All-solid-state Batteries with Thick Electrode Configurations

Transcription

1 All-solid-state Batteries with Thick Electrode Configurations Yuki Kato, * Shinya Shiotani, Keisuke Morita, Kota Suzuki, Masaaki Hirayama, Ryoji Kanno Toyota Motor Europe NV/SA, Hoge Wei 33, 1930 Zaventem, Belgium. Toyota Motor Corporation, 1200 Mishuku, Susono, Shizuoka, Japan. Department of Chemical Science and Engineering, School of Materials and Chemical Technology, Tokyo Institute of Technology, 4259 Nagatsuta, Midori, Yokohama, Japan.

2 1. Capacity loading and energy density of the presented cell 1.1 Capacity loading The capacity loading of the cell prepared in this study is presented in Fig. S1, together with the values reported for LiCoO 2 and graphite in all-solid-state systems. 8,10 12 Among the various all-solid-state batteries discussed to date, the capacity loading of our cell (L600) is the highest reported (i.e., 15.8 mah cm 2 ). Figure S1. Capacity loading of the L600 cell prepared in this study. The reported values for LiCoO 2 and graphite systems are also plotted. All capacities were calculated using specific capacities of 137 and 370 mah g 1 for LCO and graphite, respectively. The values listed in Table 2 (main text) were used for the density of each electrode.

3 1.2 Energy density The effect of the electrode thickness on the energy density was determined by calculating the energy density and specific energy of each configuration listed in Table 3 (main text). Both the energy density and the specific energy were found to increase with an increasing electrode thickness (Fig. S2). A maximum energy density of 435 Wh L 1 and a specific energy of 180 Wh kg 1 were obtained for the L600 configuration due to minimization of the volumic/gravitic contribution of the inactive components of the separator and current collectors. Figure S2. Evolution of (a) the energy densities, and (b) the specific energies of the cells. Values are calculated assuming a cell configuration that employs a 20 µm current collector for the cathode (Al; 2.7 g/cc) and the anode (Cu; 8.9 g/cc).

4 2. Experimental methods 2.1 Synthesis of the solid electrolytes SE1 (Li 10 GeP 2 S 12 ) For preparation of the SE1 solid electrolyte, the raw materials Li 2 S (Aldrich, 99.98%), P 2 S 5 (Aldrich, 98%), and GeS 2 (Kojundo Chemicals, 99.9%) were mixed in an appropriate molar ratio in an argon-filled glove box. The resulting mixture was placed in a ZrO 2 pot together with ZrO 2 balls (φ10 mm) and mechanically milled by planetary ball milling at 370 rpm for 40 h, prior to placing in a quartz tube and heating at 550 C for 8 h. The X-ray diffraction (XRD, MiniFlex600, Rigaku) pattern of the obtained SE1 is shown in Figure S3 (as obtained using an air-tight jig prepared in-house). The sample was confirmed to have the structure of Li 10 GeP 2 S 12 (ICSD#248307) as the main phase. An ionic conductivity of 3.2 ms cm 1 was obtained for the cold pressed pellet state at 25 C, and this value was used for subsequent investigations. In addition, a comparable conductivity to literature values 13 (i.e., 8.3 ms cm 1 ) was obtained for the sintered pellet state. The slightly lower ionic conductivity can likely be attributed to a secondary beta-li 3 PS 4 phase, whose presence may be due to the slightly higher temperature employed here than that required to obtain a single LGPS phase. 34 In addition, small (uncontrollable) differences in the synthetic conditions (impurities, natural cooling rate of the oven) may influence the temperature sensitivities of the materials. Figure S3. XRD pattern of SE1. The main phase was identified as Li 10 GeP 2 S 12. : Beta phase diffraction peaks.

5 2.1.2 SE2 (75Li 2 S 25P 2 S 5 ) For preparation of the SE2 solid electrolyte, the raw materials Li 2 S (Aldrich) and P 2 S 5 (Aldrich) were mixed in an appropriate molar ratio in an argon-filled glove box. The resulting mixture was placed in a ZrO 2 pot together with ZrO 2 balls (φ10 mm) and mechanically milled by planetary ball milling at 370 rpm for 40 h. The XRD pattern of the obtained SE2 is shown in Figure S4, where it is apparent that an amorphous specimen was obtained. Intensity (a.u.) θ / Figure S4. XRD pattern of SE SE3 (30LiI 70(0.75Li 2 S 0.25P 2 S 5 )) For preparation of the SE3 solid electrolyte, the raw materials Li 2 S (Aldrich), LiI (Aldrich), and P 2 S 5 (Aldrich) were mixed in an appropriate molar ratio in an argon-filled glove box. The resulting mixture was then placed in a ZrO 2 pot together with ZrO 2 balls (φ10 mm) and mechanically milled by planetary ball milling at 370 rpm for 40 h. The XRD pattern of the obtained SE3 is shown in Figure S5, and once again, the results indicate that an amorphous specimen was obtained.

6 Intensity (a.u.) θ / Figure S5. XRD pattern of SE Ionic conductivity measurements Each synthesized solid electrolyte was placed in an alumina cylinder and pressed at 420 MPa to form a pellet. Then the pellet was then sandwiched by a stainless steel current collector and the AC impedance was measured at 25 C in an argon atmosphere in the frequency range of 0.1 Hz to 1 MHz using a frequency response analyzer (Solarton 1260). 2.3 Battery fabrication For the purpose of this study, either Li 10 GeP 2 S 12 (SE1) or Li 2 S-P 2 S 5 glass (SE2) was used for the cathode layer, while LiI-Li 2 S-P 2 S 5 (SE3) was employed in the separator and in the anode layer because of its high ionic conductivity and good electrochemical stability against graphite. 16 Prior to cell preparation, large electrolyte particles were removed using a sieve (ASONE, nickel sieve, 5 µm mesh). LiCoO 2 particles (D5, Toda) measuring 3 5 µm in diameter were employed for preparation of the batteries. Thus, the LiNbO 3 -coated LiCoO 2, the solid electrolyte (SE1 or SE2), and acetylene black powder

7 were mixed for 5 min using a vortex mixer. The graphite anode was prepared by mixing graphite powder (Mitsubishi Chemical Corporation; ~10 µm diameter) and SE3 in an agate mortar. The powdered cathode layer, separator, and anode layer were placed in an alumina cylinder and pressed at 500 MPa to form pellets. The all-solid-state cells were prepared by connecting the cathode and the anode to stainless-steel current collectors at a pressure of 10 MPa. Finally, a pressure of 50 MPa was applied by screwing the stainless-steel housing. 35 The diameter of the resulting all-solid-state cell was mm (1 cm 2 ). 2.4 Charge-discharge examination The prepared batteries were charged and discharged at a constant current (0.5 ma cm 2 ) in a voltage range of V. Rate capability tests were conducted as follows. Initially, the cells were charged to 4.1 V at a constant current (0.5 ma cm 2 ) followed by a constant voltage (cut-off current = ma cm 2 ), and the rate capability was investigated by varying the discharge current density with a cut-off voltage of 2.5 V. All electrochemical tests were performed under an argon atmosphere at 25 C. 2.5 Cell impedance measurements The prepared all-solid-state cells were charged to 4.1 V using the CCCV (constant current constant voltage) charging mode with a constant current of 0.5 ma cm 2 and a cut-off current of ma cm 2. The AC impedance was then measured at 25 C under an argon atmosphere applying 10mV at between 0.1 mhz and 1 MHz using a frequency response analyzer (VMP3; Bio-Logic Science Instruments). 2.6 Scanning electron microscopy (SEM) The cross-sections of the cells were prepared using an ion milling system employing an Ar ion beam (E-3500; Hitachi High-Technologies) and were observed by scanning electron microscopy (SEM, JSM-6610A; JEOL).

8 3. SEM images of all-solid-state cells Figure S6. Cross-sectional SEM image of the cathode layers containing (a) SE1 and (b) SE2, and (c) the anode layer containing SE3. Scale bar=10 µm. SEM images of the electrode layers were obtained prior to cycling. In the case of the cathode layer, no significant differences were observed in the particle distribution shown in the cross-sectional SEM images of the electrodes containing SE1 and SE2. It should be noted that if large solid electrolyte particles are present in the electrode composite, both the ionic transport and the tortuosity factor may be affected. Thus, after preparation and grinding, the powder samples were sieved. In addition, since thiophosphates exhibit similar elastic properties 36 and can be easily deformed by cold pressing, the electrode morphology would be comparable especially at high SE volumic fractions. It would therefore be reasonable to state that τ c1 equals τ c2. As spherical graphite particles were employed herein, the deformation of graphite must occur during mixing using the agate mortar.

9 4. Impedance analysis Table S1. Parameters used for the impedance simulations of Figure 4 for (a) L300 and (b) L600 (a) CPE Z FSW Lelec / m A / m 2 σ eff / Sm 1 κ eff / Sm 1 R ct / Ω Q n d / m D / m 2 s 1 Cdiff / F N / m 3 Cathode 3.00* a 30 b * c 5*10 6 d 5*10 15 f *10 10 Anode 2.62* a 70 b * c 5*10 5 e 5*10 11 g *10 9 (b) CPE ZFSW L elec / m A / m 2 σeff / Sm 1 κeff / Sm 1 Rct / Ω Q n d / m D / m 2 s 1 C diff / F N / m 3 Cathode 6.00* a 30 b * c 5*10 6 d 5*10 15 f *10 10 Anode 5.24* a 70 b * c 5*10 5 e 5*10 11 g *10 9 a Obtained from rate-capacity characteristics. b Obtained by DC polarization. c Estimated from SEM measurements. d Reference 31. e Reference 30. f Obtained from the dq/dv value at 4.1 V. g Reference 23 The total impedance Z was modeled as the series connection of separator resistance (R sep ) and electrode impedance Z elec for anode and cathode (Z = Z elec_a + R sep + Z elec_c ). The electrode impedance were modeled based on the transmission line model (TLM) and the finite space Warburg element (FSW) to express the degree of mass transport in the electrode layer and in the active materials. For this purpose, we began with the analytic TLM solution developed by Tröltzch and Kanoun 26 as indicated below: = h h + (.1) where L elec, Z e, and Z i represent the electrode thickness, the impedance for an electron, and the impedance for ion migration. Z CT represents the charge transfer resistance per unit length. Z e and Z i can be expressed using the effective conductivities and the electrode area. The effective electron conductivity was obtained by DC polarization

10 using an ion-blocking cell, and the effective ionic conductivity obtained from the rate-capacity experiments in Figure 3 and Table 5 was also employed: = 1 1 and = (.2) Since the thickness dependence at low frequencies was pronounced in the Nyquist plot, the time constant of mass (ion) transfer in the electrode may be comparable to or greater than that of diffusion in the active materials. We therefore considered a model that integrates the FSW element into Z CT. The charge transfer impedance for the intercalation materials (Z particle ) can be accurately expressed by the unit element of connected FSW and RCT//CPE in series Finally, Z particle must consist of a parallel circuit of unit elements since the corresponding values per unit volume are unknown. Then ZCT can be expressed as: = (.3) The FSW impedance can therefore be expressed as: = τ (τ ). coth (τ ). (.4) where τ d, C diff, ω, and j are the characteristic diffusion time constant, the differential capacity obtained from the (dq/dv) curve, the angular frequency, and 1, respecitively. Using a particle size d and a diffusion coefficient D, the diffusion time constant can be estimated by: = (.5) With these results in hand, the impedance spectra could be simulated. Thus, the values of R CT, CPE, and N were refined by fitting, as such values must be determined for individual systems. The simulated result is shown in Figure S7. The good curve fitting indicated the validity of the model used herein. It should also be noted that the slight deviation observed at a frequency of approximately Hz likely reflects the distribution of the time constant of diffusion in the active materials 23 derived from the non-uniform particle size distribution. This can be enhanced by adding the Finite length Warburg impedance in series to the FSW or placing an additional FSW in parallel with the existing FSW; 23 however, this is beyond the scope of this study.

11 5. Estimation of battery performance Figure S7. The simulated rate-capacity characteristics of our cells plotted using Eq. 4 with (i) a low tortuosity factor (τ=ε 0.5 ) and (ii) a high ionic conductivity (κ=10 ms cm 1 ) for the solid electrolyte.

12 6. Additional references (34) Hori, S; Kato, M; Suzuki, K.; Hirayama, M.; Kato, Y.; Kanno, R. Phase Diagram of the Li 4 GeS 4 Li 3 PS 4 Quasi-Binary System Containing the Superionic Conductor Li 10 GeP 2 S 12. J. Am. Ceram. Soc. 2015, 1 9. (35) Kato, Y.; Kawamoto, K.; Kanno, R.; Hirayama, M. Discharge Performance of All-Solid-State Battery Using a Lithium Superionic Conductor Li 10 GeP 2 S 12. Electrochemistry 2012, 80, (36) Deng, Z.; Wang, Z.; Chu, I-H.; Luo, J.; Ong, S. P. Elastic Properties of Alkali Superionic Conductor Electrolytes from First Principles Calculations. J. Electrochem. Soc. 2016, 163, A67 A74.