FeNb11O29.
Figure 9 shows the ex-situ XRD patterns of various FeNb11
O27.9 electrodes with different states of charging. During the
electrochemical reaction, no XRD peaks for the impurity phases
(such as Fe2O3, Nb2O5 or FeNbO4) were detected although
several systematic variations in the peak intensities and
positions were observed. Obviously, FeNb11O27.9 is an intercalat-
Figure 8. a) First-cycle discharge�charge curves at 0.1 C (i), second-cycle discharge�charge curves at 0.1 (ii), 0.5 (iii), 1 (iv), 2 (v), 5 (vi), and 10 C (vii); b)percentage capacities at 0.1–10 C; and c) cyclic stability at 10 C of FeNb11O29/Li and FeNb11O27.9/Li cells. Identical discharge�charge rates were used.
3177ChemElectroChem 2017, 4, 3171 – 3180 www.chemelectrochem.org � 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Articles
Wiley VCH Dienstag, 05.12.20171712 / 99586 [S. 3177/3180] 1
123456789
101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657
ing anode material, the same as FeNb11O29 (Figure S1). This
desirable characteristic also contributes to the advanced cyclic
stability of FeNb11O27.9.
Electrochemical impedance spectroscopy (EIS) tests were
performed to further investigate the beneficial effects of the
crystal-structure modification based on the production of O2�
vacancies and lower-valence cations. The resultant Nyquist
plots of the FeNb11O29/Li and FeNb11O27.9/Li cells are depicted in
Figure 10. As can be seen, each plot consists of two depressed
semicircles and one slope. According to a previous study,[17] the
depressed semicircle in the high-frequency region is associated
with the reactions involving the electron transfer and Li+-ion
adsorption/desolvation, which are denoted as R1 and CPE1 in
the equivalent circuit (the inset of Figure 10). The other
depressed semicircle in the medium-frequency region can be
ascribed to the Li+-ion insertion at the particle surfaces (R2 and
CPE2). The slope in the low-frequency region can correspond to
the Warburg resistance (W), reflecting the Li+-ion diffusion
within the solid matrix. Rb in the equivalent circuit represents
the Ohmic resistance of the cell. The fitted R1 and R2 values for
the FeNb11O29/Li cell are 148 and 317 W, respectively. In
contrast, those for the FeNb11O27.9/Li cell are significantly
decreased to 67 and 117 W, revealing its much faster electron-
transfer, Li+-ion adsorption/desolvation and Li+-ion insertion at
particle surfaces. This EIS result is in good agreement with the
smaller polarization (Figure 6a, 6b and 6c) and better rate
capability (Figure 8a and 8b) of the FeNb11O27.9 sample.
3. Conclusions
In this study, a crystal-structure modification has been carried
out to modify FeNb11O29, and the resultant FeNb11O27.9 with a
defective crystal structure has been successfully fabricated by a
one-step solid-state reaction in N2. The crystal-structure mod-
ification of FeNb11O29 does not obviously change the ortho-
rhombic shear ReO3 crystal structure (Amma space group), the
particles size or the redox behavior, but enlarges the unit-cell
volume, produces the O2� vacancies and the Nb4 + ions, and
improves the electrical and electrochemical properties. As a
result of the enlarged unit-cell volume and the O2� vacancies,
the Li+-ion diffusion coefficient of FeNb11O27.9 is increased by a
factor of 88.3 %. Its electronic conductivity is significantly
increased by three orders of magnitude due to the free 4d
electrons in the Nb4 + ions. These improved electrical properties
in FeNb11O27.9 lead to its improved pseudocapacitive behavior.
Consequently, outstanding electrochemical properties of FeNb11
O27.9 have been achieved. At 0.1 C, FeNb11O27.9 delivers a high
first-cycle reversible capacity of 270 mAh g�1 with a safe work-
ing potential of ~1.65 V and a high Coulombic efficiency of
90.6 %. At 10 C, FeNb11O27.9 still retains a high capacity of
145 mAh g�1 with high retention of 93.1 % after 200 cycles,
whereas the corresponding values obtained for FeNb11O29 are
only 99 mAh g�1 and 88.9 %. Clearly, FeNb11O27.9 exhibits similar
merits to Li4Ti5O12 but has a remarkably higher capacity, thereby
fulfilling the four key requirements of high safety performance,
power density, energy density and cyclic stability for EVs.
Therefore, FeNb11O27.9 can provide potential applications in
high-performance LIBs for EVs.
Experimental Section
The defective FeNb11O29 compound was prepared through a one-step solid-state reaction by using Fe2O3 (Aladdin, 99.0 %) and Nb2O5
(Sinopharm, 99.9 %) powders as raw materials. Stoichiometricamounts of these chemicals (Fe2O3 : Nb2O5 = 1 : 11) were mixed andthen ball-milled for 1 h by using a high-energy ball-milling machine(SPEX 8000 M, USA). The uniform mixture was calcined in a high-temperature tube furnace at 1250 8C for 4 h in N2. For comparison,the pristine FeNb11O29 compound was also prepared by the sameprocess but calcined in air.
To prepare the specimens for electronic conductivity tests, theabove ball-milled powders were pelletized at a pressure of 20 MPa.The obtained pellets were first calcined at 900 8C for 5 h, and thenat 1250 8C for 4 h in air (for the pristine FeNb11O29) or N2 (for thedefective FeNb11O29). The average diameter and thickness of the
Figure 9. Ex situ XRD patterns of FeNb11O27.9 electrodes after i) as-fabricated,ii) discharged to 0.8 V in 1st cycle, and charged to 3.0 V in the iii) 1st and iv)10th cycles at 0.1 C.
Figure 10. Nyquist plots of FeNb11O29/Li and FeNb11O27.9/Li cells; the insetexhibits the equivalent circuit to fit the plots, in which R, CPE and Wrepresent resistance, constant phase element and Warburg resistance,respectively.
3178ChemElectroChem 2017, 4, 3171 – 3180 www.chemelectrochem.org � 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Articles
Wiley VCH Dienstag, 05.12.20171712 / 99586 [S. 3178/3180] 1
123456789
101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657
calcined pellets were ~0.92 and ~0.16 cm, respectively. Finally, theirtwo sides were uniformly coated by gold films, forming symmetricion-blocking cells.
Detailed crystal structures of the as-calcined products wereanalyzed by powder X-ray diffraction (XRD) and Rietveld refine-ments using an X-ray diffractometer (Bruker D8 Advance, Germany,Cu-Ka radiation source l= 1.5406 A) and the GSAS software,respectively.[18] The high-quality XRD data were recorded in anangle interval 158–1308 using a step width of 0.038. During therefinements, the following instrumental/structural parameters weresuccessively refined: background parameters, zero-shift, latticeparameters, profile parameters, atomic fractional coordinates,atomic isotropic temperature factors and atomic occupancies. Allthe isotropic temperature factors were fixed to be the same.Particle sizes, microstructures and morphologies were investigatedby using a field-emission scanning electron microscopy (FESEM,Hitachi S-4800, Japan) and a high-resolution transmission electronmicroscopy (HRTEM, FEI Tecnai G2 F20 S-TWIN, USA).Brunauer�Emmett�Teller (BET) specific surface areas were obtainedfrom standard N2 adsorption�desorption isotherms measured by asurface area analyzer (ASAP 2020, USA). Chemical valences wereanalyzed by an X-ray photoelectron spectrometry (XPS, ThermoEscalab 250Xi, USA). Electronic conductivities were determined by atwo-probe direct current method. Electronic conductivity measure-ments were carried out on the ion-blocking cells by using anelectrochemical workstation (CHI660, China) under a constantvoltage of 50 mV until the currents became stabilized.
Electrochemical properties were evaluated by using CR2016-typecoin cells assembled in a dry argon-filled glove box. 65 wt% activematerials (pristine or defective FeNb11O29), 10 wt% polyvinylidenefluoride (PVDF) and 25 wt% conductive carbon (Super P�) werethoroughly mixed in N-methylpyrrolidone (NMP), forming a homo-geneous slurry. The slurry was uniformly cast onto Cu foils, whichwas then vacuum-dried at 120 8C overnight and finally pressed byan automatic roller to form the working electrodes. The loadingdensity of the active materials was ~1.4 mg cm�2. Microporouspolypropylene films (Celgard 2325) were employed as separators. Lifoils acted as reference and counter electrodes. 1 mol L�1 solutionof LiPF6 (DAN VEC) in ethylene carbonate (EC)/dimethyl carbonate(DMC)/diethylene carbonate (DEC) in a volume ratio of 1 : 1 : 1 wasused as electrolyte.
Galvanostatic discharge�charge tests were performed at variouscurrent rates (C-rates) by using an eight-channel battery tester(Neware CT-3008, China) in a potential range of 3.0–0.8 V. The C-rate of 1 C equals 400 mA g�1 (corresponding to the theoreticalcapacity of FeNb11O29). Cyclic voltammetry (CV) and electrochemicalimpedance spectroscopy (EIS) measurements were carried out byusing the aforementioned electrochemical workstation. The set CVscan speeds were in a range of 0.2–1.1 mV s�1. The impedancespectra were recorded in a frequency range of 105–10�2 Hz.
Acknowledgement
This study is supported by the National Natural Science
Foundation of China (51502064), Provincial Natural Science
Foundation of Hainan (20165184), and Technology Foundation
for Selected Overseas Chinese Scholar.
Conflict of Interest
The authors declare no conflict of interest.
Keywords: FeNb11O29 · crystal structure modification · defects ·electrical properties · lithium-ion batteries
[1] a) M. Armand, J. M. Tarascon, Nature 2008, 451, 652–657; b) B. Dunn, H.Kamath, J. M. Tarascon, Science 2011, 334, 928–935; c) J. B. Good-enough, Y. Kim, Chem. Mater. 2010, 22, 587–603; d) J. Liu, J. G. Zhang,Z. G. Yang, J. P. Lemmon, C. Imhoff, G. L. Graff, L. Y. Li, J. Z. Hu, C. M.Wang, J. Xiao, G. Xia, V. V. Viswanathan, S. Baskaran, V. Sprenkle, X. L. Li,Y. Y. Shao, B. Schwenzer, Adv. Funct. Mater. 2013, 23, 929–946; e) L. M.Suo, Y. S. Hu, H. Li, M. Armand, L. Q. Chen, Nat. Commun. 2013, 4, 1481;f) M. T. McDowell, S. W. Lee, W. D. Nix, Y. Cui, Adv. Mater. 2013, 25, 4966–4984; g) J. Huang, Y. Cao, Q. Shao, X. Peng, Z. Guo, Ind. Eng. Chem.Res., 2017, DOI: 10.1021/acs.iecr.7b02835.
[2] a) S. S. Zheng, J. Power Sources 2006, 161, 1385–1391; b) K. Sun, P. Xie,Z. Wang, T. Su, Q. Shao, J. Ryu, X. Zhang, J. Guo, A. Shankar, J. Li, R. Fan,D. Cao, Z. Guo, Polymer 2017, 125, 50–57; c) C. Wang, Y. Wu, Y. Li, Q.Shao, X. Yan, C. Han, Z. Wang, Z. Liu, Z. Guo, Polym. Adv. Technol. 2017,DOI: 10.1002/pat.4105.
[3] a) B. Zhao, R. Ran, M. L. Liu, Z. P. Shao, Mater. Sci. Eng., R. 2015, 98, 1–71;b) Y. Li, X. Wu, J. Song, J. Li, Q. Shao, N. Cao, N. Lu, Z. Guo, Polymer2017, 124, 41–47.
[4] a) C. F. Lin, X. Y. Fan, Y. L. Xin, F. Q. Cheng, M. O. Lai, H. H. Zhou, L. Lu, J.Mater. Chem. A 2014, 2, 9982–9993; b) C. F. Lin, B. Ding, Y. L. Xin, F. Q.Cheng, M. O. Lai, L. Lu, H. H. Zhou, J. Power Sources 2014, 248, 1034–1041; c) C. F. Lin, M. O. Lai, L. Lu, H. H. Zhou, Y. L. Xin, J. Power Sources2013, 244, 272–279; d) Z. J. Yao, X. H. Xia, Y. Zhong, Y. D. Wang, B. W.Zhang, D. Xie, X. L. Wang, J. P. Tu, Y. Z. Huang, J. Mater. Chem. A 2017, 5,8916–8921; e) Y. J. Cai, Y. D. Huang, W. Jia, Y. Zhang, X. C. Wang, Y. Guo,D. Z. Jia, W. K. Pang, Z. P. Guo, L. S. Wang, J. Mater. Chem. A 2016, 4,17782–17790; f) Z. Y. Sun, L. Zhang, F. Dang, Y. Liu, Z. Y. Fei, Q. Shao, H.Lin, J. Guo, L. C. Xiang, N. Yerra, Z. H. Guo, CrystEngComm. 2017, 19,3288–3298; g) C. Wang, S. Wang, L. K. Tang, Y. B. He, L. Gan, J. Li, H. D.Du, B. H. Li, Z. Q. Lin, F. Y. Kang, Nano Energy 2016, 21, 133–144; h) T.Liu, K. Yu, L. Gao, H. Chen, N. Wang, L. Hao, T. Li, H. He, Z. Guo, J. Mater.Chem. A 2017, 5, 17848–17855; i) L. Zhang, W. Yu, C. Han, J. Guo, Q.Zhang, H. Xie, Q. Shao, Z. Sun, Z. Guo, J. Electrochem. Soc. 2017, 164,H651-H656.
[5] I. Pinus, M. Catti, R. Ruffo, M. M. Salamone, C. M. Mari, Chem. Mater.2014, 26, 2203–2209.
[6] a) X. M. Chen, X. F. Guan, L. P. Li, G. S. Li, J. Power Sources 2012, 210,297–302; b) C. F. Lin, S. Yu, H. Zhao, S. Q. Wu, G. Z. Wang, L. Yu, Y. F. Li,Z. Z. Zhu, J. B. Li, S. W. Lin, Sci. Rep. 2015, 5, 17836; c) T. Takashima, T.Tojo, R. Inada, Y. Sakurai, J. Power Sources 2015, 276, 113–119.
[7] a) Q. S. Cheng, J. W. Liang, Y. C. Zhu, L. L. Si, C. Guo, Y. T. Qian, J. Mater.Chem. A, 2014, 2, 17258–17262; b) C. F. Lin, G. Z. Wang, S. W. Lin, J. B. Li,L. Lu, Chem. Commun. 2015, 51, 8970–8973; c) X. Y. Wu, J. Miao, W. Z.Han, Y. S. Hu, D. F. Chen, J. Lee, J. Kim, L. Q. Chen, Electrochem. Commun.2012, 25, 39–42; d) W. L. Wang, B. Oh, J. Park, H. Ki, J. Jang, G. Lee, H.Gu, M. Ham, J. Power Sources 2015, 300, 272–278; e) G. Y. Liu, B. Jin,R. X. Zhang, K. Y. Bao, H. Q. Xie, J. L. Guo, M. Wei, Q. Jiang, Int. J.Hydrogen Energy 2016, 41, 14807–14812.
[8] R. D. Shannon, Acta Crystallogr., Sect. A: Found. Crystallogr. 1976, 32,751–767.
[9] a) H. S. Li, L. F. Shen, G. Pang, S. Fang, H. F. Luo, K. Yang, X. G. Zhang,Nanoscale 2015, 7, 619–624; b) M. S. Balogun, Z. P. Wu, Y. Luo, W. T. Qiu,X. L. Fan, B. Long, M. Huang, P. Liu, Y. X. Tong, J. Power Sources 2016,308, 7–17.
[10] a) C. F. Lin, S. Yu, S. Q. Wu, S. W. Lin, Z. Z. Zhu, J. B. Li, L. Lu, J. Mater.Chem. A 2015, 3, 8627–8635; b) C. Yang, C. F. Lin, S. W. Lin, Y. J. Chen,J. B. Li, J. Power Sources 2016, 328, 336–344.
[11] X. Lu, Z. L. Jian, Z. Fang, L. Gu, Y. S. Hu, W. Chen, Z. X. Wang, L. Q. Chen,Energy Environ. Sci. 2011, 4, 2638–2644.
[12] X. F. Wang, G. Z. Shen, Nano Energy 2015, 15, 104–115.[13] A. D. Robertson, H. Tukamoto, J. T. S. Irvine, J. Electrochem. Soc. 1999,
146, 3958–3962.[14] a) J. Bard, L. R. Faulkner, Electrochemical Methods: Fundamentals and
Applications, second ed., Wiley, New York, 2001; b) Y. Ma, L. Lyu, Y. Guo,
3179ChemElectroChem 2017, 4, 3171 – 3180 www.chemelectrochem.org � 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Articles
Wiley VCH Dienstag, 05.12.20171712 / 99586 [S. 3179/3180] 1
123456789
101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657
Y. Fu, Q. Shao, T. Wu, S. Guo, K. Sun, X. Guo, E. K Wujcik, Z. Guo, Polymer2017, DOI: 10.1016/j.polymer.2017.09.009.
[15] S. F. Lou, X. Q. Cheng, Y. Zhao, A. Lushington, J. L. Gao, Q. Li, P. J. Zuo,B. Q. Wang, Y. Z. Gao, Y. L. Ma, C. Y. Du, G. P. Yin, X. L. Sun, Nano Energy2017, 34, 15–25.
[16] V. Raju, J. Rains, C. Gates, W. Luo, X. Wang, W. F. Stickle, G. D. Stucky, X.Ji, Nano Lett. 2014, 14, 4119–4124.
[17] a) M. Nakayama, H. Ikuta, Y. Uchimoto, M. Wakihara, J. Phys. Chem. B2003, 107, 10603–10607; b) K. Zhang, G. Li, L. Feng, N. Wang, J. Guo, K.Sun, K. Yu, J. Zeng, T. Li, Z. Guo, M. Wang, J. Mater. Chem. C 2017, DOI:10.1039/C7TC02948 A.
[18] a) A. C. Larson, R. B. Von Dreele, General Structure Analysis System(GSAS), Los Alamos National Laboratory Report LAUR 86-748, L. A. N.L,Los Alamos, 1994; b) B. H. Toby, J. Appl. Crystallogr. 2001, 34, 210–213.
Manuscript received: August 5, 2017Version of record online: September 20, 2017
3180ChemElectroChem 2017, 4, 3171 – 3180 www.chemelectrochem.org � 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Articles
Wiley VCH Dienstag, 05.12.20171712 / 99586 [S. 3180/3180] 1
Top Related