Fluorine-free water-in-ionomer electrolytes for sustainable lithium-ion batteries

The continuously increasing number and size of lithium-based batteries developed for large-scale applications raise serious environmental concerns. Herein, we address the issues related to electrolyte toxicity and safety by proposing a “water-in-ionomer” type of electrolyte which replaces organic solvents by water and expensive and toxic fluorinated lithium salts by a non-fluorinated, inexpensive and non-toxic superabsorbing ionomer, lithium polyacrylate. Interestingly, the electrochemical stability window of this electrolyte is extended greatly, even for high water contents. Particularly, the gel with 50 wt% ionomer exhibits an electrochemical stability window of 2.6 V vs. platinum and a conductivity of 6.5 mS cm−1 at 20 °C. Structural investigations suggest that the electrolytes locally self-organize and most likely switch local structures with the change of water content, leading to a 50% gel with good conductivity and elastic properties. A LiTi2(PO4)3/LiMn2O4 lithium-ion cell incorporating this electrolyte provided an average discharge voltage > 1.5 V and a specific energy of 77 Wh kg−1, while for an alternative cell chemistry, i.e., TiO2/LiMn2O4, a further enhanced average output voltage of 2.1 V and an initial specific energy of 124.2 Wh kg−1 are achieved.

As can be seen in Fig. 1b, the 30 % LiPAA and the 40 % LiPAA gels exhibit rather similar behavior, with their loss modulus higher than their storage modulus (i.e. a tan δ lower than 1) at low shear strain and above 1 at higher shear strain. Moreover, values close to each other are obtained for the 30 % and 40 % electrolytes and the storage modulus clearly decreases at higher strains. This behavior is less marked for the 40% LiPAA gel already. For the 50 % electrolyte, there is a clear jump in loss and storage modulus and tan δ is above 1 on the whole investigated range, which indicates a more 'elastic' behavior. Moreover, the storage modulus increases slightly with the strain contrary to all the other gels. The 70 % gel shows a further increases for both modulus. However, the loss modulus is also, in this case, above the storage modulus at low deformations, although both converge to the same values at higher deformation, and decrease. Thus, it seems that the 50 % LiPAA gel exhibits a more elastic behavior and peculiar mechanical properties as compared to the gels with both lower and higher LiPAA contents, which might correspond to different local structure. The viscosity of the 15 % LiPAA electrolyte was measured in rotation mode with the same cone geometry and the results are shown in Supplementary Fig. 13. As can be seen, the solution exhibits a non-newtonian behavior, with clear shear thinning with increasing sheer rate, which also indicates a structured electrolyte in this case.

Supplementary Note 2. Determination of Vogel-Tammann-Fulcher (VTF) paramaters
The conductivity curves were fitted with the following version of the Vogel-Tamman-Fulcher (VTF) equation to extract the parameters σ0 and Ea (of the same dimension as the Arrhenius' activation energy and sometimes considered as such 1 ) and T0 = 0 The conductivity curves were first linearized by plotting The T0 values were varied by 0.1 K increments until reaching a maximum for the R 2 of the linear fit. The parameters as well as the R 2 values are reported in Supplementary Table 2.
As can be seen, as the concentration of LiPAA decreases, so does σ0 (which depends on the concentration of mobile charged species participating to conduction). The T0 value, which usually scales with Tg and the mobility of the electrolytes are in a rather narrow range, with only a marked decrease from 70% LiPAA to 50%. The values are then surprisingly similar (given the differences in mechanical properties (see Fig. 1b), which usually translate into differences in Tgs and T0 for 25 homogeneous gels). It is thus likely that the mechanical properties of the gels are linked to the local structure of the polymer domains rather than to the mobility within the conductive domains. Ea and σ0 increase rather linearly with the LiPAA content, in accordance with the increasing density of charge carrier (for σ0) and a more difficult movement of ions.

Supplementary Note 3. Determination of Li + transference numbers
The 1 H and 7 Li diffusion coefficients (DH and DLi) of LiPAA samples were measured at 293 K using pulsed-field-gradient NMR (PFG-NMR) with a diffBB probe, employing a stimulated echo pulse sequence (ledgp2s), on a 600 MHz Bruker NMR spectrometer with a permanent field strength of 14.1 T. The results are given in Supplementary Table 5. The maximum gradient strength was 20 T/m. Lithium transference number T + was obtained by the ratio of DLi/(DLi + DH). The diffusion coefficient was determined by measuring the decay of the signal intensity I in dependence of the gradient strength g Where I is the observed intensity, I0 the reference intensity, D the diffusion coefficient, γ is the gyromagnetic ratio of the observed nucleus, g the gradient pulse length, δ the length of the gradient, and Δ the diffusion time. As can be seen, the transference number are rather high (well above 0.5), especially for the gels, in which the polymer movement is more limited (T + = 0.74-0.77). Further increase would require either crosslinking of the polymer or using block copolymer (i.e. adding a hydrophobic block) to obtain phase separation (rather than local structuration), and complete immobilization of the polymer chains. 26

Supplementary Note 4. Details of Quantum Mechanical Modeling
For modeling we considered a monomer of LiPAA and up to 8 water molecules. We carried out full geometry optimizations for LiPAA(H2O)n (n = 0-8) by Kohn-Sham density functional theory with the M08-HX global-hybrid meta-GGA density functional 2 and the MG3S basis set 3 . Vibrational frequencies were scaled by a factor of 0.973 to improve the accuracy of the calculated zero point energy 4 . All the calculations included the effect of the surrounding environment by the SMD solvation model 5 . Because the dielectric constant ε of the environment in which the cluster model is located should be between that of water and that of the superabsorbent LiPAA polymer, we used 1,2-ethanediol (with ε = 40.245) as the solvent surrounding the LiPAA(H2O)n (n = 0-8) complexes. Density functional integrals were computed using a grid of 974 angular points per shell and 99 radial shells. "Tight" convergence criteria were used in the optimization. All of the quantum mechanical calculations of LiPAA(H2O)n (n = 0-8) were carried out with the Gaussian 09 software package 6 . The partial atomic charges on LiPAA(H2O)n (n = 0-8) were calculated using the CM5PAC package, 7 which utilizes Hirshfeld atomic charges 8 to obtain partial atomic charges by Charge Model 5 (CM5 ) 9 .
The binding free energy of the nth water molecule in LiPAA(H2O)n is defined as the standardstate Gibbs free energy change in the reaction LiPAA(H2O)n(solv) LiPAA(H2O)n-1(solv)+ H2O(solv) where "(solv)" denotes that the species is dissolved in 1,2-ethanediol. This free energy change was calculated using the equation where the Gibbs free energy is 10 , and where Ue°[X] is the solution-phase potential energy species X, defined as the equilibrium value on the potential of mean force (also known as the free energy surface) 54 including the standard-state free energy of solvation, ZPE[X] is the local zero-point vibrational energy, and Gint[X] is the vibrational-rotational free energy of species X with local zero of energy at Ue° [X] (where X is one of the species in eq. (3)) 10 . In the article and in the rest of this supplementary material we refer to Ue° values simply as energies as a substitute for the long phrase "equilibrium potential energy of mean force." The superscript " ° " denotes the standard state; all free energies are standard-state free energies with a solution-phase standard state of 1 mol L -1 . For each complex, the conformer with the lowest energy in 1,2-ethanediol was chosen to calculate the binding free energies. We also calculated binding energies by   ,bind° and U0,bind°). The calculated binding energies and binding free energies are given in Supplementary Table 4.
There are four sites per carboxylate group in LiPAA that can bind to water molecules, as shown in Supplementary Fig. 12. Thus there are at least four conformers of LiPAA(H2O). The conformations of LiPAA(H2O)2 isomers were obtained by inserting a water molecule at four sites in the lowest energy structure of LiPAA(H2O). Similarly the structures of LiPAA(H2O)n were obtained by starting with the lowest energy structure of LiPAA(H2O)n-1. The isomers of LiPAA(H2O)n (n = 1-8) obtained by this method are only a subset of the myriad of conformations existing in solution, and they need not contain the lowest-energy conformations for each n, but we use them as representative low-energy structures. As mentioned above, the lowest-energy ones for each n were used to calculate binding energies and binding free energies. By adopting this systematic method, we are able to gain insight into the effect of solvent in the LiPAA electrolyte.
The first and second water bind to the Li + cation and do not form hydrogen bonds with the carboxylate oxygens. Their bindings increase the Li-O bond length in the LiPAA unit. The third water binds to Li + and in this structure (structure 32 in Supplementary Fig. 14, which names and shows all the structures), there are hydrogen bonds to both carboxylate oxygens. The most stable conformation of LiPAA(H2O)4 (structure 45) has Li + combining with three water molecules, and O1 from LiPAA binding one water molecule; in this way a hydrogen bonding network is formed among the four water molecules. The conformation with four water molecules being bound by the lithium ion (structure 44) has an energy 0.6 kcal mol -1 higher than three water molecules bound (structure 45), but the free energy is only 0.2 kcal mol -1 higher. These calculation results show that Li + preferentially combines three waters in addition to its binding to O2 from LiPAA and thereby it has a four-coordinated conformation. The fifth water binds to the other oxygen of the carboxylate group. Starting with LiPAA(H2O)6, some of the isomers contain outer water molecules that are only indirectly connected to the LiPAA, such as 64, 711, and 714. Outer water appears in all isomers of LiPAA(H2O)8, indicating that the LiPAA cluster is saturated and that outer waters finally form free water.  7 Li NMR were measured on a Bruker Avance III HD 500MHz Smart Probe spectrometer with a 3 mm NMR tube filled with ~0.1 ml sample solution. 1M LiOH in H2O and 15 wt % LiPAA in H2O are prepared in a Ar-filled glovebox. 7 Li shifts was referenced to LiCl solution (0 ppm). No deuterated solvent was used here (and hence no locking was performed) so the 1 H shifts are not accurate and cannot be readily compared.

Rheological properties
The rheological behavior of the gels from 30% to 70% LiPAA was measured in oscillation mode with a MCR102 rheometer (Anton Paar), using a cone plate (Cp50-1, Diameter-49.966mm) with a gap of 0.1 mm, at 20°C, with controlled shear strain (via deflection angle) at 10 rad s -1 .
Small angle X-ray Scattering (SAXS): SAXS measurements were carried out at the GALAXI diffractometer equipped with a Dectris Pilatus 1M detector and a BRUKER AXS Metaljet X-ray source at JCNS 11 . The applied wavelength and sample to detector distance are 1.34 Å and 831 mm, respectively. Samples were placed in borosilicate glass capillaries and each measurement counted 1200s.

Wide angle neutron scattering (WANS):
Neutron scattering data at wide angles were collected with diffuse scattering neutron time-of-flight spectrometer (DNS) from Jülich Centre for Neutron Science (JCNS) at the FRM-2 in Garching 12 . A combination of large double-focusing PG (002) (d = 3.355 Å) monochromator and a highly efficient supermirror-based polarizer provide a polarized neutron flux of about 10 7 n cm -2 s -1 . The coherent and incoherent scattering could be distinguished by polarization analysis of the scattered neutrons. We followed the analysis described in Supplementary Ref. 13.