Understanding why replacing I3−/I− by cobalt(ii)/(iii) electrolytes in bis(diimine)copper(i)-based dye-sensitized solar cells improves performance

The performances of dye-sensitized solar cells (DSCs) comprising heteroleptic bis(diimine)copper(I) based dyes combined with either [Co(bpy)3] , [Co(phen)3] 2+/3+ or I3 /I redox mediators (bpy 1⁄4 2,20bipyridine, phen 1⁄4 1,10-phenanthroline) have been evaluated. The copper(I) dyes contain the anchoring ligand ((6,60-dimethyl-[2,20-bipyridine]-4,40-diyl)bis(4,1-phenylene))bis(phosphonic acid), 1, and an ancillary ligand (2, 3 or 4) with a 2,9-dimethyl-1,10-phenanthroline metal-binding domain. Ligands 2 and 3 include imidazole 20-functionalities with 4-(diphenylamino)phenyl (2) or 4-(bis(4-n-butoxy) phenylamino)phenyl (3) domains; in 4, the phen unit is substituted in the 4,7-positions with holetransporting 4-(diphenylamino)phenyl groups. The photoconversion efficiency, h, of each of [Cu(1)(2)], [Cu(1)(3)] and [Cu(1)(4)] considerably improves by replacing the I3 /I electrolyte by [Co(bpy)3] 2+/3+ or [Co(phen)3] , and after a change of electrolyte solvent (MeCN to 3-methoxypropionitrile). Due to the faster charge transfer kinetics and more positive redox potential, the cobalt-based electrolytes are superior to the I3 /I electrolyte in terms of open-circuit voltage (VOC), short-circuit current (JSC) and h; values of VOC 1⁄4 594 mV, JSC 1⁄4 9.58 mA cm 2 and h 1⁄4 3.69% (relative to h 1⁄4 7.12% for N719) are achieved for the best performing DSC which contains [Cu(1)(4)] and [Co(bpy)3] . Corresponding values for [Cu(1)(4)] and I3 /I DSCs are 570–580 mV, 5.98–6.37 mA cm 2 and 2.43–2.62%. Electrochemical impedance spectroscopy (EIS) has been used to study DSCs with [Cu(1)(4)] and the three electrolytes. EIS shows that the DSC with I3 /I has the highest recombination resistance, whereas the [Co(phen)3] 2+/3+ electrolyte gives the highest chemical capacitance and VOC and, between [Co(bpy)3] 2+/3+ and [Co(phen)3] , the higher recombination resistance. The [Co(phen)3] 2+/3+


Introduction
The Grätzel dye-sensitized solar cell (DSC) 1,2 converts solar energy into electrical power.][5][6][7][8][9][10][11][12][13][14][15] While the sun provides an environmentally acceptable source of energy, and the concept of DSCs for energy conversion is laudable, a vision of a sustainable future demands that the components of DSCs should also originate from maintainable sources.In 1994, Sauvage and coworkers 16 demonstrated the possibility of combining bis(diimine)copper(I) dyes ([Cu(N^N) 2 ] + ) with wide band-gap n-type semiconductors (TiO 2 or ZnO) for photoconversion.While the photophysical properties of [Cu(N^N) 2 ] + complexes continued to be a focus of attention, 17,18 no real progress in applying this family of complexes as sensitizers in DSCs was made until Sakaki in 2002 19 and our own reports in 2008. 20Since then, there have been signicant advances in the design both of ligands to anchor [Cu(N^N) 2 ] + sensitizers to an FTO/TiO 2 surface and promote electron injection from dye to semiconductor, and of ancillary ligands to improve light absorption and optimize electron transfer between electrolyte and dye. 21,22To date, the best photoconversion efficiencies, h, in copper-based DSCs have been achieved using heteroleptic dyes.Efficiencies passed 3% (fully masked DSCs 23,24 ) in 2014, 25 and the current record is 4.66% reported by Odobel and coworkers; 26 both values are with respect to h $ 7.5% for the reference dye N719.Heteroleptic dyes have the advantage over homoleptic dyes in being able to combine optimally designed anchoring and ancillary domains.
While ligand design is of crucial importance, appropriate combinations of dye and electrolyte are also key to optimizing DSC performance.The maximum voltage generated under illumination (the open-circuit voltage, V OC ) is the difference between the redox potential of the electrolyte and the Fermi level of the TiO 2 semiconductor.The I 3 À /I À redox shuttle is ubiquitous among liquid electrolytes in DSCs and is used in many studies of copper-based DSCs. 21However, the I 3 À /I À electrolyte composition has been optimized for ruthenium(II)-based cells. 27,28A way to enhance the photoconversion efficiencies of DSCs containing copper-dyes is to increase V OC , and this can, in principle, be achieved by employing an electrolyte with a redox potential that is more positive than that of I 3 À /I À .Among the range of iodine-free redox mediators that have been investigated, 29,30 those based on the Co 3+ /Co 2+ redox couple 31,32 are superior.The use of a [Co(phen) 3 ] 3+/2+ (phen ¼ 1,10phenanthroline) electrolyte contributed to the current record DSC photoconversion efficiency of 14.3%. 15The redox potentials of [Co(phen) 3 ] 3+/2+ and [Co(bpy) 3 ] 3+/2+ (bpy ¼ 2,2 0 -bipyridine) are +0.61V and +0.56 V, respectively, 29 compared to +0.35 V for I 3 À /I À (in MeCN and vs. NHE); 28,29 values are solvent dependent and values of +0.72, +0.65 and +0.31 V (vs.NHE), respectively, are also tabulated. 32Drawbacks of cobalt-based electrolytes are the larger size of the [Co(diimine) 3 ] n+ ions which leads to mass transport problems, less efficient charge transfer at the Pt counter electrode compared to that with an I 3 À /I À redox shuttle, and fast recombination of electrons between the photoanode and the oxidized (Co 3+ ) redox couple. 33,34hile the use of cobalt-based electrolytes with rutheniumcontaining and organic dyes is well established, combining the Co 3+ /Co 2+ redox couple with copper-based sensitizers has been little explored.Both we 35 and Ashbrook and Elliott 36 have demonstrated the compatibility of bis(diimine)copper(I) dyes with [Co(4,4 0 -R 2 bpy) 3 ] 3+/2+ (R ¼ H 35 or t Bu 36 ) redox mediators, but no investigations have addressed the optimization of the electrolyte composition.We now report the results of a study of the performances of DSCs containing three heteroleptic [Cu(L anchor )(L ancillary )] + sensitizers in the presence of [Co(bpy) 3 ] 3+/2+ or [Co(phen) 3 ] 3+/2+ electrolytes of varying compositions and reinforce the fact that Co 3+/2+ redox mediators are compatible with copper(I)-based DSCs.We also demonstrate the benets of changing the solvent in the electrolyte from MeCN to the less volatile 3-methoxypropionitrile.were prepared as previously described.

DSC fabrication
The FTO/Pt counter electrodes were purchased from Solaronix (Solaronix Test Cell kit).The FTO/TiO 2 electrodes were prepared in-house by screen printing as previously described; 35 the annealed TiO 2 lm was post-treated with 60 mM aqueous TiCl 4 solution. 35Each working electrode was heated to 500 C and allowed to cool to 80 C prior to dipping into a 1 mM solution of the anchoring ligand 1 (Scheme 1) for 1 day.The electrodes were washed with DMSO and EtOH, and then dried in a stream of N 2 .The electrodes were soaked in a 0.1 mM MeCN solution of [Cu(L ancillary ) 2 ][PF 6 ] (L ancillary ¼ 2, 3 or 4, Scheme 1) for 3 days aer which it was washed with MeCN and dried in N 2 .The DSCs were assembled by combining the working electrode and counter electrode (Solaronix Test Cell Platinum Electrode, heated on a heating plate at 450 C for 30 min to remove volatile organic impurities) by using a thermoplast hot-melt sealing foil (Solaronix, Meltonix 1170-60), then lled with the respective electrolyte by vacuum-backlling.The DSC was closed using thermoplast hot-melt sealing foil and a cover glass.N719 reference electrodes were made by immersing screen-printed 35 FTO/TiO 2 electrodes (post-treated with 40 mmol dm À3 aqueous TiCl 4 ) in a 0.3 mM EtOH solution of N719 (Solaronix) for 3 days.The electrodes were removed from the dye-bath, washed with EtOH, and dried in a stream of N 2 .

DSC measurements
J-V measurements were performed using a SolarSim 150 (Solaronix) or LOT Quantum Design LS0811 sun simulator, which was calibrated with a Si-reference cell to 1000 W m À2 prior to the measurements.All cells were completely masked. 23,24Voltage decay was measured on a ModuLab XM electrochemical system.Electrochemical impedance spectroscopy (EIS) measurements were carried out on a ModuLab® XM PhotoEchem photoelectrochemical measurement system setup from Solartron Analytical.The impedance was measured at steady state close to the open-circuit potential of the cell at different light intensities (LED, 590 nm) in the frequency range 0.05 Hz to 400 kHz using an amplitude of 10 mV.The impedance data was tted to an equivalent circuit model and analysed using ZView® soware from Scribner Associates Inc.

Choice of sensitizers
The heteroleptic copper(I) dyes chosen for the present study are three of a series of sensitizers that we have previously studied in combination with I 3 À /I À electrolyte in DSCs. 38All contain the phosphonic acid anchoring ligand 1 (Scheme 1); for copper(I) dyes, we have found phosphonic acids to be superior over analogous carboxylic acid anchors. 41Phosphonic acid anchors are also used in ruthenium-based dyes, 42 and the strong surface binding is advantageous with respect to carboxylic acid anchors.However, we note that in terms of the rate of electron injection, carboxylic acid anchors may be favoured over phosphonic acids. 43The ancillary ligands 2-4 (Scheme 1) contain a phen metal-binding unit and one or two triphenylaminodomains to enhance light-absorption.Ligands 2 and 3 feature a long alkyl chain to militate against dye-molecule aggregation by intermolecular p-stacking. 44Ligand 3 contains n-butoxy chains on the triphenylamino-unit which are known to improve the performance of DSCs employing triphenylamine-containing organic sensitizers in combination with cobalt electrolytes. 31ue to the steric hindrance of the n-butoxy chains, electron recombination with the electrolyte, which previously was reported to be one of the main issues with this kind of mediator, 39 is signicantly reduced.All ligands contain methyl substituents adjacent to the N,N 0 -coordination site to stabilize the copper(I) dyes (tetrahedral, Scheme 2) against irreversible oxidation to copper(II) (square planar).
The copper(I) dyes were assembled in situ on FTO/TiO 2 electrodes using a 'surfaces-as-ligands' approach which we have established as being effective for complexes which are labile in solution. 21Screen-printed mesoporous TiO 2 electrodes with scattering layer were post-treated with aqueous TiCl 4 solution using conditions that we have previously optimized. 35Each electrode was immersed in a solution of the phosphonic acid  4)] + .By eye, the orange colour of the electrodes was consistent with the presence of adsorbed dye, and this was quantied by solid-state absorption spectroscopy using electrodes assembled as described in the Experimental section but without the scattering layer.The observed absorption maxima (l max $ 470 nm) were consistent with those already reported. 38O/TiO 2 /1 + [Cu(2) 2 ] + / FTO/TiO 2 /[(1)Cu(2)] + + 2 (1) The compositions of the Co 2+ /Co 3+ electrolytes are given in Table 1.Initially, we used MeCN as the solvent and the compositions of electrolytes E1 and E2 are typical of Co 2+ /Co 3+ electrolytes employing bpy or phen ligands.The rst sensitizer to be investigated was [Cu(1)(3)] + .The open-circuit voltage (V OC ), short-circuit current density (J SC ), ll factor (ff) and photoconversion efficiency (h) were measured immediately aer sealing the DSCs; cells were masked, and reproducibility of the data was checked by using duplicate cells (Table 2).Table 2 also gives performance characteristics of a reference DSC containing the standard ruthenium dye N719 combined with an I 3 À /I À electrolyte.The right-hand column in Table 2 gives a relative h, setting the value for an N719 reference DSC to 100%.We routinely use two sun simulators (LOT Quantum Design LS0811 and SolarSim 150 instruments) and, as previously discussed, 45 DSC characteristics for the same cell recorded on the same day on these instruments (both under irradiation of 1 sun) lead to similar values of V OC but different values of J SC .By using relative h values, we can justiably compare data measured on different instruments. 45able 2 shows that there is an increase in both J SC and V OC on changing from [Co(bpy) 3 ] 2+/3+ to [Co(phen) 3 ] 2+/3+ (from electrolyte E1 to E2).The relative efficiencies of 33.8 and 38.1% for the [Co(bpy) 3 ] 2+/3+ -containing DSCs, and 42.5 and 45.0% for the [Co(phen) 3 ] 2+/3+ -containing DSCs are appreciably higher than the 22.5 and 25.7% observed for [Cu(1)(3)] + combined with a standard I 3 À /I À electrolyte (Table S1 †). 38Although the data were extremely promising, we experienced difficulties with the use of MeCN as solvent.Firstly, the DSCs tended to be unstable, performing poorly aer several days.When the dye [Cu(1)(4)] + was combined with electrolyte E1, orange crystals rapidly formed in the sealed DSC (Fig. 1).Rapid crystal growth was a persistent problem in DSCs containing a combination of [Cu(1)(4)] + and [Co(bpy) 3 ] 2+/3+ .Attempts to analyse the crystals by mass spectrometry and X-ray crystallography did not provide denitive identication of the crystalline material.Crystals were also observed in DSCs containing [Cu(1)(4)] + and [Co(phen) 3 ] 2+/3+ , although their growth was slower than in electrolyte E1; a substantial fall in J SC was observed aer two or more days.

À
) (MBI ¼ N-methylbenzimidazole) from electrolyte components in DSCs has previously been reported, 46 but of course, this salt cannot be responsible for the crystals observed in the cobalt-based electrolytes.Other than this latter report, precipitation or crystal formation from liquid electrolytes in DSCs does not appear to have been discussed in detail in the literature. 47 2+ /Co 3+ electrolytes in 3-methoxypropionitrile The problems encountered with the combination of [Cu(1)(4)] + and electrolytes E1 or E2 led us to change the solvent to 3-methoxypropionitrile (MPN) which is widely employed in DSCs. 6,47Electrolyte E3 differs from E1 only in the solvent variation (Table 1).However, the lower solubility of [Co(phen) 3 ][PF 6 ] 3 in MPN compared to MeCN resulted in the use of a 0.044 M solution of [Co(phen) 3 ][PF 6 ] 3 in electrolyte E4 rather than the preferred 0.05 M. The concentrations of the other electrolyte components were reduced to maintain the same molar ratios in E3 and E4 (Table 1).Table 3 gives the performance parameters of masked, duplicate cells for each dye/electrolyte combination, and  the corresponding J-V curves for the better performing DSC from each pair are depicted in Fig. 2. Irrespective of the cobalt complex used in the electrolyte, DSCs containing sensitizers [Cu(1)(3)] + or [Cu(1)(4)] + exhibit higher values of J SC and V OC than those containing [Cu(1)(2)] + .The superior performance of [Cu(1)(4)] + is consistent with the results obtained using an I 3 À /I À electrolyte 38 (Table S1 †).Most importantly, the observed values of J SC for DSCs with cobalt electrolytes are substantially higher than for the same dyes with an I 3 À /I À electrolyte; values of J SC ¼ 9.58 and 9.06 mA cm À2 for [Cu(1)(4)] + with electrolyte E3 (Table 3) compare with a range of 5.98-6.81mA cm À2 for [Cu(1)(4)] + with I 3 À /I À (Tables S1 and  S2 †).For the dyes [Cu(1)(3)] + and [Cu(1)(4)] + , higher values of V OC are observed for electrolyte E4 than for E3, consistent with the more positive redox potential of the [Co(phen) 3 ] 2+/3+ couple versus [Co(bpy) 3 ] 2+/3+ . 29,32On the other hand, higher values of J SC are obtained for both [Cu(1)(3)] + and [Cu(1)(4)] + when the redox mediator is [Co(bpy) 3 ] 2+/3+ .In terms of DSC efficiencies (Table 3), the use of [Co(bpy) 3 ] 2+/3+ is superior to [Co(phen) 3 ] 2+/3+ .Fig. 3 shows the EQE spectra of DSCs containing the dyes combined with the [Co(bpy) 3 ] 2+/3+ redox shuttle.The rst point to note is the differences in the shapes of the EQE spectra compared to the corresponding spectra for the same dyes in combination with I 3 À /I À electrolyte (Fig. S1 †).The gain in EQE in the region between 370 and 420 nm reects the competing light absorption 34,48 of I 3 À which reduces the number of photons being harvested by the copper dyes.Values of EQE max show a marked increase on going from an I 3 redox mediator (Table 4).A comparison of Fig. 3 and S1 † also reveals enhanced quantum efficiency at higher wavelengths (570-620 nm) for all three dyes.The mass transport problem that is known to affect [Co(diimine) 3 ] n+/(nÀ1)+ redox mediators 47 can be investigated by measuring the dependence of J SC for a given DSC on the incident light intensity. 34The DSC parameters for duplicate cells containing [Cu(1)(3)] + or [Cu(1)(4)] + and electrolyte E4 measured under different light intensities are given in Table 5.In keeping with the use of the I 3 À /I À electrolyte in the N719 reference DSC,  (see Table 1).).The spectrum for the better performing DSC of each pair measured for each dye is shown; see also Table 3.
there is an approximately linear dependence of J SC on light intensity and the photoconversion efficiency is essentially constant.However, the mass transport problem associated with [Co(phen) 3 ] 2+/3+ (electrolyte E4) manifests itself in the nonlinear dependence between J SC and light intensity seen in Table 5 for Signicantly, the values of h ¼ 3.47 and 3.69% are the highest efficiencies achieved for a heteroleptic copper(I) sensitizer using the 'surfaces-as-ligands' approach, the previous record being 3.16% with respect to 7.63% for N719. 25ectrochemical impedance spectroscopy (EIS) EIS was used to evaluate the electrochemical behaviour of the DSCs in more detail.This technique allows one to measure the internal impedances of the DSC, and key parameters including the recombination resistance (R rec ), chemical capacitance (C m ) and the electron lifetime (s) (which is the product of R rec and C m ), can be obtained. 49,50Three sets of duplicate DSCs were used in the investigation.All DSCs contained the sensitizer [Cu(1)(4)] + which was shown in the studies described above to be the best performing of [Cu(1)(2)] + , [Cu(1)(3)] + and [Cu(1)(4)] + .Each set of DSCs used a different redox mediator ([Co(bpy) 3 ] 2+/3+ , [Co(phen) 3 ] 2+/3+ or I 3 À /I À ).The electrolyte compositions for the cobalt-based electrolytes are given in Table 1.The DSCs in the EIS study correspond to those in Table 3 with electrolytes E3 and E4, and to the DSCs from a set made using [Cu(1)(4)] + and a standard I 3 À /I À electrolyte (Table S2 †) which showed 2.43 and 2.46% efficiencies.The performances of these I 3 À /I À -containing DSCs are consistent with previously published data. 38Measured EIS data at two different light intensities are presented in Tables 6 and 7. Table 6 shows the impedance spectra of the DSCs, with one cell of each conguration measured at a bias light intensity of 22 mW cm À2 .At this light intensity, the transport resistance (R t ) of the electrons in the semiconducting TiO 2 is negligible.Furthermore, at this light intensity, the V OC values follow the same trend as those found in the J-V measurements under 1 sun (see Tables 3 and S2 †), where electrolyte E4 gives the highest V OC values for [Cu(1)(4)] + .Table 7 shows the parameters obtained at low light intensity, where the transport resistance, R t , can be studied.L d which is the length of electron diffusion "backwards" in the semi-conductor should, in a well performing DSC electrode, be at least as long as the thickness of the porous TiO 2 layer L to minimize back reactions. 50Hence, L d /L > 1 must be fullled, since this implies that transit time in the semi-conductor is shorter than the electron lifetime.Pleasingly, in all DSCs measured in this study (Table 7), L d (which is calculated as the square root of R rec /R t ) is larger than L, and signicantly so for the I 3 À /I À electrolyte cells, due to their higher R rec and lower transport resistance.From Table 6, Fig. 4 and 5c, it can be seen that the diffusion resistance, R d (which is represented as the third arc from the le in each spectrum in Fig. 4a for DSCs with cobalt-electrolytes) is considerably pronounced.This particular arc is, however, barely seen in the spectrum for the DSC containing the I 3 À /I À based redox mediator.The relatively bulky cobalt complexes restrict the mass transport and lower the diffusion coefficient compared to the iodide system. 34,51Furthermore, it is clearly seen that R d is larger for the [Co(phen)] 2+/3+ than for the [Co(bpy) 3 ] 2+/3+ .The recombination resistance, R rec , is generally lower for the DSCs having cobalt electrolytes compared to I 3 À /I À (Table 6, Fig. 5b).This is also seen in the Nyquist plot in Fig. 4a, as the magnitude of R rec (which is represented by the second arc from the le in each spectrum) is signicantly larger for the DSC having the I 3 À /I À electrolyte.This is in agreement with the fast electron transfer from I À to regenerate the dye compared to the lower rate of reduction of I 3 À which minimizes the back reaction interfacial process (higher R rec ). 52For the cobalt-based electrolytes, the consequence is that the faster back reaction kinetics of the cobalt mediators decrease R rec in accordance with the discussion above.However, the simpler outer sphere electron transfer of the cobalt redox mediators (in contrast to the iodide species, which involve the creation and breaking of chemical bonds as opposed to simple electron transfer), 40 give the cobalt DSCs their higher J SC and chemical capacitance, C m .C m is related to the total density of electrons in the semiconductor and typically rises exponentially at higher light intensities; 49 this is also seen in Fig. 5a.R t , which is strongly   dependent on the electrolyte, 49 is higher in the cobalt-based DSCs than in the I 3 À /I À -containing DSCs (Table 7).As the voltage of the cell is higher and R t becomes insignicant, the more positive redox potential of the cobalt mediators and simpler charge transfer kinetics result in a higher C m .The relatively high R rec of the I 3 À /I À -based DSCs results in the larger values of electron lifetime presented in Fig. 5b and Tables 6 and  7; this decreases exponentially upon increasing the light intensity.However, the high C m of the cobalt-based DSCs results in the larger J SC seen for these DSCs (Table 3).
If one considers V OC , R rec and C m for DSCs with [Co(bpy) 3 ] 2+/3+ and [Co(phen) 3 ] 2+/3+ , it appears that the [Co(phen) 3 ] 2+/3+ electrolyte should be superior of the two in terms of DSC efficiency.However, its higher resistance to mass transport is detrimental, limiting the J SC to a greater extent, and resulting in the [Co(bpy) 3 ] 2+/3+ electrolyte having the highest DSC efficiency.On the other hand, in the iodide systems, R d is insignicant, but here, the much lower V OC and, despite the high R rec , the moderate chemical capacitance result in the more modest overall DSC performance.redox mediator; value of h ¼ 3.47 and 3.69% for duplicate DSCs relative to h ¼ 7.12% for an N719 reference DSC are the highest efficiencies achieved for a heteroleptic copper(I) sensitizer using our 'surfaces-as-ligands' approach.The results of an EIS study demonstrate that the improved performance of the [Co(bpy) 3 ] 2+/3+ or [Co(phen) 3 ] 2+/3+ -containing DSCs compared to the devices containing the I 3 À /I À electrolyte is mainly due to the more rapid charge exchange kinetics exhibited by the cobalt redox mediators and their more positive redox potentials.This results in an improved chemical capacitance and V OC .However, on going from I 3 À /I À to cobalt-based cells, the recombination rate increases as a consequence of the faster kinetics.In the present investigation which uses 3-methoxypropionitrile as solvent, [Cu(1)(4)] + combined with the [Co(bpy) 3 ] 2+/3+ electrolyte leads to the highest DSC efficiency.However, given the EIS parameters and the high V OC for the [Co(phen) 3 ] 2+/3+ system, it seems likely that a solvent with a lower viscosity should render the [Co(phen) 3 ] 2+/3+ electrolyte the more advantageous, and we are currently investigating this option.

Fig. 4
Fig. 4 (a) Nyquist plot from EIS measurements of three DSCs having the electrolyte systems E3, E4 and I 3 À /I À , and (b) a zoom-in of the high frequency region.The larger R rec of the I 3 À /I À -based system as well as the pronounced R d of the cobalt-based electrolytes are clearly seen.The graphs presented here are the ones from each of the duplicate cells denoted 'cell 1'.

Table 2
Performance parameters a on the day of sealing the DSCs for duplicate cells (numbered 1 and 2) combining dye [Cu(1)(3)] + with electrolytes E1 or E2 (MeCN solvent, see Table1).A reference N719 DSC was also measured a Measurements were made on a LOT Quantum Design LS0811 sun simulator.b Relative h is relative to N719 set at 100%.

Table 6
Impedance data obtained during measurements at 22 mW cm À2 .EIS measurements are carried out 1 day after DSC assembly.See Table1for E3 and E4 electrolyte compositions a The cell efficiency is that of the particular cell measured with EIS.

Table 7
Impedance data obtained during measurements at 0.44 mW cm À2 .EIS measurements are carried out 1 day after DSC assembly.See Table1for E3 and E4 electrolyte compositions