钠离子电池电解液文献

In search of an optimized electrolyte for Na-ion batteriesAlexandre Ponrouch,*ac Elena Marchante,ac Matthieu Courty,bc Jean-Marie Tarascon bc and M.Rosa Palac ın *acReceived 17th May 2012,Accepted 12th June 2012DOI:10.1039/c2ee22258bElectrolytes are essential for the proper functioning of any battery technology and the emerging Na-ion technology is no exception.Hence,a major focus on battery research is to identify the most appropriate formulation so as to minimize interface reactions and enhance both cell performances and safety aspects.In order to identify suitable electrolyte formulations for Na-ion chemistry we benchmarked various electrolytes containing diverse solvent mixtures (cyclic,acyclic carbonates,glymes)and Na-based salts having either F-based or perchlorate anions and measured viscosity,ionic conductivity,and thermal and electrochemical stability.The binary EC:PC solventmixture has emerged as the best solvent formulation and has been used to test the performance of Na/hard carbon cells with both NaClO 4and NaPF 6as dissolved salts.Hard carbon electrodes having reversible capacities of 200mA h g À1with decent rate capability and excellent capacity retention (>180cycles)were demonstrated.Moreover,DSC heating curves demonstrated that fully sodiated hard carbon cycled in NaPF 6–EC:PC exhibits the highest exothermic onset temperature and nearly the lowest enthalpy of reaction,thus making this electrolyte most attractive for the development of Na-ion batteries.1.IntroductionOwing to their success in the domain of portable electronics,lithium batteries are currently considered as the key technology for electric vehicle propulsion and in parallel are starting to be evaluated for stationary applications.However,the imple-mentation of a lithium-based technology for mass storage faces an important challenge linked to both lithium availability and cost.We face the risk that lithium resources cannot cope over thenext decades with the foreseen staggering energy storage demands dictated by both automotive transportation and grid-related applications.In anticipation of such a scenario,besides promoting recycling,new sustainable chemistries must be developed,and the most appealing alternative is to use sodium instead of lithium.There are several reasons:the intercalation chemistry is similar to that of lithium,sodium resources are in principle unlimited and it is an element easy to recycle.Sodium technology has already been successfully implemented in high temperature Na/S cells for MW storage and Na/NiCl 2ZEBRA-type systems for electric vehicles,both of which take advantage of the highly conducting solid beta-alumina ceramics at temperatures of ca.300 C.Mindful of these considerations,within the current knowledge gained in Li-ion technology,a room temperature Na-ion cell should be feasible and is a realisticaInstitut de Ci encia de Materials de Barcelona,ICMAB-CSIC,Campus de la UAB,08193Bellaterra,Spain.E-mail:rosa.palacin@icmab.es;aponrouch@icmab.es;Fax:+34935805729;Tel:+34935801853#279bLaboratoire de R eactivit e et Chimie des Solides,Universit e de Picardie Jules Verne,CNRS UMR7314,33rue Saint Leu,80039Amiens,France cALISTORE-ERI European Research Institute,Dynamic Article Links CEnergy &Environmental ScienceCite this:Energy Environ.Sci.,2012,5,/eesPAPERD o w n l o a d e d b y S h a n g h a i I n s t i t u t e o f C e r a m i c s , C A S o n 07 S e p t e m b e r 2012P u b l i s h e d o n 13 J u n e 2012 o n h t t p ://p u b s .r s c .o r g | d o i :10.1039/C 2E E 22258BView Online / Journal Homepage / Table of Contents for this issuetarget.This technology would be inherently safer than the high temperature cells that contain molten sodium and corrosive liquids and would not suffer from long start up times (issued from the need to warm these cells slowly to their operation temperatures).While sodium-based cells will always fall short of matching the specific energy of the corresponding lithium tech-nology at the cell level,they are certainly an option for large scale applications in which cost instead of energy density is the over-riding factor.1,2Historically,progress in sodium intercalation chemistry is parallel to that of lithium.3,4Therefore,Na-based systems rapidly fall into oblivion due to the rapid advent and success of the lithium ion technology for portable electronics and of its positive attributes for a wide range of applications requiring energy densities from m W h to MW h.However,for sustainability reasons the Na-ion concept,which requires both positive and negative Na-based insertion compounds,has recently re-captured the scientific community’s attention.5–10While a wide variety of Na-based positive electrodes have been reported,11–21hard carbon constitutes presently the most viable option for the negative electrode.When it comes to transition metal oxides,the choice is rather limited owing to the competi-tion between insertion vs.conversion 22reactions at low poten-tials,the former being only favoured for early 3d metal oxides.While Na x VO 2has been proved to reversibly react with sodium at moderate potentials,23,24we recently identified Na 2Ti 3O 7as an effective low potential insertion sodium compound.Indeed,it exhibits the reversible uptake of 2Na ions per formula unit (200mA h g À1)at an average potential of 0.3V vs.Na +/Na 0through a two-phase redox mechanism.25,26The cycling of this electrode vs.Na +/Na 0is still problematic due to a poorly stable electrode/electrolyte interphase (SEI)and calls for further studies to elucidate both the structural changes concomitant to the redox processes and the complex growth mechanism of the SEI together with its composition and nature.While a number of efforts are being directed to the search for new electrode materials for sodium ion batteries,studies dealing with the electrolyte are much scarcer.Still,some available reports demonstrate that the SEI formed on carbonaceous electrodes is markedly different for sodium-and lithium-based electrolytes even on using the same solvent.5,27Moreover,vinylene carbonate (VC),a common additive in Li-based electrolytes,appears to be detrimental to the performance of Na cells.28Presently,most studies are being performed using NaClO 4in PC as the electro-lyte with only a few reports on the use of NaPF 6and other solvents,5,27,29but no systematic studies have been reported on their physico-chemical properties.Yet,this is a must if we ever want the Na-ion technology to become a commercial reality and a serious contender for electrochemical storage.As in any other battery system,a good electrolyte should exhibit:(i)good ionic conductivity,(ii)a large electrochemical window (i.e.,high and low onset potential for electrolyte decomposition through oxidation and reduction at high and low voltages,respectively),(iii)no reactivity towards the cell components,and (iv)a large thermal stability window (i.e.melting point and boiling point lower and higher than the stan-dard temperatures for the cell utilization,respectively).Finally,it should be intrinsically safe,have as low toxicity as possible and meet cost requirements for the targeted applications.All thesefeatures are intrinsically dependent on the nature of the salt and the solvent(s)and the possible use of additives.To address the aforementioned issues we embarked on a comparative survey of electrolytes prepared using several solvents (PC,EC,DMC,DME,DEC,THF and Triglyme)and solvent mixtures (EC:DMC,EC:DME,EC:PC and EC:Tri-glyme)in combination with different Na salts,namely NaClO 4,NaPF 6and NaTFSI.This paper is mainly divided into two parts.The first part deals with the intrinsic physico-chemical properties of electrolytes:viscosity,ionic conductivity and electrochemical and thermal stability,while the second part addresses their low voltage electrochemical performance using hard carbon (HC)as the electrode plementary differential scanning calorimetry (DSC)and electrochemical impedance spectroscopy (EIS)analysis were performed in the hope of grasping some information about the SEI growth and thermal stability.2.Experimental section2.1.Electrolyte preparationAll tested electrolytes consist of a 1M solution of the salt (NaClO 4,NaPF 6or NaTFSI)in propylene carbonate (PC hereafter,Aldrich,anhydrous,99.7%),ethylene carbonate (EC,Aldrich,anhydrous,99.0%),dimethyl carbonate (DMC,Aldrich,anhydrous,99.0%),diethyl carbonate (DEC,Aldrich,anhydrous,99.0%),dime-thoxyethane (DME,Aldrich,anhydrous,99.5%),tetrahydrofuran (THF,Aldrich,anhydrous,99.9%)and triethylene glycol dimethyl ether (Triglyme,Aldrich,99.0%)or binary solvent mixtures (50/50wt%).Solvents were used as received.Salts were vacuum dried at moderate temperatures prior to being used.The water content in all as-prepared electrolytes was measured by Karl-Fisher titration and found to be lower than 50ppm in all cases.Thus,they were used without any additional drying step.2.2.Conductivity and viscosity measurementsThe ionic conductivity of the electrolytes was measured with an Oakton CON 11standard conductivity meter.Their viscosities were measured with a RheoStress RS600Rheometer (HAAKE)at 20 C.All measurements were done at room temperature and were performed simultaneously in order to avoid any bias due to changes in ambient conditions.2.3.Differential scanning calorimetryDSC measurements were performed using a DSC 204F1Netzsch calorimeter,with a heating rate of 10 C min À1.The sample,either solid or liquid,was placed in an aluminum capsule which was sealed in a dry glove box filled with argon and pierced once inside the DSC apparatus.Melting points of the solvents and the electrolytes were measured from the DSC heating curves after cooling the sample down to À120 C,using liquid nitrogen.Boiling points and phase transition temperatures were assessed by heating the sample up to 350 C.Thermal stability and heat generated from fully sodiated HC were measured from electrodes reduced to 3mV vs.Na +/Na 0at C /40,and recovered inside the argon filled dry box without any washing or drying.Afterwards,the samples were placed in stainless steel capsules which were not pierced.D o w n l o a d e d b y S h a n g h a i I n s t i t u t e o f C e r a m i c s , C A S o n 07 S e p t e m b e r 2012P u b l i s h e d o n 13 J u n e 2012 o n h t t p ://p u b s .r s c .o r g | d o i :10.1039/C 2E E 22258B2.4.ElectrochemistryThe electrochemical stability of the electrolytes was evaluated inthree-electrode Swagelok cells 30using a sodium metal cube slice (Aldrich,99.95%)as counter and reference electrodes,and an aluminum plunger as the working electrode (WE).The standard cycling protocol consists of three subsequent steps hereafter called Protocol A,B and C (cf.Fig.1).Protocol A is mainly aimed at determining the electrolyte electrochemical stability window,via the collection of CV curves at a sweep rate of 10mV s À1and of EIS spectra prior and after each CV (denoted EIS-A1,EIS-A2and EIS-A3).The first CV is carried out between 0.01and 3V vs.Na +/Na 0in order to evaluate the electrochemical stability of the electrolyte at low potentials.In subsequent CVs,the potential window is being enlarged every 3cycles to gradually attain 5.3V vs.Na +/Na 0with 400mV steps.Protocol B is mainly geared towards the determination of possible electrolyte decomposition upon various cycling conditions via EIS measurements which were performed between several linear polarization (LP)sequences (at a 10mV s À1sweep rate).Typically,the WE potential was left at the open circuit potential (OCP)for at least 2h after protocol A,and EIS-B1was recorded.Then,several LP steps were applied to the WE with EIS measurements between them (denoted as EIS-B2up to EIS-B7).These were made at three different potential values,namely 0.01,3and 5V vs.Na +/Na 0.Protocol C is devoted to WE performance when the potential is being held either at high (5V vs.Na +/Na 0)or low (0.01V vs.Na +/Na 0)values for various periods of time.More specifically,the WE was left at OCP for at least 2h after protocol B,and EIS-C1was measured as to establish an initial state of the electrode.Then a LP (at a 10mV s À1sweep rate)up to 5V vs.Na +/Na 0was applied,the potential being held at 5V vs.Na +/Na 0for various periods of time (ca.30min,1h 30min,3h and 5h)between which EIS measurements were performed (ca.EIS-C2up to EIS-C6).Finally,LP (at a 10mV s À1sweep rate)down to 0.01V vs.Na +/Na 0was applied and the potential was held at 0.01V vs.Na +/Na 0for ca.30min,1h 30min,3h and 5h,between which EIS measurements were also performed (EIS-C7up to EIS-C11).EIS measurements (10mV perturbation amplitude,with frequencies ranging from 1000kHz to 50mHz)were also per-formed upon cycling of HC tape-cast electrodes using three-electrode Swagelok cells (after a 2h OCP period).Cycling experiments were also performed in two-electrode Swagelok cells in galvanostatic mode with potential limitation (GCPL)at different rates ranging from C /20to C (1C being one Li +inserted in one hour)to monitor capacity evolution upon cycling.All electrochemical tests were performed using a Bio-Logic VMP3potentiostat and twin cells were assembled in order to ensure reproducibility of results.Hard carbon (HC)was prepared following the procedure described in ref.31by pyrolysis of sugar at 1100 C for 6h under argon flposite electrodes mimicking industrial technol-ogies were prepared from slurries (90wt%HC,5wt%poly-vinylidene fluoride binder (PVDF,Arkema)and 5wt%Super P carbon (Csp hereafter from Timcal)in N -methylpyrrolidoneFig.1Schematic representation of the standard electrochemical protocols applied for electrolyte electrochemical characterization in three-electrode Swagelok cells.D o w n l o a d e d b y S h a n g h a i I n s t i t u t e o f C e r a m i c s , C A S o n 07 S e p t e m b e r 2012P u b l i s h e d o n 13 J u n e 2012 o n h t t p ://p u b s .r s c .o r g | d o i :10.1039/C 2E E 22258B(NMP,Aldrich)).Mixing was done using either a PM100Retsch Planetary Ball Miller (stainless steel container with 3stainless steel balls of 1cm diameter at 500rpm for 2h with change of the rotating way every 30min)or a high energy SPEX 800ball miller for 30min,the latter resulting in higher capacities.These slurries were tape-cast on a 20m m thick copper foil (Goodfellow)with a 250m m Doctor-Blade and further dried at 120 C under vacuum.Once dried,0.8cm 2disk electrodes were cut and pressed at 8t prior to testing.Typically,electrode loading is 2mg corre-sponding to about ca.15m m thick deposits.3.Results and discussion3.1.Conductivity and viscosity measurementsThe electrolyte ionic conductivity is one of the most crucial factors for cell performance as it ‘‘defines how fast the energy can be released by controlling the rate of mass flow within the battery’’.32It depends on several factors,such as (i)the degree of dissociation of the salt (which is dictated by the dielectric constant of the solvent),(ii)the viscosity of the electrolyte (which influences the ion mobility),and (iii)the transport number of the Na +cation and its counter anion.The ionic conductivity values for the investigated electrolytes are reported in Fig.2and benchmarked against LP30(1M LiPF 6in EC:DMC),which is presently the most widely used electrolyte for lithium ion cells.To separate the relative influences of the solvent and the salt,PC was chosen as the solvent and the conductivities of 1M solutions of different salts were evaluated,and afterwards the conductiv-ities of 1M NaClO 4in different solvent mixtures were measured.Each reported value here is an average of three measurements,the deviation between them being less than 5%,which gives us confidence that our trends are robust.The conductivity values for PC-based electrolytes (cf.Fig.2a)are relatively similar,ranging from 7.98for NaPF 6to 6.2mS cm À1for NaTFSI and being intermediate for NaClO 4.This dependence upon the nature of the anion is consistent with what was previously reported for Li-based electrolytes 33which was ascribed to the lower polarizing character of the PF 6Àanions,which would improve salt dissociation and enhance ionic mobility (see ref.32,p.4315).In contrast,larger ionic conduc-tivity variations were found by fixing the salt (NaClO 4)and changing the nature of the solvents with s values following the trend EC:DME >EC:DMC >EC:PC >EC:Triglyme >EC:DEC >PC >Triglyme \DME,DMC,DEC (Fig.2b).These results confirm that single solvent-based electrolytes show much lower conductivity than binary solvent-based elec-trolytes,this being true even for PC,which presents a very high dielectric constant value of 64.92.The addition of EC as a co-solvent strongly improved the ionic conductivity in all cases.The lowest ionic conductivity for binary mixtures is measured for EC:DEC (ca. 6.35mS cm À1)and the highest for EC:DMEFig.2Conductivity (black bars and left hand side y axis)and viscosity (green bars and right hand side y axis)values of (a)PC based electrolytes with 1M of various Na salts and (b)electrolytes based on 1M NaClO 4dissolved in various solvents and solvent mixtures.D o w n l o a d e d b y S h a n g h a i I n s t i t u t e o f C e r a m i c s , C A S o n 07 S e p t e m b e r 2012P u b l i s h e d o n 13 J u n e 2012 o n h t t p ://p u b s .r s c .o r g | d o i :10.1039/C 2E E 22258B(ca.12.55mS cm À1),which is fully comparable to LP30(dashed line in Fig.2b).Such differences in the values of ionic conduc-tivity were found to be proportional to the dielectric constant of the EC co-solvent (PC >Triglyme >DME >DMC >DEC,see Table 1).The only exceptions to this trend are PC and Triglyme-based electrolytes which present lower ionic conductivity,most certainly due to their higher viscosity values as compared to other co-solvents (PC >Triglyme >DEC >DMC >DME,seeTable 1).As a conclusion,the nature of the co-solvent can enhance ionic conductivity through improving the dissociation of the salt (if the dielectric constant is high)and/or by lowering the viscosity of the resulting electrolyte and thus improving the ionic mobility.Viscosity measurements were performed in order to better understand this issue and results are displayed in Fig.2.Similar values of about 6.8cP are observed for PC-based electrolytes with 1M NaTFSI,NaClO 4or NaPF 6(cf.Fig.2a),in agreement with the nature of the counter anions not having a significant effect on the viscosity.On comparing the values measured for various solvents and solvent mixtures with 1M NaClO 4(cf.Fig.2b)we can see that the viscosity of single solvent-based electrolytes (i.e.1M NaClO 4in DMC,Triglyme or PC)follows the same trend as for the solvent alone (i.e.PC >Triglyme >DMC,see Table 1and Fig.2b).Viscosities for pure DMC,Triglyme and PC are 0.5,1.6and 2.2cP,respectively;while higher values are recorded for 1M NaClO 4solutions (ca.0.65,5.8and 6.8cP,respectively).The same observation is made for binary solvent EC-based electrolytes (with 1M NaClO 4),their viscosities being proportional to that of the co-solvent,thus following the trend:EC:PC >EC:Triglyme >EC:DEC >EC:DMC >EC:DME (cf.Fig.2b).Amongst single solvent-based electrolytes (i.e.1M NaClO 4in DMC,Triglyme or PC)the highest ionic conductivity and the highest viscosity are both observed for PC,which confirms that dissociation of the salt is the key parameter (i.e.high dielectric constant value).In contrast,for binary solvent-based electrolytes (i.e.1M NaClO 4in EC:PC,EC:Triglyme,EC:DEC,EC:DMC or EC:DME)the highest ionic conductivity is recorded for EC:DME even though DME does not exhibit the highest dielectric constant value (cf.Table 1),yet it presents the lowest viscosity among all binary solvent-based electrolytes (cf.Fig.2b).This confirms that ion mobility is the most important factor here (i.e.low viscosity value),the presence of EC already allowing good dissociation of the salt.3.2.Differential scanning calorimetry measurementsMelting point is a common issue for EC-based electrolytes in lithium ion batteries,since it determines the operatingtemperature range.Although EC presents a very high dielectric constant (ca.89.78),which enables a high dissociation of the salt and is recognized to be key in the formation of a stable SEI on the most common anode in lithium ion cells (graphite),its main drawback is nested in its relatively high melting point (ca.36.4 C).Such temperature can of course be lowered by the addition of the salt,but the changes are not sufficient for battery applications.This is the reason why over the years,numerous co-solvents were added to EC in the quest to lower the melting point of electrolytes.34,35Within this context,we performed DSC measurements on the various Na-based electrolytes studied herein in order to determine their solidification temperature and hence their temperature utilization range.The DSC curves for electrolytes previously cooled down to À120 C and then heated up to room temperature at a scan-ning rate of 10 C min À1are shown in Fig.3.They significantly differ from one binary EC-based solvent to another as witnessed by the temperature of the first endothermic peak,related to the first crystallization,which shifts from À25,À50to À75 C for DMC,DEC and DME co-solvents,respectively.Interesting and somewhat puzzling is the absence of an endothermic peak in the DSC heating curve of the PC-based electrolyte.We could solely spot a vitreous transition for temperatures near À95 C.In short,no electrolyte solidification is observed when PC is added as the co-solvent,which is ofTable 1Dielectric constants and measured viscosities of the solvents Solvents Dielectric constant Viscosity (cP)EC 89.78##PC64.92 2.2Triglyme 7.5 1.6DME 7.20.7DMC 3.1070.5DEC2.8050.3Fig.3DSC heating curves of electrolytes after cooling the sample down to À120 C.(a)Electrolytes based on 1M NaClO 4dissolved in various solvent mixtures and (b)PC based electrolytes with 1M of various Na salts.D o w n l o a d e d b y S h a n g h a i I n s t i t u t e o f C e r a m i c s , C A S o n 07 S e p t e m b e r 2012P u b l i s h e d o n 13 J u n e 2012 o n h t t p ://p u b s .r s c .o r g | d o i :10.1039/C 2E E 22258Bprimary importance for low temperature application since the electrolyte remains in liquid form at very low temperatures behaving seemingly like ionic liquid-based electrolytes.36This does not come as a total surprise as PC is frequently used as a co-solvent in Li-ion battery electrolytes to lower their operating temperatures.The vitreous transition is observed for all PC-based electrolytes studied in this work (cf.Fig.3b),whatever the nature of the salts,and was never reported to our knowledge for Li-based electrolytes.This finding is somewhat intriguing and efforts are currently underway to understand the origin of this phenomenon.The thermal stability of the solvents and electrolytes at high temperature was also evaluated by heating from room temper-ature to 350 C (Fig.4).Two different endothermic peaks are observed for EC:DME;EC:DMC and EC:DEC mixtures without any sodium salt (cf.Fig.4b)with onset temperatures equal to that of each solvent heated separately (cf.Fig.4a).This indicates that no eutectics are formed within the composition ranges studied.Yet,the EC:PC mixture shows a single exothermic peak around 250 C with the highest thermal stability amongst all the solvent combinations that we have tested.Similar DSC heating curves were recorded for EC:DEC alone and 1M NaClO 4in an EC:DEC electrolyte,the onset temper-ature for the endothermic peak (Fig.4d)being slightly shifted to lower temperatures as expected.The additional tiny endothermic peak appearing at ca.310 C corresponds to the melting point of the salt after evaporation of the solvents as confirmed by DSC experiments performed on salts alone (Fig.4c),which addition-ally indicate that amongst the various Na _based salts NaClO 4is the most thermally stable:NaClO 4(310 C)>NaPF 6(280 C)>NaTFSI (250 C).A similar thermal stability trend was reported for the homologous Li-based salts as a function of the anion withLiClO 4(236 C)being more stable than LiPF 6(200 C)and LiTFSI (234 C),and therefore a lower overall thermal stability owing to their lower Magdelung energy as compared to that of the Na-based salts.3.3.Electrochemical potential windowElectrochemical tests were performed using three-electrode Swagelok cells with Na as counter and reference electrodes.Only high purity aluminum plungers,free of lead,could be used as the working electrode to probe a wide potential window (0to 5.3V),otherwise the formation of Na 15Pb 4(cf.Fig.5)is observed.37The CVs collected for (i)single solvent based electrolytes with 1M NaClO 4,(ii)binary solvent based electrolytes with 1M NaClO 4,and (iii)PC with different salts are reported in Fig.6a–f,respectively.Out of them,three electrolyte combinations stand out as exhibiting poorer electrochemical stability than the others either upon oxidation or reduction.Indeed,a high reduction current is measured for 1M NaClO 4in EC:Triglyme with an onset potential of ca.1.1V vs.Na +/Na 0(cf.Fig.6b)while important oxidation currents are recorded for 1M NaClO 4in THF and 1M NaTFSI in PC with onset potentials of ca.4.26V and 3.3V vs.Na +/Na 0,respectively (cf.Fig.6a and c).In the latter case the observed oxidation current most likely corre-sponds to the well-known corrosion of the Al plunger in contact with TFSI based electrolytes.38This issue can be alleviated by adding a small quantity of NaPF 6(5wt%)to the electrolyte as previously reported in the case of Li-based electrolytes 39(see Fig.6f).When dealing with the remaining single solvent based elec-trolytes with 1M NaClO 4,the electrochemical stability (Fig.6d)decreases following the trend DEC >PC >Triglyme >DME >Fig.4DSC heating curves up to 350 C of (a)solvents alone,(b)solvent mixtures,(c)PC based electrolytes with 1M of various Na salts and (d)EC:DEC solvent mixture with or without a sodium salt (1M NaClO 4).D o w n l o a d e d b y S h a n g h a i I n s t i t u t e o f C e r a m i c s , C A S o n 07 S e p t e m b e r 2012P u b l i s h e d o n 13 J u n e 2012 o n h t t p ://p u b s .r s c .o r g | d o i :10.1039/C 2E E 22258BDMC >THF.Under similar experimental conditions,the elec-trochemical stability of binary solvent-based electrolytes (cf.Fig.6e)decreases following the trend EC:PC >EC:DMC >EC:DME >EC:DEC >EC:Triglyme (Fig.6e).The electro-chemical stability trend of EC-based binary solvents thus differs from the one observed for the co-solvent alone.This indicates,as already pointed out by Tarascon and Guyomard,33that the electrochemical stability of a solvent mixture does not strictly depend on that of each co-solvent since synergetic effects do occur when solvents are mixed.Finally,for a defined solvent mixture,similar results were observed independent of the nature of the salt (Fig.6f).The electrochemical stability window was further studied by performing EIS measurements according to Protocols B and C defined in our standard characterization procedure (cf.Fig.1).Similar trends are observed for all electrolytes tested,and the results obtained for 1M NaClO 4–EC:DMC are shown in Fig.7.The Nyquist plot obtained for Protocol B with EIS measure-ments at 5V vs.Na +/Na 0is similar to the one obtained before any polarization at the OCP (cf.Fig.7a).In contrast,the Nyquist plots for EIS measurements at 0.01V vs.Na +/Na 0present a growing charge transfer resistance contribution upon cycling (semi-circular feature at high frequencies).This appears together with a smaller Warburg diffusion tail.These features are char-acteristic of the growth of a passivation layer (SEI)at the surface of the working electrode.40The fact that these features dis-appeared at each potential sweep of up to 5V vs.Na +/Na 0is consistent with the concomitant disappearance of this SEI layer at this potential.These observations are further confirmed by the EIS measurements within Protocol C,where the potential was held at either 5V or 0.01V vs.Na +/Na 0(cf.Fig.7b and c,respectively).No change was observed when the potential was held at 5V vs.Na +/Na 0(cf.Fig.7b)whereas the growth of the SEI layer at 0.01V vs.Na +/Na 0is clearly observable from the appearance of the above-mentioned characteristic features on the Nyquist plots (cf.Fig.7c).It is also worth mentioning that a fast decrease in the reduction current intensity is observed upon cycling at low potential values (cf.Fig.5b),confirming the formation of a passivation layer which limits the electrolyte reactivity at the WE interface.3.4.Electrolyte selectionFig.8depicts the electrochemical stability window for all elec-trolyte formulations studied,which,as mentioned above,isFig.5CVs (10mV s À1,1M NaClO 4in PC)obtained in three-electrode Swagelok cells of a high purity Al plunger.Inset displays CVs obtained in similar conditions using low purity Al plunger containing lead impurities resulting in Na–Pb alloyformation.Fig.6CVs (10mV s À1,high purity Al plungers)obtained in three-electrode Swagelok cells with (a)1M NaClO 4dissolved in single solvents,(b)1M NaClO 4dissolved in solvent mixtures and (c)various salts (1M)in PC.(d),(e)and (f)are low current zoom plots of (a),(b)and (c),respectively.D o w n l o a d e d b y S h a n g h a i I n s t i t u t e o f C e r a m i c s , C A S o n 07 S e p t e m b e r 2012P u b l i s h e d o n 13 J u n e 2012 o n h t t p ://p u b s .r s c .o r g | d o i :10.1039/C 2E E 22258B。