Measurement of concentration profiles during electrodeposition of Li metal from LiP F6 -PC electrolyte solution

Article abstract of DOI:10.1149/1.2767404

During Li metal electrodeposition from a 0.5 M LiP F6 -PC electrolyte solution onto a horizontal Li metal electrode, the refractive index profile corresponding to the concentration profile of Li+ ion near the cathode was measured in situ by holographic interferometry. The Li+ concentration gradient around the rapidly growing dendrite arms is steeper than at the cathode plane, clearly reflecting the local current density convergence at the dendrite tips and arms. As in a LiCl O4 -PC solution, an incubation period was observed between the start of current passage and the onset of the refractive index fringe shift. It increases with decreasing applied current density. The incubation period in LiP F6 -PC is shorter than that in LiCl O4 -PC at current densities greater than 1.0 mA cm-2; however, at 0.5 mA cm-2 in LiP F6 -PC electrolyte solution it is appreciably longer than in LiCl O4 -PC. This complicated behavior is apparently due to the solution chemistry of LiP F6 -PC electrolyte, which produces HF and oxyfluoride impurities that are lacking in the LiCl O4 -PC system. Thus, the different current dependence of the incubation time in these two systems may yield clues for an elucidation of the dynamics of solid electrolyte interphase formation.

Full text of DOI:10.1149/1.2767404

Journal of The Electrochemical Society, 154 ͑10͒ A943-A948 ͑2007͒
A943
0013-4651/2007/154͑10͒/A943/6/$20.00 © The Electrochemical Society
Measurement of Concentration Profiles during Electrodeposition
of Li Metal from LiPF6-PC Electrolyte Solution
The Role of SEI Dynamics
K. Nishikawa,a, Y. Fukunaka,
and
*
a,**,z
b,**
T. Sakka,
b,**
Y. H. Ogata,
c,**
J. R. Selman
aGraduate School of Energy Science, Kyoto University, Sakyoku Kyoto, Japan
bInstitute of Advanced Energy, Kyoto University, Gokasyo, Uji, Japan
cCenter for Electrochemical Science and Engineering, Illinois Institute of Technology, Chicago,
Illinois 60616, USA
During Li metal electrodeposition from a 0.5 M LiPF6-PC electrolyte solution onto a horizontal Li metal electrode, the refractive
index profile corresponding to the concentration profile of Li+ ion near the cathode was measured in situ by holographic interfer-
ometry. The Li+ concentration gradient around the rapidly growing dendrite arms is steeper than at the cathode plane, clearly
reflecting the local current density convergence at the dendrite tips and arms. As in a LiClO4-PC solution, an incubation period
was observed between the start of current passage and the onset of the refractive index fringe shift. It increases with decreasing
applied current density. The incubation period in LiPF6-PC is shorter than that in LiClO4-PC at current densities greater than
1.0 mA cm−2; however, at 0.5 mA cm−2 in LiPF6-PC electrolyte solution it is appreciably longer than in LiClO4-PC. This
complicated behavior is apparently due to the solution chemistry of LiPF6-PC electrolyte, which produces HF and oxyfluoride
impurities that are lacking in the LiClO4-PC system. Thus, the different current dependence of the incubation time in these two
systems may yield clues for an elucidation of the dynamics of solid electrolyte interphase formation.
© 2007 The Electrochemical Society. ͓DOI: 10.1149/1.2767404͔ All rights reserved.
Manuscript submitted February 27, 2007; revised manuscript received May 1, 2007. Available electronically August 10, 2007.
Li-ion batteries are widely used as power sources for small-size
electronic devices. It is expected that, if safety issues can be re-
solved, they will also become a power supply for hybrid electric
vehicles ͑HEVs͒ and purely electric vehicles ͑EVs͒. The latter tech-
nologies are considered essential options in the worldwide effort to
reduce the production of greenhouse gases and thereby contribute to
the solution of a key global environmental problem. The develop-
ment of electric vehicles powered by advanced secondary batteries
or fuel cells has been proposed by some prominent automobile com-
panies as the most promising strategy to ensure an optimal power
train technology for the future.
At present, graphite and graphitic materials are predominant as
negative electrode ͑NE͒ materials for the Li-ion battery. However,
the capacity of this type of anode has almost reached the theoretical
limit. Therefore, it is necessary to develop an alternative negative
electrode material. Many researchers are focusing on Si alloys and
Sn alloys as candidate NE materials.1-6 Dahn et al. have discussed
the electrochemical properties of SiM ͑M = Fe, Sn, Zn͒, alloys as
possible NE materials.1-3 Thackeray et al. investigated the possibil-
ity of using Cu-Sn alloy prepared by heat-treatment as NE
material.4-6
model of the SEI between Li metal and some kind of electrolyte
based on results of impedance spectroscopy.13-15 Kanamura et al.
undertook an in situ X-ray photoemission spectroscopy ͑XPS͒
analysis of the composition of the surface layer in some kind of
electrolyte, and reported that HF addition to the electrolyte improves
the flatness of the electrode surface.16-18
In related work we have shown that holographic interferometry
is a very powerful technique to determine and analyze the concen-
tration profile in the electrolyte during electrodeposition and electro-
chemical dissolution of Li metal in 0.5 M LiClO4-PC electrolyte
solution.19-21 A remarkable result of this research is that in situ ho-
lographic interferometry clearly indicates the occurrence of an incu-
bation period in the initial stage of electrodeposition when the cur-
rent density is small ͑less than 2.0 mA cm−2͒.19,20 We have proposed
an explanation that links the incubation period to the formation or
modification of an SEI layer, and we have provided a model quan-
tifying this explanation. It is based on the assumption that impurities
in the electrolyte, and among these mainly H2O, interact preferen-
tially with the active Li metal, thereby causing an impurity mass-
transfer limitation that shows up as a current-density dependence of
incubation period in the fringe shift.
The occurrence of an incubation period has thus far been ob-
served only in LiClO4-PC electrolyte solution. In this study, we
investigate the effect of electrolyte chemistry on this phenomenon.
The same organic solvent ͑PC͒ but a different Li-salt ͑LiPF6͒ is
used. There are several reasons for this choice. The LiPF6-PC is
closer to Li-ion battery practice. Also, the presence of fluorine in the
anion may lead to a different SEI layer composition, for example
containing mainly LiF as an outer skin, which helps keep the Li-
electrode macroscopically smoother. The effect of HF addition as
mentioned above16-18 is significant in this respect. Finally, the choice
of LiPF6, which appears to generate HF as an additional impurity
when H2O is present, would contribute further insight into the cou-
pling of mass transfer phenomena in the electrolyte with SEI forma-
tion and Li morphology development.
In spite of all this activity focusing on Li alloys, Li metal itself
remains of interest as an NE material because of its very high energy
density. Its theoretical capacity ͑3760 mAh g−1͒ is about ten times
higher than that of graphite materials. Nevertheless, this material has
not been applied as NE in commercial batteries yet, mainly because
it is difficult to maintain a flat electrode surface after several charge-
discharge cycles due to dendritic growth of occurring at the Li metal
substrate.7-9 This phenomenon causes fatal problems, namely, pro-
gressive loss of reversibility, and eventually short circuit.
The passivating surface layer called SEI ͑solid electrolyte inter-
phase͒ on Li metal in electrolyte solution may play a crucial role in
dendritic Li formation. The SEI layer is believed to be formed as a
result of Li metal contact with the organic electrolyte solution even
before any current passage, but the SEI composition and thickness is
readjusted as soon as current passage starts to deposit Li metal, in an
extremely active form.10-12 Aurbach et al. describe an interfacial
Experimental
The electrolyte solution is 0.5 M LiPF6 dissolved in propylene
carbonate ͑PC͒ purchased from Kishida Kagaku Ltd. The chemical
analysis reported by the supplier indicates an HF concentration of
less than 10 ppm and H2O less than 20 ppm. The working electrode
is Li foil of 200 ␮m thickness, and the effective electrode surface
*
Electrochemical Society Student Member.
Electrochemical Society Active Member.
**
z E-mail: fukunaka@energy.kyoto-u.ac.jp

A944
Journal of The Electrochemical Society, 154 ͑10͒ A943-A948 ͑2007͒
Figure 1. Concentration dependence of refractive index in LiPF6-PC elec-
trolyte solution.
area is 0.6 cm2 ͑1.0 ϫ 0.6 cm͒. The counter electrode and the ref-
erence electrode are Li foils of 200 ␮m thickness, too. Details of the
electrolytic cell for holographic interferometry measurement were
described in previous reports.16-18 The electrolytic cell was as-
sembled in a glove box filled with pure Ar gas circulating continu-
ously through molecular sieves 4 Å 1/16 ͑Nacalai Tesqu, Co. Ltd.͒
to remove water contaminant, and H2O concentration is several tens
ppm. The electrodeposition of Li metal was conducted galvanostati-
cally, at current densities between 0.5 and 5.0 mA cm−2. After ho-
lographic interferometry measurement, the water content was mea-
sured by the Carl-Fischer titration ͑AQ-200, Hiranuma Co. Ltd.͒.
This measurement indicates the water content is about several tens
ppm after interferometry measurement. The concentration depen-
dence of the refractive index of the LiPF6-PC solution was mea-
sured by an Abbe refractometer ͑type PRA-B, Hitachi, Ltd.͒ in a
glove bag filled with pure Ar gas.
Figure 2. Interference fringes image near the Li metal electrode.
Results and Discussion
Holographic interferometry.— The holographic interferometry
technique detects only the refractive index profile. Thus, the concen-
tration dependence of refractive indexes was at first measured by an
Abbe refractometer at room temperature in a glove bag. Figure 1
demonstrates the concentration dependence of the refractive index
of LiPF6-PC solution. Unlike the LiClO4-PC electrolyte,19 the
LiPF6-PC electrolyte has a refractive index that decreases with in-
creasing LiPF6 concentration. From the regression line through the
data, the relationship can be expressed simply as follows
trodeposition. These images demonstrate the transient variation of
interference fringes near the horizontally installed downward-facing
Li cathode at 1.0 mA cm−2. Before electrolysis, the interference
fringe pattern is perpendicular to the horizontal cathode because of
uniform concentration in the electrolyte. Once the electrolysis starts,
the interference fringes near the cathode surface start to shift hori-
zontally. When the applied current density is less than 1.0 mA cm−2
,
the interference fringe starts to shift after a certain incubation pe-
riod, similarly as is the case in the LiClO4-PC electrolyte. This
incubation period becomes shorter with increasing current density.
The LiPF6 system yields a clear image of the interference fringes
not only at the planar deposit but also around dendritic Li deposit. In
a previous publication we pointed out the similarity of the interfer-
ence fringe shifts of LiClO4-PC with those in CuSO4–H2O.20
In the above cases a positive refractive index is observed, which
means that the refractive index is proportional to the concentration
of electrolyte solution. When a positive refractive index profile is
observed adjacent to the cathode, the measured concentration profile
is slightly shifted toward the cathode surface. This occurs because
the incident beam is deflected toward the higher refractive index
field due to the propagation of the wave front within the diffusion
layer. The degree of shift can be calculated by the refractive index
gradient of a measured refractive index profile.
n = 1.422 − 0.0071 ϫ C
͓1͔
where n is the refractive index and C is the LiPF6 concentration.
The basic interferometry equation,22 combined with this relation-
ship, yields the concentration profile near the electrode surface. It is
expressed by
S
⌬C = 0.089 ϫ
͓2͔
d
where the concentration difference with respect to bulk concentra-
tion is ⌬C, the shift number of the interference fringe is S, and d is
the optical path length ͑here 6 mm͒.
The planar Li metal working electrode is facing downward hori-
zontally, in order to suppress as much as possible any natural con-
vection caused by the concentration difference generated
during electrodeposition. The constant cathodic current density
͑0.5 to 5.0 mA cm−2͒ was applied as a step function. Figure 2
shows photographs of the interference fringe pattern during elec-
In the present case, the fringe shift is different from that in both
of the above-mentioned systems, because the concentration depen-
dence of the refractive index is negative for LiPF6-PC. In this case,

Journal of The Electrochemical Society, 154 ͑10͒ A943-A948 ͑2007͒
A945
Figure 4. Transient variation of concentration profile formed by the elec-
trodeposition of Li metal in LiPF6-PC at 1.0 mA cm−2
Figure 3. Interference fringe pattern and real image of Li cathode.
.
the direction of deflection of the laser beam passing through the
transparent electrolyte with refractive index gradient perpendicular
to the electrode surface is the opposite of that in the LiClO4-PC and
CuSO4–H2O systems.
Transient variations of surface concentration of Liϩ ion and of
boundary layer thickness.— The solid line in Fig. 4a represents the
surface concentration predicted by the transient one-dimensional
diffusion equation
Previous discussion of the effect of optical deflection19 on the
concentration profile indicates that, in the case of a negative refrac-
tive index gradient, the laser beam directed into the diffusion layer
deflects gradually toward the electrode surface. This means that the
refractive index cannot be measured at the cathode surface. That is,
in the case of LiPF6-PC, the concentration values apparently mea-
sured at the electrode surface may be significantly smaller than the
actual surface concentration. In such a case, the concentration pro-
file must be outward shifted to the bulk electrolyte.
ץC1
ץt
ץ2C1
ͩ ͪ
ץx2
= D
͓3͔
with the initial condition
and boundary conditions
C1͑x,0͒ = C01
͓4͔
ץC1͑x,t͒
i͑1 − t*͒
D
= −
at x = 0
͓5͔
͓6͔
ͩ ͪ
Morphological variations.— Moreover, the surface morphology
of the electrodeposited Li metal appears, even when it is actually
rough, smoother than is the case in the LiClO4-PC system. This is
because any roughness does not interfere with the propagation of the
laser beam near the cathode surface, or at least less so than in a
system with positive concentration dependence of the refractive in-
dex. The onset of dendritic growth, therefore, can be distinguished
much more sharply. Figure 3 demonstrates this, showing the time
evolution of the image ͑real͒ of the electrode surface when dendritic
growth occurs during electrodeposition of Li metal.
It is observed that Li dendrite growth in some cases resembles a
swinging pendulum. Yamaki et al. reported a similar movement of
growing Li dendrite in LiClO4-PC electrolyte.23 Further investiga-
tion of dendritic growth of Li metal is now in progress in our labo-
ratory, using in situ scanning confocal laser microscopy.
ץx
zF
C1͑ϱ,t͒ = C01
Here, C1 is the concentration of Li+ ion, D is the diffusion coeffi-
cient of Li+ in LiPF6-PC solution, i is the current density, t* is the
Li+ transference number,24 z is the Li+ valence, F is the Faraday
constant, and C01 is the bulk concentration of Li+ ion ͑0.5 M͒.
Figure 4a shows the surface concentration Cs1 calculated by ap-
plying Eq. 2 to the measured interference fringe shift, while Fig. 4b
demonstrates the development of the concentration boundary layer
thickness ␦ at 1.0 mA cm−2. The open circles represent the surface
concentration Cs1 measured around the growing Li dendrite, and the
triangles show Cs1 at the electrode surface without dendrite growth.
It is evident that the interference fringes do not shift immediately

A946
Journal of The Electrochemical Society, 154 ͑10͒ A943-A948 ͑2007͒
Figure 6. Relationship between the incubation period and applied current
density.
trodeposition in 0.5 M LiPF6-PC solution. Once the interference
fringes start to shift, ␦ grows roughly proportional to the square root
of time at either current density. However, the t1/2 regression lines
clearly do not go through the origin as the transient diffusion model
predicts. Intersecting the t1/2 regression lines with the abscissa ͑time
axis͒ in either figure defines an incubation period ͑␶͒, specific for
that current density. Similarly to that in LiClO4-PC solution, the
incubation period increases with decreasing applied current density
͑Fig. 6͒.
Current density dependence of incubation period.— At current
densities greater than 1.0 mA cm−2, the incubation period in
LiPF6-PC is shorter than in LiClO4-PC. However, at 0.5 mA cm−2
in LiPF6-PC it is considerably longer than in LiClO4-PC. The gen-
erality of this behavior at lower current density must be further
confirmed. Nevertheless, this complicated dependence of incubation
period on current density is not unexpected. It may be understood
from the different composition and properties of the SEI formed in
electrolyte solutions having a different solution chemistry.
Figure 5. Transient variation of concentration profile formed by the elec-
trodeposition of Li metal in LiPF6-PC at 2.0 mA cm−2
.
Incubation period in LiClO4-PC system.— In earlier work19,20
we have explained the incubation phenomenon in the case of
LiClO4-PC solution by simultaneous diffusion of Li+ ions and H2O
molecules which are always present as a contaminant in the organic
solution. The chemical reaction of the electrodeposited Li metal
with H2O and other reducible contaminants results in an SEI
layer7-18 through which transport of Li+ ion is established. In the
electrolyte solution near the electrode, this causes diffusion-limited
H2O transport to the SEI, in tandem with further reduction of Li+
ions migrating toward the electrode. A steady state of SEI composi-
tion and thickness is eventually established, controlled by the rate of
H2O diffusion. According to this simple model, the incubation pe-
riod is directly linked to the bulk concentration of H2O. In properly
prepared LiClO4-PC this concentration is slight, and the relative
stability of the ClO−4 anion does not give rise to reactive decompo-
sition products interacting with H2O except within the confines of
the SEI, upon current passage.
after the electrolysis starts. This incubation period is similar to that
noticed in the initial stage of the electrodeposition of Li metal in
LiClO4-PC electrolyte. At all times the concentration difference ⌬C
around the Li dendrite appears to be larger than that at the area
apparently without dendrite.
The transient surface concentration measured near the Li den-
drite apparently agrees with the predicted 25 s after the start of
electrodeposition. Strictly speaking, a deviation from the predicted
values would be expected anyway because the one-dimensional
model Eqs. 3-6 can be applied only in the case of uniform current
distribution at a planar surface. The deviations from prediction for
the “around dendrite” data in Fig. 4a seem quite compatible with a
much increased local current density once the dendrites start grow-
ing.
Figure 5 shows the transient surface concentration Cs1 and con-
centration boundary layer thickness ␦ measured at 2.0 mA cm−2. In
this case, the incubation period is appreciably shorter than that at
1.0 mA cm−2. The surface concentration Cs1 decreases much more
slowly than the model of Eq. 3-6 predicts. This is partly due to the
optical deflection effect, discussed above, which is of course stron-
ger when measuring in a steeper refractive gradient field. Another
reason may be found in the more extensive dendritic growth of Li
metal. The extensive dendritic growth may interfere with precise
observation of the electrode surface location.
Solution chemistry in LiPF6-PC system.— However, in the case
of LiPF6-PC, the solution chemistry is different and leads to mul-
tiple contaminant species because LiPF6 reacts with water easily.
Aurbach et al. proposed the sequence of reactions below, Eq. 7-9,
between LiPF6 and H2O molecule.15 Each of these generates reac-
tive intermediates that may contribute to the SEI but are not neces-
sarily confined to it
Incubation period.— Figures 4b and 5b show the development
of the concentration boundary layer thickness ␦ during Li elec-
Li+ + PF−6 ꢀ LiPF6
͓7͔

Journal of The Electrochemical Society, 154 ͑10͒ A943-A948 ͑2007͒
A947
1. Why at high current density the incubation period for both
LiClO4-PC and LiPF6-PC decreases with increasing current density
level, and does so roughly inversely. Note also that, at high current
density, the incubation period in LiPF6-PC is roughly one-half or
one-third of that in LiClO4-PC. This corresponds reasonably well
with the total level of contaminants ͑H2O + HF͒ in the respective
electrolytes as mentioned in the Experimental section.
2. At low current density the incubation period in LiClO4-PC
becomes quite long as expected, according to ͑1͒. But, the incuba-
tion period in LiPF6-PC becomes even longer, as shown in Fig. 6,
because the current density level is evidently below the critical value
for rapid breakdown of the SEI surface structure.
LiPF6 ꢀ LiF + PF5
͓8͔
͓9͔
PF5 + H2O → PF3O + 2HF
Moreover, it is known that the addition of HF to the LiClO4-PC
electrolyte system reforms the surface layer on Li metal to generate
a LiF layer.16-18 In our work on LiPF6-PC, we found that after the
interferometry measurements, water content increased to about sev-
eral tens ppm. We believe this was caused by the inevitable exposure
of the cell to the atmosphere, in spite of tight sealing. Therefore,
especially in the case of LiPF6-PC, it is possible that during the
experiment the increase of water content contributes to grow LiF
layer in an SEI layer on the Li metal. This is as a consequence that
a more complicated incubation process occurs than at LiClO4-PC. In
the following we try to visualize this process, making particular
assumptions, and explain how it can result in the characteristic dif-
ference in behavior shown in Fig. 6.
SEI dynamics.— Further research is necessary to understand
quantitatively, and in more detail, the coupling between the mass
transfer of chemical species in the electrolyte and the surface mor-
phological variation ͑SEI dynamics, smooth deposition, dendritic
deposition͒ at a Li metal electrode. This study has shown some
connections that may form a useful basis for further work, especially
on the LiPF6-PC electrolyte solution. In any case, it appears that
holographic interferometry can make a perhaps unexpected contri-
bution to understanding SEI dynamics, i.e., SEI formation and
reorganization
Incubation period in LiPF6-PC system.— At the interface be-
tween a Li metal electrode and an electrolyte solution such as
LiClO4-PC, an SEI layer with a dense structure is frequently found
during the electrodeposition of Li metal at a relatively low current
density.12-18 We believe that the dense structure is characteristic for
the SEI when its formation or reorganization is dominated by chemi-
cal processes and not by high-rate Li+ ion transfer to the electrode
͑that is, high current density͒. As proposed in previous papers,19 SEI
formation ͑or rather reorganization͒ upon current passage occurs at a
rate limited by H2O transfer to the electrode, typically by diffusion
and possibly Li+ -coupled migration.
In the LiPF6-PC electrolyte there is the additional possibility that
chemical reactions such as Eq. 7-9 contribute to a LiF-rich surface
layer which decreases the ionic conductivity ͑as well as the porosity͒
and which is rather stable mechanically and chemically. Neverthe-
less, modification ͑reorganization͒ of this layer will occur upon cur-
rent passage, that is, application of a sufficient cell potential. This
requires in our opinion a progressive ͑but rapid͒ breakdown of the
LiF-rich surface structure, for example by migration of the constitu-
ent Li+ ions within the SEI to the SEI/Li interface and reduction to
Li atoms. This means that the very first stage of current passage ͑SEI
reorganization͒ does not require ionic mass transfer in the electro-
lyte. Hence, no fringe shift occurs, initially, in the LiPF6-PC elec-
trolyte until the breakdown process is sufficiently advanced.
Once this initial stage is past, the reorganization and growth of
the SEI occurs similarly as in LiClO4-PC, that is by reaction of
highly active Li atoms formed within or at the SEI by current pas-
sage with H2O or other reducible impurities supplied from the bulk
electrolyte to the SEI as described above. This second stage there-
fore triggers formation of a concentration gradient, though not nec-
essarily a refractive-index gradient ͑as discussed in previous work
on LiClO4-PC19,20͒.
Conclusion
During Li metal electrodeposition from a 0.5 M LiPF6-PC elec-
trolyte solution onto a horizontal Li metal electrode, the refractive
index profile corresponding to the concentration profile of Li+ ion
near the cathode was measured in situ by holographic interferom-
etry. As in a LiClO4-PC solution, an incubation period was observed
between the start of current passage and the onset of the refractive
index fringe shift. Similarly to LiClO4-PC case, the incubation pe-
riod increases with decreasing applied current density. The incuba-
tion period in LiPF6-PC is shorter than that in LiClO4-PC at current
densities greater than 1.0 mA cm−2, however, at 0.5 mA cm−2 in
LiPF6-PC electrolyte solution it is appreciably longer than in
LiClO4-PC. This complicated behavior is apparently due to the so-
lution chemistry of LiPF6-PC electrolyte, which produces HF and
oxyfluoride impurities that are lacking in the LiClO4-PC system.
Thus, the different current dependence of the incubation time in
these two systems may, in principle, yield clues for the elucidation
of SEI dynamics, that is, SEI formation and reorganization. A con-
ceptual model was presented that qualitatively explains the main
incubation characteristics observed in this work.
Growth of dendritic Li metal was also investigated during the
holographic measurements. The Li+ concentration gradient around
the rapidly growing dendrite arms is steeper than at the cathode
plane, clearly reflecting the local current density convergence at the
dendrite tips and arms.
Further experiments must be conducted to understand the inter-
ference fringe behavior when dendritic growth occurs during Li
Effect of current density on incubation period.— The role of the
applied current density level is now seen to be twofold
electrodeposition. This requires
a more complicated, two-
dimensional model than used in this work.
1. If the SEI structure does not have a protective surface struc-
ture, as is the case in LiClO4-PC, or if it has lost that structure by
breakdown, as we assume in 2 below, the current density level de-
termines the rate at which H2O is supplied to the changing SEI, and
thereby the time period needed to reach a steady-state SEI. This time
period may, or may not, coincide with the incubation time of the
fringe shift, depending on the mode of H2O transport to the SEI. In
the case of LiClO4-PC, it does coincide.
Acknowledgments
The authors would like to thank Professor K. Kanamura for his
encouragement and suggestions. Part of this work was carried out
with financial support to Y.F. by the Ministry of Education, Science
and Technology of Japan ͑project no. 15360402͒, which is gratefully
acknowledged.
Kyoto University assisted in meeting the publication costs of this article.
2. If the SEI structure does have a protective surface structure,
as in LiPF6-PC, a critical current density level is needed to break up
that structure rapidly. Below this threshold, both breakdown and
H2O supply are slow, and the duration of the incubation period is the
compounded effect of this.
References
1. L. Y. Beaulieu, K. C. Hewitt, R. L. Turner, A. Bonakdarpour, A. A. Addo, L.
Christensen, K. W. Eberman, L. J. Krause, and J. R. Dahn, J. Electrochem. Soc.,
150, A149 ͑2003͒.
2. T. D. Hatchard and J. R. Dahn, J. Electrochem. Soc., 151, A1749 ͑2004͒.
3. T. D. Hatchard, M. N. Obrovac, and J. R. Dahn, J. Electrochem. Soc., 151, A1749
͑2004͒.
This conceptual framework explains

A948
Journal of The Electrochemical Society, 154 ͑10͒ A943-A948 ͑2007͒
4. K. D. Kepler, J. T. Vaughey, and M. M. Thackeray, J. Power Sources, 81–82, 383
͑1999͒.
5. J. T. Vaughey, K. D. Kepler, R. Benedek, and M. M. Thackeray, Electrochem.
Commun., 1, 517 ͑1999͒.
6. R. Benedek and M. M. Thackeray, J. Power Sources, 110, 406 ͑2002͒.
7. D. Aurbach and Y. Cohen, J. Electrochem. Soc., 143, 3525 ͑1996͒.
8. Z. Takehara, J. Power Sources, 68, 82 ͑1996͒.
9. D. Aurbach, E. Zinigrad, H. Teller, and P. Dan, J. Electrochem. Soc., 147, 1274
͑2000͒.
10. E. Peled, J. Electrochem. Soc., 126, 2047 ͑1979͒.
11. M. Odziemkowski and D. E. Irish, J. Electrochem. Soc., 139, 3063 ͑1992͒.
12. D. Rahner, J. Power Sources, 81–82, 358 ͑1996͒.
13. D. Aurbach and A. Zaban, J. Electroanal. Chem., 348, 155 ͑1993͒.
14. D. Aurbach and A. Zaban, J. Electroanal. Chem., 367, 15 ͑1994͒.
15. A. Zaban, E. Zinigrad, and D. Aurbach, J. Phys. Chem., 100, 3089 ͑1996͒.
16. K. Kanamura, H. Tamura, S. Shiraishi, and Z. Takehara, J. Electrochem. Soc., 142,
340 ͑1995͒.
17. K. Kanamura, S. Shiraishi, and Z. Takehara, J. Electrochem. Soc., 143, 2187
͑1996͒.
18. S. Shiraishi, K. Kanamura, and Z. Takehara, Langmuir, 13, 3542 ͑1997͒.
19. M. Ota, S. Izuo, K. Nishikawa, Y. Fukunaka, E. Kusaka, R. Ishii, and J. R. Selman,
J. Electroanal. Chem., 559, 175 ͑2003͒.
20. K. Nishikawa, M. Ota, S. Izuo, Y. Fukunaka, E. Kusaka, R. Ishii, and J. R. Selman,
J. Solid State Electrochem., 8, 174 ͑2004͒.
21. K. Nishikawa, Y. Fukunaka, T. Sakka, Y. H. Ogata, and J. R. Selman, J. Electroa-
nal. Chem., 584, 63 ͑2005͒.
22. Y. Fukunaka, T. Minegishi, N. Nishioka, and Y. Kondo, J. Electrochem. Soc., 128,
1274 ͑1981͒.
23. J. Yamaki, S. Tobishima, K. Hayashi, K. Saito, Y. Nemoto, and M. Arakawa, J.
Power Sources, 74, 219 ͑1998͒.
24. L. O. Valoen and J. N. Reimers, J. Electrochem. Soc., 152, A882 ͑2005͒.