This is the accepted version of the journal article: Knipping, Etienne; Aucher, Christophe; Guirado López, Gonzalo; [et al.]. «Suit- ability of blended ionic liquid-dimethylsulfoxide electrolyte for lithium-oxygen battery». Batteries and Supercaps, Vol. 2, Issue 3 (March 2019), p. 200-204. DOI 10.1002/batt.201800078 This version is available at https://ddd.uab.cat/record/274562 under the terms of the license Suitability of blended ionic liquid-dimethylsulfoxide electrolyte for lithium-oxygen battery E. Knipping,*[a,b] C. Aucher,[a] G. Guirado[b] and L. Aubouy[a] Electrolytes composed of dimethylsulfoxide (DMSO), 1-ethyl-3- containing different amount of EMI TFSI, a RTIL known for its methylimidazolium bis(trifluoromethylsulfonyl)imide (EMI TFSI) and thermal resistance, high conductivity and electrochemical LiClO4 are characterized for lithium-oxygen battery. The increase of stability.[17] Hence, four mixed electrolytes ET-0, ET-30, ET-50 EMI TFSI content in the electrolyte enables to improve its thermal and ET-100 are prepared by adding 5w% of LiClO4 to the mixed stability and to decrease the overpotential of the cell. The optimum solution of DMSO and EMI TFSI, the weight ratio of EMI addition of EMI TFSI enables to reduce the overpotential from 1.43 V TFSI/DMSO being 0/100, 30/70, 50/50 and 100/0 respectively. to 1.06 V, with a cyclability of 69 cycles with 600 mAh g-1 of limited The results are summarized in the Table 1. capacity. The effect of the use of ionic liquids on the thermal stability of the above-mentioned electrolytes are analysed by performing DSC The Li-O2 battery has received much interest in the last few years, tests. Before the heating ramp, the samples are cooled and as its theoretical capacity reaches 5-10 times the current Li-ion equilibrated at -90˚C to allow their crystallization. In the case of battery capacity.[1] The main electrochemical process in an aprotic ET-0 and ET-30, the DSC profiles (Figure 1) show endothermic Li-O2 battery is the formation and decomposition of the Li2O2, peaks at 13 and 2˚C respectively, corresponding to the melting during discharge and recharge respectively. However, this point. The ET-0 shows another endothermic peak at -7˚C that may reaction suffers from low reversibility leading to a low cyclability of correspond to a solid-solid phase transition. For the samples with the battery. Recently, Room Temperature Ionic Liquids (RTILs) more EMI TFSI content, i.e. ET-50 and ET-100, no crystalline attracted much attention for Li-O2 battery electrolyte.[2–12] Namely, phase is detected, showing that for the procedure used, the Nakamoto et al.[13] demonstrated the advantage of the RTILs electrolytes remained amorphous. Moreover, thermal stability is stability against electrochemical oxidation vs. Li+/Li and O2 redox improved with the addition of EMI TFSI. Without RTIL, an reversibility. Cai et al.[14] proved that ionic liquid could show higher endothermic peak occurred at 105˚C, corresponding to the specific capacity compared to carbonate solvent electrolytes. evaporation of the solvent. However, when 30% of EMITFSI is Moreover, Cechetto et al.[15] observed a decrease of 0.4 V of the added, the endothermic peak is postponed to 182˚C. For ET-50 overpotential by mixing ionic liquids with an ether-based and ET-100, no endothermic peak is observed, showing that the electrolyte. This could be explained by a higher ionic conductivity degradation of the electrolyte occurs at higher temperature. The of the electrolyte containing the ionic liquid, a less insulating Li2O2 safety of the battery can be then improved with the addition of an particle interface and a better Li2O2 solubility resulting to improved RTIL, as it can enhance the thermal stability of the electrolyte. kinetics of the Li2O2 oxidation. On the other hand, pure RTIL electrolytes were tested in previous experiments, but an adequate Ionic conductivity and viscosity of the blended electrolytes are reversibility of the Li-O2 battery reaction is not attainable probably measured and represented in function of the DMSO/EMI TFSI due to their high viscosity that affects the ion diffusion and the ratio (Figure 2). The ionic conductivity initially increases with kinetics of the battery reaction.[16] Moreover, poor battery increasing EMI TFSI content, reaching a maximum of 13.1 mS capacity[13] and a cyclability lower than 30 cycles is generally cm-1 at 30% by weight. Because EMI TFSI itself is a salt, the obtained.[5,14] In the present work, we study the influence of the 1- number of ions in the electrolyte increases with increasing EMI ethyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide (EMI TFSI content and then the ionic conductivity too. However, adding TFSI) as co-solvent of Li-O2 battery electrolyte in the final aim of designing an adequate solution for this technology. Thermal Table 1. Ionic conductivity, viscosity, degradation temperature, capacity, overpotential and cycling life of the four studied electrolytes. ET-0 ET-30 ET-50 ET-100 Ionic conductivity (mS cm-1) 7.88 13.1 10.1 0.49 Viscosity (cP) 6.5 6.3 12 81 Degradation temperature (˚C) 105 182 >200 >200 Capacity full discharge (mAh g-1) 5651 7618 4631 3060 Overpotential 1st cycle (V) 1.43 1.06 0.90 1.16 Cycling life (cycles) 98 69 18 7 properties, stability with lithium metal and electrochemical performance are evaluated for DMSO-based electrolyte Figure 1. DSC profiles of the ET-0, ET-30, ET-50 and ET-100 electrolytes. [a] Leitat Technological Center, Carrer de la Innovació, 2 08225 Terrassa, Spain. E-mail: eknipping@leitat.org [b] Departament de Química, Universitat Autònoma de Barcelona, E- 08193 Bellaterra, Barcelona, Spain. molecule adsorption at the surface within lower voltage amplitude with respect to blended electrolytes. In all cases, the electrolyte enables the stripping/plating of the lithium. The full discharge capacities of the Li-O2 cells were measured at a constant specific current of 200 mA g-1. The first full discharge curves obtained with the different electrolytes are shown in the Figure 4. The cell with ET-30 presents the highest capacity with 7618 mAh g-1, whereas the cell with RTIL-free electrolyte (ET-0) has a capacity of 5651 mAh g-1. This improvement by adding 30% of EMI TFSI can be explained by the higher ionic conductivity and a better wettability of the ET-30. Indeed, the EMI TFSI, which is hydrophobic, may have more affinity with the hydrophobic carbon.[8] However, adding more EMI TFSI in the electrolyte increases the viscosity, which leads to lower capacity. An optimization of the electrolyte content is then needed to achieve Figure 2. Ionic conductivity and viscosity of blended electrolytes as function of high battery performance. The four electrolytes are then tested in EMI TFSI content in the EMI TFSI/DMSO solution. All the solutions have Li-O2 cells to obtain discharge/recharge cycles at a controlled LiClO4 content of 5wt%. depth of discharge of 600 mAh g-1 and constant current of 200 mA g-1. On the first cycle (Figure 5), the oxidation potential decreases with increased content of EMI TFSI. Indeed, the higher ionic EMI TFSI increases also the viscosity of the electrolyte from 6.5 conductivity of ET-30 and ET-50 compared to ET-0 enables faster cP for ET-0 to 81.4 for ET-100, due to the increased ion-solvent kinetics of the oxidation reactions. The difference between ET-30 interactions and Coulombic interactions between ionic species[8]. and ET-50, 3.86 and 3.68 V vs. Li+/Li respectively, may be Thus, the decrease of the ionic conductivity from 50% of EMI TFSI ascribed to a better wettability of the ET-50 electrolyte.[8] The content can be attributed to the increase of the viscosity. It is higher EMI TFSI content makes the electrolyte more hydrophobic, shown then that the addition of a small amount of RTIL in the wetting then more effectively the hydrophobic carbon cathode. electrolyte can increase its ionic conductivity, which may be Concerning the ET-100, the discharge plateau has a lower beneficial for the battery. voltage (2.66 V vs. Li+/Li) than for the electrolytes containing DMSO (2.80 V vs. Li+/Li). Despite its high viscosity, the ET-100 To verify the compatibility of these new electrolytes with the show low oxidation potential of only 3.82 V vs. Li+/Li, which may lithium anode, Li/Li stripping/plating test is realized within the be due its hydrophobicity.[8] However, this oxidation potential is different electrolyte at 10 mA cm-2 and 25 ˚C. The cell is charged higher than for the ET-50 as the ionic conductivity of the latter is at constant current for 1 hour at 10 mA cm-2 and subsequently 20 times higher. Moreover, this potential rapidly increases with discharge for 1 hour at -10 mA cm-2. A typical voltage profile is the cycles (Figure 6-D). The evolution of the overpotential with the shown in the Figure 3-A. Since the cell is symmetric, it is not cell cycling is shown in the Figure 6. Although it shows the highest affected by parasitic chemical or electrochemical reactions and overpotential, ET-0 has better cycling stability with 98 cycles the cycles are symmetrical.[18,19] The evolution of the overpotential whereas the ET-30 and ET-50 work for 69 and 18 cycles of the Li/Li cell is shown for each electrolyte (Figure 3-B). With the respectively. The overpotential of ET-0 is the highest observed at ET-0 as electrolyte, the cell voltage is seen to be very low, around the first cycle and the charge voltage increases rapidly to 4.35 V 9 mV vs. Li+/Li and stays constant upon cycling for at least 200 vs. Li+/Li after only 12 cycles. On the other hand, the ET-30 shows cycles. This low voltage amplitude confirms that no stable charge voltage below 4.35 V vs. Li+/Li until the 54th cycle. As passivation layer is formed.[20] Concerning both blended shown previously, the addition of EMI TFSI enables to increase electrolytes (ET-30 and ET-50), the cell voltage varies between the ionic conductivity of the electrolyte but it may also enhance 0.2 and 0.5 V vs. Li+/Li, which is due to the formation of a solid the O2 solubility.[21,22] Thus a better reversibility is observed for the electrolyte interphase (SEI) layer. A cell voltage around 0.1 V vs. electrolytes containing the ionic liquid for the first cycles. Li+/Li for ET-100 is observed, showing that the lithium surface is passivated in the presence of the EMI TFSI, probably due to Figure 3. A) Typical voltage behavior of Li-Li cell during galvanostatic lithium stripping/plating cycle test (case of ET-0). A current of 10 mA cm-2 was applied for 1 hour in each direction. B) Evolution of the cell-voltage cutoff with the number of cycles for the different electrolytes studied Figure 4. Initial full discharge curves of Li-O2 cells at 25ºC with ET-0, ET-30, Figure 5. First cycle galvanostatic discharge and recharge of Li-O2 cell at ET-50 and ET-100 electrolytes. 25ºC with the different electrolytes studied, at the applied current of 200 mA g- 1 and limited capacity of 600 mAh g-1. Indeed, the charge potential increases with the number of cycles until the battery fails, which occurs earlier when the electrolyte In summary, it is important to remark that for blended DMSO/EMI contains EMI TFSI. This can be ascribed to the degradation TFSI electrolytes, significant improvements are observed products formation which was proved in previous work, that it is concerning the thermal stability and cell overpotential. In faster in presence of EMI TFSI than in DMSO without RTIL.[23] It particular, by adding 30% in weight of EMI TFSI, the was observed indeed that with cycling, LiOH concentration degradation/evaporation occurs at a temperature 80˚ higher and increases more rapidly in presence of ionic liquids, due to the high the ionic conductivity increasing from 7.88 to 13.1 mS cm-1. The reactivity of the C2 proton of imidazolium cations. Moreover, Das Li-O2 battery performance is also improved with a 35% higher full et al.[7] exlpained that imidazolium-based cations are unstable in discharge capacity and an overpotential reduced to 1.06 V versus presence of radical peroxide.[24] Then, even with a reduced 1.43 V for the DMSO electrolyte. Reasonable cycling overpotential, so a more reversible battery reaction, the performance are also obtained, although the electrolyte without electrolyte degradation is more important in presence of the RTIL. RTIL shows higher cycling life (98 cycles), 69 cycles are obtained Concerning the DMSO, Sharon et al.[25] observed the formation of with the blended electrolyte. The use of a RTIL more stable side products when exposed to superoxide. This would explain against peroxide radical attack and in presence of lithium may the battery cycling being limited at 98 cycles when using the ET- enable a longer cycle life. Besides, this study shows that the 0. Figure 6. Cycling performance (blue curves) and corresponding cell-voltage cut-off (red and black curves) of Li-O2 cells at 25ºC with A) ET-0, B) ET-30, C) ET-50 and D) ET-100 electrolytes at a constant current of 200 mA g-1 and 600 mAh g-1 capacity limit. Keywords: Li-O2 battery; Ionic liquids [1] G. Girishkumar, B. D. McCloskey, A. C. Luntz, S. Swanson, W. bulk electrolyte parameters (ionic conductivity, viscosity) have Wilcke, J. Phys. Chem. Lett. 2010, 1, 2193–2203. more influence on the battery performance than the cathode [2] C. J. Allen, J. Hwang, R. Kautz, S. Mukerjee, E. J. Plichta, M. a. surface parameters (cell overpotential, wettability) and the anode surface parameters (Li/Li overpotential). Hendrickson, K. M. Abraham, J. Phys. Chem. C 2012, 116, 20755– 20764. [3] S. Ferrari, E. Quartarone, C. Tomasi, M. Bini, P. Galinetto, M. Experimental Section Fagnoni, P. Mustarelli, J. Electrochem. Soc. 2014, 162, A3001– A3006. EMI TFSI was purchased from Solvionic and DMSO from Sigma [4] M. Olivares-marín, A. Sorrentino, E. Pereiro, D. Tonti, J. Power Aldrich. Both are used after drying with 4 Å molecular sieves (Aldrich) Sources 2017, 359, 234–241. for two days. Lithium perchlorate salt was purchased from Sigma [5] G. A. Elia, J. Hassoun, B. Scrosati, F. Mueller, D. Bresser, S. Aldrich, dry with purity of 99.99%. Four mixed electrolytes ET-0, ET- Passerini, P. Oberhumer, N. Tsiouvaras, J. Reiter, Nano Lett. 2014, 30, ET-50 and ET-100 are prepared by adding 5w% of LiClO4 to the mixed solution of DMSO and EMI TFSI in a glove box filled with 14, 6572–6577. purified argon. The weight ratio of EMI TFSI/DMSO in ET-0, ET-30, [6] D. Bresser, E. Paillard, S. Passerini, J. Electrochem. Sci. Technol. ET-50 and ET-100 are 0/100, 30/70, 50/50 and 100/0 respectively. 2014, 5, 37–44. The water content of the electrolytes prepared is measured by Karl [7] S. Das, J. Højberg, K. B. Knudsen, R. Younesi, P. Johansson, P. Fischer (Coulometer KF 831) titration and was determined to be less Norby, T. Vegge, J. Phys. Chem. C 2015, 119, 18084–18090. than 50 ppm. The ionic conductivity is measured with a titanium tip (Ref. 50 73) on [8] S.-M. Han, J.-H. Kim, D.-W. Kim, J. Electrochem. Soc. 2015, 162, an EC Meter Basic 30+ conductimeter (Crison Instrument) at room A3103–A3109. temperature. The viscosities are measured using a Malvern Bohlin [9] Y. Li, Z. Zhang, D. Duan, Y. Sun, G. Wei, X. Hao, S. Liu, Y. Han, W. CVO 100-901 rheometer in air, at room temperature. Shear rates are varied between 0.1 s-1 and 10 s-1. No shear thinning behavior due to Meng, J. Power Sources 2016, 329, 207–215. the variation of the shear rate is observed. Hence, the viscosities are [10] P. Lou, C. Li, Z. Cui, X. Guo, J. Mater. Chem. A 2016, 4, 241–249. determined from the plateau value at all shear rates. Differential [11] W. Ni, S. Liu, Y. Fei, Y. He, X. Ma, L. Lu, Y. Deng, ACS Appl. Mater. scanning calorimetry (DSC) results are obtained using a TA Instruments Q20 apparatus. Samples weighting 5-14 mg are Interfaces 2017, 9, 14749–14757. hermetically sealed in aluminum pans and cooled to -90˚C followed [12] S. Matsuda, K. Uosaki, S. Nakanishi, J. Power Sources 2017, 356, by heating to 200˚C applying a scan rate of 10 ˚C min-1. Finally the 12–17. samples are cooled back to -90˚C at the same scan rate. [13] H. Nakamoto, Y. Suzuki, T. Shiotsuki, F. Mizuno, S. Higashi, K. Lithium/lithium stripping/plating tests are carried out in CR2032 coin cells using lithium chip (Ref: EQ-Lib-LiC25, MTI Corp.) as anode and Takechi, T. Asaoka, H. Nishikoori, H. Iba, J. Power Sources 2013, cathode and a glass fiber membrane (Whatman GF/A) as separator. 243, 19–23. A quantity of 150 µL of each electrolyte is used to perform the [14] K. Cai, H. Jiang, W. Pu, Int. J. Electrochem. Sci. 2014, 9, 390–397. galvanostatic cycling test with an applied current of 10 mA cm-2 with [15] L. Cecchetto, M. Salomon, B. Scrosati, F. Croce, J. Power Sources limited charge/discharge time of one hour and a cut-off voltage limit of 2012, 213, 233–238. +/- 1 V vs. Li+/Li. Li-O2 batteries are assembled inside an Ar-filled glove box using ECC- [16] C. O. Laoire, S. Mukerjee, K. M. Abraham, E. J. Plichta, M. a. air cell (EL Cell). The cathode for the Li-O2 battery are prepared from Hendrickson, J. Phys. Chem. C 2009, 113, 20127–20134. a mixture of 90wt% TIMCAL Super C65 and 10wt% polyvinylidene [17] P. Bonhôte, A.-P. Dias, N. Papageorgiou, K. Kalyanasundaram, M. fluoride (PVdF Kynar ADX 161, Arkema) in N-methyl-2-pyrrolidone Grätzel, Inorg. Chem. 1996, 35, 1168–1178. (NMP, 99.5%, Sigma-Aldrich). The slurry is tape casted on a Gas [18] L. Grande, E. Paillard, G. T. Kim, S. Monaco, S. Passerini, Int. J. Diffusion Layer (GDL SIGRACET 24BC – SGL Company) using the doctor blade technique up to an active loading of 1.5 mg cm-2. Mol. Sci. 2014, 15, 8122–8137. Although this GDL may participate to the overall capacity of the [19] G. B. Appetecchi, S. Scaccia, S. Passerini, 2000, 147, 4448. battery [26], only the mass of TIMCAL is taken into account to normalize [20] M. Roberts, R. Younesi, W. Richardson, J. Liu, J. Zhu, K. Edstr, the specific capacity calculated. A circular lithium chip (Ref: EQ-Lib- Electrochem. Lett. 2014, 3, A62–A65. LiC25, MTI Corp.) is used as the anode. The separator is a glass fiber [21] Y.-F. Hu, Z.-C. Liu, C.-M. Xu, X.-M. Zhang, Chem. Soc. Rev. 2011, membrane (Whatman GF/A) soaked with the tested electrolyte. Prior applying any current, the cell is maintained under O2 flow (8 mL min- 40, 3802. 1 ) up to stabilize the open circuit voltage after 10 hours. Galvanostatic [22] S. Monaco, A. M. Arangio, F. Soavi, M. Mastragostino, E. Paillard, cycling tests are performed at 25ºC using a VMP3 Biologic instrument S. Passerini, Electrochim. Acta 2012, 83, 94–104. with a current of 200 mA g-1 applied between 2.15 V and 4.35 V vs. [23] E. Knipping, C. Aucher, G. Guirado, F. F. Fauth, L. Aubouy, New J. Li+/Li as cut-off voltage with an oxygen flow rate of 8 mL min-1 and Chem. 2017, 41, 7267–7272. discharge/recharge cycle duration of 6h each. [24] M. Hayyan, F. S. Mjalli, M. A. Hashim, I. M. Alnashef, J. Mol. Liq. 2013, 181, 44–50. Acknowledgements ((optional)) [25] D. Sharon, M. Afri, M. Noked, A. Garsuch, A. a Frimer, D. Aurbach, J. Phys. Chem. Lett. 2013, 4, 3115–3119. The STABLE project received funding from the European [26] J. Zeng, J. R. Nair, C. Francia, S. Bodoardo, N. Penazzi, Solid State Union’s FP7 research and innovation program under grantagreement Nº 314508. G. G. thanks financial support Ionics 2014, 262, 160–164. from project CTQ2015-65439-R from the MINECO. Entry for the Table of Contents Layout 1: COMMUNICATION Beneficial effect of ionic liquid co- E. Knipping*, C. Aucher, G. Guirado solvent in lithium-oxygen battery López, L. Aubouy. electrolyte Page No. – Page No. Suitability of blended ionic liquid- dimethylsulfoxide electrolyte for lithium-oxygen battery