Cation‐Assisted Lithium‐Ion Transport for High‐Performance PEO‐based Ternary Solid Polymer Electrolytes

Abstract N‐alkyl‐N‐alkyl pyrrolidinium‐based ionic liquids (ILs) are promising candidates as non‐flammable plasticizers for lowering the operation temperature of poly(ethylene oxide) (PEO)‐based solid polymer electrolytes (SPEs), but they present limitations in terms of lithium‐ion transport, such as a much lower lithium transference number. Thus, a pyrrolidinium cation was prepared with an oligo(ethylene oxide) substituent with seven repeating units. We show, by a combination of experimental characterizations and simulations, that the cation's solvating properties allow faster lithium‐ion transport than alkyl‐substituted analogues when incorporated in SPEs. This proceeds not only by accelerating the conduction modes of PEO, but also by enabling new conduction modes linked to the solvation of lithium by a single IL cation. This, combined with favorable interfacial properties versus lithium metal, leads to significantly improved performance on lithium‐metal polymer batteries.


Introduction
Thecurrent electrification of transport and decarbonizing of electricity production pushes battery researchers to explore new battery systems beyond the currently and dominant lithium-ion technology. [1] Lithium metal is considered as the "holy grail" of negative electrodes due to its ultrahigh theoretical specific capacity (i.e.3 860 mAh g À1 vs. 372 mAh g À1 for state-of-the-art graphite electrodes) and its very low standard reduction potential (À3.04 Vv s. standard hydrogen electrode). [2] Many challenges,h owever,l imit aw idespread deployment of rechargeable lithium-metal batteries (LMBs), including inhomogeneous electrodeposition of lithium metal, leading to the formation of high surface area lithium (HSAL). [3,4] Theformation of HSAL in the form of dendrites present serious safety hazards,e specially when highly flammable organic liquid electrolytes are utilized as dendrites might readily penetrate the separator and induce internal short circuits. [3][4][5] Therefore,a lternative electrolytes with high ionic conductivity,y et with better mechanical, chemical, electrochemical, and thermal stability than liquid electrolytes must be developed to facilitate the adoption of LMBs at large scale.P oly(ethylene oxide) (PEO)/lithium salt complexes are promising candidates in this respect, although they have been studied for more than 50 years as solid polymer electrolytes (SPEs). [6,7] Theg ood mechanical stability of cross-linked PEO systems,t heir wide electrochemical stability window (ESW) and the excellent ability of PEO chains to dissolve lithium salts are all suitable for au se in LMBs. [6] PEO/lithium bis(trifluoromethanesulfonyl)imide (LiTFSI) salt complexes are able to reach ionic conductivities up to 10 À3 Scm À1 at 80 8 8C, but the ionic conductivity drops below useful values at low temperatures, [8] (i.e. % 10 À4 Scm À1 at 40 8 8Cf or the best amorphous complexes [9] ), thereby preventing the use of "dry" PEO-based SPEs at room temperature so far. Af easible solution consists in adding plasticizers to increase ionic mobility. [10] Ionic liquids (ILs) are promising in this respect due to their ultra-low vapor pressure, broad ESW,h igh thermal and chemical stability and nonflammability. [11] PEO/Li salt/IL ternary solid polymer electrolytes (TSPEs) can reach ionic conductivities of up to 10 À3 Scm À1 at 40 8 8Cw ith ILs based on N-alkyl-N-methylpyrrolidinium (Pyr 1,x TFSI, x being the number of carbon atoms in the longer alkyl chain). [12] Molecular dynamic (MD) simulations by Diddens et al. [13][14][15] showed that the enhanced ionic conductivity results from the increased segmental mobility of PEO chains.A sfor binary PEO/salt complexes, the actual lithium transport still occurs mainly along the polymer chains [13][14][15] rather than involving solvation sphere including the anion or oxygen from other polymer chains, since PEO preferentially solvates Li + cations via consecutive oligo(ethylene oxide) units.L i + ion-transport modes can be differentiated between "structural" transport, such as that occurring along the PEO chain by exchanges of Li + ion solvating units in its dynamic solvation sphere,a nd "vehicular" transport, corresponding to the transport of the Li + ion with its solvation sphere before any exchange occurs (more predominant, for instance,i no ligo(ethylene oxide) liquid electrolytes). [16] Alkyl pyrrolidinium-based ILs do not interact much with Li + ions since the anion is far less coordinating than PEO.A saresult, PEO/Li salts conductive paths are "diluted" by the IL addition and the Li-PEO interaction is in some cases increased, slowing down Li + ion transport. [13][14][15] This,t ogether with the addition of IL ions that do not participate in the Li + ion transport, results in ar educed lithium transference number (t Li + ), which limits the potential performance gains when employing these electrolytes in LMBs.
Therefore,todecouple Li + ion transport, at least partially, from the segmental mobility of PEO chains,solvating ILs are apriori more suitable for enabling further transport modes as illustrated in Figure 1( e.g. structural or vehicular modes involving IL species in Li + ion solvation spheres). In particular,t he presence of solvating cations could advantageously accelerate Li + ion transport via the formation of complexes with two positive charges overall. We reported previously on improved interaction of Li + ion with short oligo(ethylene oxide) chains-substituted pyrrolidinium-based ILs in liquid electrolytes.T he data suggested that the first oxygen from the nitrogen center is not interacting with lithium and that longer oligo(ethylene oxide) chains are necessary for reaching the full solvation of Li + ions by asingle cation considering that Li + ion complexes have ap referred coordination number of 6. [17][18][19] Aiming at the Li + ion solvation by as ingle pyrrolidinium cation and thus enabling vehicular transport by the formation of charged complex with two positive charges,w er eport here on the synthesis of N-methyl-N-oligo(ethylene oxide) pyrrolidinium TFSI with amedian oligo(ethylene oxide) chain length of seven repeating units (noted Pyr 1,(2O)7 TFSI;S upporting Information). We compared the designed IL with Pyr 1,4 TFSI in terms of physiochemical properties and Li + ion-IL interaction in binary liquid electrolyte as well as in TSPEs.Lithium mobility and its conduction mode in TSPEs were analyzed by electrochemical measurements,p ulsed field gradient nuclear magnetic resonance (PFG-NMR) spectroscopy and MD simulation. Practical improvements were then verified by testing the rate performance and cycling stability of Li k Li and LiFePO 4 (LFP) k Li cells.

Results and Discussion
Physicochemical properties and Li + ion coordination in liquid binary Pyr 1,(2O)7 TFSI-based and PEO-based TSPEs Fort he fundamental understanding of the coordination process and its evidence,P yr 1,(2O)7 TFSI was characterized as liquid binary electrolyte LiTFSI:IL (mol:mol) prior to membrane preparation and characterization.
Thei onic conductivity and the viscosity of the pure Pyr 1,(2O)7 TFSI is compared with that of Pyr 1,4 TFSI (Figure 2a and inset). DSC thermograms of the liquid electrolytes with LiTFSI are shown in Figure 2b.T he large size of Pyr 1,(2O)7 TFSI leads expectedly to ah igher viscosity and consequently to al ower ionic conductivity than Pyr 1,4 TFSI.
Interestingly though, when mixed with LiTFSI at high mole fractions,t he binary mixtures are all amorphous (Figure 2b), whereas the 1:1LiTFSI:Pyr 1,4 TFSI complex exhibits two melting transitions corresponding to crystalline phases. [20] As for the glass transition point (T g ), although it increases with the salt content as aresult of increased interactions and decreased ion mobility,t he 1:1m ixture still exhibits a T g of À48 8 8Ct hat is remarkably low for such ah igh salt concentration and hinting at aw ell-preserved ion mobility as compared with the Pyr 1,4 TFSI complex that exhibits a1 28 8C higher T g ,e ven though Pyr 1,4 TFSI has a T g of À86 8 8C [21] (vs. -69.0 8 8Cfor Pyr 1,(2O)7 TFSI). At high temperature,TGA results show aslightly earlier onset of Pyr 1,(2O)7 TFSI weight loss.The TSPE 20:2:1 O7 ,c ross-linked or linear, exhibits at hermal stability very close to that of Pyr 1,4 TFSI TSPEs (Supporting Information, Figure S2a).
Raman spectroscopy allows deriving the Li + ion solvation in TFSI-based electrolytes,s ince the TFSI À bands are sensitive tools for analyzing the ionic coordination. Theband at 742-744 cm À1 is attributed to "free" (uncoordinated) TFSI À whereas ab and at 747-750 cm À1 is associated with coordinated TFSI. [22] Ther elative ratio of the two bands (for the corresponding Raman spectra, Figure S1a) allows quantifying the amount of TFSI À coordinated to Li + (Figure 2c)a nd deriving the number of coordinating TFSI À per Li + solvation shell ( Figure 2d). ForL iTFSI mole fractions lower than 0.33 (i.e.c orresponding to the first high melting crystalline phase of the LiTFSI/Pyr 1,4 TFSI system [20] ), there is only as mall fraction of coordinated TFSI À anion in the LiTFSI/ Pyr 1,(2O)7 TFSI mixtures up to 1:1. In contrast, the Pyr 1,4 TFSI electrolytes show alinear increase of coordinated TFSI À from 1:10 to 1:2( am etastable liquid in the experimental conditions) where 60 %o fT FSI À anions are coordinated to Li + ions.I nt erms of TFSI À per Li + ions coordination shell, the Pyr 1,4 TFSI electrolytes shows approximately two TFSI À per Li + ,similarly to previous reports, [18] slightly decreasing at the highest concentrations but limited to % 1.7 (TFSI À /Li + )f or the 1:2electrolyte.For Pyr 1,(2O)7 TFSI, on the other hand, even at avery high 1:1mole ratio we calculate only 0.24 TFSI À per Li + ,which shows that most Li + do not have any TFSI À in their solvation spheres (only 12 %c oordinated TFSI À ) and are thus mostly coordinated by the cations. Then umber of TFSI À per solvation sphere only significantly increases above 1:1m olar ratio (i.e. above the saturation of the oligo(ethylene oxide) solvating sites) with arather linear trend, corroborating our initial hypothesis that seven repeating units are required for the pyrrolidinium side chain to fully solvating one Li + ion.
Thee ffect of Pyr 1,(2O)7 TFSI cationic solvation on the physicochemical properties of TSPE membranes was then investigated. Thep reparation of the TSPEs is described in the Supporting Information. In the following,t he TSPEs are labeled using their PEO:LiTFSI:IL molar stoichiometry (the number for PEO corresponds to the number of repeating -(CH 2 ) 2 O-units) and the IL used is indicated as subscript (O7 for Pyr 1,(2O)7 TFSI and 1,4 for Py 1,4 TFSI). Thep refix (cl-) is used for TSPEs that have been cross-linked by UV irradiation. Cross-linking is typically introduced used to improve the mechanical properties of "dry" and plasticized polymer electrolytes,p reventing creeping under pressure, while also hindering membrane crystallization. [23] Theionic conductivities and DSC thermograms of cross-linked polymer membranes (those of the linear TSPEs can be found in Figure S2b,c) are shown in Figure 3a,b.Ahigher IL content increases crystallinity,e ven though TSPEs exhibit only very limited crystallinity (very small peaks compared to the T g steps). In fact, only the membrane with the highest IL content (cl-20:2:4) shows atransition in the ionic conductivity curves (while still exhibiting aw ell-maintained ionic conductivity (for detailed values see the Supporting Information, Table S1) below the melting transition). Cross-linking allows increasing the IL content in the TSPEs to lift the ionic conductivities even higher (6.6 10 À4 Scm À1 and 1.4 10 À3 Scm À1 at, respectively 40 8 8Cand 60 8 8Cfor cl-20:2:4 O7 ). Even though higher ionic conductivities can be reached for the cl-20:2:4 O7 membrane and IL phase separation might be favorable to improve the performance and wetting [24] in LFP k Li cells,T SPEs with lower IL content were used in the following to avoid partial crystallization at an operating temperature of 40 8 8C.

Lithium-ion mobility and conduction modes in PEO-based TSPEs
Them obility of each ionic species based on the selfdiffusion coefficients derived from PFG-NMR experiments ( Figure 4a).
Switching from Pyr 1,4 TFSI to Pyr 1,(2O)7 TFSI in 20:2:1 TSPEs results in am oderate relative increase of TFSI À mobility,w ith as ignificantly higher relative increase of lithium diffusion coefficient in (i.e.a98 %i ncrease from  Angewandte Chemie Forschungsartikel 0.292 10 À12 to 0.578 10 À12 m 2 s À1 ), whereas the ion mobility of the larger pyrrolidinium cation is lower. This shows that the interaction of Li + with the immobile PEO matrix is lowered and suggests that in TSPEs,I Lc ation-Li + complexes are formed assisting lithium-ion transport. Cross-linking does not affect considerably the results (decrease of less than 7% for cl-20:2:1 O7 vs.the linear TSPE). Increasing the IL content to 20:2:2 yields af urther increase of diffusion coefficients,i n accordance with ionic conductivity.T he Li + transference numbers were estimated by using either the self-diffusion coefficients or the "Bruce and Vincent" electrochemical technique [25] (Figure S3a). Theresults (Table S2) vary slightly depending on the utilized method used but are consistent within the error margin as shown in Figure 4b.Bysubstituting Pyr 1,4 TFSI with Pyr 1,(2O)7 TFSI in 20:2:1 TSPE, ad oubling of t Li + can be reached at 40 8 8C( i.e.f rom 0.05 AE 0.01 to 0.10 AE 0.01) and even higher for the cross-linked TSPEs.T he increase of t Li + is likely due to the enhanced solvation by the Pyr 1,(2O)7 + cation, affording as imilar t Li + compared to the 20:2:0 "dry" SPE. Increasing the IL content in the 20:2:2 and cl-20:2:2 membranes keeps t Li + on asimilar level to 20:2:1 but boosts the Li + ion conductivity even higher because of the higher ionic conductivity.
To unravel the molecular lithium-ion transport mechanism, MD simulations were performed. We extended our previous study [19] by taking into account longer oligo(ethylene oxide) side chains and explicitly focusing on the correlated motion between Li + and pyrrolidinium ions.Inparticular,we simulated TSPEs with oligo(ethylene oxide)-based ILs with side chain lengths of one,f our,a nd eight ethylene oxide monomers at amolar ratio of 20:2:1.
In the MD simulations,w en ote that 1.0 %, 6.2 %a nd 28.2 %ofall lithium ions are not coordinated to PEO chains for 20:2:1 O1 ,2 0:2:1 O4 ,a nd 20:2:1 O8 ,r espectively (p IL in Table 1). This is in good agreement with the observations from Figure 2c,r eflecting that for sufficiently long oligo(ethylene oxide) side chains,the IL cations preferentially coordinate the Li + ions.F or 20:2:1 O1 and 20:2:1 O4 ,the values in Table 1a re slightly lower than reported previously for the corresponding systems with ac oncentration of 20:2:4 due to the lower amount of oligo(ethylene oxide)-based IL. [19] In conventional PEO-based polymer electrolytes,t hree different transport mechanisms are generally identified: [26] First, motion of ac oordinated lithium ion along the backbone of aP EO chain, second, the cooperative motion of polymer segments with the coordinated Li + ions,and third, the transfer of Li + ions between two different PEO chains.P reviously,w ed eveloped an analytical lithium-ion transport model, in which the significance of each of the above mentioned transport mechanism is characterized by specific time scales t 1 , t 2 ,a nd t 3 ,w hich can be extracted from MD simulations.Here, t 1 is the time agiven Li + requires to explore the PEO chain by diffusing along its backbone, t 2 denotes the relaxation time of the polymer segments bound to Li + ions, while t 3 indicates the average residence time of aLi + ion at agiven chain. Theparticular benefit of our model comprises the extrapolation of the D Li to the experimentally important limit N !1 .F urther details on the transport model, the extraction of the time scales and the calculation of the D Li can be found in the Supporting Information.
From Table 1, we observe that t 1 decreases with increasing oligo(ethylene oxide) side chain length of the IL cation, that is,t he Li + ion motion along the backbone becomes more efficient as the Li + ions become progressively coordinated to the IL cations (see discussion of p IL ), rendering the ions that remain coordinated to PEO more mobile.I nterestingly,t he segmental dynamics expressed by the Rouse time t R [27] (all monomers) and t 2 (bound monomers) are approximately constant for all side chain lengths.I nt his context, we demonstrated that ILs act as plasticizers in PEO membranes, which increases the dynamics of the lithium ions attached to the polymer chains as the polymer motion itself is enhanced by the plasticizer. [13][14][15] The t 2 values from Table 1suggest that such ap lasticizing effect is comparable for all simulated ILs, and that increasing the oligo(ethylene oxide) chain length only marginally increases t 2 .W eobserve that t 3 decreases by almost af actor of two when going from 20:2:1 O1 to 20:2:1 O8 , illustrating that, for sufficiently long oligo(ethylene oxide) chains at the IL cation, Li + is structurally and dynamically Figure 4. a) Self-diffusionc oefficients of different ionic species in polymer electrolytes ( 1 Hfor pyrrolidinium-based IL, 7 Li for lithium ion, and 19 Ffor TFSI À )measured by PFG-NMR with an estimated error of AE 2% relative to calibration. Dashed arrows indicate the percentage increase between values. b) Overviewo fLi + ion transference numbers determined from either electrochemical data or from PFG-NMR for Pyr 1,4 TFSI-, Pyr 1,(2O)7 TFSI-basedT SPEs, and for the 20:2:0 SPE. All samples were measured at 40 8 8C.  (Table 1). These observations demonstrate that the ether-functionalized IL cations decouple the lithium ions from the PEO chains and serve as molecular shuttles in this way. As in our previous work, [19] we used our transport model to compute D Li in the limit of long chains (Table 1; Supporting Information). We observe that when going from the essentially non-coordinating IL cation with one ethylene oxide monomer only (i.e.2 0:2:1 O1 )t o2 0:2:1 O4 or 20:2:1 O8 , D Li increases by factors of 1.7 and 3.7, respectively.A lthough the absolute D Li values in Table 1c annot be directly compared to the experimental PFG-NMR values due to higher temperature in the MD simulations,t hese findings nonetheless clearly confirm the trends for D Li observed in Figure 4a,w here similar factors are found.
So far,wefocused on the self-diffusion of the lithium ions. However,f or quantities such as the ionic conductivity or the transference number (when determined via the electrochemical method), the cooperative motion of distinct ions-as expressed by dynamical ion correlations-is important as well. [28][29][30][31] Thei onic conductivity s can then be derived from equilibrium simulations via the expression [28][29][30][31] s ¼ lim where M is the total number of ion pairs in the simulation box with volume V,eis the elementary charge, Dt the observation or lag time, k B T the thermal energy, z i and z j the (integer) charge numbers and Dr i and Dr j the displacement vectors of ions i and j. Thet erms with i = j correspond to the selfdiffusion contribution, whereas the dynamical ion correlations are expressed by the terms with i ¼ 6 j. Atypical example for the impact of these correlations includes the cooperative motion of cation-anion pairs,w hich decreases s because the product z i z j is negative and the dot product hDr i Dr j i positive due to the cooperative motion of the ions.Inthe present case, however, one would expect that the IL cations carry the lithium ions over larger distances,r esembling as huttling mechanism and thus an increase of the ionic conductivity and the Li + transference number. Unfortunately,extracting absolute values for ionic conductivities or transference numbers from MD data is challenging for statistical reasons,especially for ap olymer host with large intrinsic relaxation times (this contrasts the single-ion diffusion, which can be averaged over individual ions).
To nonetheless probe the correlated motion of IL cations and Li + ions,w ec omputed hDr Li + (Dt)Dr Pyr (Dt)i c for all initially coordinated pairs of lithium ions and IL cations (i.e., when their initial center-of-mass separation was not larger than 7 )a saf unction of Dt (Figure 5a,c ompare to illustration in Figure 5c). Them ean squared displacement (MSD) hDr 2 Li þ i averaged over all lithium ions is also shown for comparison (dashed curves). We find that hDr Li + Dr Pyr i c increases substantially when increasing the oligo(ethylene oxide) side chain length and becomes even comparable to the lithium ion MSD for al ength of eight monomers.T his suggests that the correlated motions between Li + ions and IL cations significantly contributes to the overall lithium-ion conductivity,a so bserved from the experimentally determined transference numbers of 20:2:1 1,4 and 20:2:1 O7 (Figure 4b).
Figure 5b also shows hDr Li + Dr Pyr i c normalized by the geometric mean of the MSDs of the lithium ions and the IL cations.W ithin this representation, ap erfectly correlated motion between initially coordinated ions of either species would result in avalue of one.From Figure 5b,wefind values of up to 0.6 on short time scales for 20:2:1 O8 .V alues below unity presumably occur due to the rotation of the Li + /IL cation complexes and the internal degrees of freedom of the side chain, nonetheless,t he correlation is significant. This is also reflected by the fact, that the relative correlation for shorter oligo(ethylene oxide) chains is substantially lower. On larger time scales,t he correlation diminishes due to the exchange of Li + /IL cation pairs.T his decay is slowest for 20:2:1 O8 ,reflected by an average pair lifetime of about 100 ns, followed by 20:2:1 O4 ( % 10 ns) and 20:2:1 O1 (0.07 ns;S upporting Information).

Forschungsartikel
Our simulations confirm our approach to employ functionalized pyrrolidinium cations with sufficiently long oligo(ethylene oxide) side chains,w hich can indeed carry Li + ions over larger distances (nanometers), significantly enhancing both the single-ion transport (D Li )a nd the cooperative motion between Li + and IL cations in this way.
Practical properties and electrochemical performance of crosslinked TSPEs for Li k Li and LFP k Li cells Electrochemical requirements for any electrolyte,e ither polymer or liquid, are its ability to facilitate sufficient ion transport at the electrode j electrolyte interface while being subjected to as little electrochemical degradation (anodic/ cathodic) as possible.T he cross-linked samples,c l-20:2:1 O7 and cl-20:2:1 1,4 were explored in terms of ESW on copper for cathodic stability and on stainless steel and LiNi 1.5 Mn 0.5 O 4 (LNMO) for anodic stability (Figure 6). An important property of an electrolyte includes its partial electrochemical reduction at the Li j electrolyte interface to ensure an effective solid electrolyte interphase (SEI) [32] formation, to inhibit further electrolyte reduction and to enable stable longterm cycling. Thef irst reductive peak appears at % 1.6 V (vs.L i j Li + ), which is commonly observed. [33] Thep eak appears for both TSPEs,i ndicating an egligible influence of oligo(ethylene oxide) side chain (Figure 6i nsert) and no significant peaks are seen before lithium electrodeposition. On the anodic scan, the voltamperograms are comparable with overlapping curves and an anodic stability up to 5.2 V (vs.Lij Li + )close to the ESW of pure Pyr 1,(2O)7 TFSI (5.0 Vvs. Li j Li + ; Figure S3a) as well, considering a0.1 mA cm À2 limit in both cases.N evertheless,c ommon lithium-metal polymer batteries (LMPB) composite electrodes,such as LFP,exhibit higher surface area and av astly different surface reactivity and the TSPEs stability was thus probed using ahigh voltage spinel LNMO composite electrode as well. It resulted in as imilar oxidative stability for both TSPEs of % 4.6 V (vs.L ij Li + ).
TheTSPEs performance in terms of galvanostatic lithium plating/stripping,w hich depends strongly on their interfacial properties toward Li metal, was investigated in symmetrical Li k Li cells and is reported in Figure 7a for the crosslinked TSPEs and in Figure S6 for 20:2:0 and 20:2:1 O7 ). The decrease in voltage over time is likely due to some evolution of SEI resistance during cycling. At ac urrent density of 0.1 mA cm À2 ,t he cell voltage is reduced by % 35 %f or cl-20:2:1 O7 compared to cl-20:2:1 1,4 (0.15 vs.0.23 V) at 40 8 8Cand the effect is even more marked at 60 8 8C ( Figure S4a,b). A closer look at the voltage profile in Figure 7b shows that the voltage can be divided into two main contributions.T he IR drop (DV IR drop )atthe beginning of each step is assigned to the initial resistance,d ominated by the SEI, the contact and electrolyte resistance.T he second part is an asymptotical voltage increase during each step,w hich is attributed to the establishment of ionic concentration gradients (DV polarization ) that become steeper and steeper until as teady-state is reached (or the lithium concentration reaches 0atthe plated electrode,a tw hich point fast dendrite growth becomes inevitable [34] ). Since gradient formation depends on Li + ion mobility,they are strongly influenced by t Li + . Thedifferences in terms of initial IR drop are rather limited and do not show any obvious trend which was also verified by afollowing cell impedance evolution either during galvanostatic cycling or at rest ( Figure S4c,d). Asignificantly lower voltage plateau due to al ower DV polarization can be observed for cl-20:2:1 O7 (Figure 7b dashed lines). It is half compared to cl-20:2:1 1,4 (that does not allow reaching steady-state in one hour). These improvements result in no short circuit after 1000 hf or cl-20:2:1 O7 .I nf act, cl-20:2:1 O7 also allows maintaining higher steady-state currents and, as ar esult, obtaining more homogenous plating onto Cu ( Figure S7). To verify that the prevention of as hort circuit does not result from increased mechanical stability,d ynamic shear rheometry experiments were performed (Figure 7c). Thedetermination of the linear viscoelastic range (LVE) is reported in Figure S5a,b.The two TSPEs cl-20:2:1 1,4 and cl-20:2:1 O7 show comparable storage and loss moduli values.I ti ndicates that the mechanical stability differences have,atbest, aminor influence on the Li cycling behavior.I nc omparison to linear TSPEs,afurther advantage of the cross-linking is the much improved elasticity and limited deformation at low frequencies as seen Figure S5c,d. Finally,the effect of the improved Li + mobility of the new TSPE on LMBs performance was verified by cycling in LFP k Li cells.F igure 7d compares the cycling stability of LFP k Li cells with either ac ross-linked Pyr 1(2O)7 TFSI TSPE or aP yr 1,4 TFSI analog.T he initial Coulombic efficiency (CE) increased from 91.6 %for cl-20:2:1 1,4 to 95.7 %for cl-20:2:1 O7 and from 99.93 %t o9 9.96 %i nt he following cycles,w hich would have ah ighly beneficial effect on cycle life with as maller Li metal electrode.B eside the excellent CEs,t he specific discharge capacities are higher for cl-20:2:1 O7 (151 mAh g À1 in the 5 th cycle,). Most importantly,t he longterm capacity retention is considerably improved. After 200 cycles,t he cl-20:2:1 O7 cell retains 99.3 %o fi ts initial capacity (referred to the 5 th cycle) vs.6 7.2 %f or cl-20:2:1 1,4 ). Ther easons for this can be attributed to the slower Li + ion transport in cl-20.2:1 1,4 .Although LFP is known for avery flat voltage plateau at 3.4 V(vs.Lij Li + ), [35] the plateau of the cl-20:2:1 1,4 cells is more sloped than that of the cl-20:2:1 O7 cell (Figure 7e), which results in alower capacity at 0.5C in the 3 rd cycle.Inaddition, the plateau becomes increasingly sloped as an effect of the increasing cell polarization over cycling which explains the capacity decay.I tl ikely results from the formation of HSAL at ac urrent density of % 0.1 mA cm À2 , as seen in Li k Li experiments and, over cycling,toadegraded transport at the Li j electrolyte interface.I nc ontrast, the cl-20:2:1 O7 cell exhibits flat plateaus with constant cell voltage over cycling.D ue to the increasingly slower lithium-ion transport in the cl-20:2:1 1,4 cell, the cut-off voltage is reached faster and faster, and the capacity decays.This is also visible in the discharge rate performance of the cells (Figure 7f). The capacity retention already decreases to 92.4 %from 0.05 Cto 0.1 C(related to 1 st cycle) for cl-20:2:1 1,4 compared to 99.8 % with cl-20:2:1 O7 .Even at 1C,acapacity retention of 96.8 %is reached for cl-20:2:1 O7 .A t2C, the differences in capacity retention are even stronger with 50.0 %f or cl-20:2:1 1,4 vs. 93.5 %f or cl-20:2:1 O7 .T his clearly illustrates the improved rate performance induced by Pyr 1,(2O)7 TFSI. In both cells,the initial specific discharge capacity at 0.1 Ccan be reached after the rate test, showing that at constant charge rate,t he cells (i.e.the lithium-metal electrode) were not overly affected by the discharge (i.e.l ithium electrodissolution). It shows, however, that the much faster lithium transport results in much higher rate performance capability of the LMPB cells, regardless of the degradation of the lithium-metal interface.

Conclusion
Pyr 1,(2O)7 TFSI was specifically designed with an oligo(ethylene oxide) side chain on its pyrrolidinium cation to overcome the limitations of N-alkyl-N-alkyl pyrrolidiniumbased ILs in terms of t Li + when used as plasticizers for PEObased TSPEs.T his IL is not only highly promising for formulating super-concentrated binary liquid electrolytes, since it allows reaching a3:1 (LiTFSI:Pyr 1,(2O)7 TFSI mol:mol) liquid composition, but it also enables PEO-based polymer membranes with far higher performance than Pyr 1,4 TFSI analogues,a lthough having similar physicochemical properties.Inparticular, the cross-linked TSPE cl-20:2:1 O7 possesses as imilar ionic conductivity as the state-of-the-art cl-20:2:1 1,4 electrolyte but exhibits aL i + ion conductivity three times higher with a t Li + of 0.10 AE 0.01 at 40 8 8C(vs.0.03 AE 0.01 for cl-20:2:1 1,4 ). This increase in t Li + reflects the excellent solvating properties of Pyr 1,(2O)7 TFSI that enable fast lithium-ion transport, especially via enabling an additional "vehicular" transport mode since the IL cation is able to solvate one Li + ion. Our MD simulations confirmed the cooperative motion of Li + ions and IL cations both via asimplified model for the Li + ion transport and an explicit analysis of dynamical ion correlations.T hese insights show that, the use of ionic shuttle molecules opens up new avenues to improve the ion-transport properties of TSPEs.P ractically,t hermally stable and elastic membranes allow significantly higher rate performance of LFP k Li LMPB cells.L ong-term cycling of symmetrical Li k Li and LFP k Li cells show that the faster lithium transport results in well-maintained cell performance with no deterioration over cycling of the lithium transport, contrary to what happens with Pyr 1,4 TFSI-based TSPEs in the same conditions. In the latter case,t he increasingly slower lithium transport over cycling is attributed to the evolution of the lithium-metal interface as slow lithium transport favors the formation of HSAL. On the contrary,a99.3 %c apacity retention is reached with cl-20:2:1 O7 (vs.6 7.2 %f or cl-20:2:1 1,4 )a fter 200 cycles.T his demonstrates that faster lithium transport results not only in higher power capability but also in safer and longer living LMPB cells.