Sn

EJa

db

c

d

a

ARRAA

KLnSim

1

Ianml[lsgut

c

h0

Electrochimica Acta 137 (2014) 751–757

Contents lists available at ScienceDirect

Electrochimica Acta

j ourna l ho me page: www.elsev ier .com/ locate /e lec tac ta

tudy and modeling of the Solid Electrolyte Interphase behavior onano-silicon anodes by Electrochemical Impedance Spectroscopy

tienne Radvanyia,1, Kristof Van Havenberghb,c,1, Willy Porchera, Séverine Jouanneaua,ean-Sébastien Bridelb, Stijn Putb, Sylvain Frangerd,∗

French Commissary of Atomic and Alternative Energies (CEA), Laboratory of Innovation for New Energy Technologies and Nanomaterials (LITEN), 17 ruees Martyrs, 38054 Grenoble Cedex 9, FranceUmicore Research, Watertorenstraat 33, B-2250 Olen, BelgiumEMAT, University of Antwerp, Groenenborgerlaan 171, B-2020 Antwerp, BelgiumICMMO-ERIEE (UMR CNRS 8182), University Paris Sud, 15 avenue G. Clemenceau F-91405 Orsay, France

r t i c l e i n f o

rticle history:eceived 8 April 2014eceived in revised form 26 May 2014ccepted 12 June 2014vailable online 19 June 2014

eywords:ithium batteriesano-silicon

a b s t r a c t

The instability of the Solid Electrolyte Interphase (SEI) at the surface of nano-silicon electrodes has beenrecognized as one of the key issues to explain the rapid capacity fading of theses electrodes. In this paper,two distinct Si-based systems are studied by using Electrochemical Impedance Spectroscopy (EIS). First,several EIS spectra are recorded along the second electrochemical cycle. Although the active material,the electrode formulation, and the experimental conditions are different for the two systems, the samephenomena are observed in both cases: (i) the SEI deposit around 50 kHz, (ii) the charge transfer (CT) witha characteristic frequency varying from 300 to 1 500 Hz, and (iii) an inductive loop at ∼1 Hz which appearsonly when the potential of the electrode is below 0.35 V vs Li. As the latter has never been reported for

EImpedance spectroscopy

odeling

Si-based electrodes, the second step of the work consists in understanding this phenomenon. Thanks tothe results obtained in a set of several complementary experiments, we finally attribute the inductiveloop to the constant formation/deposition of SEI products, in competition with the CT process. In addition,we propose a mechanism for this specific phenomenon and the equivalent circuit to fit the recorded EISspectra.

© 2014 Elsevier Ltd. All rights reserved.

. Introduction

In the past two decades, the demand for high-capacity Lithium-on Batteries (LIBs) for portable electronics, hybrid electric vehicles,nd large scale energy storage has stimulated the research forew electrode materials [1–3]. Concerning negative electrodes:etals and semi-metals that can electrochemically form alloys with

ithium represent a relevant alternative to carbon-based anodes4,5]. Silicon is considered as one of the leading candidates: the fullyithiated state of silicon at room temperature is Li15Si4 which corre-ponds to a capacity of 3 580 mAh.g−1, ten times larger than that of

raphite [6]. Nevertheless, silicon undergoes huge volume changespon alloying and de-alloying with lithium: up to +300% comparedo the initial volume for a full lithiation [7]. These volume changes

∗ Corresponding author.E-mail address: sylvain.franger@u-psud.fr (S. Franger).

1 The first two authors contribute equally to this work and are both equallyonsidered as first author.

ttp://dx.doi.org/10.1016/j.electacta.2014.06.069013-4686/© 2014 Elsevier Ltd. All rights reserved.

are responsible for strong mechanical strains inside both the sili-con particles and the bulk composite electrode and subsequent lossof electrical contacts between the active material, the conductivematrix and the current collector resulting in poor cyclability forSi-based electrodes [4,5]. To improve these electrochemical per-formances, many studies concerning the silicon material itself [8],the binder [9,10], and/or the formulation [11,12] have been carriedout and the performances have been very much improved in recentyears. As an example, thanks to the use of nanoparticles, Si parti-cles integrity can be maintained even when they are cycled at highcapacities [8,13]. However, despite these promising results, cyclingperformances of Si electrodes, particularly in terms of coulombicefficiency, remain so far unsatisfactory for a use in practical LIBs.

Indeed, one of the strong limitations has been recognized to bedirectly linked to electrode/electrolyte interfaces [14–16]. Uponcycling, the electrolyte standardly made of LiPF6 dissolved in

organic carbonates is reduced and the resulting reduction prod-ucts constitute a Solid Electrolyte Interphase (SEI) at the Si electrodesurface. By using X-ray Photoelectron Spectroscopy (XPS), SEI com-position which forms on Si electrodes has been found to be similar

7 himica Acta 137 (2014) 751–757

tc[imsaae

bt[t

reears

2

2

udliva

2

SPsu1Fse1Srecls(%

MB

2

utsa2

52 E. Radvanyi et al. / Electroc

o that observed on graphite: it is mainly made of lithium alkylarbonates, lithium carbonate, and other lithium inorganic salts14–16]. Nevertheless, volume changes of Si particles can cause anmportant instability of this SEI. At each cycle, SEI reconstruction

ay consume lithium, electrolyte components, and electrons, con-equently contributing to the aforementioned irreversible capacitynd low coulombic efficiency. The behavior of this SEI upon cyclingppears then as one of the key parameters governing the practicallectrochemical performances of Si electrodes.

Electrochemical impedance spectroscopy (EIS) has proven toe a powerful technique to study the different aspects of elec-rode kinetics, interfacial changes and especially the SEI formation17–25]. However, only few studies with detailed EIS characteriza-ions on silicon anodes are available in literature.

As EIS measurements can be very specific to active mate-ials (morphology, surface treatment, etc.) and also depend onxperimental conditions (electrochemical cell, reference electrode,lectrodes alignment. . .), we choose in this paper to carry out EISnalyses on two different systems. For both cases, the active mate-ial, the formulation and the experimental conditions to acquire thepectra are different.

. Experimental

.1. Materials

Two different silicon nanopowders, called here Si1 and Si2, weresed. Si1 is a commercial nano-Si from Umicore, with an averageiameter of 50 nm, a specific surface area of 23 m2.g−1 and an oxide

ayer thickness of ∼3 nm. Si2 is made of 150 nm diameter particles;ts specific surface area is close to 13 m2.g−1 and the oxide layer isery thin (< 1 nm). SEM and TEM characterizations of both materialsre available in ESI.

.2. Electrode fabrication and electrochemical testing

Si1 and Si2 based slurries were used to realize the electrodes.i1-slurry was an aqueous mixture of Si1 silicon (80 wt %), super

(12 wt %), and CMC (8 wt %), whereas Si2-slurry was made of Si2ilicon (65 wt %), carbon fibers (25 wt %) and CMC (10 wt %). Bysing a doctor blade, both slurries were coated at a thickness of00 �m on a copper foil current collector and dried 24 h at 80 ◦C.or both electrodes, the final loading was ∼2 mg of Si per cm2,uggesting a porous nature of these as-prepared anodes. For Si1,lectrodes were cut into 11 mm diameter disk shape, dried 3 h at50 ◦C and transferred into an argon-filled glove box. 3-electrodeswagelok®-type cells were then assembled. For Si2, 17 × 35 mm2

ectangular electrodes were cut, dried 48 h and assembled in 3-lectrodes pouch-cells in an anhydrous room. For both types ofells, the counter and the reference electrodes are made of metal-ic lithium. The electrodes are separated by glass fiber separatorsoaked with a liquid electrolyte (1 M LiPF6 in ethylene carbonateEC): diethyl carbonate (DEC) (wt % of 1:1)). In specific cases, 10 wt

of FEC was also added.Galvanostatic charges and discharges were carried out with a

accor® battery tester for the Swagelok® cells (Si1) and a VMP3iologic® system for the pouch cells (Si2).

.3. EIS measurements

All the EIS measurements were performed after relaxing the cellntil the deviation of the potential was inferior to 1 mV.h−1. For Si1,

he perturbation amplitude was 5 mV and the frequency range wascanned from 100 kHz to 0.1 Hz whereas for Si2, the perturbationmplitude was 10 mV and the frequency range was scanned from00 kHz to 0.01 Hz.

Fig. 1. Galvanostatic curve of the first cycle of Si1 (full line) and Si2 (dashed line) atC/20.

A table summarizing the experimental conditions for Si1 and Si2is reported in ESI.

3. Results and discussion

3.1. Electrochemical signatures

Even though the two studied silicon electrodes are made ofcomparable materials, they differ in some distinct ways. Si1 hasa smaller average diameter, larger surface area and thicker nativeoxide layer than Si2. Fig. 1 shows the galvanostatic curve of cycle1 of both electrodes, Si1 and Si2. Both galvanostatic curves showcomparable features. Indeed, Si1 and Si2 present a plateau duringlithiation around 0.10 V. The fully lithiated capacity for Si1 is 3 850mAh.g−1, compared to 3 800 mAh.g−1 for Si2. Both recorded valuesexceed the theoretical capacity for silicon of 3 579 mAh.g−1. Thiscan be easily explained by the additional consumption of electronsto form the SEI layer at the surface of the electrode [26]. Actually,Si1 only reaches a coulombic efficiency of 83%, whereas Si2 reaches92%. The difference in coulombic efficiency can be explained by thelarger production of SEI, possibly due to a larger surface area andthicker oxide layer, which makes the surface of Si1 more chemicallyreactive [27].

To identify undoubtedly the different phenomena as ChargeTransfer (CT) or SEI, which can be observed by EIS, severalimpedance spectra were recorded along representative cycles.Since the very first electrochemical cycle, which results in a com-plete amorphization of the active material, is very specific forSi-based electrodes [14,28], these measurements were actually car-ried out upon the second electrochemical cycle and/or subsequentcycles, since it has already been shown that nano silicon-based elec-trodes are stable from the 2nd to the 50th cycle, in standard cyclingconditions [16].

Fig. 2 compares both the galvanostatic curves of cycle 2 as well asthe evolution of the impedance spectra for Si1 and Si2. Even thoughboth electrodes differ in active material, electrode composition, andsurface area, the shape of the impedance signals is comparable. Allthe spectra show two semi-circles at High Frequencies (HF), i. e.f > 1 Hz and straight lines in the Low Frequencies (LF) region, i. e.f < 1 Hz. The HF region is typical of interfacial phenomena: the twosemi circles observed for our system, in that domain, can be asso-ciated with the presence of the SEI at the electrode surface and thecharge transfer (CT) process. The characteristic frequency associ-ated with the first HF semi-circle remains constant upon cycling

(∼50 kHz for both Si1 and Si2) suggesting that this semi-circle cor-responds to a passivation surface layer and is here related to theSEI, of which the composition is almost not dependent on the stateof charge of the electrode [15,16]. The characteristic frequency of

E. Radvanyi et al. / Electrochimica Acta 137 (2014) 751–757 753

F 1 and

r

tSocaaeAbteftapat

Fec

ig. 2. Impedance spectra obtained at different stages of the second cycle, for (a) Sielaxation step until dE/dt < 1 mV/h.

he second HF semi-circle evolves with the voltage: for both Si1 andi2 it varies from 300 to 1 500 Hz. thus, we can attribute this sec-nd semi-circle to the CT process. The straight lines, in LF region,an be easily related to the Li+-diffusion processes in the Si bulkctive material. However, in addition to these relatively standardnd already observed semi-circles, the EIS spectra systematicallyxhibit, for both systems, an inductive loop at lower potentials.s far as we know, this kind of inductive phenomenon has nevereen reported in the literature for silicon electrodes. Moreover, ashe experimental conditions for Si1 and Si2 are completely differ-nt, it is very hard to believe that the presence of this loop comesrom an instrumental artefact. On both recordings, we can see thathis inductive loop is always observed below 0.35 V vs Li. Addition-

lly, Fig. 3 shows the impedance spectra obtained with differenterturbation signals: 5, 10, 20 and 30 mV. Changing the excitationmplitude has a direct influence on the inductive part of the spec-rum, since the different curves are not superimposable at these

ig. 3. Impedance spectra recorded at different perturbation amplitudes for a Si2lectrode, lithiated until a capacity of 1 200 mAh.g−1 in the third electrochemicalycle, at C/20.

(b) Si2. The spectra in red (lithiation) and green (delithiation) were acquired after a

medium frequencies. Conversely, we note that the HF and LF partof the different spectra are similar, indicating that the change inperturbation amplitude only has an influence on the inductive part.Consequently, the inductive phenomenon is surely a parasitic reac-tion, intrinsic to the electrochemical system under investigation.

An inductive loop has rarely been reported for EIS spectra in theLIB research field. Two main hypotheses are proposed in the liter-ature to explain this particular phenomenon: (i) a phase change ofthe active material during the measurement or (ii) heterogeneityof the lithium distribution inside the electrode which creates con-centration cells between the different particles. Since the lithiationand delithiation of silicon in cycle 2 are of a solid solution type[28,29], the first hypothesis can hardly explain the inductive loopand can consequently be eliminated. However, in addition to thelast hypothesis, it is also known that the SEI layer on the surface ofsilicon electrodes is very unstable and continuously consumes elec-trons and lithium ions. It is thus possible for this process to changethe current flowing through the system during the EIS measure-ment, causing an electromotive force superimposed to the lithiumintercalation and potentially explaining the inductive loop [30].In the following, several complementary experiments have beenperformed on both Si1 and Si2 systems to discriminate betweenthe most probable hypotheses: heterogeneity of the lithiation orinstability of the SEI layer.

3.2. Influence of the C-rate

If the inductive loop in the impedance spectra, present at lowerpotential states, is caused by heterogeneity in the lithiation, alteringthe cycling rate should have an influence on the presence of the

inductive phenomena. Since a slower lithiation favors the homo-geneity of the lithiation, the disappearance or, at least, a decreaseof the phenomenon is expected. On the other hand, if the induc-tive loop is caused by instability of the SEI layer, changing the

754 E. Radvanyi et al. / Electrochimica Acta 137 (2014) 751–757

F(

ctwfpti

3

mepwcmopca

dtiafedbtt

Fm

ig. 4. Impedance spectrum recorded with a Si1 electrode at fully lithiated state40 mV vs Li) in the second cycle, at a slow cycling rate of C/50.

ycling rate should only have a minimal effect. To illustrate this,he impedance spectrum of a cell with a slower cycling rate (C/50)as recorded, using a Si1 electrode. Fig. 4 shows the spectrum at

ully lithiated state in cycle 2. The inductive effect is still clearlyresent on the spectrum, suggesting its source does not depend onhe cycling rate. This first experiment is rather in favor of certainnstability of the SEI layer.

.3. Influence of the temperature

The temperature is one of the key parameters for an impedanceeasurement, since it has a great influence on the viscosity of the

lectrolyte, the ionic conductivity and the electrochemical kineticsrocesses [31]. Multiple measurements at different temperaturesere performed on a quasi-steady state Si2 electrode, after 6

harge-discharge cycles (Fig. 5). The equilibrium potential beforeeasurement was 0.302 V vs Li, which corresponds to a capacity

f 1 200 mAh.g−1. The first measurement was performed at a tem-erature of 25 ◦C. The cell was than heated to 50 ◦C and afterwardsooled again to 25 ◦C and further to 0 ◦C. Afterwards it was heatedgain to 25 ◦C.

After heating to 50 ◦C, the global impedance of the systemecreases. The higher temperature improves both the viscosity andhe ionic conductivity of the electrolyte. Furthermore, the kinet-cs of the charge transfer and the diffusion in the active materialre enhanced. On the other hand, the formation of SEI is alsoavored. Studies on graphite electrodes have indeed shown thatlevated temperature increases the kinetics of electrolyte degra-ation [32,33]. Subsequently, since the high temperature increases

oth the ionic mobility (favoring homogeneity of the lithiation) andhe formation of the SEI, it is difficult, at that stage, to conclude onhe origin of the inductive loop which is still clearly present.

ig. 5. Impedance spectra recorded with a Si2 electrode after a lithiation of 1 200Ah.g−1 after 6 cycles.

Fig. 6. Impedance spectra recorded with a Si2 electrode after a lithiation of 1 200mAh.g−1 at the 10th cycle with and without FEC.

However, the second measurement at 25 ◦C, after heating at50 ◦C, is not superimposable with the original spectrum, indicatingthat an irreversible phenomenon has occurred at 50 ◦C. Concern-ing the repartition of lithium in the silicon electrode: increasingthe temperature has surely favored a more homogeneous distri-bution of lithium. Taking Fick’s law into account, it would be verysurprising, that the lithium distribution will return to its possibleheterogeneous state after lowering the temperature. If the hetero-geneity of the lithiation is indeed at the origin of the inductive loop,the stay of the electrode at 50 ◦C would have caused the inductivephenomenon to now disappear or, at least, to decrease. Since theinductive loop has increased, the hypothesis of the instability of theSEI is more relevant. Moreover, the first HF semi-circle (related tothe SEI layer) is bigger, suggesting again the electrolyte degradationis clearly activated at high temperature.

Furthermore, lowering the temperature to 0 ◦C results, asexpected, in an increase of the total impedance, since the electrolyteviscosity increases whereas the ionic mobility in both the elec-trolyte and the active material decreases [34]. Note also the perfectsuperposition of the spectra recorded at 25 ◦C, before and after thestay at 0 ◦C indicating that no irreversible (electro)chemical processoccurs at 0 ◦C (in contrast to what happens at 50 ◦C).

3.4. Influence of the electrolyte

The influence of the electrolyte on the inductive phenomenonin the impedance spectrum has also been investigated. Since theelectrolyte mainly contributes to the composition of the SEI layer,it will also play an important role on its behavior during cycling.The influence of fluoroethylene carbonate (FEC) as an electrolyteadditive was thus investigated here, since it is known to produce asmoother and more stable SEI layer [35–37].

Fig. 6 compares two impedance spectra, with and without FECadditive, at the 10th cycle, after a lithiation of 1 200 mAh.g−1.The addition of FEC results in an increase of the characteristic fre-quency associated with the SEI, from 50 kHz to 100 kHz, indicatinga noticeable modification of its chemical composition (change ofthe dielectric permittivity), in good agreement with XPS resultsobtained elsewhere [36]. Note also that the size of the inductiveloop has decreased; the latter being related to the better stabil-ity of the SEI in presence of FEC. We can also observe a noticeabledecrease of the potential at which the inductive loop now appears(< 0.2 V vs Li). This last statement, together with the decrease ofthe inductive effect observed in the EIS spectra, confirms that theSEI formation is both improved and stabilized in presence of FEC

additive in the electrolyte.

Once again, these results confirm that the inductive phe-nomenon is more likely caused by the instability of the SEI at thesurface of silicon particles.

E. Radvanyi et al. / Electrochimica Acta 137 (2014) 751–757 755

ation

3

entsistichtrHltaateaoi

[ot

Ft

Fig. 7. Schematic represent

.5. Proposed (electro)chemical mechanism

The results from the three complementary aforementionedxperiments indicate that the main origin of the inductive phe-omenon on the impedance spectra is probably the instability ofhe SEI on the Si particle surface. During the impedance mea-urement, part of the electrons and Li+-ions would be involvedn the irreversible reaction of SEI formation. Fig. 7 shows achematic representation of the mechanism for such a reaction athe silicon-electrode/electrolyte interface. The lithiation of silicons an electro-insertion reaction, associated with the double layerharge together with the charge transfer process, which would beere in competition with the reduction of the electrolyte. To explainhe appearance of possible inductive impedance, this reductioneaction shall contain a two-step process based on a Volmer-eyrovsky-type reaction [38]. The first step is an adsorption of

ithium ions together with reduction products of the electrolyte athe electrode surface. The second step is a combination of the latterdsorbed species to form the final SEI layer. The detailed equationsnd their mathematical treatments are attached in appendix. Fromhe analytical resolution of these kinetic equations, an electricalquivalent circuit could be deduced to model the proposed mech-nism and finally to fit the recorded impedance data. The HF partf the equivalent circuit (related to interfacial reactions) is shownn Fig. 8.

As already extensively described in literature

11,17,22,24,25,39], the high frequency part contains the resistancef the electrolyte (Re) and the RSEI//CSEI system corresponding tohe impedance of the SEI deposit. The mid frequency part contains

ig. 8. Equivalent electrical circuit for the high frequency region (interface elec-rode/electrolyte), used to model the presented data.

of the reaction mechanism.

the double layer capacitance (Cdl), the charge transfer resistance(Rct) and an RSEI formation//LSEI formation couple corresponding to theinductive loop, related to the SEI formation/deposition. Note alsothat the equivalent circuit in Fig. 8 was only used to model spectracontaining the inductive loop. Obviously, for other spectra, RSEI

formation//LSEI formation was left out.One has to bear in mind that it is delicate to interpret the physical

meaning of R and L individually. From their analytical expressions,both are linked to the formation process of the SEI layer. How-ever, RSEI formation could be seen as the ease to form the SEI. Themeaning of LSEI formation is slightly more delicate since it containsboth contributions of the faradic current. Actually, the faradic cur-rent (IF), shown in Fig. 8, is divided between two processes: chargetransfer (CT) and SEI building (passivation). If the contribution ofSEI formation is small, IF will be determined by the CT resistancesolely. Hence, no inductive loop will be observed. During the pro-cess of SEI-formation, IF will also have a RSEI formation//LSEI formationcontribution, resulting in the presence of an inductive loop, atmid frequencies, in the impedance spectrum. For example, Table 1presents the potential at which an inductive loop appears for Si2in cycle 2. As already mentioned, the inductive phenomenon onlyappears at low potentials. Consequently, the process of SEI forma-tion, and accordingly the adsorption of lithium ions, takes place atlow potentials on silicon electrodes. According to literature, withgraphite electrodes the decomposition of electrolyte and formationof SEI occurs at higher potentials, generally between 1 V and 0.5 Vvs Li and separated from the lithiation of the electrode. Literaturealso described this to be the case for silicon electrodes [17,40–42].Different values for the potential range have been reported, butgenerally the SEI formation takes place between 1.5 V and 0.25 Vvs Li, after which lithiation occurs between 0.25 V and 0 V vsLi.

The results presented in this paper indicate that both processes(SEI formation and lithiation of the silicon electrode) do take placesimultaneously at low potentials. One can probably argue that dur-ing the SEI formation between 1.5 V and 0.25 V vs Li, as no–or

little–lithiation occurs, the volume of Si particles do not change,whereas, during the lithiation process (between 0.25 V and 0 V vsLi), the volume of the particles increases markedly, increasing thesurface of the Si particles which induces a second SEI formation

756 E. Radvanyi et al. / Electrochimica Acta 137 (2014) 751–757

Table 1Values of the parameters obtained for the EIS spectra for cycle 2 of the Si2 electrode. At certain (higher) potentials RSEI formation and LSEI formation were not observed.

Decreasing Potential (V)(upon lithiation)

0.40 0.38 0.33 0.31 0.29 0.25 0.22 0.20 0.19 0.17 0.12

Re (�.cm2) 9.3 8.7 8.4 8.5 7.7 8.2 8.4 8.5 8.8 8.7 8.8RSEI (�.cm2) 5.5 3.1 1.2 1.1 0.96 1 1.1 1.4 2.2 3 2.8RTC (�.cm2) 6.6 4.5 3.3 3.4 1.8 1.5 1.4 1.7 2.8 4 35RSEI formation (�.cm2) - - 0.234 0.6 0.9 1 1.2 1.32 1.6 2.4 7.2LSEI formation (H.cm2) - - 4.7.10−4 1.3.10−3 1.6.10−3 1.9.10−3 2.3.10−3 2.4.10−3 3.1.10−3 5.6.10−3 1.1.10−2

Increasing Potential (V)(upon delithiation)

0.2 0.23 0.26 0.28 0.31 0.34 0.38 0.40 0.43 0.47 0.52

Re (�.cm2) 8.8 10 9.2 8.6 8.8 9 8.9 9.2 8.8 8.9 9.1RSEI (�.cm2) 2.8 2.6 1.9 1.4 1.3 1.4 3.1 4.6 6.6 5.9 5.2RTC (�.cm2) 7.8 6.6 4.3 3.2 3.5

RSEI formation (�.cm2) 1.5 1.3 0.9 0.7 0.44

LSEI formation (H.cm2) 1.9.10−3 2.6.10−3 1.4.10−3 9.6.10−4 7.2.10

Fo

sp

fea(cLabcitoe

4

eai

(

vlb

[

ig. 9. Experimental and modelled spectra obtained during cycle 2 after a lithiationf 2 400 mAh.g−1 and a potential of 0.21 V.

tep (new mechanism), different from that encountered at higherotentials and evidenced here by EIS recordings.

Values for the different parameters in Fig. 8 have been fittedor cycle 2 of the Si2 electrode and are shown in Table 1. As anxample, Fig. 9 shows the experimental and the modeled spectrafter a lithiation of 2 400 mAh.g−1 and a potential of 0.21 V vs LiFig. 2). Rct increases when approaching full lithiation. This indi-ates that it is more difficult to further lithiate the highly lithiatedixSi alloys. During delithiation Rct decreases again as the delithi-tion proceeds. The evolution of RSEI contains a minimum duringoth lithiation and delithiation, indicating (for an almost chemi-ally constant layer) either a slight dissolution or a surface changen the SEI deposit. Nevertheless, it suggests a dynamical behavior ofhis latter upon cycling, completely consistent with the instabilityf the passivation layer on the Si-particles claimed in this work andnlightened by the recorded EIS experiments presented.

. Conclusion

The electrochemical impedance study, performed on the secondlectrochemical cycle, on two distinct nano silicon-based materi-ls with 2 different 3-electrodes cell configurations, has alloweddentifying the different characteristic frequencies associated to:

(i) the presence of the SEI layer on the surface of the silicon parti-cles around 50 kHz;

ii) the charge transfer process between 300 and 1 500 Hz (varyingwith the working potential)

This work has also highlighted the presence, below a specificoltage (depending of the electrolyte composition), of an inductiveoop: a particular phenomenon, rarely observed in the field of Li-ionatteries, and to the best of our knowledge, never mentioned for

[

4.3 4.5 5.6 5.8 7.2 7.81.4 - - - - -

−4 3.2.10−4 - - - - -

silicon-based electrodes. Several complementary and independentexperiments, performed on both nano Si-based systems, suggestthat it is mostly the instability of the SEI on the surface of thesilicon particles that causes the appearance of the inductive loopon the corresponding EIS spectra. Finaly, a reaction mechanism andan equivalent electrical circuit, associated to this phenomenon, areproposed to model this effect. As a conclusion, it clearly appearshere that the EIS technique is particularly interesting to study thestability/coverage of the SEI on the surface of silicon electrodes.

Acknowledgements

Dr. F. Berthier (ICMMO-SP2 M) is gratefully acknowledged forfruitful discussion on EIS modeling.

K. Van Havenbergh gratefully acknowledges IWT Flanders forthe financial support.

E. Radvanyi acknowledges CEA-INSTN for supporting part of thisstudy.

Appendix A. Supplementary data

Supplementary data associated with this article can be found,in the online version, at http://dx.doi.org/10.1016/j.electacta.2014.06.069.

References

[1] B.L. Ellis, K. Town, L.F. Nazar, New composite materials for lithium-ion batteries,Electrochimica Acta 84 (2012) 145–154.

[2] J.B. Goodenough, K.S. Park, The Li-Ion Rechargeable Battery: A Perspective,Journal of the American Chemical Society 135 (2013) 1167–1176.

[3] T.H. Kim, J.S. Park, S.K. Chang, S. Choi, J.H. Ryu, H.K. Song, The Current Move ofLithium Ion Batteries Towards the Next Phase, Advanced Energy Materials 2(2012) 860–872.

[4] D. Larcher, S. Beattie, M. Morcrette, K. Edstroem, J.C. Jumas, J.M. Tarascon,Recent findings and prospects in the field of pure metals as negative electrodesfor Li-ion batteries, Journal of Materials Chemistry 17 (2007) 3759–3772.

[5] W.J. Zhang, A review of the electrochemical performance of alloy anodes forlithium-ion batteries, Journal of Power Sources 196 (2011) 13–24.

[6] M.N. Obrovac, L. Christensen, Structural changes in silicon anodes duringlithium insertion/extraction, Electrochemical and Solid State Letters 7 (2004)A93–A96.

[7] L.Y. Beaulieu, T.D. Hatchard, A. Bonakdarpour, M.D. Fleischauer, J.R. Dahn,Reaction of Li with alloy thin films studied by in situ AFM, Journal of theElectrochemical Society 150 (2003) A1457–A1464.

[8] H. Wu, Y. Cui, Designing nanostructured Si anodes for high energy lithium ionbatteries, Nano Today 7 (2012) 414–429.

[9] C. Erk, T. Brezesinski, H. Sommer, R. Schneider, J. Janek, Toward Silicon Anodesfor Next-Generation Lithium Ion Batteries: A Comparative Performance Studyof Various Polymer Binders and Silicon Nanopowders, Acs Applied Materials &Interfaces 5 (2013) 7299–7307.

10] J. Li, R.B. Lewis, J.R. Dahn, Sodium carboxymethyl cellulose - A potential binder

for Si negative electrodes for Li-ion batteries, Electrochemical and Solid StateLetters 10 (2007) A17–A20.

11] J.S. Bridel, T. Azais, M. Morcrette, J.M. Tarascon, D. Larcher, In Situ Observa-tion and Long-Term Reactivity of Si/C/CMC Composites Electrodes for Li-IonBatteries, Journal of the Electrochemical Society 158 (2011) A750–A759.

himica

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[[

[

[

[

[4

E. Radvanyi et al. / Electroc

12] B. Lestriez, S. Desaever, J. Danet, P. Moreau, D. Plee, D. Guyomard, Hierar-chical and Resilient Conductive Network of Bridged Carbon Nanotubes andNanofibers for High-Energy Si Negative Electrodes, Electrochemical and SolidState Letters 12 (2009) A76–A80.

13] J.R. Szczech, S. Jin, Nanostructured silicon for high capacity lithium batteryanodes, Energy & Environmental Science 4 (2011) 56–72.

14] C.K. Chan, R. Ruffo, S.S. Hong, Y. Cui, Surface chemistry and morphology of thesolid electrolyte interphase on silicon nanowire lithium-ion battery anodes,Journal of Power Sources 189 (2009) 1132–1140.

15] B. Philippe, R. Dedryvere, J. Allouche, F. Lindgren, M. Gorgoi, H. Rensmo, D.Gonbeau, K. Edstrom, Nanosilicon Electrodes for Lithium-Ion Batteries: Inter-facial Mechanisms Studied by Hard and Soft X-ray Photoelectron Spectroscopy,Chemistry of Materials 24 (2012) 1107–1115.

16] E. Radvanyi, E. De Vito, W., Porcher, S. Jouanneau Si Larbi, An XPS/AEScomparative study of the surface behaviour of nano silicon anodes forLi-ion batteries, Journal of Analytical Atomic Spectrometry, (2014) DOI:10.1039/C1033JA50362 C.

17] T.L. Kulova, A.M. Skundin, Y.V. Pleskov, E.I. Terukov, O.I. Kon’kov, Lithium inser-tion into amorphous silicon thin-film electrodes, Journal of ElectroanalyticalChemistry 600 (2007) 217–225.

18] W.R. Liu, J.H. Wang, H.C. Wu, D.T. Shieh, M.H. Yang, N.L. Wu, Electrochemi-cal characterizations on Si and C-coated Si particle electrodes for lithium-ionbatteries, Journal of the Electrochemical Society 152 (2005) A1719–A1725.

19] E. Pollak, G. Salitra, V. Baranchugov, D. Aurbach, In situ conductivity, impedancespectroscopy, and ex situ raman spectra of amorphous silicon during theInsertion/Extraction of lithium, Journal of Physical Chemistry C 111 (2007)11437–11444.

20] R. Ruffo, S.S. Hong, C.K. Chan, R.A. Huggins, Y. Cui, Impedance Analysis of SiliconNanowire Lithium Ion Battery Anodes, Journal of Physical Chemistry C 113(2009) 11390–11398.

21] H. Schranzhofer, J. Bugajski, H.J. Santner, C. Korepp, K.C. Moller, J.O. Besen-hard, M. Winter, W. Sitte, Electrochemical impedance spectroscopy study ofthe SEI formation on graphite and metal electrodes, Journal of Power Sources153 (2006) 391–395.

22] S.S. Zhang, K. Xu, T.R. Jow, EIS study on the formation of solid electrolyte,Electrochimica Acta 51 (2006) 1636–1640.

23] X.W. Zhang, P.K. Patil, C.S. Wang, A.J. Appleby, F.E. Little, D.L. Cocke, Electro-chemical performance of lithium ion battery, nano-silicon-based, disorderedcarbon composite anodes with different microstructures, Journal of PowerSources 125 (2004) 206–213.

24] Q.C. Zhuang, Z.F. Chen, Q.F. Dong, Y.X. Jiang, L. Huang, S.G. Sun, Studies of the firstlithiation of graphite materials by electrochemical impedance spectroscopy,Chinese Science Bulletin 51 (2006) 1055–1059.

25] P.J. Zuo, G.P. Yin, Y.L. Ma, Electrochemical stability of silicon/carbon compositeanode for lithium ion batteries, Electrochimica Acta 52 (2007) 4878–4883.

26] P.B. Balbuena, Y., Wang, Lithium-Ion Batteries: Solid-Electrolyte Interphase,

Imperial College Press: London, UK, (2004).

27] S. Xun, X. Song, L. Wang, M.E. Grass, Z. Liu, V.S. Battaglia, G. Liu, The Effectsof Native Oxide Surface Layer on the Electrochemical Performance of SiNanoparticle-Based Electrodes, Journal of the Electrochemical Society 158(2011) A1260–A1266.

[4

Acta 137 (2014) 751–757 757

28] M.N. Obrovac, L.J. Krause, Reversible cycling of crystalline silicon powder, Jour-nal of the Electrochemical Society 154 (2007) A103–A108.

29] M.T. McDowell, S.W. Lee, J.T. Harris, B.A. Korgel, C.M. Wang, W.D. Nix, Y. Cui,In Situ TEM of Two-Phase Lithiation of Amorphous Silicon Nanospheres, NanoLetters 13 (2013) 758–764.

30] J.S. Gnanaraj, R.W. Thompson, S.N. Iaconatti, J.F. DiCarlo, K.M. Abraham, For-mation and growth of surface films on graphitic anode materials for Li-ionbatteries, Electrochemical and Solid State Letters 8 (2005) A128–A132.

31] D. Aurbach, B. Markovsky, G. Salitra, E. Markevich, Y. Talyossef, M. Koltypin, L.Nazar, B. Ellis, D. Kovacheva, Review on electrode-electrolyte solution interac-tions, related to cathode materials for Li-ion batteries, Journal of Power Sources165 (2007) 491–499.

32] A.M. Andersson, K. Edstrom, Chemical composition and morphology of the ele-vated temperature SEI on graphite, Journal of the Electrochemical Society 148(2001) A1100–A1109.

33] T. Zheng, A.S. Gozdz, G.G. Amatucci, Reactivity of the solid electrolyte interfaceon carbon electrodes at elevated temperatures, Journal of the ElectrochemicalSociety 146 (1999) 4014–4018.

34] S.S. Zhang, K. Xu, T.R. Jow, Electrochemical impedance study on the low tem-perature of Li-ion batteries, Electrochimica Acta 49 (2004) 1057–1061.

35] N.S. Choi, K.H. Yew, K.Y. Lee, M. Sung, H. Kim, S.S. Kim, Effect of fluoroethy-lene carbonate additive on interfacial properties of silicon thin-film electrode,Journal of Power Sources 161 (2006) 1254–1259.

36] V. Etacheri, O. Haik, Y. Goffer, G.A. Roberts, I.C. Stefan, R. Fasching, D. Aur-bach, Effect of Fluoroethylene Carbonate (FEC) on the Performance and SurfaceChemistry of Si-Nanowire Li-Ion Battery Anodes, Langmuir 28 (2012) 965–976.

37] Y.M. Lin, K.C. Klavetter, P.R. Abel, N.C. Davy, J.L. Snider, A. Heller, C.B. Mullins,High performance silicon nanoparticle anode in fluoroethylene carbonate-based electrolyte for Li-ion batteries, Chemical Communications 48 (2012)7268–7270.

38] J.-P. Diard, B. Le Gorrec, C., Montella, Cinétique électrochimique, 1997.39] J.C. Guo, A. Sun, X.L. Chen, C.S. Wang, A. Manivannan, Cyclability study of

silicon-carbon composite anodes for lithium-ion batteries using electrochem-ical impedance spectroscopy, Electrochimica Acta 56 (2011) 3981–3987.

40] L.L. Du, Q.C. Zhuang, T. Wei, Y.L. Shi, Y.H. Qiang, S.G. Sun, ElectrochemicalImpedance Spectroscopic Studies of the First Lithiation of Si/C Composite Elec-trode, Acta Chimica Sinica 69 (2011) 2641–2647.

41] I. Ryu, J.W. Choi, Y. Cui, W.D. Nix, Size-dependent fracture of Si nanowire batteryanodes, Journal of the Mechanics and Physics of Solids 59 (2011) 1717–1730.

42] J.J. Wu, W.R. Bennett, Fundamental Investigation of Si anode in Li-ion cells,NASA Glenn Research Center, Electrochemistry Branch.

Further reading

3] J. Zhou, P.H.L. Notten, Development of reliable lithium microreference electrodes

for long-term in situ studies of lithium-based battery systems, J. Electrochem.Soc. 151 (12) (2004) A2173–A2179.

4] M. Ender, A. Weber, E. Ivers-Tiffée, Analysis of three-electrode setups forAC-impedance measurements on lithium-ion cells by FEM simulations, J. Elec-trochem. Soc. 159 (2) (2012) A128–A136.