You are on page 1of 7

Solid State Ionics 303 (2017) 154–160

Contents lists available at ScienceDirect

Solid State Ionics

journal homepage: www.elsevier.com/locate/ssi

Study of structural changes that occurred during charge/discharge of


carbon-coated SiO anode by nuclear magnetic resonance
Takakazu Hirose a,b,⁎, Masanori Morishita a, Hideya Yoshitake a, Tetsuo Sakai a
a
Storage Device Development Research Center, Yamagata University, Yonezawa, Yamagata 992-0119, Japan
b
Silicone-Electronics Materials Research Center, Shin-Etsu Chemical Co., Ltd, Annaka, Gunma 379-0125, Japan

a r t i c l e i n f o a b s t r a c t

Article history: To improve the cycling performance of the carbon-coated SiO (SiO-C) as a high capacity anode material, the
Received 28 January 2017 charge–discharge mechanism and the structural changes in the bulk during long-term cycling were investigated
Received in revised form 6 March 2017 by the solid-state nuclear magnetic resonance (NMR). The electron conductivity of the SiO-C was improved fol-
Accepted 7 March 2017
lowing the charging state, whereas the electrical resistivity tended to be low. It suggested that from the beginning
Available online 11 March 2017
of the discharge to its half state (at the low potential side), the silicide reaction is predominant, whereas at the
Keywords:
terminal phase of the discharge (at the high potential side), the release of Li from the silicate is predominant.
Lithium-ion battery Upon designing a negative cutoff potential of 1.1 V, the cell's cycling performance was checked. The results sug-
Negative electrode material gested that during repeated charge-discharge cycling, Li release from the silicate decreased and a disproportion-
Silicon monoxide ation reaction occurred which produced Si and Li4SiO4.
Solid-state NMR It was confirmed that a part of the cycle performance was improved by designing a battery utilizing the low po-
tential side which is supposed to increase the electron conductivity and suppressed the disproportionation.
© 2017 Elsevier B.V. All rights reserved.

1. Introduction carbon layer needs to be formed on the surface by the CVD process or
other methods to be electrically conductive [13,14].
Li-ion secondary batteries with a high energy density have been de- Heat of ca. 1000 °C was applied to the SiO material during the pro-
veloped mainly for use in small mobile devices. In recent years, due to cess of layer formation. The disproportionation into Si and SiO2 in the
the increasing variety of applications even in the automotive industry, SiO bulk during this process has been reported. Eventually a fine Si dot
miniaturization and high-capacity of the battery is a MUST factor to ex- structure is formed [15,16].
tend the traveling distances. Even though the SiO material has been improved for the bulk dete-
A high capacity anode using tin or silicon has been developed as a rioration during the charging and discharging, only a small amount is
candidate for increasing the energy density of the battery [1–7]. Several still used (1–5 wt%) in combination with the carbon as the main
high capacity anode materials that include metal systems based on the material.
Li alloy have a principal issue in the use of the active materials upon re- In order to increase the amount of SiO in the anode the following
peated charging and discharging [8–10]. two factors need to be mainly evaluated.
SiO materials have now been developed affording a significant re- 1. Improvement of the initial battery coulombic efficiency (ICE).
duction in the problems mentioned above; moreover, in 2010, batteries 2. Improvement of the cycle retention rate.
using the SiO anode in conbination with a carbonaceous active material Compared to the carbon materials, which have a high ICE like 90%,
were commercialized. The SiO material has a structure featuring Si par- the SiO material's ICE is as low as ca. 65% (around 1.0 V).
ticles of several nm in size finely dispersed in amorphous SiO2, in which Many institutes and R&D organizations have studied the ICE im-
refinement of the active material is suppressed during the charge and provement of batteries and found some technical solutions [17,18]. On
discharge cycles [11,12]. the other hand, improvement of the cycling performance is not suffi-
Furthermore, the SiO material is categorized as an insulating materi- cient and still needs to be improved. From the viewpoint of automotive
al, therefore, it is difficult to use as it as a battery material. Therefore, a use, a further improvement is also required.
In this study, we report the investigation of the bulk structural
⁎ Corresponding author at: Silicone-Electronics Materials Research Center, Shin-Etsu
changes accompanying the charge and discharge long-term cycling of
Chemical Co., Ltd, Annaka, Gunma 379-0125, Japan. the SiO material using the Solid-state NMR. It suggests a way of improv-
E-mail address: t_hirose@shinetsu.jp (T. Hirose). ing the cycling using the SiO material.

http://dx.doi.org/10.1016/j.ssi.2017.03.004
0167-2738/© 2017 Elsevier B.V. All rights reserved.
T. Hirose et al. / Solid State Ionics 303 (2017) 154–160 155

2. Experimental the electrodes was 4.3–2.5 V or 3.0 V. The discharge was terminated at
0.07C.
2.1. Production of SiO-C For the second and subseguent cycles, the rates of the charge and
discharge were 0.7C and 0.5C, respectively. The C rate was used in the
The mixture of silicon (Shin-Etsu Chemical Co., Ltd, Purity 99%, Size range of 4.3–2.5 V.
D50 = 10 μm) and silicon dioxide (Shin-Etsu Chemical Co., Ltd, Purity
99.9%, Size D50 = 10 μm) was vaporized in the reactor at 10 Pa and 2.4. Analysis and evaluation conditions
1340 °C, then cooled and accumulated on the plate inside as silicon-
mono-oxide (SiO). The SiO bulk was pulverized by milling to make For evaluation of the anode using metal Li, the state of the anode ma-
the SiO powder with the median particule size around 5 μm. The SiO terial in the charging process and discharging process was investigated
powder was then processed for 3 h in methane at ca. 1000 °C to be coat- by 29Si-magic angle spinning (MAS)-NMR, 7Li-MAS-NMR, and laser-
ed with a conductive carbon layer of ca. 50-nm thickness on the surface assisted atom probe tomography (La-APT).
of the SiO powder (SiO-C). For the MAS-NMR, each anode was removed in a GB (glove box) and
washed with DMC. The active material was separated from the elec-
trode and packed into the rotor. For reference, Li4SiO4 was measured.
2.2. Production of electrodes using the SiO-C and Si materials
The measuring equipment was a Bruker Avance700 operating at the
observation frequencies of 272 MHz (Li) and 139 MHz (Si). The mea-
The SiO-C was mixed with the polyamic acid (UBE INDUSTRIES, LTD,
surement method was comprised of a single-pulse irradiation under
U-varnish A) in the weight ratio of 87.5:12.5 to make the anode mixture.
MAS with an irradiation pulse intensity of 30° and relaxation times of
The slurry was then applied to both surfaces of a roughened (Rz =
100 s (Li) and 30 s (Si). The scan numbers were 32 (Li) and 12,000
2 μm) 18-μm-thick current collector using a die head coating device
(Si). LiCl was used as a secondary standard for the chemical shifts, and
and dried in hot air. The electrode density was 2.5 mg/cm2. Finally, it
its shift value was obtained from 1 M LiCl.
was processed for 3 h at 680 °C in an Ar atmosphere. The active material
After the cycling deterioration, the state of the electrodes was ana-
layer became electrically conductive by partial carbonization after
lyzed using 100 Cycle and 419 Cycle electrodes at 2.5–4.3 V and
imidization of the polyamic acid during the process.
100 Cycle and 473 Cycle electrodes at 3.0–4.3 V. The capacity retention
For comparison, an electrode was produced using Si (Kojundo
of the 419 Cycle electrode was 70%. Furthermore, the accumulative ca-
Chemical Laboratory Co., Ltd. Product number: SIE23PB, Purity 99.9%,
pacity of the 473 Cycle was the same as that of the 419 Cycle
Size D50 = 5 μm) by the same method to obtain an electrode with
(2.5–4.3 V). A coin cell was made by combining the extracted electrode
the density of 1.5 mg/cm2.
with Li metal. The charged and discharged electrodes were analyzed.
The capacity and the charge–discharge curve of the anodes were
The discharged electrode was discharged until the voltage became the
measured using a 2032-type coin cell with metallic Li as the counter
same as the open circuit voltage (OCV) 10 min after the end of discharge
electrode and 1.2 M LiPF6, 0.1 M LiBF4 EC:FEC:DMC = 25:5:70 (vol.%)
of the electrode, resulting in an 82.6% initial discharge, to obtain the
as the electrolyte. The evaluation was conducted under the Following
electrode for analysis and the active material.
conditions: current rate of 0.2 mA/cm2, constant current and constant
La-APT was used for the SiO-C material before the charge–discharge
voltage charging, and constant current discharging in the potential
and the 419 Cycle electrode was discharged until the OCV equaled 2.0 V.
range between the electrodes of 0–2.5 V.
The analysis was conducted at the Toshiba Nanoanalysis Corporation.
The measuring equipment included a needle sample preparation appa-
2.3. Production and evaluation of the Al laminate battery ratus (FEI Co.), Helios450 3DAP measuring device (Ametek Co., Ltd.),
and LEAP4000XSi tomograph (laser wavelength: 355 nm). The mea-
For the cyclic performance assessment, a wound-type cell of 500 surement conditions were as follows: 40 pJ pulse energy, 30 K sample
mAh class (4.3 V–2.5 V)packaged in Al laminate cladding was used. temperature, 500 kHz pulse frequency, and 90 mm ion flight distance
The cathode was a mixture of NCA (JFE MINERAL CO.,LTD, Product (Flight Length).
number:503LP), a cathode conductive auxiliary agent, and polyvinylidene The Si area density of 40 atoms/nm3 is a curved surface obtained by
fluoride in the weight ratio of 95:2.5:2.5. connecting the points containing 40 at.% Si in a three-dimensional
The mixture was similarly coated on the anode, pressed by roll space. In particular, at a voxel size (nm) of 0.5, 0.5, 0.5, the portions
pressing, then finally treated at 120 °C in a vacuum for 12 h. This time, are connected where at least 40 at.% Si was detected [19].
an Al cathode current collector with a 20-um thickness was used. The
cathode area density was 23 mg/cm2. 3. Results and discussion
Using two electrodes, secondary batteries was made by the follow-
ing process. First, an aluminum tab was ultrasonically welded to one 3.1. Structural changes during the first charge and discharge
end of the cathode current collector and a nickel tab was welded to
the anode current collector. Subsequently, the cathode, separator, A structural analysis was conducted after the initial charge-
anode, and separator were sequentially laminated to obtain a longitudi- discharge to determine the structural changes before/after the cycling
nally wound type cell. The separator was a laminated film (12-μm thick- deterioration. First, the charge–discharge behavior of the SiO-C material
ness) comprised of porous polyethylene as the main component was investigated.
sandwiched between films containing porous polypropylene as the Fig. 1 shows the charge–discharge curve of the SiO-C half-cell. The
main component. Subsequently it was packaged in an Al laminate film full charging capacity was 2260 mAh/g, and 1866 mAh/g was the fully
and the pre-processed electrolyte was injected through the opening. discharged capacity at 2.5 V. The initial coulombic efficiency in this
Impregnation of the electrolyte was performed under vacuum then case was 82.6%.
sealed. Figures C and D show the charge and discharge curves, respectively.
The electrolyte used for the cycling performance was 1.2 M LiPF6, C100% means a state of full charge and C20%–C90% is the 20% to 90% ca-
0.1 M LiBF4 EC:FEC:DMC = 25:5:70 (vol%) containing 1 wt% VC. pacity when the fully charged state was 100%. Similarly, D20%–D82.6%
The cycling performance was evaluated under the following condi- denotes the 20% to 82.6% of the discharge capacity when the full charge
tions: A 0.2C charge and discharge was performed once at 25 °C under was 100%.
atmospheric conditions. For the constant current–constant voltage Fig. 2-(1) shows the 29Si-MAS-NMR spectra of SiO-C in the non-
charging and constant current discharging, the potential range between discharged state, and those of the SiO-C from the electrodes are
156 T. Hirose et al. / Solid State Ionics 303 (2017) 154–160

Fig. 1. The 1st charge and discharge curve of Li/SiO-C at 0–2.5 V and 25 °C. The 1st
coulombic efficiency was 82.6%. C and D indicate each charge and discharge state,
respectively. C100% denotes a full charge. D20% indicates a 20% discharge state of the
full charge capacity.

described in Fig. 1. The resulting SiO2 area at the 20% charge has signif-
icantly decreased, thus the reaction with SiO2 was preferentially occur-
ring during the initial stage of charging. As there are no major changes in
the Si peak up till 20% charge, the bulk structural contribution to the ini-
tial electrochemical reaction is from the SiO2 area. Subsequently, at the
50% charge along with the increasing Li4SiO4 peak [20,21], the peaks of
Si and SiO2 disappeared. We see that between 20% and 50% charge, Li
becomes inserted in the Si as well as SiO2 areas. That is to say, the Li sil-
icide reaction and Li silicate reaction occur simultaneously. The peak of
the generated Li4SiO4 gradually sharpened to the end of the charging.
Fig. 2-(2) shows the 29Si-MAS-NMR spectra of SiO-C removed from
the electrodes during the discharging process. No significant changes
were observed up to the 50% discharge; however, at the 70% discharge,
a shoulder peak, which is attributed to amorphous Si, was observed
around −70 ppm [22]. Proceeding to higher potentials, at the 82.6% dis-
charge, which induces release of the Li, the peak for Li4SiO4 significantly
Fig. 2. The 29Si-MAS-NMR spectra of SiO-C in non-charged state. Samples for
diminished and a small peak was observed in the range of − 60 to measurement were taken from the electrodes described in Fig. 1.
− 100 ppm centered around − 78 ppm. This result suggests a certain
variety of Si states. The discharge curve shown in Fig. 1 has an inflection
point at ca. 0.7 V. At a lower potential level than the inflection point only
a small change in the spectral shape of the Li silicate was observed, sug-
gesting that Li is released from the Li silicide. In contrast to the high po-
tential level above the inflection point, the Li silicate spectrum collapsed
suggesting that Li release from the Li silicate was the dominant reaction.
Fig. 3 shows the 7Li-MAS-NMR spectra of the anode material taken
from the electrodes during the charging process as described in Fig. 1.
The fact that SiOx-C shows a substantial shift to a low magnetic field
during the charging and presents a peak as low as 0 ppm indicates that
many Li compound states exist. The peak around 0 ppm is considered to
be Li silicate [20]. The peak shift to the low magnetic field from the ini-
tial stage of discharge moved to a higher magnetic field that during the
middle stage of discharge, a shift was observed in the area close to
0 ppm. Although the peak of Li silicate at D82.6% as described in
Fig. 2-(2) disappeared, the peak corresponding to the 7Li NMR spectrum
is observed. This suggests that Li silicate was not decomposed, but the
regular directional state of the Li silicate changed to a non-directional
state.
The following discussion relates to the shift in the low magnetic field Fig. 3. The 7Li-MAS-NMR spectra of SiO-C measured using the electrodes described in
area during the charging process. When the way in which Si and SiO2 Fig. 1.
T. Hirose et al. / Solid State Ionics 303 (2017) 154–160 157

react with Li is considered, the peaks present at the low magnetic field change during charging using X-ray photoelectron spectroscopy (XPS)
are assumed to be due to Li silicide. [24]. In that report, Si 0+, which contributes to the Li silicide at the ini-
Therefore, we further examined the reaction of the Si material and Li. tial full charge, showed the strongest peak, but the spectra also sug-
The 1st coulombic efficiency was 92.3%. (2) The 7Li-MAS-NMR spec- gested the existence of compounds with oxidized silicon states (Si1+
tra of Si measured using the electrodes described in Fig. 4-(1). to 4 +). Considering the XPS results, the peak shift in the vicinity of
Fig. 4-(1) shows the charge–discharge curve of a half-cell using the 47 ppm could then be assigned to compounds with Si oxidized to low
Si material. The capacity at the time of full charging was 3580 mAh/g, valence states (Si1+ to 3+).
whereas during the 2.5 V discharge, it became 3304 mAh/g. The initial These results suggest that the reaction mechanisms of SiO-C and Si
coulombic efficiency of this sample was 92.3%. with Li are significantly different.
The results obtained by subjecting the Si material to the same 7Li
NMR analysis are shown in Fig. 4-(2). During the initial period of dis- 3.2. Peak shift and theoretical calculation
charge (C10%), multiple states exist, indicating that the Li diffusion in
the bulk is nonhomogeneous. It had been reported that in this case, Thus, how the battery characteristics are influenced by this phenom-
the peak near 18.5 ppm is Li12Si7, and that near 0 ppm is diamagnetic enon will be evaluated.
Li [23]. The σ shielding constant related to the NMR chemical shift values
From 50% to 100% charge, a shift toward a high magnetic field is ob- has been formulated by Ramsey [25]. This formula gives a shielding con-
served. These results are different from those obtained for SiO-C. More- stant concerning the chemical shift of a non-magnetic insulator (includ-
over, since at C100%, there is a peak at a higher magnetic field ing a diamagnetic term and a paramagnetic term), and by adding the
(− 4 ppm), this may indicate that part of the Si-Li compound is ionic. Knight shift term due to the conduction electrons, Formula (1) is obtain-
This Si-Li reaction revealed by 7Li-MAS-NMR is similar to that reported ed [26].
in the literature [23]. *   +
e2  
As SiO-C is being charged, there is a corresponding peak shift to the  −1 
σ¼ 0  ∑ r 0
3mc2  j j

lower magnetic field, and at 100% charge the maximum peak is located  2 (*   +*   + *   +*   +)
       
at 47 ppm. Since at 20% charge Li4SiO4 is created in the vicinity of 0 ppm, 2 eℏ 1        
− ∑ 0∑ L j n n∑ L j r −3 0 þ 0∑ L j r −3 n n∑ L j 0
3 2mc n En −E0  j   j   j j   j 
we speculate that the peak in the vicinity of 47 ppm would be Li silicide. 8π
However, Fig. 4-(2) indicates, the maximum value of the Si-Li com- − χ se ρð0ÞE F
3
pound in 7Li-NMRis near 18 ppm, which could not be explained by ð1Þ
the Li silicide reaction. Miyachi et al. have reported on the Si valence
Furthermore, the 1.2 term of Formula (1) can be approximated by
Eq. (2) [27,28].
   
e2 1 2e2 ℏ2 2 1 1
−λ ap ð2Þ
3mc2 r 3m2 c2 ΔE r 3 p

where the first term shows the effect of shielding by the electron densi-
ty, and when considering values of 17.8 ppm for a hydrogen atom per
unit electron density and 25 ppm for a carbon atom [29], it is presum-
ably ca. 20 ppm for the lithium atoms. In other words, when the charge
density of the lithium ions changes by 0.1, the chemical shift experi-
ences a deviation of ca. 2 ppm. However, in practice, even if the ionic
bonding power of the lithium ions is weakened and the covalent bond-
ing nature becomes stronger, electrons flowing to the cation side as-
sume the role of covalent bonding electrons, causing a slight
distancing from the nucleus. Thus, a change in the chemical shift de-
pending on b1/rN is believed to be smaller than the above-mentioned
value.
The second term indicates the degree of covalency.
Assuming ΔE of about 8 eV [28] and using Eq. (3), the shift range is
ca. 5 ppm. Here, the Bohr radius was used for ao [30].
 
1 0:0387
¼ ð3Þ
r3 p a0 3

It should be noted that the contribution of the first term and second
term is in the order of a few ppm at most, and thus the contribution in
the 7Li NMR is small.
In fact, however, an ca. 48 ppm shift to the low magnetic field occurs
at C100% for SiO-C. The cause of this large peak shift is presumably due
to the dominance of the Knight shift from the conduction electrons of
the third term. The shift toward the low magnetic field indicates that
the electron conductivity is improved.
These results indicated that the electronic resistance of SiO-C de-
creases as the charging progresses. In contrast to charging, the shift oc-
curs toward the high magnetic field during discharging, suggesting that
the electronic resistance increases, particularly beyond D50%. After the
Fig. 4. (1) The 1st charge and discharge curve of Li/Si at 0–2.5 V and 25 °C. middle stage of the charging, the resistance is also likely to increase in
158 T. Hirose et al. / Solid State Ionics 303 (2017) 154–160

the Si material. Furthermore, at C10%, peaks appear in the low magnetic The charge–discharge curves of the designed laminate battery using
field and in the vicinity of 0 ppm, suggesting the co-existence of a region the conditions stated in Fig. 5-(1) are shown in Fig. 5-(2). The solid line
with a high electron conductivity and a portion of high resistance, as represents the cut-off at 3.0 V and the dotted line at 2.5 V. Furthermore,
well as indicating a distribution in the Li concentration. the battery charge cut-off potential was 4.3 V.
These results indicated that the cell characteristics can be further im- Fig. 6-(1) shows the capacity retention rate of the laminate battery.
proved using SiO-C as the battery material and applying it in the poten- At this time, the charge and discharge rate, because it is correlated
tial range at D50% or less, which exhibits a lower electronic resistance. with the current density per electrode area, is the C rate calculated
To date, the structural changes occurring during the initial charge from the capacitance obtained at the 2.5 V cut-off (1C = 500 mAh)
and discharge have been described. and also in the batteries during the 3.0–4.3 V cycling. The charge and
As already mentioned, by using SiO-C that exhibits peaks in the low discharge were conducted at the same current density.
magnetic field, i.e., in the low potential region, it becomes possible to In the case of performing cycles in the 3.0–4.3 V range, a cycling eval-
use this material with a high conductivity. uation was conducted until 473 Cycle, but was aligned with the inte-
In this respect, the relationship between the potential range of the grated capacity at the time of 419 Cycle for the 2.5–4.3 V cycles. In
anode application and the degradation behavior has been verified other words, the amounts of Li desorbed from the anode material
using a prototype lab-fabricated laminate cell. were similar.
Based on an evaluation of the cycling characteristics, the results con-
firmed that the 3.0–4.3 V cycling shows a high capacity retention rate.
3.3. Structural changes after charge and discharge This is consistent with the results inferred from the solid-state NMR
experiments.
Fig. 5-(1) shows the simulated curves of the positive and negative In order to determine the phenomenon of deterioration, the anode
electrodes used in the cell design. A half-cell charge and discharge was removed from the battery after completion of the cycling evalua-
were performed with the anode and cathode, and the simulated curves tion and the structure of the active material was investigated.
were calculated from the results. At the battery cut-off voltages of 3.0 V Accordingly, the batteries obtained aften each cycle, as shown in
and 2.5 V, the anode potentials were 0.66 V and 1.1 V, respectively, close Fig. 6-(2) (100 Cycle, 419 Cycle and 473 Cycle), were dismantled, a
to those at D50% and D70%, as shown in Fig. 1.

Fig. 6. (1) Discharge capacity of the laminated batteries using NCA/SiO-C electrodes at 2.5
Fig. 5. (1) The simulated discharge curve based on the possible anode/cathode or 3.0 to 4.3 V. The bold line denotes the 3.0 V cut-off and the other line is at 2.5 V. (2) The
combination and conditions. (2) Charge-discharge curves of the laminated batteries 29Si-MAS-NMR spectra of SiO-C samples at several cycle points in 2.0 V/OCV same leveled
using NCA/SiO-C electrode at 2.5 or 3.0 to 4.3 V at 25 °C and 0.2C. by reassembling with half coin cells.
T. Hirose et al. / Solid State Ionics 303 (2017) 154–160 159

half-cell was assembled, and a discharge was conducted until the OCV determine the extent of the disproportionation, the structure was ana-
became equal to that at D82.6%, then the 29Si-MAS-NMR spectra were lyzed before and after the cycling deterioration using the atom probe
acquired, as illustrated in Fig. 2. The reason for the OCV alignment was method. La-APT is an analytical tool for understanding nanometer-
to take into account the extent of the load deterioration of the degraded scaled structures in the SiO-C materials at the atomic level. Before the
electrode. The results showed that for the electrode discharged in Cycle La-APT measurements using a LEAP4000XSi (Ametek Co.) system,
1 (D82.6%), the peak for Li silicate disappears and an indistinct peak ap- needle-like specimens were fabricated from the SiO-C material by a fo-
pears in the Si area, whereas for the 100 Cycle (2.5–4.3 V), the peak for Li cused ion beam (FIB) machine. In this case, the apex curvatures of
silicate remains even after discharging to the same OCV. This suggests sharp-needle specimens were less than a few tens of nanometers. Dur-
that by performing charging and discharging, the Li silicate assumes a ing measurement, each atom was field-evaporated in the direction nor-
higher regularity. As the charge-discharge cycle is continued, the mal to the hemispherical surface on the tip of the needle-like specimen
Li4SiO4 peak becomes sharp in the 419th cycle, and the shoulder of upon the application of an extremely high electric field. These field-
the amorphous-Si peak at −70 ppm becomes even smaller. By repeat- evaporated atoms were detected by a position-sensitive detector, thus
edly charging and discharging in the 2.5–4.3 V range, it seems that enabling us to retrace their original locations on the needle tip. At the
Li4SiO4 has lost some of its reactivity for charging and discharging to same time, the time of flight for each evaporated atom was recorded
the initial capacity. to assign the element. After the measurements, the IVAS software
Conversely, in the case of charging and discharging in the voltage (Ametek Co.) was used to reconstruct a 3D image of the needle tip
range of 3.0–4.3 V for the 100 Cycle, the regularity of the Li silicate de- from the raw data, and to analyze the detailed structures. The
creases and the peak for amorphous Si appears. isoconcentration-surface application in IVAS can display regions with
When the cathode undergoes repeated charge and discharge in the high concentrations of a chosen element in 3D image.
low potential region, it is more difficult for Li4SiO4 to form, suggesting Fig. 8 shows an illustration obtained by connecting the portions
that amorphous-Si is active. After 473 cycles, the Li4SiO4 peak is stronger where at least 40 at% Si atoms were detected inside a cube of 0.5 nm
and sharper. These results are similar to the spectra obtained after per side. Based on this result, it was confirmed that before the charge
100 cycles at 2.5–4.3 V. Therefore, it can be speculated that not only and discharge, Si is randomly scattered into islands; however, by re-
the structure of the active material remains similar, but the degradation peating the charge and discharge, the scattered Si islands become con-
of the charge-discharge cycle is also delayed. solidated. Thus, the disproportionation phenomenon of the separation
We investigated the phenomenon responsible for the regularity of into Si and other components is confirmed.
the Li silicate observed in the solid-state NMR experiments. Fig. 7
shows the discharge curves before and after the cycling evaluation. 4. Conclusions
Upon repeated cycling, SiO-C shows a curve behavior similar to that of
Si. In order to achieve an improvement in the cycling characteristics of
SiO-C as a high-capacity anode material, the changes in the bulk struc-
3.4. Visualization of the bulk structure using the La-APT method ture due to the insertion and release of Li were investigated by solid-
state NMR.
During repeated charge and discharge, the effect of disproportion- It has been suggested that SiO-C reacts with Li derived from the SiO2
ation reaction is tentatively attributed to the separation of Si and other regions to generate Li4SiO4.·The resulting Li4SiO4 was shown to mainly
components (in this case, presumably Li silicate). release Li at the high potential side.
The disproportionation of SiO-C is due to phase separation into Si It has been further presumed that the electron conductivity of SiO-C
and SiO2 by application of thermal energy. At this time, since the crystal- improves with the progress of charging, involving a decrease in the
linity of Si is enhanced by thermal energy, growth of the Si grains can be resistivity.
checked by a TEM analysis, and the structure can be easily determined
[16].
Conversely, because the amorphization of Si progresses with the in-
sertion and release of Li, the analysis is difficult. Therefore, in order to

Fig. 7. The discharge curve of Li/Si (1st) anode and Li/SiO-C (1st and after 419 Cycle) anode
at 0–2.5 V and 25 °C. The bold line denotes the 1st discharge curve of SiO-C. The dotted line
denotes the 419 Cycle discharge curve of SiO-C. The dashed line denotes the 1st discharge Fig. 8. Si atomical proof method (40 atoms/nm3). Domain size: 20 nm × 20 nm × 50 nm.
curve of Si. Voxel size: 0.5 nm × 0.5 nm × 0.5 nm.
160 T. Hirose et al. / Solid State Ionics 303 (2017) 154–160

Upon repeated battery charge–discharge cycling, the occurrence of [9] T.D. Besenhard, J. Yang, M. Winter, J. Power Sources 68 (1997) 87–90.
[10] C.J. Wen, R.A. Huggins, J. Solid State Chem. 37 (1976) 271–278.
the disproportionation reaction has been suggested to occur by separa- [11] A. Hohl, T. Wieder, P.A. van Aken, T.E. Weirich, G. Denninger, M. Vidal, S. Oswald, C.
tion into Si and Li4SiO4. Deneke, J. Mayer, H. Fuess, J. Non-Cryst. Solids 320 (2003) 255.
The results of the battery charge–discharge tests confirmed that the [12] Y. Ohzuku, H. Tomura, K. Sawai, J. Electrochem. Soc. 151 (2004) 1572.
[13] M. Yamada, A. Ueda, K. Matsumoto, T. Ozuka, J. Electrochem. Soc. 158 (2011)
cycling characteristics were significantly improved by suppressing the A417–A421.
disproportionation reaction associated with the charge and discharge. [14] Masayuki Yamada, Kazutaka Uchitomi, Atsushi Ueda, Kazunobu Matsumoto,
This was made possible by the intelligent battery design, and its applica- Tsutomu Ohzuku, J. Power Sources 225 (2013) 221–225.
[15] V. Kapaklis, J. Non-Cryst. Solids 354 (2008) 612.
tion in a range of anode discharge potentials up to 0.66 V. [16] Mikito Mamiya, Humihiko Takei, Masae Kikuchi, Chiaki Uyeda, J. Cryst. Growth 229
(2001) 457.
Acknowledgment [17] Toru Tabuchi, Hideo Yasuda, Masanori Yamachi, J. Power Sources 146 (2005)
507–509.
[18] Hye Jin Kim, Sunghun Choi, Seung Jong Lee, Myung Won Seo, Jae Goo Lee, Erhan
We would like to express our gratitude to Chinami Matsui, general Deniz, Yong Ju Lee, Eun Kyung Kim, Jang Wook Choi, American Chemical Society,
manager of the Analytical Technology Group of Shin-Etsu Chemical Nano Lett. 16 (2016) 282–288.
Co., Ltd., and Reiko Sakai, assistant at the Silicone-Electronics Materials [19] T. Sasaki, S. Ishimura, M. Kato, H. Akutsu, M. Tomita, T. Kinno, H. Uchida, SISS-16,
2016 Abstract.
Research Center for support with the experimentations. [20] Taeahn Kim, Sangjin Park, Seung M. Oh, J. Electrochem. Soc. 154 (2007) A1112–A1117.
[21] Masanari Takahashi, Hiroshi Toyuki, Masahiro Tatsumisago, Tsutomu Minami, Solid
References State Ionics 86-88 (1996) 535–538.
[22] W.L. Shao, J. Shinar, B.C. Gerstein, Phys. Rev. B 41 (1993) (Number 13).
[1] Y. Idota, Tadahiko Kubota, Akihiro Matsufuji, Yukio Maekawa, Tsutomu Miyasaka, [23] Baris Key, Rangeet Bhattacharyya, Mathieu Morcrette, Vincent Seznec, Jean-Marie
Science 276 (1997) 1395–1397. Tarascon, Clare P. Grey, J. Am. Chem. Soc. 131 (2009) 9239–9249.
[2] H. Inoue, T. Takata, Y. Kudo, Electrochemistry 76 (2008) 358. [24] Mariko Miyachi, Horonori Yamamoto, Hidemasa Kawai, J. Electrochem. Soc. 154
[3] A.D.W. Todd, R.E. Mar, J.R. Dahn, J. Electrochem. Soc. 154 (2007) A597. (2007) A376–A380.
[4] A.M. Wilson, J.R. Dahn, J. Electrochem. Soc. 142 (2) (1995) 326–332. [25] N.F. Ramsey, Phys. Rev. 78 (1950) 699.
[5] J. Yin, M. Wada, K. Yamamoto, Y. Kitano, S. Tanase, T. Sakai, J. Electrochem. Soc. 153 [26] A. Carrington, A.D. Mclachlan, Magnetic Resonance of Chemists, 1970 (in Japanese).
(3) (2006) A472–A477. [27] T. Kanda, J. Phys. Soc. Jpn. 10 (1955) 85.
[6] G.X. Wang, L. Sun, D.H. Bradhurst, S. Zhong, S.X. Dou, H.K. Liu, J. Power Sources 88 [28] K. Yoshida, T. Moriya, J. Phys. Soc. 11 (1956) 33.
(2000) 278–281. [29] Yonezawa, Nagata, Katou, Imamura, Morokuma, Quantum chemistry introduction
[7] K. Candace, Hailin Peng Chan, Gao Liu, Kevin McIlwrath, Xiao Feng Zhang, Robert A. (last volume), Chemistry Coterie, 1983 (in Japanese).
Huggins, Yi Cui, Nat. Nanotechnol. 3 (2008) 31–35. [30] R.G. Barnes, W.V. Smith, Phys. Rev. 93 (1954) 95.
[8] J.O. Besenhard, P. Komenda, M. Josowiez, Solid State Ionics 18&19 (1986) 823.

You might also like