KFSA/glyme electrolytes for 4 V-class K-ion batteries † Journal of Materials Chemistry A COMMUNICATION

Highly concentrated electrolytes of KN(SO 2 F) 2 (KFSA)/diglyme and KFSA/triglyme are developed for high-voltage K-ion batteries. They demonstrate high oxidation stability up to 4.7 V vs. K/K + , which is considerably higher than that of the KFSA/monoglyme electrolyte. Moreover, these electrolytes can improve the cycling stability of high-voltage K-ion batteries.


Introduction
Recently, K-ion batteries (KIBs), which do not contain minor metals like Li and Co, have attracted considerable attention as potential high-voltage and high-power secondary batteries. [1][2][3] The high-voltage and high-power characteristics are attributed to the low standard electrode potential of K/K + in carbonateester-based electrolytes and the fast ionic diffusion of K + ions in electrolytes, respectively. 1 One of the main challenges in the development of high-voltage KIBs is the requirement of highperformance electrolytes that have the passivation ability of Al, high oxidation stability, and enabling the formation of stable solid electrolyte interphase (SEI) on the surface of the negative electrodes. [4][5][6][7] To date, aprotic solutions containing KPF 6 salt have been mainly utilized for developing KIBs as its counterpart Li electrolyte is broadly used in commercialized LIBs. 1 According to our investigation and knowledge, 1 KPF 6 salt has some advantages such as passivation of Al current collector similar to LIB cases due to the formation of AlF 3 as a stable passivation layer on the Al surface. 8,9 Although the second common K salt of KN(SO 2 F) 2 (KFSA) normally corrode Al current collector, highly concentrated KFSA electrolytes can passivate it. 5 Moreover, previous studies have indicated that an appropriate SEI is formed in highly concentrated electrolytes of imide salts in glymes such as KFSA/monoglyme (G1). 5,10,11 However, the low oxidation stability of G1 causes electrolyte decomposition at high-potential positive electrodes, thereby limiting the application of the electrolytes for high-voltage KIBs. Indeed, a previous study of the electrochemical performance of positive and negative electrodes in 7 mol kg À1 KFSA/G1 electrolyte reported a slightly lower coulombic efficiency for the K 2 Mn[Fe(CN) 6 ] positive electrode at a working potential ca. 3.8 V than that for the graphite negative electrode. 5 The mismatch of coulombic efficiency would result in capacity degradation of full cells. 12 To solve the low oxidation stability of G1 electrolytes, we utilized longer glymes such as diglyme (G2) and triglyme (G3) based on Yoshida et al.'s report that the oxidation stability of glyme molecule increases when they form solvate complexes with Li + ions. This occurs because of electron pair donation to Li + and consequent lowering of the highest occupied molecular orbital energy level. 13 Most of the glyme molecules are coordinated to alkali-metal cations in highly concentrated electrolytes, and the oxidation stability of the electrolyte is improved. 13 The degree of oxidation stability improvement depends on the stability of the solvate complexes in electrolytes. Electrolytes consisting of highly stable solvate complexes demonstrate higher oxidation stability 14 because the glyme molecules are tightly bound to the cation. Therefore, an appropriate combination of cation size and glyme length can enhance the oxidation stability of the highly concentrated electrolytes, and longer glymes are suitable for the large K + ions. 15 In this study, we have evaluated the physical and electrochemical properties of highly concentrated KFSA/G2 and KFSA/G3 solutions, which are liquid at room temperature. We have also investigated the electrochemical properties of K 2 Mn[Fe(CN) 6 ] and graphite electrodes in these electrolytes.
($5.5 mol kg À1 for KFSA/G1, $5.6 mol kg À1 for KFSA/G2, and G3). Physicochemical properties of the electrolytes were examined in terms of their ionic conductivities and viscosities ( Table  1). The 2.2 mol kg À1 KFSA/G1 solution showed high ionic conductivity and low viscosity. The highly concentrated (5.5 mol kg À1 ) KFSA/G1 solution also showed a high ionic conductivity of 8.60 mS cm À1 and a relatively low viscosity of 20.2 mPa s. The high conductivity and low viscosity could be attributed to the low viscosity of the G1 solvent. Compared with the G1 solutions, the highly concentrated KFSA/G2 and KFSA/G3 solutions exhibited lower ionic conductivity and higher viscosity. The 5.6 mol kg À1 KFSA/G3 (K : G3 ¼ 1 : 1 mol : mol) solution showed a low ionic conductivity of 0.89 mS cm À1 and a high viscosity of 229 mPa s. The viscosities of these solutions tend to be lower than the Na counterparts. 16 For example, the NaFSA/G3 (Na : G3 ¼ 1 : 1 mol : mol) solution exhibits a considerably higher viscosity of 497 mPa s. 16 The 5.6 mol kg À1 G3 solution was a transparent liquid without any precipitate at room temperature, whereas the 2.2 mol kg À1 G3 mixture was a solid at room temperature (see Fig. S1 in ESI †). This phenomenon is oen observed in alkali metal-glyme mixtures. This is caused by the lower melting point of the solvates in higher concentration solutions compared to that in lower concentration solutions. 15,17 We also prepared KFSA/tetraglyme and KFSA/pentaglyme solutions with a molar ratio of 1 : 1, but both were solid at room temperature and, therefore, excluded from the investigation of electrochemical behavior in this study.
The solution structures were investigated using Raman spectroscopy. Fig. 1a-c display the Raman spectra over the range of 780-900 cm À1 , where important Raman bands assigned to mixed modes of CH 2 rocking and C-O-C stretching vibrations in glymes are known to appear. 18 Pure G1 exhibited two broad peaks around 817 and 844 cm À1 . The intensity of these peaks decreased as the molality of KFSA in the solution increased. In contrast, the intensity of a new peak at 854 cm À1 , which can be assigned to G1 bound to a K + ion, monotonously increased. The Raman spectra were deconvoluted (see Fig. S2 †); the intensities of each peak are shown in Fig. 1d-f. Peak deconvolution revealed that the intensity of a peak at 829.5 cm À1 also increased with increasing molality up to 3.7 mol kg À1 (Fig. 1d), implying that this peak could also be attributed to the bound G1. Similarly, the G2 and G3 solutions exhibited a decrease in the intensity of the peaks assigned to free glyme and an increase in the intensity of the peaks corresponding to bound glyme peaks with an increase in the molality of KFSA. The area fractions of the bound glyme peaks were 70%, 67%, and 78% for the highly concentrated G1, G2, and G3 solutions, respectively. It should be noted that the Raman scattering coef-cients of free and bound glyme are almost identical and independent of the salt concentration. This is because the normalized area intensities of the FSA band between 680 and 780 cm À1 (I FSA / I Glyme ) for the G1 and G2 solutions exhibited a linear relationship with the molality (Fig. S3 †). 19 This indicates that the ratios of peak areas of free and bound glyme would reect their molar ratio. Therefore, 70-80% of glyme molecules are bound to K + ions in the highly concentrated electrolytes, which is similar to the case of a highly concentrated NaTFSA/G1 solution ($68% at a molar ratio of Na : G1 ¼ 1 : 2.5). 19 These solution structures will contribute to the enhancement of oxidation stability if the solvate complexes are stable.
The coordination structure of the FSA À anion was also examined by Raman spectroscopy. In the 2.2 mol kg À1 KFSA/G1 solution, the Raman band of the FSA À anion was observed around 720 cm À1 (Fig. S4a †), which can be assigned to free FSA À or weakly interacting FSA À weekly interacting with K + ions (solvent separated ion pairs, SSIPs). 20 The peak blue-shied with increasing molality. The shi indicates the formation of ion pairs, such as contact ion pairs and aggregates (AGGs). 20 The G2 solutions also exhibited the same trend (Fig. S4b †). The peak positions of the highly concentrated G1, G2, and G3 solutions were almost identical (Fig. S4c †), revealing that, in these solutions, ion pairing is not solvent dependent. The signicant degree of ionpair formation in these solutions is reasonable because the number of ether groups per K + ion is lower than the typical solvation number of K + ions in aprotic solutions, i.e., 6 to 8. 15,21 Thermal stabilities of these highly concentrated solutions were further tested by thermogravimetric analysis (Fig. S5 †). The 5.5 mol kg À1 KFSA/G1 solution showed signicant weight loss above 30 C, indicating high volatility. This could be attributed to the weak interaction between K + and G1 as well as the high volatility of G1. In contrast, 5.6 mol kg À1 KFSA/G2 and KFSA/G3 solutions showed a signicant improvement in thermal stability as compared to the pure solvent. The pure G2 solvent showed weight loss from ca. 50 C, whereas the 5.6 mol kg À1 G2 solution showed almost no weight loss up to 80 C. Most notably, the 5.6 mol kg À1 KFSA/G3 solution exhibited almost no weight loss up to 140 C and 10 wt% loss at 185 C, demonstrating the highest thermal stability among the solutions studied. The thermal stability of the KFSA/G3 solution is similar to that of the LiFSA/G3 (molar ratio of 1 : 1) solution, which exhibited a 10 wt% loss at 152 C at the same scan rate used in this study. 22 The improved thermal stability of the G2 and G3 solutions indicates a strong interaction between K + ions and the glymes. The Al corrosion behavior and anodic stability were electrochemically evaluated using voltammetry. Fig. 2a shows CV curves of the Al working electrodes in three-electrode cells lled with the highly concentrated electrolytes. Also shown, for comparison, is the CV curve with the 2.2 mol kg À1 KFSA/G1 solution. In the 2.2 mol kg À1 KFSA/G1 electrolyte, the CV curve delivered a potential hysteresis loop showing continuous current rise even in the cathodic scan, which indicates localized corrosion of Al foil. 23 A microscopic image of this Al working electrode showed signicant pitting corrosion (Fig. S6 †). However, in case of the highly concentrated electrolytes, the anodic current was negligible up to 5 V vs. K/K + , and no pitting corrosion was observed in the microscopic images (Fig. S6 †). This result revealed that Al foil was successfully passivated in highly concentrated solutions, which is consistent with our previous data with highly concentrated KFSA/DME solution. 5 Although all the highly concentrated electrolytes demonstrated sufficient passivation ability and anodic stability with an Al electrode, the electrolytes can be oxidized on active materials or conductive carbons when a composite electrode is used instead of an Al electrode. Therefore, we further investigated the oxidation stability by linear sweep voltammetry using Ketjen black (KB) electrodes consisting of KB : PVdF ¼ 80 : 20 (wt%). As shown in Fig. 2b, the 2.2 mol kg À1 KFSA/G1 and 2.2 mol kg À1 KFSA/G2 electrolyte cells exhibited an increase in anodic current from ca. 4.3 V vs. K/K + . More importantly, the 5.5 mol kg À1 KFSA/G1 electrolyte exhibited a curve similar to that of 2.2 mol kg À1 KFSA/G1, indicating a negligible enhancement in oxidation stability. In contrast, the 5.6 mol kg À1 KFSA/G2 and KFSA/G3 solutions exhibited an increase in anodic current from $4.7 V vs. K/K + . The anodic stability is higher than that of the 5.5 mol kg À1 KFSA/G1 electrolyte and similar to that of highly concentrated NaFSA/G2 ($4.6 V vs. Na/Na + ) reported by Geysens et al. 16 The high oxidation stability can be attributed to the higher stability of the solvate complexes in G2 and G3 compared to that in G1. 14,15 Fig. 3a-c show the galvanostatic charge-discharge curves of KkK 2 Mn[Fe(CN) 6 ] cells 24 lled with the highly concentrated  KFSA/G1, KFSA/G2, and KFSA/G3 solutions. The KFSA/G1 cell exhibited an initial charge capacity of 164 mA h g À1 and a discharge capacity of 101 mA h g À1 , resulting in a low initial coulombic efficiency of 61%. By contrast, the KFSA/G2 cell delivered a discharge capacity of 111 mA h g À1 and an initial coulombic efficiency of 78%. The higher discharge capacity and higher coulombic efficiency of the KFSA/G2 than those of the KFSA/G1 cell are likely due to lower electrolyte decomposition and self-discharge in the former. Moreover, the KFSA/G3 cell achieved the largest discharge capacity of 115 mA h g À1 and the highest initial coulombic efficiency of 87%. Aer the initial cycle, the KFSA/G1 cell maintained a low coulombic efficiency of around 90% (Fig. 3d). In contrast, the G2 and G3 cells demonstrated higher coulombic efficiencies of ca. 99.0% and 99.5%, respectively (Fig. 3d), and large reversible capacity (Fig. 3b and  c). Furthermore, even when the upper cut-off voltage was changed from 4.3 V to 4.5 V, the G3 electrolyte cell delivered good cycle performance and high coulombic efficiency. In contrast, a cell lled with a highly concentrated G1 electrolyte exhibited huge irreversible capacity attributed to the continuous anodic decomposition of the electrolyte (see Fig. S7 †). These data reveal excellent compatibility between the K 2 Mn [Fe(CN) 6 ] composite electrodes and the G3 electrolytes because of the high oxidation stability of the electrolytes.
In addition to the K 2 Mn[Fe(CN) 6 ] positive electrodes, graphite negative electrodes 25 were evaluated using highly concentrated KFSA/glyme electrolytes. Fig. S8a-c † show the charge and discharge curves, corresponding to K intercalation and deintercalation, respectively, of Kkgraphite cells lled with the KFSA/G1, KFSA/G2, and KFSA/G3 solutions. These cells showed initial reversible capacities of >270 mA h g À1 and coulombic efficiencies of 82-85%. The reversible capacity is almost the theoretical capacity of KC 8 (279 mA h g À1 ), suggesting the insertion of desolvated K + ions and the formation of binary graphite intercalation compounds. It has been reported that, in case of highly concentrated electrolytes, desolvated K + ions are intercalated into graphite 5 instead of K + -glyme cointercalation that occurs in conventionally concentrated electrolytes. 26 In subsequent cycles, all the cells exhibited excellent cycle performance with high coulombic efficiencies of >99% (Fig. S8 †). These results are consistent with good electrochemical performances of negative electrodes in imide salt/ glyme electrolytes 5,10,11 and prove that these KFSA/glyme electrolytes are similarly effective in forming a stable SEI, so as to achieve highly reversible K intercalation into graphite. To characterize the SEI formed in a highly concentrated KFSA/ glyme electrolyte, we conducted hard X-ray photoelectron spectroscopy (HAXPES) on the graphite electrode tested in a highly concentrated KFSA/G1 electrolyte. Fig. S9a-e † display HAXPES spectra of the pristine graphite electrode and that tested in the highly concentrated KFSA/G1 electrolyte. The obtained C 1s spectra of the cycled electrode can be deconvoluted into six peaks, which are assigned to sp 2 C, -CH x -, -CH  27 The organic components are mainly attributed to the solvent-derived decomposition products. Moreover, S 1s and F 1s spectra showed several peaks assigned to FSA À -derived components, such as SO x F, -SO 2 -, -S(]O)-, as well as KF. These decomposition products would be a key component of the SEI enabling high coulombic efficiency, as reported before. [28][29][30] Fig. S9f † shows the relative atomic amount of carbon (except for sp 2 carbon), oxygen, potassium, phosphorus, nitrogen, uorine, and sulfur elements relative to that of sp 2 carbon estimated from the HAXPES data of the tested electrode in comparison to the one tested in a conventional KPF 6 /EC : DEC electrolyte. 31 In the KFSA/G1 electrolyte, the relative amount of carbon was much lower than in the KPF 6 /EC : DEC electrolyte, indicating less organic species were deposited in the KFSA/G1 electrolyte. Furthermore, the total atomic amount on the graphite surface in the highly concentrated KFSA/G1 electrolyte was $60% of that in the KPF 6 one, suggesting that a relatively thin SEI layer was formed on the graphite electrode in the KFSA/G1 electrolyte.
Since both the K 2 Mn[Fe(CN) 6 ] positive electrode and graphite negative electrode exhibited good electrochemical performance with high coulombic efficiency in 5.6 mol kg À1 KFSA/G3, we fabricated a K-ion full cell lled with the electrolyte, and its electrochemical performance was compared to that of the KFSA/G1 cell. Fig. 4a and b display the charge-discharge curves of graphitekK 2 Mn[Fe(CN) 6 ] full cells lled with 5.5 mol kg À1 KFSA/G1 +0.5 wt% vinylene carbonate (VC) and 5.6 mol kg À1 KFSA/G3 +0.5 wt% VC, respectively. We note that the addition of 0.5 wt% VC into electrolyte solution is effective to suppress the initial irreversible capacity of full cells (see Fig. S10 †). The irreversible capacity of the full cell is larger than the expected one from the half cell results. The different electrochemical performance or SEI characteristics between half and full cells have been reported in several studies on Na-ion and K-ion batteries, and it has been explained by limited active Na/K content in the full cell or reaction between alkali metal and electrolyte in a half cell. [31][32][33][34] The detail of the difference between half and full cells is currently under investigation and will be reported elsewhere. Although the irreversible capacity was observed, the G1 and G3 cells delivered reasonable initial discharge capacities of 86 and 79 mA h g À1 , respectively (Fig. S11a †). In subsequent cycles, the G1 cell showed a signicant decrease in discharge capacity, which became as low as 14 mA h g À1 aer 100 cycles ( Fig. 4a and S11b †). The capacity degradation would be due to irreversible reactions mainly caused by electrolyte decomposition because the cell exhibited low and unstable coulombic efficiency over 100 cycles (Fig. 4c). In the G1 cell, the large irreversible capacity of the positive electrode would induce overcharge of the negative electrode in repeated cycles as the cut-off voltage (not potential) is xed ( Fig. S12a and b †). The overcharge causes K metal plating, which further drops the coulombic efficiency as observed around the 10th and 50th cycle. In contrast, the G3 full cell showed a discharge capacity of 106 mA h g À1 aer 100 cycles and exhibited considerably higher cycle performance than the G1 cell (Fig. 4b). It should be mentioned that the G3 cell exhibited a capacity increase, especially during the initial 30 cycles (Fig. 4b and S11b †). In the G3 cell, the positive electrode showed higher coulombic efficiency than the negative electrode at the initial cycle, but the order was contrary aer the second cycle as is shown in the half cell data (Fig. S12c †). Thus, the negative electrode would not be fully charged at the initial cycle and charged to a deeper state of charge (SOC) in each cycle aer the second cycle (Fig. S12d †). The SOC increase would induce the increase in capacity during cycling. Moreover, the G3 cell maintained high coulombic efficiency over 100 cycles (Fig. 4c). Thus, the use of a 5.6 mol kg À1 KFSA/G3 electrolyte with high oxidation stability enhances the cycling stability of high-voltage KIBs by avoiding electrolyte decomposition.

Conclusions
This study reveals that the highly concentrated KFSA/G2 and KFSA/G3 electrolytes are promising for high-voltage KIBs. These electrolytes passivate the Al current collector at high potentials and have better oxidation stability than the G1 counterpart. Their high oxidation stability allows highly reversible operation of high-potential positive electrode materials such as K 2 Mn [Fe(CN) 6 ]. This high oxidation stability is attributed to the strong interaction of K + ions with long glymes. The KFSA/G2 and KFSA/G3 electrolytes facilitate higher cycling stability of 4 V-class potassium-ion batteries than the KFSA/G1 electrolyte.

Conflicts of interest
There are no conicts to declare.