Evolution of The Electrochemical Interface in Sodium Ion Batteries With Ether Electrolytes

Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Corrected: Author Correction

ARTICLE
https://doi.org/10.1038/s41467-019-08506-5 OPEN

Evolution of the electrochemical interface


in sodium ion batteries with ether electrolytes
Kaikai Li1,2,3, Jun Zhang1, Dongmei Lin3, Da-Wei Wang 4, Baohua Li5, Wei Lv5, Sheng Sun6, Yan-Bing He5,
Feiyu Kang1,5, Quan-Hong Yang1,7, Limin Zhou2,3 & Tong-Yi Zhang6
1234567890():,;

Ether based electrolytes have surfaced as alternatives to conventional carbonates allowing


for enhanced electrochemical performance of sodium-ion batteries; however, the primary
source of the improvement remains poorly understood. Here we show that coupling titanium
dioxide and other anode materials with diglyme does enable higher efficiency and reversible
capacity than those for the combination involving ester electrolytes. Importantly, the elec-
trolyte dependent performance is revealed to be the result of the different structural evolution
induced by a varied sodiation depth. A suit of characterizations show that the energy barrier
to charge transfer at the interface between electrolyte and electrode is the factor that
dominates the interfacial electrochemical characteristics and therefore the energy storage
properties. Our study proposes a reliable parameter to assess the intricate sodiation
dynamics in sodium-ion batteries and could guide the design of aprotic electrolytes for next
generation rechargeable batteries.

1 Shenzhen Environmental Science and New Energy Technology Engineering Laboratory, Tsinghua-Berkeley Shenzhen Institute (TBSI), Tsinghua University,
Shenzhen 518055, China. 2 Interdisciplinary Division of Aeronautical and Aviation Engineering, The Hong Kong Polytechnic University, Hong Kong, China.
3 Department of Mechanical Engineering, The Hong Kong Polytechnic University, Hong Kong, China. 4 School of Chemical Engineering, The University of New

South Wales, Sydney 2052 NSW, Australia. 5 Shenzhen Key Laboratory for Graphene-based materials and Engineering Laboratory for Functionalized Carbon
Materials, Graduate School at Shenzhen, Tsinghua University, Shenzhen 518055, China. 6 Materials Genome Institute, Shanghai University, 333 Nanchen
Road, 200444 Shanghai, China. 7 Nanoyang Group, State Key Laboratory of Chemical Engineering, School of Chemical Engineering and Technology, Tianjin
University, 300072 Tianjin, China. These authors contributed equally: Kaikai Li, Jun Zhang. Correspondence and requests for materials should be addressed
to Q.-H.Y. (email: qhyangcn@tju.edu.cn) or to L.Z. (email: mmlmzhou@polyu.edu.hk) or to T.-Y.Z. (email: mezhangt@ust.hk)

NATURE COMMUNICATIONS | (2019)10:725 | https://doi.org/10.1038/s41467-019-08506-5 | www.nature.com/naturecommunications 1


ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-08506-5

S
odium-ion batteries (SIBs) have attracted more attention in based anodes can better illustrate the generality and origin of
recent years particularly for large-scale energy storage due ether-enhanced interfacial electrochemical characteristics.
to the natural abundance of sodium compared to lithium1,2. In the present work, we investigate ether-induced interfacial
However, their performance including specific capacity, rate electrochemical characteristics for SIBs by coupling anatase TiO2
capability, and cycle life are severely hindered by the larger dia- with a diglyme-based electrolyte. Such a TiO2 anode shows
meter of sodium ions than lithium ions3,4, and as a result there excellent capacity retention and rate capability, compared to
have been extensive efforts in searching for suitable electrode those coupled with an EC/DEC-based electrolyte. Insights into
materials, in particular on the anode side5,6. In order to improve the structural evolution and sodiation dynamics obtained by in
the electrochemical performance of the present anode materials, operando Raman, X-ray diffraction (XRD), and electrochemical
modifying the electrolyte is an important and effective approach. kinetic studies reveal that the charge transfer characteristics of the
Most commonly used non-aqueous electrolytes inevitably electrolyte/electrode interface play a vital role in determining the
decompose forming a solid electrolyte interphase (SEI) on the performance, which is also confirmed for Sn, rGO, and CMK-3
anode materials because the operating voltage is lower than their anodes. Comprehensive X-ray photoelectron spectroscopy (XPS)
stable voltage range. Conventional electrolytes in SIBs are for- analyses show that the SEI formed in different electrolytes is
mulated using ester-based solvents similar to those used in composed of different organic species and exhibits distinct
lithium-ion batteries (LIBs)7,8, and these produce SEIs that are composition changes along the SEI depth with more inorganic
often reported to be not sufficiently stable and likely to cause species in the inner region when formed in an ether-based elec-
severe polarization9. In this context, ether-based electrolytes are trolyte. This is the origin of the electrolyte-dependent sodiation
believed to be less useful in LIBs owing to inferior passivation on dynamics and battery performance. Our results indicate a general
both anodes and cathodes, but they have recently been revived in and fundamental explanation of the improved electrochemical
SIBs because of their ability to trigger the highly reversible co- performance of different anodes for SIBs induced using an ether-
intercalation of sodium ions with an ether solvent into gra- based electrolyte. This study might suggest an avenue for the
phite10–14. The improved sodium storage performance has design of efficient electrolyte/electrode interfaces for improved
been observed on different anode materials like CuS nanosheets3, performance of non-aqueous rechargeable batteries.
ZnS nanospheres11, bismuth15, and reduced graphene oxide
(rGO)9 when ester-based electrolytes are replaced with ether
analogs. It is proposed that this improvement is due to the sup- Results
pression of dissolved polysulfide intermediates3, faster Na+ Sodium ion storage performance. Rhombic anatase TiO2
transport, good electrode wettability11,15, and the formation of nanocrystals were synthesized using a solvothermal technique
a thinner SEI9. Su et al.11 proved that Na ions can be easily similar to that reported elsewhere38. The average size of the
absorbed on an ether-based solvent and there is a low energy nanocrystals was around 16 nm as shown in Supplementary
barrier for Na ion diffusion in it, and Seh et al.16 proved that a Figures 1 and 2. Supplementary Figure 2b is a HRTEM image of
glyme-based electrolyte (a typical type of ether-based electrolyte) an individual TiO2 nanocrystal, clearly showing the TiO2 anatase
is superior to an EC/DEC-based electrolyte (a typical ester-based crystal lattice with (101) planes and a d-spacing of 0.35 nm. The
electrolyte; here ethylene carbonate is denoted as EC and diethyl XRD pattern and Raman spectrum (Supplementary Figure 2c and
carbonate is denoted as DEC) for a Na metal anode due to d) confirm that the anatase TiO2 nanocrystals are monophasic
the formation of low resistivity SEI, making a large advance (JCPDS 21-1272) with a high crystallinity. The SIB anode was
on the research of ether-based electrolytes for SIBs. However, prepared using these TiO2 nanocrystals (70 wt%) as the active
the interfacial electrochemical characteristics, especially charge material, carbon black (15 wt%) as the conductive agent, and
transfer in different types of electrolyte has not yet been sys- polyvinylidene fluoride (PVDF) (15 wt%) as the binder. Coin cells
tematically investigated. A deeper understanding of the issue were then assembled using two different solvent-based electro-
is urgently needed so as to clarify the intrinsic origin of the lytes, i.e., NaCF3SO3 dissolved in EC/DEC and NaCF3SO3 in
improvement produced by ether-based electrolytes in SIBs. This diglyme. Figure 1a shows the discharge/charge voltage profiles of
is a key point in the revival of this technology using an ether- SIBs prepared with this TiO2 anode and the EC/DEC electrolyte,
based electrolyte17,18. measured at current densities ranging from 0.1 to 2 A g−1. The
Selecting a typical and ideal anode material is crucial to address voltage profiles are linear with no obvious plateau regions, which
this issue and TiO2 is preferably chosen in this study considering is different from the flat and prolonged voltage plateau observed
the following points. Firstly, the sodiation of TiO2 results in in LIBs39. This indicates that the sodiation of TiO2 proceeds by a
the formation of an amorphous sodium titanate phase so different mechanism from lithiation, which might indicate a
that the structural evolution during sodiation can be easily single-phase Na-intercalation process40. As clearly demonstrated
detected19–22. Secondly, the intercalation mechanism and the by others, the sodiation of anatase TiO2 is a solid-solution-like
small volume change produced by pseudocapacitive sodium sto- reaction21,41. The operation potential of these anatase nanocrys-
rage in amorphous sodium titanate23,24 also guarantee the for- tals as the SIB anode is averaged to be ~0.8 V, which is
mation of a mechanically stable interface between electrolyte and significantly lower than that for Li intercalation (~1.75 V vs. Li/
electrode, which is difficult to achieve in other anodes that involve Li+)39, making TiO2 more promising for practical applications.
large volume changes or conversion reactions upon sodiation. When EC/DEC is replaced by diglyme, a significant improvement
Thirdly, unlike the long flat plateau at ~ 1.75 V vs Li/Li+ in is observed in the anode performance, as shown in Fig. 1b. For
lithium-ion batteries (LIBs), the operating potential of TiO2 in instance, the specific capacity at 0.2 A g−1 is improved from 114.8
SIBs decreases to 0.005 V, which inevitably results in the forma- to 199.4 mAh g−1. Increasing the current rate from 0.1 to 2 A g−1
tion of SEI layer on the electrode25. Lastly, optimizing electrolyte is causes a capacity loss in both cases, but it is more severe in the
also promising to address the bottleneck of practical application EC/DEC case (decreasing from 137.5 to 42.2 mAh g−1) than in
of TiO2 anodes including low initial coulombic efficiency (ICE) diglyme (decreasing from 257.9 to 102.1 mAh g−1). It is also
and fast capacity decay18,26–28, although current efforts have worth noting that using the diglyme-based electrolyte obviously
mostly targeted the improvement of the electronic conductivity in improves the ICE of TiO2 as an SIBs anode, which is one of the
TiO229–33,34–36,37. Further extending the knowledge and experi- critical issues for TiO2 SIB anodes. As shown in Supplementary
ence of above TiO2-based research to other metal and carbon- Figure 3, the ICE is as high as 69% in a diglyme-based electrolyte,

2 NATURE COMMUNICATIONS | (2019)10:725 | https://doi.org/10.1038/s41467-019-08506-5 | www.nature.com/naturecommunications


NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-08506-5 ARTICLE

a 3 c
2 1 0.5 0.2 0.1 A g–1

Scapacity (mAh g–1)


EC/DEC
300 0.1 A g–1 Diglyme
0.1 A g–1

Voltage (V)
2 0.2 A g–1
200
0.5 A g–1
1 A g–1
1 2 A g–1
100

0 0
0 100 200 300 0 10 20 30 40 50 60
Scapacity (mAh g–1) Cycle number

b d Pure anatase
3 Carbon coated
–1 Carbon coated

Scapacity (mAh g–1)


2 1 0.5 0.2 0.1 A g
300 Hollow spheres
Hollow nanospheres
Voltage (V)

2 Nanorod
200 Carbon-coated nanorod

1
100
Diglyme
EC/DEC
0 0
0 100 200 300 0.01 0.1 1 10
–1)
Scapacity (mAh g i (mA g–1)

e 300
Scapacity (mAh g–1)

EC/DEC
Diglyme
200 0.08 A g–1
–1
0.0335 A g
0.01 A g–1 0.168 A g–1 0.173 A g–1 0.335 A g–1
100

0.1 A g–1
0
0 100 200 300 400 500
Cycle number

Fig. 1 Battery performance. a Voltage profiles of the TiO2 anode at different rates using the EC/DEC-based electrolyte (scapacity: specific capacity).
b Voltage profiles of the TiO2 anode at different rates using the diglyme-based electrolyte. Note that the profiles at 0.1 A g−1 in a and b are for the third
cycle while the others are for the first cycle at corresponding rates. c Rate performance. d Performance compared with anatase TiO2-based SIB anodes
reported in the literature, i.e., pure anatase TiO256, carbon-coated anatase57,58, hollow spheres23 and hollow nanospheres59, anatase nanorods60,
and coated anatase nanorods60. e Cyclic performance at 0.1 A g−1, and comparison with cyclic performance reported in the literature

much higher than the 33% in an EC/DEC-based electrolyte, much lower than those with the diglyme counterpart. After 500
although a further increase is still needed for the real applications. cycles at 0.1 A g−1, a reversible capacity of 165 mAh g−1 was
Figure 1c, d shows the rate capabilities of the TiO2 anode using maintained for the cell based on the diglyme electrolyte, while the
two different electrolytes. The cells using the diglyme-based value was only 87 mAh g−1 for the case with EC/DEC. Our
electrolyte show an exceptional high rate capability, much higher previous work9 demonstrated that carbon anodes such as rGO
than that for the EC/DEC system. Remarkable reversible and CMK-3 showed much better efficiency and reversible
capacities of ~252, 197, 153, 124, and 99 mAh g−1 were recorded capacity in the diglyme-based electrolyte than in EC/DEC system.
for the diglyme case at current rates of 0.1, 0.2, 0.5, 1, and Here, we further studied an Sn anode and found that the
2 A g−1, respectively. The corresponding values for EC/DEC are efficiency and reversible capacity in a diglyme-based electrolyte
only ~165, 139, 113, 86, and 66 mAh g−1. When the current rate were superior to what was observed for EC/DEC (Supplementary
is decreased to 0.1 A g−1, the cell with diglyme retains its high Figure 4).
capacity while the cell based on EC/DEC does not, indicating
better reversibility when the diglyme-based electrolyte is used.
Compared to the anatase TiO2 SIB anodes reported in the Structural analysis. To gain an insight into the excellent per-
literature, the performance of the present TiO2 anode coupled formance from the diglyme-based electrolyte and particularly into
with diglyme is among the best available. It is worth noting that the Na-ion diffusion dynamics during sodiation/de-sodiation, we
the present TiO2 active materials are not carbon-modified or conducted in operando Raman spectroscopy (Supplementary
doped with other elements, implying a potential for further Figure 5) and XRD (Supplementary Figure 8), and ex situ XRD to
optimization by compositing with carbons or other elements. reveal the structural evolution, and scanned rate-dependent cyclic
Both cells using diglyme and EC/DEC electrolytes have a quite voltammetry (CV) as well as temperature-dependent electro-
stable long-life cyclic performance at 0.1 A g−1 as shown in chemical impedance spectroscopy (EIS). The sodiation of anatase
Fig. 1e. But the absolute capacities of the cells using EC/DEC are TiO2 is an intercalation reaction described as follows: the sodium

NATURE COMMUNICATIONS | (2019)10:725 | https://doi.org/10.1038/s41467-019-08506-5 | www.nature.com/naturecommunications 3


ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-08506-5

Voltage (V)
EC/DEC 2
0.01 V
0
200 TiO2 + Na+ + e– → Nax(TiO2) +TiO2 + Ti + NaO2 Nax(TiO2)

R.S. (cm–1)
+TiO2
Amorphous
150
Eg,TiO2
100
20 40 60 80 100 120 26.6
SCapacity (mAh g–1) –40.00 2000

Voltage (V)
Diglyme 2
0.01 V
0
200 TiO2 + Na+ + e– → Nax(TiO2) + Ti + NaO2 Nax(TiO2)
R.S. (cm–1)

Amorphous
150
Eg,TiO2
100
50 100 150 200 250 83.5
SCapacity (mAh g–1)

Fig. 2 In operando Raman spectroscopy. Raman spectra of the TiO2 anode during the first cycle of electrochemical sodiation/de-sodiation at a rate of 100
mA g−1 (R.S.: Raman shifts), a using the EC/DEC-based electrolyte and b using the diglyme-based electrolyte21

is intercalated into the TiO2 lattice, distorts the lattice, and par- dependent CV measurements, by which the effect at the Na
tially reduces the TiO2 to form metallic titanium, sodium oxide, counter electrode was eliminated. The CV results obtained from
and amorphous sodium titanate21, which acts as the host for the these tests are quite similar to those observed in the 2-electrode
Na ion storage in the following cycles. Taking the above measurements (Supplementary Figure 11), as shown in Fig. 3,
mechanism into consideration, complete sodiation of TiO2 results except for the smaller polarizations. It can be seen from Fig. 3c
in the disappearance of the Raman bands of crystalline TiO2, that the TiO2 working electrode coupled with the EC/DEC-based
which will not reappear after de-sodiation. As shown in Fig. 2b, electrolyte has a much higher polarization than when coupled
the Raman spectra evolution characteristics of the TiO2 anode with the diglyme-based electrolyte, and the polarization becomes
coupled with a diglyme-based electrolyte during the first cycle more severe as the scan rate increases. The potential polarization
conform to this mechanism. In sharp contrast, when using the with an increased scan rate is extremely small in the case of
EC/DEC-based electrolyte, the TiO2 cannot be fully sodiated even diglyme, indicating quite favorable sodiation dynamics of the
when the voltage is 0.01 V in the first cycle, as shown in Fig. 2a TiO2 electrode. Additionally, CV shows that the sodium storage
where the Raman signal of TiO2 can still be observed, indicating a of TiO2 is controlled by both capacitance and diffusion, and using
lower accessible capacity than when using diglyme in the first the diglyme-based electrolyte increases the contribution from
cycle, and only after ~4 cycles is the TiO2 completely transformed capacitance and facilitates the rapid (de)intercalation of sodium
to the amorphous phase (Supplementary Figure 6). This incom- ions. This is very similar to that observed in graphene-modified
plete sodiation of anatase TiO2 has also been observed in an EC/ TiO2 where the polarization decreases43 and the Na ion inter-
PC (propylene carbonate)-based electrolyte42. The ex situ XRD calation pseudocapacitive process increases24.
results (Supplementary Figure 7) further prove the amorphization The total charge stored in a CV scan can be separated into
of the TiO2 during the first discharge in the diglyme-based faradaic contributions from Na-ion intercalation and pseudo-
electrolyte and the incomplete sodiation in the EC/DEC-based capacitance, and a non-faradaic contribution from the double
electrolyte. Ex situ XRD results of the Sn anode (Supplementary layer effect that depends on the surface area44. These capacitive
Figure 10) show the formation of the NaxSn phase when dis- effects (pseudo-capacitance and double layer charging) can be
charged to 0.1 V in the diglyme-based electrolyte, which is not distinguished from Na-ion intercalation by analyzing the CV data
observed for the EC/DEC case. The in operando Raman spec- at different sweep rates according to the relationship between
troscopy and XRD (Supplementary Figure 9) along with the the measured current (I) and the scan rate (v)45:
ex situ XRD results show that the sodiation of TiO2 and Sn is
much easier in the diglyme-based electrolyte than in the EC/ I ¼ avb
DEC-based electrolyte. In our previous work9, we showed that
carbon materials like rGO and CMK-3 can be sodiated to a higher and the b-value is determined by the slope of the log(v)–log(I)
extent in the diglyme-based electrolyte than in the EC/DEC-based plot. In particular, a b-value of 0.5 indicates total diffusion-
electrolyte. controlled behavior, whereas 1.0 indicates a capacitive process.
The log(v)–log(Ip) plots and corresponding b-values are shown in
Fig. 3d (note that here we use the peak current Ip rather than the
Dynamic analysis. To reveal the true nature of the polarization at current I at a given potential because of the large polarization),
the TiO2 working electrode, we designed a coin cell-type 3-elec- showing that in both cases the (de)sodiation is a combination of
trode setup (Supplementary Figure 12) and performed scan rate- capacitance and diffusion. However, the b-values in the EC/DEC

4 NATURE COMMUNICATIONS | (2019)10:725 | https://doi.org/10.1038/s41467-019-08506-5 | www.nature.com/naturecommunications


NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-08506-5 ARTICLE

a c
2 0.6
1

Polarization (V)
Current (A g–1)
0
0.4 Diglyme
–1 1 mV/s EC/DEC
–2 0.1 mV/s 2 mV/s
0.2 mV/s 5 mV/s 0.2
–3
0.5 mV/s 10 mV/s
–4
0.0 0.5 1.0 1.5 2.0 2.5 0 2 4 6 8 10
E (V, vs. Na/Na+)  (mV s–1)

b 6 d 1.0
0.1 mV/s
4 0.2 mV/s 0.5 Slope: 0.94
Current (A g–1)

Log (Ip, A g–1)


0.5 mV/s
2
0.0
0 Slope: 0.85
1 mV/s –0.5
–2 2 mV/s
–1.0 Diglyme
–4 5 mV/s
EC/DEC
10 mV/s
–6 –1.5
0.0 0.5 1.0 1.5 2.0 2.5 –1.0 –0.5 0.0 0.5 1.0

E (V, vs. Na/Na+) Log (, mV s–1)

Fig. 3 Scan rate-dependent cyclic voltammetry (CV) from the 3-electrode tests (see the experimental setup in Supplementary Figure 12). a, b CV curves of
the TiO2 anode at different scan rates with the EC/DEC-based electrolyte (a) and the diglyme-based electrolyte (b). c Separation (potential polarization)
between the cathodic and anodic peaks in the CV curves as a function of scan rate. d b-values (slopes) obtained from the current peaks, Ip = avb, of the CV
curves

case are smaller than in the case of diglyme, meaning that for the the SEI at the Na counter electrode. The second semicircle at
EC/DEC the (de)sodiation is more diffusion-controlled and has lower frequencies in the 2-electrode study is very similar to the
slower diffusion kinetics than with diglyme. When using diglyme, semicircle observed in the 3-electrode study. The DC-bias
a higher contribution of capacitive charge causes an increased current-dependent EIS study using the 3-electrode setup confirms
intercalation pseudocapacitive behavior of Na+ in the electrode that this semicircle may be assigned to the charge-transfer process
(considering the double layer charging is sensitive to the surface (Supplementary Figure 13a). Therefore, in the following discus-
area and thus is the same no matter the electrolyte), which sion, we use the 2-electrode EIS study. The Arrhenius plots in
contributes to the excellent rate capability and high capacity24. Fig. 4c and data in Supplementary Table 2 represent the
The increased intercalation pseudo-capacitance triggered by individual contributions of the solid electrolyte interphase RSEI
the diglyme-based electrolyte indicates faster sodium ion and the charge transfer resistance Rct, obtained from data fitting
diffusion and electron transport across the whole cell. To gain the temperature-dependent impedance spectra based on the
more fundamental insight into the faster dynamics, we conducted equivalent circuit given in Supplementary Figure 14. For both the
temperature-dependent EIS to examine the dynamic properties SEI and the SEI/electrode interface, an Arrhenius equation is used
(the resistance and activation energy Ea) of individual compo- to determine the activation energies47:
nents in the cell system. According to Barsoukov’s model46, the
sodium ions diffuse across the liquid electrolyte (LE) and SEI σT ¼ AexpðEa =kB T Þ:
layer, followed by a charge transfer process at the SEI/electrode
interface and diffusion inside the electrode material. Here, we Here, A is the pre-exponential factor, Ea is the apparent activation
assume an SEI/electrode interface extending from the SEI layer to energy for ion transport, kB is the Boltzmann constant, σ is the
the electrode (as illustrated in Supplementary Figure 14). ionic conductivity, and T is the absolute temperature. As
Different resistances and activation energies are involved. The expected, transport across the SEI/electrode interface has higher
resistances can be determined by fitting the EIS spectra, and the activation energy (Ea, ct) than that across the SEI layer (Ea, SEI) for
activation energies can be determined by fitting the temperature- both cases. When the EC/DEC-based electrolyte is used, the
dependent resistances. Figure 4a, b, respectively, shows the activation energy across the interface (Ea,ct, EC/DEC) is the highest
temperature-dependent EIS spectra of 2-electrode cells assembled with a value of ~410 meV (which is very close to the value of
using EC/DEC-based and diglyme-based electrolytes after 10 ~400 meV from the 3-electrode EIS study, as shown in
discharge–charge cycles at a rate of 100 mA g−1. Two semicircles Supplementary Figure 13b, c and Supplementary Table 1, further
and a straight line are identified in all these spectra, and in order verifying the effectiveness of the analysis in the 2-electrode study),
to assign them, we compared the EIS spectra measured using both and is higher by a factor of 2.4 than the corresponding value for
2-electrode and 3-electrode setups (Supplementary Figure 13). It the diglyme-based electrolyte. This means that at any tempera-
was found that the small semicircle at higher frequencies ture, transport of Na ions is faster across the SEI/electrode
observed in the 2-electrode study almost disappears in the 3- interface formed in diglyme than that formed in EC/DEC. The
electrode study, which implies that the first semicircle at higher activation energy of Na+ diffusion in the SEI layer Ea, SEI is almost
frequencies in the 2-electrode study includes the information of the same for the two cases. This is probably due to the fact that

NATURE COMMUNICATIONS | (2019)10:725 | https://doi.org/10.1038/s41467-019-08506-5 | www.nature.com/naturecommunications 5


ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-08506-5

a b

0 0

–Im (Z) (Ω cm2)


200

C)

–Im (Z) (Ω cm2)


30
C) 10


re (°10 150

ture
20
20 20
ratu
100

pera
50 10
e

30 30
p

Tem
Tem

40 0 40 0
0 100 200 300 400 500 600 0 5 10 15 20 25 30
Re (Z) (Ω cm2) Re (Z) (Ω cm2)

c d 500
Diglyme EC/DEC
4 400
In (R–1T(SK ))

Ea, ct (meV)
Ea,SEI,diglyme = 157 meV
300
Ea,ct,diglyme = 172 meV
2
Ea,SEI,EC/DEC = 157 meV 200

0 100
Ea,ct,EC/DEC = 410 meV

0
3.2 3.3 3.4 3.5 3.6 3.7 TiO2 CMK-3 rGO Sn
1000/T (K –1) Anode material

Fig. 4 Electrochemical impedance spectroscopy analysis after ten discharge–charge cycles. a, b Temperature-dependent Nyquist plots of the TiO2 anode
for the EC/DEC-based electrolyte (a) and the diglyme-based electrolyte (b). c Arrhenius plot of the resistance contributions of the solid electrolyte
interphase RSEI and the charge transfer resistance Rct, with the derived activation energies Ea,SEI and Ea,ct for the TiO2 anode in the two electrolytes.
d Comparison of the charge transfer energy barriers in the diglyme-based electrolyte and EC/DEC-based electrolytes for various electrode materials

Na+ diffuses in the SEI layer by hopping migration and the quantitative analysis of the XPS spectra. The most significant
activation energy for this is not significantly different in different feature is that in the SEI formed in the diglyme-based electrolyte,
SEI compositions48,49. It is also worth noting that the interfacial the atomic fraction of C decreases distinctly with increasing
ionic conductivity in the diglyme-based system is much higher etching depth (~28% at the surface and ~5% at 23 nm depth),
than that in the EC/DEC-based system, which implies a major while the atomic fraction of Na shows a completely opposite
effect of the use of different solvent-based electrolytes in SIBs. If trend, and the atomic fraction of O increases slightly with etching
transport through the SEI/electrode interface is sluggish, this depth. This means that the organic species are mainly distributed
interfacial resistance will have a detrimental effect on the rate near the surface while the interior of the SEI is mainly composed
capability47. This finding provides crucial insight into the of inorganic species. Although a slight decrease in the atomic
improved interfacial characteristics induced by ether-based fraction of C in the SEI formed in the EC/DEC-based electrolyte
electrolytes, which is also demonstrated for other electrode with etching depth is also observed, its value is much higher
materials, as shown in Fig. 4d. We also examined the than with the diglyme counterpart. This indicates that there are
temperature-dependent EIS of a CMK-3 anode (Supplementary more organic species in the SEI formed in the EC/DEC-based
Figure 15 and Table 3), an rGO anode (Supplementary Figure 16 electrolyte and they are spread over the whole thickness of
and Table 4), and an Sn anode (Supplementary Figure 17 and the SEI. Analyses of the Na Auger Parameter confirms these
Table 5) after cycling. In all cases, the charge transfer energy conclusions (Supplementary Figure 21 and Supplementary
barrier (Ea, ct) across the interface in the EC/DEC-based Table 6)50. The inorganic species are generally localized and
electrolyte is much higher than that in the diglyme-based remain in a stable thin layer during cycling, leading to a low
electrolyte, by a factor of around 6 for a CMK-3 anode, 2.7 for charge transfer resistance and energy barrier16,49. The presence of
an rGO anode, and 4 for an Sn anode. organic products, which are usually porous, makes the SEI much
thicker16, leading to a much higher charge transfer resistance
SEI analysis. SEM examination of the electrode surface showed and energy barrier.
that an SEI had formed after cycling (Supplementary Figure 18). The SEI composition was also analyzed by high-resolution XPS
It seems that the formation of the SEI in a diglyme-based using a Mg X-ray source. First, we analyzed the composition of
electrolyte is more uniform than that formed in an EC/DEC- the SEI formed in the diglyme-based electrolyte (Fig. 5d,
based electrolyte. We therefore carried out XPS to examine Supplementary Figures 22–26 and Supplementary Table 7). The
the SEIs formed in the two electrolytes, as shown in Fig. 5a–d. C 1s spectrum was fitted using four peaks with binding energies
XPS depth profiling was done using Ar ion sputtering. As shown of ~284.9 eV (C–C), ~286.0 eV (C–O), ~289.9 eV (O–C=O), and
in Supplementary Figure 19, C, O, F, and Na signals were ~292.0 eV (–CF2–), which are consistent with sodium alkoxides
detected on the electrode surface in both cases and also at (RCH2ONa) being the main reduction products of diglyme and
different etching depths. Figure 5a shows the elemental compo- Na2CO3 and PVDF9,16,50. The O 1s spectrum contained three
sitions of the SEI layer at different depths obtained by peaks at ~532.6 eV (R–O–Na), ~531.1 eV, and ~530.0 eV

6 NATURE COMMUNICATIONS | (2019)10:725 | https://doi.org/10.1038/s41467-019-08506-5 | www.nature.com/naturecommunications


NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-08506-5 ARTICLE

a b
60 SEI layer SEI/electrode interface Anode

50 EC/DEC
Charge

Atomic fraction %
+ transfer e–
40 Organic [Na ]
Diglyme EC/DEC
30 C C Inorganic
O O
F F Ea,ct,EC/DEC Na+[Lattice] –
20 Na Na

Diglyme
10

Organic
Ea,ct,diglyme
0
Inorganic
0 5 10 15 20 25
Reaction coordinate
Etching depth (nm)

c 50 d 50
C 1s 0 nm C 1s
0 nm
40 40
ROCO2Na RCH2ONa
CPS (×103)

CPS (×103)
30 30 C–O C–C
O–C=O C–O C–C
(ROCO2Na,
20 Na2CO3) 20
O–C=O
10 10
-CF2– -CF2– Na2CO3)

0 0
292 288 284 292 288 284
Binding energy (eV) Binding energy (eV)

Fig. 5 SEI chemistry and the charge transfer process. a The atomic fractions of C, O, F, and Na as a function of SEI depth, calculated from the XPS spectra
depth profiling (Supplementary Figure 19). b Schematic of the SEI compositions formed in different electrolytes and a comparison of the charge transfer
energy barriers. [Na+] means that the Na+ may be solvated. There is a smaller amount of organic species in the SEI formed in the diglyme-based
electrolyte and the inner part of the SEI is mainly composed of inorganic species. c XPS C 1s spectrum before etching in the EC/DEC-based electrolyte.
d XPS C 1s spectrum before etching in the diglyme-based electrolyte

(Na–O–Ti), belonging to RCH2ONa, Na2CO3, and NaxTiO2, Discussion


respectively. The O 1s spectrum change with etching depth also The dynamic properties of the SEI/electrode interface play a vital
proves that the organic species decrease in the inner region of the role in the sodium storage. Modification of this interface can be
SEI (Supplementary Figure 24a–c). The F 1s spectrum showed a achieved by changing the SEI chemistry, which depends on
peak at ~684.9 eV (Na–F) and the Na 1s spectrum showed a peak whether an ether-based or an ester-based electrolyte is used. But
at ~1072.2 eV (Na–F, Na–O), indicating the existence of NaF and how the SEI chemistry works behind-the-scenes has not yet been
other sodium compounds (RCH2ONa, Na2CO3). Overall, we fully determined, and undoubtedly needs further in situ experi-
concluded that the SEI formed in the diglyme-based electrolyte mental investigation and theoretical analysis. To support the
contains both organic (RCH2ONa) and inorganic (Na2CO3, NaF) governing role of the SEI/electrode interface, we measured
components. In the case of the EC/DEC-based electrolyte, we saw the electrochemical performance and analyzed the sodiation
distinct XPS peaks corresponding to sodium alkyl carbonate dynamics of an anatase TiO2 anode composed of 84 nm nano-
(ROCO2Na) and inorganic products (Na2CO3, NaF), as shown in crystals (Supplementary Figure 26). First, as shown in Supple-
Fig. 5c, Supplementary Figures 22–26, and Supplementary Table 8 mentary Figure 27, this anode shows similar voltage profiles to
(refs. 9,16,50–54). Therefore, in addition to the distinct composition those of the 16 nm TiO2 nanocrystal anode but has relatively
change along the SEI depth, there is also a difference between lower specific capacities in both the EC/DEC-based and diglyme-
the SEIs formed in the two electrolytes because of the organic based electrolytes, with the capacity in the latter being higher than
components formed by the decomposition of the different in the former, similar to what is observed for the 16 nm nano-
solvents, which is consistent with what is found in SIB Bi crystal anode. We hypothesize that the reasons for the inferior
electrodes15. It has been reported that sodium alkoxides capacity of 84 nm nanocrystal material are the high sodium dif-
(RCH2ONa) ensure fast Na+ transport and are essential for the fusion energy barrier and relatively small crystal surface area.
interfacial stability of the Bi electrode15. First-principles calcula- Interestingly, dynamic analysis shows that in the diglyme-based
tions conducted by Su et al. demonstrated that an ether-based electrolyte, as shown in Supplementary Figure 28, 29 and Table 9,
solvent reduces the energy barrier for sodium-ion diffusion and the activation energies of sodium diffusion through the SEI layer
thus improves the electrochemical performances11. It is reason- as well as the SEI/electrode interface are quite close for the two
able to deduce that the RCH2ONa also reduces the energy barrier materials, meaning almost the same energy barriers for sodium
for sodium-ion diffusion. The inorganic species in the SEI transport and charge transfer. On the one hand, this indicates
generally have low Na+ diffusion activation energies (e.g., the that the activation energy for charge transfer at the electrolyte/
activation energy for NaF is ~100–150 meV49). Combined with electrode interface is inherent to the nature of the interface and is
the above dynamic analysis, we conclude that the different SEI almost independent of particle size. On the other hand, the large
chemistries lead to the huge difference in the electrochemical surface area is one of the reasons accounting for the better per-
performance of the two different solvent-based electrolytes by formance of the 16 nm than the 84 nm material, because a larger
modifying the resistance of the SEI/electrode interfaces and the surface area results in a correspondingly larger interface area and
transport of Na+ and electrons in them. higher electrochemical reactivity.

NATURE COMMUNICATIONS | (2019)10:725 | https://doi.org/10.1038/s41467-019-08506-5 | www.nature.com/naturecommunications 7


ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-08506-5

Apart from the electrolyte/electrode interfacial properties, with ethanol and then dried at room temperature before being heated at 350 °C for
solvent wettability is another parameter that may influence the 5 h in air.
sodium ion surface diffusion kinetics. Better solvent wettability is
beneficial for improving the rate performance because of better Materials characterization. Transmission electron microscope (TEM) images and
selected area electron diffraction (SAED) patterns of the TiO2 nanocrystals were
ion availability as a result of the easier penetration of the elec- obtained using a JEOL 2010F transmission electron microscope operated at 200 kV.
trolyte, since the surface ion availability may play an essential role Raman spectra were collected using a Raman spectrometer (LabRAM HR spec-
in providing sufficient Na ions for intercalation into the TiO255. trometer, Horiba) with a 532.05-nm Ar-ion laser. XRD patterns were recorded over
The solvent wettability is measured by the contact angle. As the range 20–80° on a Rigaku diffractometer equipped with a Cu Kα radiation
source operated at 40 kV and 120 mA.
shown in Supplementary Figure 30, both the EC/DEC-based and
diglyme-based electrolytes effectively spread over the electrode Cell assembly and testing. The active materials (70 wt%) were homogeneously
surface composed of 16 nm TiO2 nanocrystals, super P and PVDF mixed with conductive carbon black (15 wt%) and PVDF (15 wt%) in NMP to
within 5 s, suggesting that the electrolyte wettability is not a form a slurry, which was stirred for 6 h and coated onto a Cu foil. After vacuum-
decisive factor that influences the battery performance of the drying at 110 °C for 10 h, the electrodes were cut into circular pieces with a dia-
16 nm TiO2 anode. While for the electrode consisting 84 nm meter of 12 mm, which were used as anodes for the assembled cell. The mass
loading of the electrodes was 1.5 mg cm−2. The electrolytes for comparison were 1
nanocrystals, the wettability is distinctly different for the two M NaCF3SO3 dissolved in a mixture of EC and DEC (volume ratio = 1:1), and 1 M
electrolytes (Supplementary Figure 30) with that of the diglyme- NaCF3SO3 dissolved in diglyme. Using these electrodes and electrolytes with pure
based electrolyte being better than that of the EC/DEC-based sodium foils as counter electrodes and Whatman GF/D glass fibers as separators,
electrolyte. Except for the larger electrochemical active surface 2032-type coin cells were assembled in an argon-filled glove box with oxygen and
moisture contents below 0.1 ppm. Galvanostatic charge–discharge tests were con-
area, the better electrolyte wettability should also contribute to the ducted on a LAND CT2001 battery program controlling system at different current
superior electrochemical performance of the 16 nm TiO2 com- densities with a voltage window of 0.005–2.5 V vs. Na/Na+.
pared to the 84 nm TiO2.
In summary, as a superior model anode, the 16 nm nanocrystal Cyclic voltammetry. CV measurements were conducted using an electrochemical
anatase TiO2 coupled with the diglyme-based electrolyte shows workstation (VMP3, Bio Logic, France) at scanning rates from 0.1 mV s−1 to
10 mV s−1 in the voltage range of 0.005–2.5 V vs. Na/Na+ at room temperature.
a reversible capacity of 257.9 mAh g−1 at 100 mA g−1 and more
than 100 mAh g−1 at 2000 mA g−1, both of which are much
Electrochemical Impedance Spectroscopy. All EIS measurements were carried
better than in the EC/DEC-based electrolyte and are among out using a PARSTAT 4000 electrochemical workstation, using a frequency range
the best values ever reported for anatase-based anodes for of 105 Hz to 100 mHz and an AC voltage amplitude of 10 mV. The cell temperature
SIBs. The Sn, rGO, and CMK-3 anodes also show superior Na+ was controlled in a temperature-controlled chamber (Linkam Scientific) and was
storage performance in the diglyme-based electrolyte than in varied between 0 and 40 °C (±0.1 °C) in steps of 10 °C using a nitrogen-gas flow
and electronic heaters (Linkam Scientific) for the temperature-dependent
the EC/DEC-based electrolyte. In operando Raman spectra experiments.
and XRD show that the ether-based electrolyte facilitates the
sodiation-induced structural transitions of the TiO2 and Sn In operando Raman spectroscopy and in situ XRD. Raman spectra acquisition
anodes. We have also shown that charge transfer dynamics at during cell operation has been described in detail elsewhere and is schematically
the SEI/electrode interface play a crucial role in the rate-cycling shown in Supplementary Figure 5 (ref. 39). For the measurements, a special elec-
performance and are nearly size-independent. Charge transfer trode was used which was prepared by mixing the active materials with super P and
5 wt% PTFE in water to obtain a homogenous slurry. The final content of the active
has an energy barrier of 172 meV for the interface formed in materials was 80 wt%. The slurry was rolled into a thin film, which was then
the diglyme-based electrolyte (1.4 times lower than for the pressed onto a stainless-steel mesh. A delicate battery cell with a quartz window on
interface in the EC/DEC-based electrolyte) for the TiO2 anode, the top was used (Supplementary Figure 5). The Raman spectra were recorded on a
providing faster charge transfer across the electrolyte/electrode MicroRaman system (LabRAM HR spectrometer, Horiba) with an Olympus BX
microscope and an argon ion laser (532.05 nm). Each spectrum was acquired for
interface. Significant reductions in the charge transfer energy 20 s. The galvanostatic discharge of the cell was controlled by an electrochemical
barrier are also demonstrated in Sn, CMK-3, and rGO electrodes workstation (PARSTAT 4000). A similar battery cell to that used for the in
when the EC/DEC-based electrolyte is replaced by the diglyme- operando Raman spectroscopy was designed, replacing the quartz window with a
based electrolyte. The faster dynamics increase the Na+ inter- Kapton film window to perform the in situ XRD measurements (Supplementary
Figure 8).
calation pseudocapacitive behavior, which is highly beneficial
to fast charge storage and long-term stability. The chemistry
Surface characterization. XPS survey scan analyses were conducted on a PHI
and composition distribution in the SEI layer are of great rele- 5000 VersaProbe II spectrometer using a monochromatic Al Kα X-ray (1486.6 eV)
vance for the charge transfer, with the organic species formed in source. High-resolution XPS analyses were carried out using an ESCALAB 250Xi
the ether-based electrolyte together with the inorganic species spectrometer with a Mg Kα X-rays source, in order to eliminate the interferences of
favoring the sodiation dynamics. The information obtained in the Na Auger peak with the oxygen 1s photoelectron line and the Ti Auger peak
with the Na 1s peak. The batteries were disassembled in a glove box after 10
this work provides a general framework for understanding the discharge–charge cycles at a rate of 100 mA g−1, and the anodes were washed
origin of the extraordinary performance and improved interfacial several times with the related solvent, i.e., dimethyl carbonate (DMC) for EC/DEC
characteristics of an ether-based electrolyte for SIB anodes, and and dimethoxyethane (DME) for diglyme. After drying, the anodes were trans-
can guide the future design and matching between non-aqueous ferred to a vacuum box and then transferred into the XPS chamber. Samples were
ion-etched using 2 kV Ar ions over 2 × 3 mm2 with an etching rate of approxi-
electrolytes and novel electrode materials for rechargeable mately 4.3 nm min−1. Spectra were charge corrected to the main line of the carbon
batteries. 1s spectrum (adventitious carbon) and set to a BE of 284.8 eV (Supplementary
Figure 21).

Methods Data availability


Synthesis of anatase TiO2 nanocrystals. Titanium(IV) butoxide (TB, 97%), oleic
The data that support the findings of this study are available within the article and
acid (OA, 90%), oleylamine (OM, 70%), and absolute ethanol were purchased form
its Supplementary Information files. All other relevant data supporting the findings
Aldrich. All chemicals were used as received. The synthesis of the TiO2 nano-
of this study are available from the corresponding authors upon reasonable request.
crystals was accomplished using a solvothermal method. Typically, 5 mmol TB was
added to a mixture of 25 mmol OA, 25 mmol OM, and 100 mmol absolute ethanol.
The mixture was stirred in a 40-mL Teflon cup for 10 min before being transferred Received: 9 July 2018 Accepted: 11 January 2019
into a 100-mL Teflon-lined stainless-steel autoclave containing 20 mL of a mixture
of ethanol and water (96% ethanol, v/v). The system was then heated at 180 °C for
18 h. The white TiO2 nanocrystal precipitates obtained were washed several times

8 NATURE COMMUNICATIONS | (2019)10:725 | https://doi.org/10.1038/s41467-019-08506-5 | www.nature.com/naturecommunications


NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-08506-5 ARTICLE

References 31. Qiu, S., Xiao, L., Ai, X., Yang, H. & Cao, Y. Yolk-shell TiO2@C nanocomposite
1. Van Noorden, R. The rechargeable revolution: a better battery. Nature 507, as high-performance anode material for sodium-ion batteries. ACS Appl.
26–28 (2014). Mater. Interfaces 9, 345–353 (2017).
2. Vaalma, C., Buchholz, D., Weil, M. & Passerini, S. A cost and resource analysis 32. Shi, X. D. et al. Anatase TiO2@C composites with porous structure as an
of sodium-ion batteries. Nat. Rev. Mater. 3, 18013 (2018). advanced anode material for Na ion batteries. J. Power Sources 330, 1–6
3. Li, J. et al. Significantly improved sodium-ion storage performance of CuS (2016).
nanosheets anchored into reduced graphene oxide with ether-based 33. Chen, J. et al. Pinecone-like hierarchical anatase TiO2 bonded with carbon
electrolyte. ACS Appl. Mater. Interfaces 9, 2309–2316 (2017). enabling ultrahigh cycling rates for sodium storage. J. Mater. Chem. A 4,
4. Huang, H., Wu, H. H., Wang, X., Huang, B. & Zhang, T. Y. Enhancing sodium 12591–12601 (2016).
ionic conductivity in tetragonal-Na3PS4 by halogen doping: a first principles 34. Ni, J. et al. Self-supported nanotube arrays of sulfur-doped TiO2 enabling
investigation. Phys. Chem. Chem. Phys. 20, 20525–20533 (2018). ultrastable and robust sodium storage. Adv. Mater. 28, 2259–2265 (2016).
5. Li, Y. M. et al. Recent advances of electrode materials for low-cost sodium-ion 35. Wang, B. et al. Boron-doped anatase TiO2 as a high-performance anode
batteries towards practical application for grid energy storage. Energy Storage material for sodium-ion batteries. ACS Appl. Mater. Interfaces 8, 16009–16015
Mater. 7, 130–151 (2017). (2016).
6. Masse, R. C., Uchaker, E. & Cao, G. Z. Beyond Li-ion: electrode materials for 36. Liu, S. A., Cai, Z. Y., Zhou, J., Pan, A. Q. & Liang, S. Q. Nitrogen-doped TiO2
sodium- and magnesium-ion batteries. Sci. China Mater. 58, 715–766 (2015). nanospheres for advanced sodium-ion battery and sodium-ion capacitor
7. Slater, M. D., Kim, D., Lee, E. & Johnson, C. S. Sodium-ion batteries. Adv. applications. J. Mater. Chem. A 4, 18278–18283 (2016).
Funct. Mater. 23, 947–958 (2013). 37. Zhang, J. et al. The interplay of oxygen functional groups and folded texture in
8. Kundu, D., Talaie, E., Duffort, V. & Nazar, L. F. The emerging chemistry of densified graphene electrodes for compact sodium-ion capacitors. Adv. Energy
sodium ion batteries for electrochemical energy storage. Angew. Chem. Int. Ed. Mater. 8, 1702395 (2018).
Engl. 54, 3431–3448 (2015). 38. Dinh, C. T., Nguyen, T. D., Kleitz, F. & Do, T. O. Shape-controlled synthesis
9. Zhang, J. et al. Achieving superb sodium storage performance on carbon of highly crystalline titania nanocrystals. ACS Nano 3, 3737–3743 (2009).
anodes through an ether-derived solid electrolyte interphase. Energy Environ. 39. Li, K. et al. Ultrafast-charging and long-life li-ion battery anodes of TiO2-B
Sci. 10, 370–376 (2017). and anatase dual-phase nanowires. ACS Appl. Mater. Interfaces 9,
10. Hu, Z. et al. Pyrite FeS2 for high-rate and long-life rechargeable sodium 35917–35926 (2017).
batteries. Energy Environ. Sci. 8, 1309–1316 (2015). 40. Li, K. et al. Discovering a first-order phase transition in the Li-CeO2 system.
11. Su, D. W., Kretschmer, K. & Wang, G. X. Improved electrochemical Nano Lett. 17, 1282–1288 (2017).
performance of Na-ion batteries in ether-based electrolytes: a case study of 41. Bella, F. et al. Unveiling the controversial mechanism of reversible Na storage
ZnS nanospheres. Adv. Energy Mater. 6, 1501785 (2016). in TiO2 nanotube arrays: amorphous versus anatase TiO2. Nano Res. 10,
12. Cohn, A. P., Share, K., Carter, R., Oakes, L. & Pint, C. L. Ultrafast solvent- 2891–2903 (2017).
assisted sodium ion intercalation into highly crystalline few-layered graphene. 42. Yan, D. et al. Improved sodium-ion storage performance of TiO2 nanotubes
Nano. Lett. 16, 543–548 (2016). by Ni2+ doping. J. Mater. Chem. A 4, 11077–11085 (2016).
13. Kim, H. et al. Sodium storage behavior in natural graphite using ether-based 43. Das, S. K. et al. Graphene mediated improved sodium storage in
electrolyte systems. Adv. Funct. Mater. 25, 534–541 (2015). nanocrystalline anatase TiO2 for sodium ion batteries with ether electrolyte.
14. Zhang, J., et al. Ethers illume sodium-based battery chemistry: uniqueness, Chem. Commun. 52, 1428–1431 (2016).
surprise, and challenges. Adv. Energy Mater. 8, 1801361 (2018). 44. Wang, J., Polleux, J., Lim, J. & Dunn, B. Pseudocapacitive contributions to
15. Wang, C. C., Wang, L. B., Li, F. J., Cheng, F. Y. & Chen, J. Bulk bismuth as a electrochemical energy storage in TiO2 (anatase) nanoparticles. J. Phys. Chem.
high-capacity and ultralong cycle-life anode for sodium-ion batteries by C 111, 14925–14931 (2007).
coupling with glyme-based electrolytes. Adv. Mater. 29, 1702212 (2017). 45. Li, S. M. et al. Alternating voltage induced ordered anatase TiO2 nanopores: an
16. Seh, Z. W., Sun, J., Sun, Y. M., & Cui, Y. A highly reversible room-temperature electrochemical investigation of sodium storage. J. Power Sources 336,
sodium metal anode. ACS Cent. Sci. 1, 449–455 (2015). 196–202 (2016).
17. Larcher, D. & Tarascon, J. M. Towards greener and more sustainable batteries 46. Barsoukov, E. et al. Comparison of kinetic properties of LiCoO2 and
for electrical energy storage. Nat. Chem. 7, 19–29 (2015). LiTi0.05Mg0.05Ni0.7Co0.2O2 by impedance spectroscopy. Solid State Ion. 161,
18. Munoz-Marquez, M. A. et al. Na-ion batteries for large scale applications: a 19–29 (2003).
review on anode materials and solid electrolyte interphase formation. Adv. 47. Busche, M. R. et al. Dynamic formation of a solid-liquid electrolyte interphase
Energy Mater. 7, 1700463 (2017). and its consequences for hybrid-battery concepts. Nat. Chem. 8, 426–434
19. Xu, Y. et al. Nanocrystalline anatase TiO2: a new anode material for (2016).
rechargeable sodium ion batteries. Chem. Commun. 49, 8973–8975 (2013). 48. Soto, F. A., Marzouk, A., El-Mellouhi, F. & Balbuena, P. B. Understanding
20. Lunell, S., Stashans, A., Ojamae, L., Lindstrom, H. & Hagfeldt, A. Li and Na ionic diffusion through SEI components for lithium-ion and sodium-ion
diffusion in TiO2 from quantum chemical theory versus electrochemical batteries: insights from first-principles calculations. Chem. Mater. 30,
experiment. J. Am. Chem. Soc. 119, 7374–7380 (1997). 3315–3322 (2018).
21. Wu, L. M. et al. Unfolding the mechanism of sodium insertion in anatase 49. Choudhury, S. et al. Designing solid-liquid interphases for sodium batteries.
TiO2 nanoparticles. Adv. Energy Mater. 5, 1401142 (2015). Nat. Commun. 8, 898 (2017).
22. Li, W. et al. A reversible phase transition for sodium insertion in anatase 50. Munoz-Marquez, M. A. et al. Composition and evolution of the solid-
TiO2. Chem. Mater. 29, 1836–1844 (2017). electrolyte interphase in Na2Ti3O7 electrodes for Na-ion batteries: XPS and
23. Zhang, G. Q., Wu, H. B., Song, T., Paik, U. & Lou, X. W. TiO2 hollow spheres Auger parameter analysis. ACS Appl. Mater. Interfaces 7, 7801–7808 (2015).
composed of highly crystalline nanocrystals exhibit superior lithium storage 51. Pan, Y. et al. Investigation of the solid electrolyte interphase on hard carbon
properties. Angew. Chem. Int. Ed. Engl. 53, 12590–12593 (2014). electrode for sodium ion batteries. J. Electroanal. Chem. 799, 181–186 (2017).
24. Chen, C. et al. Na(+) intercalation pseudocapacitance in graphene-coupled 52. Bi, X. X. et al. Investigating dendrites and side reactions in sodium-oxygen
titanium oxide enabling ultra-fast sodium storage and long-term cycling. Nat. batteries for improved cycle lives. Chem. Commun. 51, 7665–7668 (2015).
Commun. 6, 6929 (2015). 53. Dahbi, M. et al. Black phosphorus as a high-capacity, high-capability negative
25. Aravindan, V., Lee, Y. S., Yazami, R. & Madhavi, S. TiO2 polymorphs in electrode for sodium-ion batteries: investigation of the electrode/interface.
‘rocking-chair’ Li-ion batteries. Mater. Today 18, 345–351 (2015). Chem. Mater. 28, 1625–1635 (2016).
26. Longoni, G. et al. Shape-controlled TiO2 nanocrystals for Na-ion battery 54. Eshetu, G. G. et al. Impact of the electrolyte salt anion on the solid electrolyte
electrodes: the role of different exposed crystal facets on the electrochemical interphase formation in sodium ion batteries. Nano Energy 55, 327–340
properties. Nano Lett. 17, 992–1000 (2017). (2019).
27. Wu, L. M., Buchholz, D., Bresser, D., Chagas, L. G. & Passerini, S. Anatase 55. Zhou, M. et al. Amorphous TiO2 inverse opal anode for high-rate sodium ion
TiO2 nanoparticles for high power sodium-ion anodes. J. Power Sources 251, batteries. Nano Energy 31, 514–524 (2017).
379–385 (2014). 56. Chen, J. et al. Black anatase titania with ultrafast sodium-storage performances
28. Dahbi, M., Yabuuchi, N., Kubota, K., Tokiwa, K. & Komaba, S. Negative stimulated by oxygen vacancies. ACS Appl. Mater. Interfaces 8, 9142–9151
electrodes for Na-ion batteries. Phys. Chem. Chem. Phys. 16, 15007–15028 (2016).
(2014). 57. Li, Y. et al. Excellent sodium storage performance of carbon-coated TiO2:
29. Tahir, M. N. et al. Extraordinary performance of carbon-coated anatase TiO2 assisted with electrostatic interaction of surfactants. J. Power Sources 361,
as sodium-ion anode. Adv. Energy Mater. 6, 1501489 (2016). 326–333 (2017).
30. Xiong, Y., Qian, J., Cao, Y., Ai, X. & Yang, H. Electrospun TiO2/C nanofibers 58. Oh, S. M. et al. High electrochemical performances of microsphere C-TiO(2)
as a high-capacity and cycle-stable anode for sodium-ion batteries. ACS Appl. anode for sodium-ion battery. ACS Appl. Mater. Interfaces 6, 11295–11301
Mater. Interfaces 8, 16684–16689 (2016). (2014).

NATURE COMMUNICATIONS | (2019)10:725 | https://doi.org/10.1038/s41467-019-08506-5 | www.nature.com/naturecommunications 9


ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-08506-5

59. Su, D. W., Dou, S. X. & Wang, G. X. Anatase TiO2: better anode material than Competing interests: The authors declare no competing interests.
amorphous and rutile phases of TiO2 for Na-ion batteries. Chem. Mater. 27,
6022–6029 (2015). Reprints and permission information is available online at http://npg.nature.com/
60. Kim, K. T. et al. Anatase titania nanorods as an intercalation anode material reprintsandpermissions/
for rechargeable sodium batteries. Nano Lett. 14, 416–422 (2014).
Journal peer review information: Nature Communications thanks the anonymous
reviewers for their contribution to the peer review of this work.
Acknowledgements
This work was supported by the National Science Fund for Distinguished Young Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in
Scholars, China (No. 51525204), the National Natural Science Foundation of China (Nos. published maps and institutional affiliations.
U1601206, 51772164, and U1710256), Local Innovative and Research Teams Project of
Guangdong Pearl River Talents Program (2017BT01N111), the Shenzhen Technical Plan
Project (Nos. JCYJ20170412171630020 and JCYJ20170412171359175), the 111 project Open Access This article is licensed under a Creative Commons
(No. D16002) from the State Administration of Foreign Experts Affairs and the Ministry Attribution 4.0 International License, which permits use, sharing,
of Education, PRC, and the General Research Fund (No. 15210718) from Hong Kong adaptation, distribution and reproduction in any medium or format, as long as you give
Research Grants Council. The authors thank Mr. Kun Qian for his assistance in the appropriate credit to the original author(s) and the source, provide a link to the Creative
in situ XRD measurements. Commons license, and indicate if changes were made. The images or other third party
material in this article are included in the article’s Creative Commons license, unless
indicated otherwise in a credit line to the material. If material is not included in the
Author contributions
article’s Creative Commons license and your intended use is not permitted by statutory
Q.H.Y., T.Y.Z., and L.M.Z. managed the project and guided the research; K.K.L. and J.Z.
regulation or exceeds the permitted use, you will need to obtain permission directly from
contributed equally to the paper; D.M.L., B.H.L., Y.B.H., and F.Y.K. participated in the
experiments; W.L., D.W.W., and S.S. were involved in the discussions; K.K.L. and J.Z. the copyright holder. To view a copy of this license, visit http://creativecommons.org/
wrote the manuscript with the supervision of Q.H.Y., T.Y.Z., and L.M.Z., and all authors licenses/by/4.0/.
have approved the final manuscript.
© The Author(s) 2019
Additional information
Supplementary Information accompanies this paper at https://doi.org/10.1038/s41467-
019-08506-5.

10 NATURE COMMUNICATIONS | (2019)10:725 | https://doi.org/10.1038/s41467-019-08506-5 | www.nature.com/naturecommunications

You might also like