A 3.5 V Lithium–Iodine Hybrid Redox Battery with Vertically Aligned

Letter pubs.acs.org/NanoLett

A 3.5 V Lithium−Iodine Hybrid Redox Battery with Vertically Aligned Carbon Nanotube Current Collector Yu Zhao,†,§,∥ Misun Hong,†,‡,∥ Nadège Bonnet Mercier,† Guihua Yu,§ Hee Cheul Choi,‡ and Hye Ryung Byon*,† †

Byon Initiative Research Unit (IRU), RIKEN, Hirosawa 2-1, Wako, Saitama 351-0198, Japan Department of Chemistry, Pohang University of Science and Technology (POSTECH), and Center for Artificial Low Dimensional Electronic Systems, Institute for Basic Science, San31, Hyoja-Dong, Nam-Gu, Pohang 790-784, South Korea § Department of Mechanical Engineering and the Materials Science and Engineering Program, The University of Texas at Austin, Austin, Texas 78712, United States ‡

S Supporting Information *

ABSTRACT: A lithium−iodine (Li−I2) cell using the triiodide/iodide (I3−/I−) redox couple in an aqueous cathode has superior gravimetric and volumetric energy densities (∼ 330 W h kg−1 and ∼650 W h L−1, respectively, from saturated I2 in an aqueous cathode) to the reported aqueous Li-ion batteries and aqueous cathode-type batteries, which provides an opportunity to construct cost-effective and high-performance energy storage. To apply this I3−/I− aqueous cathode for a portable and compact 3.5 V battery, unlike for grid-scale storage as general target of redox flow batteries, we use a three-dimensional and millimeter thick carbon nanotube current collector for the I3−/I− redox reaction, which can shorten the diffusion length of the redox couple and provide rapid electron transport. These endeavors allow the Li−I2 battery to enlarge its specific capacity, cycling retention, and maintain a stable potential, thereby demonstrating a promising candidate for an environmentally benign and reusable portable battery. KEYWORDS: Carbon nanotube, current collector, iodine, aqueous cathode, redox batteries constructing a large fixed storage system coupled with a reservoir accommodating a large volume of liquid cathode and a flow-through system but not suitable for a portable battery. Efforts to increase the operating potential in redox flow batteries have been carried out using various redox couples. A zinc−cerium redox battery, for example, holds a high theoretical cell potential of ∼2.5 V but is limited by water electrolysis, a sluggish Ce4+/Ce3+ redox reaction, high internal resistance, and short lifespan.9,16−18 Unlike these redox couples, a triiodide/iodide (I3−/I−) can be applied for a promising aqueous portable battery with high energy density, rechargeability, and rate capability. The I3− can be constructed by dissolution of active material I2 in the presence of excess of I−, as in eq 119

P

ortable secondary battery technology has been developed around solid-electrode-based Li-ion batteries with suitable energy densities (for example, ∼176 W h kg−1 from Panasonic 18650 3.6 V Li-ion battery1). These Li-ion batteries have been extensively employed in mobile electronics whereas their high cost, limited cycle durability, and ineffective battery recycle have raised the total charge of electronics. In particular, the cathode and aprotic organic electrolyte, which occupy over 50% of the total material cost in the cell-level,2 are the predominant triggers for decline of storage capability during cycling by phase transition, structural disordering, metal dissolution, and electrolyte decomposition.3 This indicates that mechanical stress and parasitic side reactions inevitably occur in the solid cathode and at the solid cathode/aprotic electrolyte interface, which reduce the battery cycle-life. A liquid-type cathode, obviating the highcost organic electrolyte and problematic solid electrode/ electrolyte ensemble, can provide promise to construct a long-lifespan storage device. In a liquid cathode comprised of a soluble redox couple and solvent, a simple redox reaction does not cause any mechanical and chemical damage.4−7 The typical liquid cathode-based storage system is, however, not fit for the portable and compact batteries due to its low operating potential and poor energy density.8−10 Representatively, a vanadium redox battery exhibits a potential of 1.1−1.6 V and gravimetric and volumetric energy densities of 10−20 W h kg−1 and 15−25 W h L−1, respectively,11−15 which is suitable for © 2014 American Chemical Society

I 2(s) + I− ↔ I3−

K ≈ 720 ± 10 (298 K)

(1)

High solubility of I3−/I− redox couple (∼8.5 mol L−1 with a potassium salt) in the aqueous cathode enables to deliver superior energy density.6,7 In addition, the aqueous media is of benefit to the I3−/I− redox reaction; low polyiodide stability allows the I3− to the predominant electroactive species,19 and the electron-transfer rate for the I3− reduction reaction is Received: December 25, 2013 Revised: January 22, 2014 Published: January 29, 2014 1085

dx.doi.org/10.1021/nl404784d | Nano Lett. 2014, 14, 1085−1092

Nano Letters

Letter

Figure 1. Schematic illustrations and images of Li−I2 cell preparation process. (a) Scheme (left) and SEM images (right) of VACNTs. (b) Dry contact transfer of VACNTs from quartz to Ti film. Optical images are VACNTs before (right top) and after transfer (right bottom). (c) Surface treatment via O2 plasma and acid leaching. Right images represent wettability of VACNTs tested by shape of water-drop before (right top) and after surface treatment (right bottom). (d) Configuration of Li−I2 cell with VACNT current collector and I3−/I− aqueous cathode (left) and optical image of cell operation for lighting LED bulbs (right).

candidates with respect to (1) being lightweight and having suitable electronic conductivity (∼10−1 S cm−1), (2) appropriate design in terms of density, area, and thickness using chemical vapor deposition (CVD) method, and (3) potential for VACNT transfer to designed substrates and surface property modulation. In addition, oxygen functional groups on the CNTs have catalytic activity for I3−/I− redox reactions.29 Accordingly, the VACNT can serve as a promising platform and current collector for the I3−/I− redox reaction, thus demonstrating superior energy densities (330 W h kg−1 and 650 W h L−1), discharge potential (∼ 3.5 V vs Li+/Li), and cyclability for a high-performance portable Li−I2 battery. The 3D VACNT current collectors were prepared by the CVD growth, contact transfer, and surface treatment, illustrated in Figure 1. The VACNT mat was grown on a quartz substrate in the absence of predeposited metallic catalyst film using a low-pressure thermal CVD method (see Methods and Supporting Information Figure S1).30,31 Transmission electron microscopy (TEM) images depict multiple walls of CNT with a diameter of 30−100 nm (Supporting Information Figure S2). These CNTs were vertically aligned with a thickness of around 1 mm (scanning electron microscopy (SEM) images of Figure 1a and Supporting Information Figure S3) and the pore size of 2.73 nm estimated by Barrett−Joyner−Halenda (BJH) method (Figure 2a). The long and dense VACNTs can serve as an appropriate 3D current collector to decrease the redox couple’s diffusion length in the aqueous cathode. The Raman spectrum

superior to those in organic medium such as acetonitrile, DMSO, DMF, and PC.20,21 Namely, the redox reaction is simply limited to the I3−/I− via fast two-electron transfer at ∼0.54 V vs standard hydrogen electrode (SHE), which is suitable to avoid water electrolysis, as in eq 222 I3− + 2e− ↔ 3I−

(2) ·

−

I2−·/I−

Other possible reactions such as I /I (1.33 V vs SHE), (1.03 V vs SHE) and I3−/I2−· (0.04 V vs SHE) are away from the I3−/I− with the potential difference of over 0.5 V.23 The combination of Li anode, that is, construction of lithium− iodine (Li−I2) cell, determines the total cell reaction as follows (eq 3) with a standard reduction potential (E0) of ∼3.58 V vs Li+/Li, which contributes to achieve appreciable energy densities I3− + 2Li ↔ 3I− + 2Li+

(3)

To further improve the Li−I2 cell performance for portable battery application, here we introduce a three-dimensional (3D) nanoarchitectured current collector and demonstrate high capacity and cycling performance. Unlike typical 2D current collectors composed of metal or porous carbon with polymer binder, binder-free 3D conducting nanostructures24−28 shorten the diffusion length of redox couples in the aqueous cathode, allowing for rapid electron transport and the avoidance of binder aging. In particular, a vertically aligned multiwalled carbon nanotube (VACNT) mat can be one of the successful 1086

dx.doi.org/10.1021/nl404784d | Nano Lett. 2014, 14, 1085−1092

Nano Letters

Letter

Figure 2. Characterizations of VACNTs. (a) Pore size distribution of as-prepared VACNTs estimated by BJH method. (b) Raman spectra of asprepared (black, bottom) and surface-treated (red, top) VACNT sidewall. ID/IG are 0.57 and 0.71, respectively. (c) SEM image of VACNTs after transfer onto Ti film. The VACNT mat has a 3D open structure. (d) I−V curves of as-prepared (blue) and surface-treated (red) CNTs. Error bars in (d) indicate standard deviations from averaged currents of three different electrodes at each sample.

resistance of individual CNTs, the VACNT mat still maintained sufficiently low sheet resistance (1.0 ± 0.2 and 2.5 ± 0.2 Ω·cm for before and after surface treatment, respectively). Finally, this VACNT mat on Ti (a geometric area of 0.2 cm2) was assembled with 135 μL aqueous cathode containing 0.1 M I2 (17.5 mg cm−2) and 1 M of KI. It is worth noting that K+ and I−, which is not involved in the I2 transformation, from the KI are nonelectroactive species for Li−I2 cell operation. In addition, no Li ion is added in the as-prepared aqueous cathode unlike other Li batteries with aqueous media,34−36 aqueous cathode-type batteries,4,5 and Li-ion batteries, which aid in cutting down the material cost of the cathode. The Li−I2 cells were completely constructed by assembly of the aqueous cathode, a Li-ion conductive ceramic separato r (Li1+x+3zAlx(Ti,Ge)2−xSi3zP3−zO12, LATP), and a metallic Li anode containing 1 M of LiPF6 in EC/DMC electrolyte (Figure 1d). The true redox potential of the I3−/I− redox couple in the Li−I2 hybrid redox cell was then determined using a cyclic voltammetry (CV). The CV curves in Figure 3a showed cathodic and anodic peak-potentials of 3.5 and 3.7 V vs Li+/Li, respectively, which are close to the effective redox potential of the I3−/I− (E ≈ 3.55 V vs Li+/Li estimated from 0.1 M of I3− and 1 M of KI using Nernst equation). Upon the following cycles, the cathodic and anodic peak-potentials were slightly shifted while the integrated redox peak-areas became identical. The aqueous cathode was stable at the sweeping potential range, manifesting neither H2 nor O2 evolution peak. Further tests using a galvanostatic method could evaluate the Li−I2 cell performance. The charge/discharge profiles in Figure 3b exhibit a discharge capacity of 206 mA h g−1, that is, ∼98% of the theoretical capacity based on the I2 (211 mA h g−1) and stable charge and discharge potentials (3.75 and 3.4 V vs Li+/Li,

of the VACNT sidewalls in Figure 2b reveals substantial graphitic structure based on the relatively small intensity of the carbon defect-derived D band (a Raman shift of 1320 cm−1) referred to graphite structure-derived G band (1570 cm−1) (ID/ IG = 0.57).32 Unlike the sidewalls, the VACNTs had more defective tip as evidenced by the higher D band intensity (ID/IG = 1.18, Supporting Information Figure S4) on the top side of the VACNT mat. The VACNT mat on the quartz was then transferred onto a carbon-coated Ti film using a dry contact transfer, which resulted in the seamless and well-aligned VACNT mat on the Ti (right optical image of Figure 1b and SEM top-view image of Figure 2c). Subsequent oxygen plasma treatment and acid leaching improved the VACNT hydrophilicity. Figure 1c shows spherical and spreading water-drops on the VACNT mat before (right top optical image) and after surface treatment (right bottom image), respectively, which demonstrate greatly increasing VACNT wettability by surface oxidation. These surface oxidation treatments modulated the chemical structure and electrical conductivity of the VACNTs. The D to G band intensity ratio of the VACNT sidewalls in Raman spectrum was increased to 0.71 after the oxygen plasma treatment (Figure 2b). X-ray photoelectron spectra exhibited slightly increasing oxygen-species peak at 531−534 eV, deconvoluted and assigned to carbonyl, hydroxyl, epoxide, ester, and carboxylic groups from the VACNTs,33 in the O 1s binding energy (BE) region and a slightly decreasing sp2 carbon hybridization in the C 1s BE region (Supporting Information Figure S5). The I−V curves of a few CNT bundles dispersed between prepatterned source and drain electrodes (Au/Cr) on SiO2/Si substrates show almost twice the electrical resistance after the surface oxidation (27 kΩ) compared with before the treatments (13 kΩ) (Figure 2d). It is noteworthy that although the surface oxidation increased the electrical 1087

dx.doi.org/10.1021/nl404784d | Nano Lett. 2014, 14, 1085−1092

Nano Letters

Letter

capacity on the aqueous cathode under the low I 2 concentration. Binder-free VACNTs could eliminate unnecessary mass and avoid parasitic corrosive reaction arising from polymer binder, thus providing stable performance. After cycling, there was no visible evidence of either CNT structure damage (Supporting Information Figure S6) or LATP separator degradation (Supporting Information Figure S7a), which was further evidenced by negligible impedance change of the Li−I2 cell (Supporting Information Figure S7b). In addition, the idle I3− concentration with the corresponding stranded Li+ after the first cycle is a half of that in the aqueous cathode linked with a 2D Super P carbon/binder current collector reported previously (Supporting Information Figure S8).7 Furthermore two 2D current collectors, Ti foil (representing a typical metal) and a pyrolytic carbon film prepared by the low-pressure thermal CVD on Ti, demonstrated poor rechargeability and decreasing discharge potentials for cycling (Supporting Information Figures S9 and S10). Consequently, higher capacity is mostly attributed to the short I3−/I− diffusion length in the long and dense 3D VACNTs. Along with the advantages of the 3D current collector, some techniques such as constant-current−constant-potential (CCCP) charging and preloading of Li salt could be investigated to decrease idle I3− concentration. Switching of the constant current mode to the constant potential at 4.2 V versus Li+/Li raised Coulombic efficiency up to ∼99% on the first cycle with normalized capacities of 0.98−0.99 (Figure 4a). Besides, the incorporation of a small concentration of LiI with the as-prepared aqueous cathode could attain ∼100% Coulombic efficiency with a normalized capacity of 0.98 (Figure 4b). In addition, increasing I2 loading concentration, which should be conducted for practical portable batteries, enhances Coulombic efficiency. Figure 4c,d demonstrates improved Coulombic efficiencies from 81 to 95% with I2 concentrations of 0.05 and 0.2 M, respectively, for the first cycle. Furthermore, the I2/KI-saturated aqueous cathode achieved almost 100% Coulombic efficiency with a capacity of 32 mA h and specific and volumetric energy densities of ∼330 W h kg−1 and ∼650 W h L−1, respectively, (Supporting Information Figure S11). These results arise from almost constant concentrations of the stranded Li+ and idle I3− regardless of the I2 amount. The calculated stranded Li+ concentrations are 19−20 mM in the variable I2 concentrations such as 0.05, 0.1, and 0.2 M (Figure 4e). Further improvement of the Li−I2 cell performance for the portable battery could be achieved by optimum operation conditions such as operating temperature and current rate. Figure 5a displays every 2nd−20th cycled charge/discharge profile of the Li−I2 cell at a current rate of 2.5 mA cm−2 and at each temperature of 278, 298, 318, and 328 K in sequence. The capacity and charge/discharge potentials were stable in the given temperature range, and obviously with increasing temperature the capacity was enhanced and the potential polarization decreased. At 328 K, the specific capacity approached 207 mA h g−1, that is, ∼98% of the theoretical capacity, and charge/discharge potentials were between 3.5 to 3.7 V versus Li+/Li. In particular, the potential polarization (0.1−0.15 V) at 328 K is considerably smaller than those of the rechargeable solid-electrode−Li-I2 battery37 and iodine−carbon composite using organic electrolyte38 recently developed. It is noted that the effective redox potential of the I3−/I− at a temperature range of 278−328 K is around 3.55 V versus Li+/ Li estimated from Nernst equation. Consequently, the

Figure 3. Electrochemical performances of Li−I2 cells with VACNT current collectors. (a) CV curves at a sweeping rate of 0.05 mV s−1. A red curve indicates the first cycle. (b) Charge/discharge profiles at a current rate of 2.5 mA cm−2 at 298 K and (c) the corresponding specific capacity and Coulombic efficiency for 200 cycles.

respectively, on average) for the first cycle at a current rate of 2.5 mA cm−2. The discharge capacity was then decreased to 185 mA h g−1 in following cycles (2−200 cycles). This behavior arises from no additional Li+ ion in the as-prepared aqueous cathode. The decreasing capacity was responsible for the 90% Coulombic efficiency of the first cycle, resulting from the considerable ionic conductivity drop associated with the low Li+ ion concentration remaining in the aqueous cathode at 90% charge.7 Upon the subsequent cycle, the stranded Li+ ions (∼0.02 M at a current rate of 2.5 mA cm−2 and 298 K) maintain reasonable ionic conductivity while they make the corresponding idle I3− (∼0.01 M at the same condition), which results in 100% Coulombic efficiency but lower capacity, respectively. The Li−I2 cells then exhibited remarkably stable cycling performance for 200 cycles: excellent capacity retention (185 ± 3 mA h g−1), steady charge/discharge potentials (potential polarization ≈ 0.35 V), and ideal Coulombic efficiency (100 ± 0.5%, Figure 3c). We point out that the usage of 3D VACNT mat can improve cycle life and specific 1088

dx.doi.org/10.1021/nl404784d | Nano Lett. 2014, 14, 1085−1092

Nano Letters

Letter

Figure 4. Effect of CCCP charging and preloading of Li salt on Coulombic efficiency for the first cycle and relationship between stranded Li+ ion and I2 loading concentration. (a) The first cycle profile using CCCP charging. The discharged cell was recharged to 4.2 V at a current rate of 2.5 mA cm−2, then a constant current mode was applied until the current rate decreased to 0.25 mA cm−2. (b) The first cycle profile with the additional 0.02 M of LiI in the I2/KI aqueous solution. (c−e) Correlation between stranded Li+ ion and active material of I2 concentration. First cycled charge/ discharge profiles of Li−I2 cells with an I2 concentration of (c) 0.05 and (d) 0.2 M in the presence of 1 M KI in the aqueous cathode. (e) Estimated stranded Li+ ion concentration (black columns) after the first cycle and I2 concentration (red columns).

LATP separator, which was estimated to be ∼520 and ∼150 Ω at 278 and 328 K, respectively. This result suggests that the battery performance can be further improved using an advanced separator. Although the ionic conductivity of a LATP separator can be increased with temperature, the development of highly ionic conductive separator having considerable mechanical and chemical stability and reasonable cost is the ultimate target for practical applications in the near future.43,44 Besides, due to safety issues metallic Li in the anode part would be replaced with alternative materials such as a commercially available graphite and small organic molecules with suitable redox potential.45 Despite these challenges, however, it is interesting that the I3−/I− aqueous cathode-based Li−I2 cell offers sufficiently high energy and power densities as required for a portable battery. The diagram in Figure 6 exhibits a summary of various redox batteries’ performance. With respect to discharge potential and Coulombic efficiency, the Li−I2 hybrid redox battery in this work shows one of the best performances over conventional redox batteries such as vanadium redox battery,46 zinc−cerium redox battery,16 zinc−bromine battery,47 molten sodium−sulfur battery,48 all-organic lithium-ion redox battery,49 iron/vanadium hybrid redox battery,50 and newly emerged promising redox batteries such as lithium-polysulfide battery,51 and Fe(CN)63−/Fe(CN)64− redox battery,4 lithium-ion batteries using aqueous media.36,52−56 The specific energy density of I3−/ I− aqueous cathode in the Li−I2 cell was estimated to ∼330 W h kg−1 thanks to high discharge potential. The projected energy

improved potential polarization cannot be fully accounted for the I3−/I− itself in the aqueous media. This behavior is mostly attributed to the LATP separator. The low ionic conductivity of the LATP separator (10−4−10−5 S cm−1 at 278−328 K),39 2 orders of magnitude lower than the aqueous and aprotic organic medium (10−2−10−3 S cm−1 at 278−328 K),40,41 logarithmically increases with temperature, thus aiding smooth Li+ ion commute and suppressing internal resistance. Figure 5b shows considerable discharge capacity retention (a variation of ±3%) and Coulombic efficiency (100 ± 0.5%) for the 2nd− 20th cycles at the given temperature. The Coulombic efficiency values somewhat varies at low temperature (for example 278 K) mostly due to low LATP conductivity whereas they are in close proximity to 100% at high temperature. This behavior can be also clearly observed for the first cycles where the limited Li+ ion concentration restricts charge process. Supporting Information Figure S12 shows improved Coulombic efficiency with increasing temperature during the initial cycles. The increment of LATP conductivity also raises power density. The polarization graph in Figure 5c demonstrates discharge potential stably retained at 3.3 V versus Li+/Li at 328 K and a current rate of 12 mA cm−2, which corresponds to a power density of 40 mW cm−2, around 4 times higher than that at 278 K and several orders of magnitude greater than the solid Li−I2 battery.37,42 The correlation between discharge potential and temperature was clearly displayed in Figure 5d. Increment of temperature-dependent discharge potential upon current rate resulted from the internal resistance of Li−I2 cell containing the 1089

dx.doi.org/10.1021/nl404784d | Nano Lett. 2014, 14, 1085−1092

Nano Letters

Letter

Figure 5. Temperature-dependent galvanostatic performances of Li−I2 cells with VACNT current collectors. (a) The 2−20 times cycled charge/ discharge profiles and (b) the corresponding cycling performance recorded from 278 to 328 K at a current rate of 2.5 mA cm−2. (c) Polarization graph and (d) the corresponding discharge potential change with respect to current rate (0.1−12 mA cm−2) and temperature (278 (black), 298 (red), and 328 K (blue)).

total mass of packaged battery.52,57 Along with the high energy density of the I3−/I− aqueous cathode-based Li−I2 cell, the employment of 3D VACNT current collector retains strikingly stable discharge potentials at even high current rate and temperature, which is distinguished from the 2D current collector-based cell presenting unsettled potentials over 5 mA cm−2 at 298 K.7 This demonstrates the VACNT mat has prominent chemical mechanical stability in the I3−/I− aqueous cathode. The millimeter-thick and dense VACNT mat is also lightweight (approximately ∼1.2 mg cm−2), which does not significantly reduce the total energy density in the battery pack. The length of VACNT can be modulated from micrometer to millimeter, which benefits in shortening I3−/I− diffusion pathways according to the design of aqueous cathode configurations. It is noteworthy that while the prototype Li− I2 hybrid redox cells demonstrated in laboratory scale show improved performance, further technical development to curtail an expensive and cumbersome process for the preparation of a 3D current collector is strongly required for practical battery applications. In summary, we demonstrated the Li−I2 hybrid redox cell with the 3D VACNT current collector for aqueous cathode. Highly dense, lightweight, and binder-free VACNT current collector was constructed via CVD synthesis and dry contact transfer method. In addition, mild surface treatment modulated the surface property of VACNT to be hydrophilic, which contributed to accommodate the aqueous cathode with short I3−/I− diffusion length, thus achieving high cycling performance and stable potential. In particular, the 3.5 V Li−I2 cell with the VACNT reached 98% of the theoretical capacity (207 mA h

Figure 6. Diagram of various redox batteries’ performance: vanadium redox battery (ref 46), zinc−cerium redox battery (ref 16), zinc− bromine battery (ref 47), sodium−sulfur battery (ref 48), all-organic lithium-ion redox battery (ref 49), and iron/vanadium hybrid redox battery (ref 50), lithium-polysulfide battery (ref 51), Fe(CN)63−/ Fe(CN)64− redox battery (ref 4), and lithium-ion battery with aqueous media (refs 36, 52, and 56). Blue and red symbols denote discharge potential and Coulombic efficiency, respectively.

density of a packaged Li−I2 battery was expected to be ∼170 W h kg−1 considering 50 − 60 wt % of the cathode mass ratio in 1090

dx.doi.org/10.1021/nl404784d | Nano Lett. 2014, 14, 1085−1092

Nano Letters

Letter

g−1) with ideal Coulombic efficiency (∼100%) and small polarization potential (0.1 V) at 328 K. The I2-saturated aqueous cathode delivered energy densities of 330 W h kg−1 and 650 W h L−1, respectively, at room temperature, which can give an opportunity to produce a new type of high-performance portable battery. In addition, reusable and environmentally benign I3−/I− aqueous cathode from I2 can make a green battery, which is another significant issue for future energy storage. Methods. Materials. All chemicals were used as received. Titanium foil (99.5%, 100 μm in thickness) and copper foil (99.9%, 30 μm in thickness) were purchased from Nilaco Corporation. Potassium iodide (KI, 99.5%), iodine (I2, 99%), and ferrous chloride (FeCl2, 98%) were from Wako Chemicals. Acetylene gas (C2H2, 98%) was purchased from Iwatani Corporation. The ceramic solid separator LATP (Li1+x+3zAlx(Ti,Ge)2−xSi3zP3−zO12) was purchased from Ohara Corporation (a thickness of 150 ± 20 μm, mass of 3.15 g cm−3, and ionic conductivity of 10−4 S cm−1 at room temperature). Lithium hexafluorophosphate (LiPF6, anhydrous) was from Aldrich. Ethylene carbonate (EC, water content