Ion-Conduction Properties of LiCsF₂-Based Fluoride Materials

Abstract

Solid electrolytes with high Li⁺ conductivity and excellent electrochemical stability are required for the realization of all-solid-state lithium-ion batteries. In this study, LiCsF₂, which has been proposed to possess a wide electrochemical stability window, was fabricated and its ion-conduction properties were investigated. LiCsF₂ and Mg²⁺–LiCsF₂ were fabricated via ball milling. The dissolution of MgF₂ in LiCsF₂ via variation of the lattice parameters of LiCsF₂ was suggested. The conductivity of LiCsF₂ was of the order of 10⁻⁸ S/cm at room temperature, and the activation energy for ion conduction was estimated as 1.3 eV. Li deposition/dissolution currents were not clearly observed in the cyclic voltammetry (CV) curves of Mg²⁺–LiCsF₂. The conductivity of Mg²⁺–LiCsF₂ significantly increased upon increasing the relative humidity of the measurement atmosphere. Based on the voltage variation in the water vapor concentration cell, it was concluded that the major conduction carrier in Mg²⁺–LiCsF₂ after exposure to moisture was H⁺.

1. Introduction

Lithium-ion batteries (LIBs) have dominated the battery market for decades owing to their high energy densities, which enables their use in high-performance portable devices. The high energy densities of LIBs primarily originate from the low operating voltage of the negative electrode. Because the charging voltage of the graphite anode is much lower than the reduction voltage of H₂O, aqueous solutions cannot be used as electrolytes. Alternatively, flammable organic liquids are used as the electrolyte solution. However, this introduces concerns regarding the safety of batteries, which should be ensured, particularly for large-scale use, such as in electric vehicles and as stationary power supply. Unfortunately, a tradeoff between high energy density and safety cannot be avoided for conventional LIBs with organic liquid electrolytes. Consequently, all-solid-state LIBs, which enable higher performances while ensuring safety, are believed to be the batteries of the future.

Inorganic ceramic electrolytes with high Li⁺ conductivities (solid electrolytes) are crucial for the realization of all-solid-state LIBs. At present, oxide, hydride, and sulfide materials are highly promising owing to their extremely high Li⁺ conductivities at room temperature.^1–4) In addition to high Li⁺ conductivity, solid electrolytes should not decompose within the working voltage range between the anode and cathode. Thus, a wide electrochemical window for solid electrolytes is necessary for stable battery performance. However, most solid electrolytes have limited electrochemical stability windows owing to their lack of tolerance to high and/or low voltages.⁵⁾ For example, the reduction limit of sulfide electrolytes makes it difficult to construct high-voltage batteries using Li metal, despite their exceptionally high Li⁺ conductivities.⁶⁾ Similarly, though hydride solid electrolytes are known to be high Li⁺ conductors,^7–9) owing to their instability under high voltage, Li/TiS₂ and Li/S cells have been mainly constructed, whose operating voltage is below 2.5 V.^10–12) In addition, to avoid direct contact with LiBH₄ and LiCoO₂, the surface of LiCoO₂ has been coated with Li₃PO₄.¹³⁾ Therefore, there is dire need for solid electrolytes with high Li⁺ conductivity and electrochemical stability.

Recently, lithium fluoride-based materials have attracted considerable attention because of their wide electrochemical stability at the anode and cathode. For example, it has been reported that Li₃AlF₆ is tolerable over 6 V (vs. Li⁺/Li), indicating that Li₃AlF₆ does not undergo oxidative decomposition (Li₃AlF₆ → 3Li⁺ + 3e⁻ + AlF₃ + 3/2F₂) at the interphase with conventional cathode materials.¹⁴⁾ Indeed, the charge–discharge reaction for an all-solid-state Li/LiMn₂O₄ cell was confirmed when Li₃AlF₆ was used as the solid electrolyte.¹⁵⁾ Li₃AlF₆ has also been used as a coating material for the cathode.¹⁶⁾ High electrochemical stability does not always indicate thermodynamic stability; however, a kinetically stable interphase is acceptable for practical batteries. Recently, we reported that a Li₃AlF₆–Li₂SO₄ composite can function as the solid electrolyte for a graphite/Li(Ni_0.3Co_0.6Mn_0.1)O₂ cell, despite Li₃AlF₆-based materials not being thermodynamically stable under 1.06 V (vs. Li/Li⁺).^14,17) The kinetically stable interphase of the Li₃AlF₆–Li₂SO₄ composite can be partially attributed to the passivation layer formed at the anode interphase.¹⁷⁾

From the perspective of electrochemical stability, LiCsF₂ would be the most promising solid electrolyte among the proposed fluoride materials because the onset voltage of its reductive decomposition (2LiCsF₂ + Li⁺ + e⁻ → Li₃CsF₄ + Cs) has been estimated as 0.09 V (vs. Li/Li⁺).¹⁴⁾ This excellent tolerability at low voltages is significantly higher than those of other solid electrolytes.⁵⁾ Furthermore, the oxidation limit of LiCsF₂ (6.38 V (vs. Li/Li⁺)) was also significant in a similar manner when compared with the other fluoride materials.¹⁴⁾ In addition to the excellent electrochemical stability, the melting point of LiCsF₂ has been reported to be a moderately low 494°C,¹⁸⁾ which suggests high deformability and, consequently, guarantees intimate contact between the active materials upon uniaxial pressing. However, despite its excellent electrochemical stability and predicted mechanical properties, there are few reports on the conductivity of LiCsF₂. Recently, LiCsF₂ was used as the F⁻ source for the conversion electrode MnO.¹⁹⁾ However, to the best of our knowledge, experimental studies focusing solely on the conduction properties of LiCsF₂ have not been reported. In addition to theoretical calculations, the Li⁺ conduction properties of LiCsF₂ should be experimentally investigated to determine whether LiCsF₂ can be utilized as a solid electrolyte.

In this study, pure and Mg²⁺-doped LiCsF₂ were fabricated by ball milling and their ion conduction properties were characterized. The dominant conduction carriers were determined by changing the water vapor atmosphere during the conductivity measurements. The results indicate that Mg²⁺-doped LiCsF₂ is highly sensitive to moisture, and that proton conduction is dominant upon exposure of the cell to air.

2. Experimental

LiF (99.98%, Aldrich) and CsF (99.9%, Aldrich) were mixed in a mortar at a molar ratio of 1:1. The equimolar LiF–CsF mixture was placed in a ZrO₂ vessel (inner volume = 45 mL) along with ten balls (ϕ10) and ball-milled 100 times for 30 min each cycle at 400 rpm, with a 5 min interval between each cycle. The total milling time was 50 h. MgF₂-doped LiCsF₂ was then fabricated by ball-milling the raw materials LiF, MgF₂ (Aldrich, 99.99%), and CsF at a molar ratio of 0.9:0.1:1. The ball-milling conditions were identical to those used for the fabrication of LiCsF₂. The crystalline phases of the obtained samples were determined by X-ray diffraction (XRD) using a MiniFlex (Rigaku Co.). To avoid air exposure, the samples were sealed using Kapton^® tape during measurement. The electrical conductivity was measured using the AC impedance method (Solartron 1260) in the frequency range of 1 MHz to 1 Hz. The sample powder was uniaxially pressed at approximately 270 MPa, resulting in pellets with a diameter and thickness of 10 mm and ∼1 mm, respectively. Stainless steel (SS) electrodes were used as the blocking electrodes. SS/SS symmetrical cells were sealed in HS cells (Hosen. Co). The cells were heated from room temperature to 80°C and their conductivities were measured at intervals of 10–15°C. The cells were then kept overnight at 80°C, and the measurements was subsequently performed during the cooling process. For comparison, Li foils were also used, and the impedances of the cells were measured. After the conductivity measurements, infrared (IR) spectra of the samples were obtained to confirm the variation in the microstructures of the samples during the measurement. To investigate the influence of cell leakage on conductivity, conductivity was measured by controlling the relative humidity (RH) of the measuring atmosphere. To evaluate the electrochemical stability of the samples, cyclic voltammetry (CV) measurements were performed at 80°C using SP-150 (Biologic Co.). An SS disk and Li foil were used as the working and counter electrodes, respectively. The SS/Li cell was sealed in the HS cell, and the CV measurement was conducted in the voltage range between −0.5 and 5.0 V (vs. Li⁺/Li). A water vapor concentration cell was fabricated to determine the conduction carriers, and the open-circuit potential of the cell was measured. Prior to the measurements, the solid electrolyte (250 mg) was pressed at 400 MPa for 5 min using a PEEK cylinder. Subsequently, the solid electrolyte pellet was fixed at 150 MPa using screws. This cell was kept at 100°C for one night in a temperature-controlled furnace where the RH was maintained at 100%. Water vapor partial pressure (WVPP) at one side of the cylinder was maintained at 0.02 kPa (dry), while the other side was maintained at 3.2 kPa (wet). The WVPPs at the dry and wet side correspond to the saturation water vapor pressures at −40°C and 25°C, respectively. Both sides of the cylinder were then sealed with rubber plugs equipped with Cu current collectors.

3. Results and Discussion

3.1 Sample preparation

Figure 1 shows the XRD patterns of pristine LiF–CsF and LiF–CsF containing MgF₂ at room temperature. The diffraction peaks of LiF–CsF were indexed to the space group C2/c, which is isostructural with that previously reported on LiCsF₂.²⁰⁾ LiCsF₂ was successfully fabricated via ball milling. The XRD pattern of LiF–CsF with MgF₂ also fitted the C2/c structure, although its crystallinity was lower than that of LiCsF₂. This result indicates the doping of MgF₂ into the LiCsF₂ lattice. The calculated lattice parameters are presented in Table 1. For the a, b, and c axes, the values increased with increasing MgF₂ doping. In the LiCsF₂ crystal lattice, the Li⁺ ions are positioned in the F⁻ tetrahedra. The ion radii of Li (0.59 Å) and Mg (0.57 Å) in the tetrahedral coordination indicates lattice shrinkage rather than an increase in the lattice parameters upon the substitution of Li⁺ with Mg²⁺.²¹⁾ For charge compensation, Li⁺ vacancies can be introduced by the partial substitution of Li⁺ with Mg²⁺. The expansion of the lattice volume can reflect the presence of cation vacancies in addition to Mg²⁺ doping. Hereafter, the LiF–CsF with MgF₂ is referred to as Mg²⁺–LiCsF₂.

Fig. 1

XRD patterns for LiCsF₂ (bottom) and Mg²⁺-doped LiCsF₂ (top) prepared by ball-milling for 50 h. The results of whole powder pattern fitting are represented by the red line.

Table 1 Calculated lattice parameters for LiCsF₂ and Mg²⁺-doped LiCsF₂.

3.2 Electrical conductivity measurement

The results of the conductivity measurements for LiCsF₂ and Mg²⁺–LiCsF₂ are presented in Fig. 2 as Arrhenius plots. The conductivity of LiCsF₂ was measured to be of the order of 10⁻⁸ S/cm at room temperature. From the slope of the Arrhenius plot, the activation energy for ion conduction was calculated as 1.3 eV. In contrast, the conductivity at room temperature increased to an order of 10⁻⁷ S/cm for the Mg²⁺–LiCsF₂. The activation energy of the heating cycle was estimated to be 1.5 eV, which is comparable to that of LiCsF₂. The XRD results suggest that Mg²⁺ was doped into LiCsF₂. One possible reason for the improved conductivity of Mg²⁺–LiCsF₂ is the introduction of Li⁺ vacancies as a result of the replacement of the host Li⁺ with Mg²⁺. Figure 2(b) shows the Nyquist plots of SS/SS and Li/SS cells of LiCsF₂. Fitting results for the SS/SS and Li/SS cells are plotted as black and red lines. The equivalent circuits used for data fitting are shown in the Figures. For SS/SS cells, a semicircle with capacitance values of 10⁻¹¹–10⁻¹⁰ F was observed, independent of the solid electrolyte. This resistance component can originate from the bulk solid electrolyte. This semicircle does not change significantly for SS/SS or Li/SS cells, indicating that the bulk resistance was not influenced by the type of electrode used. In contrast, an extra semicircle was observed in the low-frequency region for Li/SS cells. The capacitance values of these low-frequency semicircles were calculated to be in the order of 10⁻⁷ F. Considering that this resistance component was not confirmed for SS/SS cells, it is predicted that this resistance originates from the interphase with the Li metal.

Fig. 2

(a) Arrhenius plot of the conductivity of LiCsF₂ (open circles) and Mg²⁺-doped LiCsF₂ (filled circles). The heating and cooling results are plotted with red and blue circles, respectively. For the results of Mg²⁺-doped LiCsF₂, the typical heating/cooling data are presented. The reproducibility of the data is presented in Fig. A2 in Appendix. Inset: photographs of the pellet after conductivity measurement. (b) Nyquist plots of SS/SS (black) and Li/SS (red) cells using LiCsF₂. The fitting results are plotted by a red line, and the equivalent circuits used for the simulation are presented in the inset. (c) CV curves of Li/SS cell using Mg²⁺-doped LiCsF₂ at 80°C.

Although a slight increase in the conductivity was observed by Mg²⁺ addition, the conduction properties were insufficient for using LiCsF₂ and Mg²⁺–LiCsF₂ as solid electrolytes. In contrast, the conductivity of Mg²⁺–LiCsF₂ increased drastically during the cooling cycle. When the cell was maintained at 80°C, the conductivity gradually increased and reached 4.5 × 10⁻³ S/cm. During the cooling cycle, the activation energy decreased to 0.34 eV and the conductivity was measured as 8.0 × 10⁻⁴ S/cm at 28°C. Photographs of the pellets after conductivity measurements are presented in Fig. 2(a). The LiCsF₂ pellet remained white, whereas the Mg²⁺–LiCsF₂ pellet became partially transparent, indicating its densification. The cell was screwed in during the measurement, and the resultant pressure on the pellet was approximately 50 MPa. Therefore, it is possible that the density of the pellet was increased by the pressure applied during the measurement. Indeed, after the conductivity measurement, the thickness of the pellet decreased, and the densification of the pellet was confirmed by SEM (Fig. A1 in Appendix). The improved conductivity of Mg²⁺–LiCsF₂ during the cooling cycle can therefore be partially attributed to the densification of the pellet. We have tested the conductivity measurement of Mg²⁺–LiCsF₂ for five different cells, however, the conductivity values of were not unambiguously determined and cannot reproduce, although the hysteresis of the heating/cooling cycle is certainly confirmed for all tested cell (Fig. A2 in Appendix). The origins for the lack of reproducibility are described in Appendix.

From the results of the CV measurements, it could not be concluded that the dominant conduction carrier was Li⁺. Figure 2(c) shows the CV curves of the Li/SS cells with Mg²⁺–LiCsF₂ at 80°C. The voltage was swept between −0.5 and 5.0 V for 5 times. If the conductivity of 4.5 × 10⁻³ S/cm was attributable to the Li⁺ ions, the Li deposition reaction should have smoothly occurred at approximately 0 V (vs. Li⁺/Li). However, redox current peaks, which indicate Li deposition/dissolution, were not clearly observed. The overall features of these CV curves did not appear to be those of pure Li⁺ conductors. Hence, the conduction carriers in Mg²⁺–LiCsF₂ should be carefully investigated.

To clarify the variation in the microstructures, IR spectra were obtained for the pellets before and after the conductivity measurements. Figure 3 shows the IR spectrum of Mg²⁺–LiCsF₂. Although not all the peaks were identified, the most significant change was the growth of the absorption peak at approximately 3000 cm⁻¹, which was assigned to the stretching vibration mode of OH⁻.²²⁾ These IR spectra strongly indicate that H₂O was attached to the surface of the pellet during the conductivity measurement. Though the cells were sealed during the measurement, it is considered that moisture infiltrated from the atmosphere owing to a slight leakage in the cell. Mg²⁺–LiCsF₂ is exceptionally sensitive to trace amounts of moisture. The presence of the OH functional groups suggests a trigger for H⁺ or OH⁻ conduction. To specifically investigate the effect of moisture, the conductivity was measured by changing the water vapor atmosphere.

Fig. 3

IR spectra for Mg²⁺–LiCsF₂ before (black) and after (red) conductivity measurement.

Figure 4 shows the Arrhenius plots of the conductivity of Mg²⁺–LiCsF₂ at RH values of 40, 60, and 100%. The conductivity increased along with the RH, and reached to 10⁻³ S/cm at room temperature under 100% RH. However, such a conductivity enhancement was not confirmed when a series of measurement was conducted in a dry room with a dewpoint of −40°C (Fig. A3). These results indicate that Mg²⁺–LiCsF₂ no longer functions as a Li⁺ conductor after exposure to air, and that H⁺ or OH⁻ ions dominate the total ion conduction. It can be concluded that the contribution of Li⁺ was significantly less dominant than that of H⁺ (or OH⁻). The conductivities of LiCsF₂ were also measured under R.H. 100%, however, the conductivity was not largely enhanced compared to that of Mg²⁺–LiCsF₂ (Fig. A4). It is said that the conductivity of LiCsF₂ is less sensitive to moisture atmosphere compared to that of Mg²⁺–LiCsF₂.

Fig. 4

Conductivity of Mg²⁺–LiCsF₂ under relative humidities of 40% (circle), 60% (square) and, 100% (triangle). The data in Fig. 2(a) are plotted by open circles for comparison.

3.3 Open circuit potential measurement of the water vapor concentration cell

Whether the dominant conduction ion is H⁺ or OH⁻ remained unclear. Therefore, to determine which ion was the dominant conduction carrier, a water vapor concentration cell was constructed using Mg²⁺–LiCsF₂, and the open-circuit voltage was measured. For H⁺ conduction in the solid electrolyte, the cell voltage is determined as H₂O (wet side) ↔ 1/2O₂ + 2H⁺ + 2e⁻ and H₂O (dry side) ↔ 1/2O₂ + 2H⁺ + 2e⁻. Consequently, the measured voltage is expressed as follows:

  

\begin{equation*} \Delta E = \frac{RT}{2F}\frac{P_{\textit{H${_{2}}$O},\textit{dry}}}{P_{\textit{H${_{2}}$O},\textit{wet}}} \end{equation*}

The theoretical voltage under the tested condition was calculated as −65 mV. For the same reasons, the measured voltage will become +65 mV if OH⁻ is the conduction ion. As shown in Fig. 5, the cell voltage increased with time and gradually approached the theoretical value. The negative voltage value indicates that the humidified Mg²⁺–LiCsF₂ was an H⁺ conductor. Although the conduction carrier in a dry atmosphere was not determined, the conductivity was of the order 10⁻⁸ S/cm when the dewpoint of the measuring atmosphere was maintained at −40°C (Fig. A3). Even if Li⁺ ions are the main carriers in a dry atmosphere, the Li⁺ conductivity is not sufficient for use as a solid electrolyte. In addition, the vapor pressure should be strictly controlled to avoid H⁺ conduction. Therefore, it was concluded that LiCsF₂ and Mg²⁺–LiCsF₂ are difficult to handle and utilize as solid-state Li⁺ conductors, despite their excellent electrochemical stability according to theoretical calculations.¹⁴⁾

Fig. 5

Aging variation of the voltage of the water vapor concentration cell.

4. Conclusions

In this study, the ionic conductivity of LiCsF₂ was investigated. LiCsF₂ was successfully fabricated by ball milling, and dissolution of MgF₂ was suggested. The conductivity of LiCsF₂ was of the order of 10⁻⁸ S/cm at room temperature, and the activation energy was calculated as 1.3 eV. The conductivity of LiCsF₂ was increased by Mg²⁺ doping, but still remained at 10⁻⁷ S/cm at room temperature. However, upon maintaining the cell at 80°C for one night, the conductivity of Mg²⁺–LiCsF₂ drastically increased and reached 4.5 × 10⁻³ S/cm at 80°C. After cooling the cell to 28°C, the conductivity was 8 × 10⁻⁴ S/cm. However, a redox current indicating Li deposition/dissolution was not clearly observed in the CV curves. Furthermore, the conductivity of Mg²⁺–LiCsF₂ increased with increasing humidity in the measurement atmosphere. Based on the voltage of the water vapor concentration cell, it was concluded that H⁺ ions dominated the total ion conduction in humidified Mg²⁺–LiCsF₂. Hence, LiCsF₂-based materials are difficult to be used as the solid electrolytes for all-solid-state LIBs despite of its predicted wide electrochemical stability window.

Acknowledgment

This study was supported by the NGK Environment Innovation Laboratory and JSPS KAKENHI Grant Number JP22H02179.

REFERENCES

• 1)   C.  Chen,  K.  Wang,  H.  He,  E.  Hanc,  M.  Kotobuki and  L.  Lu: Small 19 (2023) 2205550. doi:10.1002/smll.202205550
• 2)   C.  Bao,  C.  Zheng,  M.  Wu,  Y.  Zhang,  J.  Jin,  H.  Chen and  Z.  Wen: Adv. Energy Mater. 13 (2023) 2204028. doi:10.1002/aenm.202204028
• 3)   V.  Thangadurai,  S.  Narayanan and  D.  Pinzaru: Chem. Soc. Rev. 43 (2014) 4714–4727. doi:10.1039/c4cs00020j
• 4)   A.  Kraytsberg and  Y.  Ein-Eli: Energy Technol. 11 (2023) 2201291. doi:10.1002/ente.202201291
• 5)   Y.  Zhu,  X.  He and  Y.  Mo: ACS Appl. Mater. Interfaces 7 (2015) 23685–23693. doi:10.1021/acsami.5b07517
• 6)   Y.  Kato,  S.  Hori,  T.  Saito,  K.  Suzuki,  M.  Hirayama,  A.  Mitsui,  M.  Yonemura,  H.  Iba and  R.  Kanno: Nat. Energy 1 (2016) 16030. doi:10.1038/nenergy.2016.30
• 7)   D.  Blanchard,  A.  Nale,  D.  Sveinbjörnsson,  T.M.  Eggenhuisen,  M.H.W.  Verkuijlen,  Suwarno,  T.  Vegge,  A.P.M.  Kentgens and  P.E.  de Jongh: Adv. Funct. Mater. 25 (2015) 184–192. doi:10.1002/adfm.201402538
• 8)   W.S.  Tang,  K.  Yoshida,  A.V.  Soloninin,  R.V.  Skoryunov,  O.A.  Babanova,  A.V.  Skripov,  M.  Dimitrievska,  V.  Stavila,  S.-i.  Orimo and  T.J.  Udovic: ACS Energy Lett. 1 (2016) 659–664. doi:10.1021/acsenergylett.6b00310
• 9)   Y.S.  Choi,  Y.-S.  Lee,  D.-J.  Choi,  K.H.  Chae,  K.H.  Oh and  Y.W.  Cho: J. Phys. Chem. C 121 (2017) 26209–26215. doi:10.1021/acs.jpcc.7b08862
• 10)   A.  Unemoto,  T.  Ikeshoji,  S.  Yasaku,  M.  Matsuo,  V.  Stavila,  T.J.  Udovic and  S.-i.  Orimo: Chem. Mater. 27 (2015) 5407–5416. doi:10.1021/acs.chemmater.5b02110
• 11)   A.  Unemoto,  S.  Yasaku,  G.  Nogami,  M.  Tazawa,  M.  Taniguchi,  M.  Matsuo,  T.  Ikeshoji and  S.-i.  Orimo: Appl. Phys. Lett. 105 (2014) 083901. doi:10.1063/1.4893666
• 12)   A.  Unemoto,  K.  Yoshida,  T.  Ikeshoji and  S.-i.  Orimo: Mater. Trans. 57 (2016) 1639–1644. doi:10.2320/matertrans.MAW201601
• 13)   K.  Takahashi,  K.  Hattori,  T.  Yamazaki,  K.  Takada,  M.  Matsuo,  S.  Orimo,  H.  Maekawa and  H.  Takamura: J. Power Sources 226 (2013) 61–64. doi:10.1016/j.jpowsour.2012.10.079
• 14)   B.  Liu,  D.  Wang,  M.  Avdeev,  S.  Shi,  J.  Yang and  W.  Zhang: ACS Sustain. Chem. Eng. 8 (2020) 948–957. doi:10.1021/acssuschemeng.9b05557
• 15)   M.  Feinauer,  H.  Euchner,  M.  Fichtner and  M.A.  Reddy: ACS Appl. Energy Mater. 2 (2019) 7196–7203. doi:10.1021/acsaem.9b01166
• 16)   H.  Kobayashi,  G.  Yuan,  Y.  Gambe and  I.  Honma: ACS Appl. Energy Mater. 4 (2021) 9866–9870. doi:10.1021/acsaem.1c01885
• 17)   R.  Miyazaki,  S.  Ozaki,  E.  Yagi,  Y.  Sato,  Y.  Kobayashi,  T.  Yoshida and  Y.  Katsuda: ACS Appl. Energy Mater. 5 (2022) 15365–15372. doi:10.1021/acsaem.2c02985
• 18)   T.V.  Gubanova,  I.K.  Garkushin and  A.V.  Mal’tseva: Russ. J. Inorg. Chem. 61 (2016) 1163–1168. doi:10.1134/S0036023616090072
• 19)   S.K.  Jung,  I.  Hwang,  I.R.  Choi,  G.  Yoon,  J.H.  Park,  K.Y.  Park and  K.  Kang: Adv. Energy Mater. 9 (2019) 1900503. doi:10.1002/aenm.201900503
• 20)   J.H.  Burns and  W.R.  Busing: Inorg. Chem. 4 (1965) 1510–1512. doi:10.1021/ic50032a041
• 21)   R.D.  Shannon: Acta Crystallogr. Sec. A 32 (1976) 751–767. doi:10.1107/S0567739476001551
• 22)   N.  Fukatsu,  N.  Kurita,  Y.  Oka and  S.  Yamamoto: Solid State Ionics 162–163 (2003) 147–159. doi:10.1016/S0167-2738(03)00218-2

Appendix

A1. SEM observations of Mg²⁺–LiCsF₂ pellet before and after conductivity measurement

Fig. A1

SEM images of the cross section of Mg²⁺–LiCsF₂ pellet (a), (b): before and (c), (d): after the conductivity measurement. (a) and (c) are the secondary electron images while in the back scattering electron images in the same observed areas are presented in (b) and (d).

A2. Reproducibility of the conductivity for Mg²⁺–LiCsF₂

We have tested the reproducibility of the conductivity of Mg-doped LiCsF₂ using HS cell for five times. As shown in Fig. A2, the conductivities were highly dispersed and different results were obtained for each trial. These results indicate that the conductivity is highly sensitive and influenced by the slight leakage of the cell, which is slightly changed by the condition of gaskets and/or screwing and cannot be strictly controlled. Hence, the conductivity results on Mg-doped LiCsF₂ using HS cell are lacking of universality and strictness of the study. Although the conductivity value was not reproducible, the general tendencies were confirmed for five different cells; the conductivity increased during cooling cycle after maintaining the cell at the maximum temperature of measurement for 12 hours. Arrhenius plots of the Mg-doped LiCsF₂ are presented in Fig. 2(a) as the typical data. Again, these results of reproducibility tests indicate that the conductivity of Mg-doped LiCsF₂ should be evaluated by strictly controlling the moisture.

Fig. A2

Arrhenius plot of the conductivity of Mg²⁺-doped LiCsF₂ (filled circles) for five different measurements. The heating and cooling results are plotted with red and blue circles, respectively. The results for each trial were plotted by different symbols.

A3. Aging variation of the impedance for Mg²⁺–LiCsF₂ in the dry room

The aging variation of the impedance of Mg²⁺–LiCsF₂ was investigated in the dry room (dewpoint: −40°C) and the results are presented below. First, the cell was heated for 100°C and the impedance was recorded for pristine cell (black circles). Subsequently, the cell was left in dry room for 20 hours at 100°C and impedance was measured again (red circles).

Fig. A3

Impedance of SS/SS cell pristine (black) and kept in the dry room (dewpoint of −40°C) at 100°C for 20 hours (red).

A4. Conductivity for LiCsF₂ under water vapor atmosphere

Figure A4 shows the Arrhenius plot of the conductivity of LiCsF₂ under relative humidity of 100%. In a similar manner with the case for Mg²⁺–LiCsF₂, the hysteresis was observed during heating/cooling cycle and conductivities were rather enhanced for the cooling cycle under R.H. 100%. However, the conductivity value after cooling cycle was still in the order of 10⁻⁷ S/cm at 30°C, indicating the moisture effects on the conductivity are less significant compared to those on the Mg–LiCsF₂. Although the influence was less significant, LiCsF₂ also showed the moisture effect.

Fig. A4

Arrhenius plot of the conductivity of LiCsF₂ (triangles) under R.H. of 100%. The heating and cooling results are plotted with red and blue circles, respectively. The results without controlling R.H., which are the same data in Fig. 2(a) are also presented by open circles.