Development of Li₁₀GeP₂S₁₂-type Solid Electrolytes for the Understanding of Superionic Conduction, and Their Application in All-solid-state Batteries

Abstract

The lithium superionic conductor Li₁₀GeP₂S₁₂ (LGPS) and its crystal structure analogues can be applied as solid electrolytes for all-solid-state lithium-ion batteries due to their exceptionally high ionic conductivities, which exceed 10 mS cm⁻¹ at room temperature. This paper reviews the author’s publications into LGPS and its derivatives. Initially, its unique crystal structure is described with a particular focus on the ionic conduction mechanism. Subsequently, the syntheses of various structural analogues of LGPS are summarized, focusing on the phase diagram containing LGPS, as well as highlighting Li_9.54[Si_0.6Ge_0.4]_1.74P_1.44S_11.1Br_0.3O_0.6, which was discovered during high-entropy material design. Finally, the development and understanding of all-solid-state batteries that incorporate the developed solid electrolytes are described. This review is notable due to the importance of discovering novel solid electrolytes for further clarification of the superionic conduction mechanism and for improving battery performances.

1. Introduction

In contrast to current lithium-ion batteries, which employ organic liquid electrolytes, all-solid-state Li-ion batteries are based on solid electrolytes possessing high ionic conductivities.^1,2 Consequently, they are expected to offer several advantages, such as a wider range of operating temperatures,³ higher power densities,³ and the utilization of novel active materials^4–6 that cannot be used in batteries based on conventional liquid electrolytes. The performances of all-solid-state batteries significantly depend on the properties of the solid electrolyte, wherein sulfide-based electrolytes have been demonstrated to exhibit higher room-temperature (RT) ionic conductivities (e.g., 10 mS cm⁻¹) and enhanced mechanical strength and flexibility compared to oxide- and halide-based electrolytes,⁷ thereby producing high-performance battery cells.^3–6

Li₁₀GeP₂S₁₂ (LGPS)⁸ and its iso-structural crystalline phases,⁹ which constitute the main research focus of the author, have been demonstrated to exhibit high bulk ionic conductivities of >10 mS cm⁻¹, potentially rendering them superior to most sulfide-based solid electrolytes. These high conductivities are ascribed to the unique crystal structures of these species, as presented in Fig. 1.¹⁰ Since no similar crystal structure had been previously reported, this unique structure was registered as the “LGPS-type” structure in the Inorganic Crystal Structure Database.¹¹ In this structure, Li ions are widely distributed over the skeletal framework, leading to a high ionic conductivity that is comparable to those of organic liquid electrolytes. A variety of iso-structural derivatives have also been synthesized by varying framework units. For example, Li_9.54[Si_0.6Ge_0.4]_1.74P_1.44S_11.1Br_0.3O_0.6 (LSiGePSBrO),¹² possessing an extremely high bulk ionic conductivity of 32 mS cm⁻¹ at room temperature, was synthesized based on the “high entropy design” guidelines for alkaline-ion conductors.¹³

Figure 1.

Crystal structure and potential–distance profiles determined for the LGPS single crystal. (a, b, c) Perspective views along the a axis (a), and the c axis to show the Li4 (b) and Li2 (c) sites. The Li atoms determined by neutron diffraction analyzes at 10 K are depicted by colored spheres, while the [P]S₄ and [P/Ge]S₄ units are represented by light- and dark-grey tetrahedra, respectively. The blue arrows indicate Li ion jumping between sites in possible conduction pathways. (d) Nuclear density distributions of the Li atoms as visualized by MEM analyses at RT. (e) Potential–distance profiles at RT for the pathway through Li sites at different potential values. The numbers give the barrier heights in eV for the corresponding paths. Reproduced from Ref. 10 with permission from the Royal Society of Chemistry.

This review covers the author’s work into the crystal structure of LGPS, its ionic conduction mechanism, the phase diagram related to the synthesis of various LGPS-type materials, and the application of these materials to all-solid-state battery cells. The current understanding of these battery cells is also described.

2. LGPS Crystal Structure and Its Role in Ionic Conduction

Figure 1 shows the crystal structure of LGPS along with its possible ionic conduction pathways. Considering previous studies regarding the crystal structure of LGPS,^8,14,15 the structural model determined using single-crystal neutron diffraction¹⁰ was employed in this figure. In the tetragonal lattice of LGPS with the P4₂/nmc space group, the structural framework is formed from a complex alignment of isolated tetrahedral units of PS₄ and [Ge/P]S₄. Each tetrahedron is composed of one center framework cation (i.e., Ge or P) and four anions (i.e., S). In the [Ge/P]S₄ tetrahedral units, Ge and P share a crystallographic site. The Li ions are distributed between the structural units, wherein five Li sites were identified by single-crystal neutron diffraction analysis at 10 K (Fig. 1a).¹⁰ In the direction along the c-axis, three Li sites (i.e., Li1, Li3a, and Li3b) connect with one another to form one-dimensional channels. Figures 1b and 1c show that this channel connects with one another over the ab-plane via Li4 or Li2 sites. In Fig. 1d, the Li sites are visualized in the form of the Li nuclear density distribution, as obtained using the maximum entropy method (MEM) of data analysis for the single-crystal neutron diffraction results obtained at RT. The distribution at the Li-triad site (i.e., the combination of Li1, Li3a, and Li3b, which are clearly distinguished at 10 K) spreads continuously and widely along the c-axis, indicating that the ions migrate along the c-axis via these sites. In contrast, the Li distribution at the Li2 and Li4 sites adopts the form of flattened ellipsoids over the ab-plane, indicating that ions migrate over the ab-plane via these sites.

The energy barriers for Li-ion migration along these possible pathways (see Figs. 1a, 1b, and 1c), were estimated using one-particle potential (OPP) analysis for the data recorded at RT.¹⁰ Figure 1e shows the profiles of the OPPs as a function of the distance along the pathway via the potential saddle points. The potential barriers for the Li2 → Li3a and Li4 → Li1 site migrations were determined to be 0.55 and 0.35 eV, respectively, for pathways over the ab-plane (Figs. 1b and 1c), while that for the migration along the c-axis (via the Li3a → Li3a → Li1 → Li1 site, Fig. 1a) was determined to be 0.09 eV. The ionic conduction in the pathway along the c-axis was therefore considered to be dominant, which is consistent with the MEM analysis results.

Quasi elastic neutron scattering (QENS) measurements were performed for LGPS to reveal whether the OPP barriers between the Li sites are related to practical Li-ion dynamics.¹⁶ Using the QENS method, where the energy transfer occurring during neutron scattering is analyzed, the dynamic behavior with a correlation time in the order of 100 ps can be captured.¹⁷ Figure 2a shows the representative measured spectra, wherein the spectrum at 614 K is broadened compared with the reference data at 150 K. The observed broadening is due to QENS, which in this case originates from the motion of Li ions in the LGPS structure. The scattering-vector dependence of the spectral broadening was analyzed using the Chudley–Elliott model¹⁸ to extract jump length and residence time. Consequently, the jump length at the measurement temperature of 338 K was determined to be 2.4 Å, which was ascribed to the estimated distance for the jump between the Li1 and Li3b sites along the c-axis. In addition, the diffusivity for uncorrelated jumps at the measurement temperature was calculated from the determined jump length and residence time (i.e., 120 ps at 338 K). Figure 2b shows the Arrhenius plots for the obtained diffusivity, giving an activation energy of 0.13 eV; this value is in a good agreement with the OPP barrier for site-to-site jumping along the c-axis (i.e., 0.09 eV).

Figure 2.

Measurement of the microscopic ionic dynamics by the QENS (a, b) and μ⁺SR (c, d) techniques. (a) Normalized scattering spectra S(Q, E) as a function of the energy transfer ΔE at a representative spatial scale with Q = 0.33 Å⁻¹. The inset shows an amplification of the quasi-elastic region. (b) Arrhenius plots of the diffusivities measured using the QENS technique. (c) Zero-field μ⁺SR spectra recorded at 100, 200, and 250 K. (d) Relationship between the fluctuation rate (ν) and the Arrhenius temperature. The solid line represents the fit result using a thermal activation process, ν = ν₀ + A exp(−E_a/k_BT), wherein E_a = 0.09 eV. Figures (a) and (b) are adapted from Ref. 16 and partly modified with permission under CC BY 4.0, Copyright 2022, the Authors (published by American Chemical Society). Figures (c) and (d) are adapted from Ref. 19 (Copyright 2018 The Physical Society of Japan).

Additionally, the Li motion characterized by the thermal activation energy of 0.09 eV has been examined using the muon-spin relaxation (μ⁺SR) measurement,¹⁹ which is a measurement technique to detect the microscopic dynamic behavior of the ions, as is the case with the QENS measurement. Figure 2c shows the representative μ⁺SR spectra recorded for LGPS. These spectra were analyzed using the dynamic Kubo–Toyabe function to determine the field fluctuation rate (ν), which is shown in Fig. 2d as a function of the measurement temperature. Assuming that the temperature dependence is fully due to Li motion, the thermal activation energy was calculated to be 0.09 eV at 250 K. The value is in line with the lowest value observed for the OPP barrier for site-to-site jumping. These results indicate that for LGPS, the OPP barrier minima, which are derived from the static structural information, are related to the Li dynamics.

The above studies into the ion dynamics and ionic conduction mechanism in the static crystal structure indicate that Li ions diffuse preferentially in the pathways that follow the c-axis. The anisotropic ionic conduction was observed by electrochemical impedance spectroscopy (EIS) measurements on single crystals.²⁰ In the c-axis and ab-plane directions, the reported conductivities were 27 and 7 mS cm⁻¹, respectively, revealing the existence of anisotropic conduction.²⁰ However, this anisotropy is very weak compared with that of the typical two-dimensional lithium conductor Li₃N, wherein the reported ionic conductivity is 2.8 mS cm⁻¹ over the conduction plane, and only 0.01 % of that value for the direction perpendicular to the conduction plane.²¹ Furthermore, it has been reported that the activation energy of the LGPS single crystal is not anisotropic.²⁰ Similar values were obtained for the direction along the c-axis and over the ab-plane, i.e., 0.3 and 0.4 eV for temperature ranges of 253–293 and 193–243 K, respectively. These results suggest that macroscopic ionic conduction is governed not by site-to-site jumps, but by other impeding factors such as the correlation of carrier ions.¹⁰ When this situation is viewed from another perspective, one expects that macroscopic ionic conduction could be enhanced by tuning correlations between the dynamic ions via elemental substitution and control of the crystal structure.

3. Synthesis

3.1 The Li₄GeS₄–Li₃PS₄ pseudo-binary phase diagram

The phase diagram for the pseudo-binary system of the (100–k) Li₄GeS₄–k Li₃PS₄ system, which belongs to the Li₂S–GeS₂–P₂S₅ pseudo-ternary system shown in Fig. 3a, was used as a basis for the synthesis of LGPS-type solid electrolytes. The corresponding phase diagram was constructed by identifying the phases in the synthesized samples, and by clarifying their changes using high-temperature X-ray diffraction (XRD) and differential thermal analysis.²² Figure 3b shows the obtained phase diagram, wherein it can be seen that Li₄GeS₄ (β′ phase) and Li₃PS₄ (β phase) are the main crystal phases for each end composition (i.e., k = 0 or 100 mol% in the phase diagram). Both phases reportedly belong to the thio-LISICON family.²³ At one end composition of Li₃PS₄ (k = 100 mol%), the γ-Li₃PS₄ and α-Li₃PS₄ polymorphs exist, which possess respective PS₄ tetrahedral alignments different from the β phase.^24,25 In addition, LGPS (G phase) exists around the middle of each end composition, forming solid solutions in the 45 ≤ k ≤ 67 mol% range with different Ge/P ratios.

Figure 3.

Phase relationships in the Li₂S–GeS₂–P₂S₅ pseudo-ternary chemical system. (a) Ternary composition diagram for the Li₂S–GeS₂–P₂S₅ system. (b) Phase diagram for the Li₄GeS₄–Li₃PS₄ pseudo-binary system. (c) XRD patterns for the material reported as thio-LISICON (top) and a solid-solutions phase of Li₁₀GeP₂S₁₂ (bottom). Adapted from Ref. 9 with minor modifications with permission (Copyright 2020 Wiley-VCH GmbH).

The phase diagram shown in Fig. 3b allows for the understanding of a complex phase relationship. A representative XRD pattern for a solid-solution phase of LGPS is shown at the bottom in Fig. 3c, wherein all peaks were indexed by the P4₂/nmc space group. Notably, transmission electron microscopy images for these highly crystalline monophasic samples indicated the absence of amorphous phases.⁹ The top pattern of Fig. 3c presents the XRD pattern for Li_3.25Ge_0.25P_0.75S₄, which was reported to belong to a group of thio-LISICONs in the early 2000s,²³ and which exhibits the highest ionic conductivity reported for sulfide crystals at that time (i.e., 2.2 mS cm⁻¹ at RT). Based on the phase diagram shown in Fig. 3b, the reported pattern was ascribed to a mixture of the LGPS and β phases.²³ Thus, the Li_3.25Ge_0.25P_0.75S₄ phase that was reported to be a thio-LISICON in the early 2000s was determined to exist as a phase mixture. Through the construction of a phase diagram, the previous synthetic results were re-interpreted.

The phase diagram is also helpful when growing single crystals.²⁰ In previous reports,^10,20 single crystals were grown to sizes >1 mm using a self-flux method for a mixture of raw materials based on a composition of k = 75 mol% from the phase diagram. After heating the mixture to form a melt, the LGPS primary crystal was precipitated and grown from the melt.

Moreover, the phase diagram is also useful when selecting compositions and temperatures to synthesize novel LGPS-type phases via chemical substitutions. For chemical substitutions with silicon or tin, single phases with maximized conductivities, as described in the following section, have been synthesized as solid solutions by tuning the Si/P and Sn/P ratios in the Li₄SiS₄–Li₃PS₄ and Li₄SnS₄–Li₃PS₄ species, respectively.²⁶

3.2 Synthesis of LGPS derivatives for enhanced electrochemical properties

After the discovery of LGPS in 2011, the substitution of Ge with low-cost Si or Sn was examined.^27,28 Consequently, the Sn-substituted Li₁₀SnP₂S₁₂ and Si-substituted Li₁₀SiP₂S₁₂ species were reported.^27,28 On the other hand, the author synthesized a series of solid-solution phases according to the composition Li_10+δM_1+δP_2−δS₁₂ (wherein M means Si and/or Sn, and δ represents a compositional parameter relating to the molar ratio of M to P) to enhance the material conductivity. Including the grain boundary resistance at particle/particle interfaces, values of 6.7 and 5.0 mS cm⁻¹ were obtained for samples of the Si and Sn derivatives, respectively, giving higher values than those previously reported for Si and Sn phases (i.e., 2.3 and 4.0 mS cm⁻¹).^27,28

The compositions for solid-solution phases were selected based on the LGPS phase diagram, which implies that single phases can be synthesized by varying the Si/P or Sn/P ratio. Successful elemental substitution was clearly confirmed by XRD, as shown in Fig. 4a. While the diffraction peaks were indexed by the same space group as LGPS, the peaks shifted toward a lower 2θ angle, depending on the ionic radii of the constituent atoms [r(Si⁴⁺) = 0.26 Å, r(Ge⁴⁺) = 0.39 Å, r(Sn⁴⁺) = 0.55 Å].²⁹ This observation indicates that the lattice was expanded by substitution. In addition, XRD identified the monophasic region wherein LGPS-type solid solution phases could be obtained without any secondary phases (e.g., β- or β′-type phases; Li_3+yM_yP_1−yS₄; M = Si, Sn). The δ value for the single phase region was found to depend on the component M atom, i.e., 0.20 < δ < 0.43 for Si, 0 < δ < 0.5 for Ge, and −0.25 < δ < 0 for Sn. These different δ values were attributed to the sizes of the [M]S₄ tetrahedral and [Li]S₆ octahedral units in the crystal structure, because the size changes depending the δ value and probably deviates from the range required to maintain the LGPS-type structure for the δ values outside the monophasic range.

Figure 4.

(a) XRD peak shift indicating chemical substitution. (b) Composition plot for the reported LGPS-type and argyrodite-type phases, in addition to that for the prepared LSiGePSBrO. (c) Arrhenius plots for LGPS and LSiGePSBrO. Figures (b) and (c) are adapted from Ref. 12 with minor modifications.

In contrast to the original LGPS, these Si- or Sn-derived phases formed by the single substitution of an M atom did not exhibit an ionic conductivity in the order of 10 mS cm⁻¹. On the other hand, upon the selection of two M species with appropriate sizes, [M]S₄ framework units were formed. Consequently, the solid solution synthesized via a double substitution using Si and Sn gave a conductivity of 11 mS cm⁻¹.³⁰ A further conductivity improvement was achieved by double substitution with Si (replacing Ge) and Cl (replacing S), wherein Cl was selected to decrease coulombic attraction between Li ions and anions; at the composition of Li_9.54Si_1.74P_1.44S_11.7Cl_0.3³ the measured ionic conductivity reached 25 mS cm⁻¹, indicating the significant effects of introducing small amounts of halogen atoms.

Moreover, an additional conductivity enhancement was challenged based on the “high entropy design” approach,^13,31–33 wherein the introduction of an appropriate quantity of local distortion into the target crystal structure flattened the potential distribution of the Li sites and lowered the ion migration barrier.^13,33 With this guideline, the LSiGePSBrO (Li_9.54[Si_0.6Ge_0.4]_1.74P_1.44S_11.1Br_0.3O_0.6) phase showing improved ionic conductivity was synthesized.¹² Pure phase construction of the LGPS-type crystal, which possess superionic conduction pathways, is the prerequisite for improving ionic conductivity. Therefore the target chemical composition was screened out so that it meets a compositional condition for the single-phase formation, which was empirically derived based on the chemical compositions previously reported for LGPS-type phases. Under this precondition, the target composition was selected so that the configurational entropy can be higher than those for reported superionic conductors of LGPS- and argyrodite-type phases. Consequently, the composition of LSiGePSBrO located in composition plots shown in Fig. 4b was selected. In this plot, the S_mix and t indices were calculated from the compositions of the LGPS- and argyrodite-type crystal phases.¹² The S_mix index mimics a configuration entropy to measure the compositional complexity, and was calculated under the assumption that anions and cations (with the exception of the Li ions) randomly occupy each site within the structure. The t value was obtained using the molar ratios and ionic radii of the constituent anions and cations to calculate the total spherical volume for the anions and cations. This index was devised to empirically judge whether the LGPS-type crystal forms from the target compositions. It was found that the t value of LSiGePSBrO (▲)¹² fell within the range of reported LGPS-type monophasic compositions (○), while its S_mix was higher than those of LGPS- and argyrodite-type phases (□). These results indicate that LSiGePSBrO meets the requirements regarding a high complexity and maintenance of the LGPS-type structure. The high phase purity of the synthesized LSiGePSBrO sample was confirmed by XRD measurements. Figure 4c shows the bulk ionic conductivity of the monophasic sample, which was measured for a pellet sintered via the hot-press method. At RT a value of 32 mS cm⁻¹ was obtained. Thus, at temperatures ranging from 223 to 328 K the conductivity values are 2.3–3.8 times higher than those of the original LGPS phase.

Although the abovementioned materials containing Si, Ge, and Sn exhibit high ionic conductivities >5 mS cm⁻¹ at RT, they are prone to reduction by Li metal. This finding was supported by charge–discharge tests, which were carried out using all-solid-state Li battery cells prepared with these solid electrolytes as separators, and the composite of the LGPS electrolyte and a LiNbO₃-coated-LiCoO₂ active material as a common type of cathode. In such cells the magnitude of the interfacial resistance between the electrode and separator depended on the separator species more significantly at the Li anode side than at the composite-cathode side. The top panel of Fig. 5a presents the charge–discharge curves for the LGPS cell, which shows a lower specific capacity than the theorical value for a LiNbO₃-coated LiCoO₂ cathode material with a 2.55–4.25 V cutoff voltage range (i.e., ∼140 mAh g⁻¹). In addition, a large overpotential was observed. The cells prepared using LGPS-type electrolytes containing Si and/or Sn showed similar charge–discharge curves. In contrast, solid electrolytes in the Li–P–S-based systems containing halogen elements were reported to have the interfacial stability to the Li metal.^34–37 Indeed, the bottom panel of Fig. 5a shows the charge–discharge curves for the cell using LGPS-type Li₁₀P₃S₁₂Br (LPSBr),³⁸ wherein it can be seen that specific capacity was close to the theorical value, and the overpotential was reduced.

Figure 5.

Electrochemical performances of the LGPS derivatives in Li metal cells. (a) Charge–discharge curves for the all-solid-state Li metal cells incorporating the LGPS (top) and LGPS-type Li₁₀P₃S₁₂Br (bottom) solid electrolytes as separators. (b) Performance plots (left panel) for the Li cells (illustrated in right panel) using LGPS derivatives as the separator electrolytes. Portions of Fig. (b) are adapted from Ref. 38 with permission (Copyright 2022 American Chemical Society).

Figure 5b plots the cell performances recorded for the various LGPS-type separator electrolytes.³⁸ The vertical axis shows the energy efficiency calculated from the ratio of the charge/discharge energy values (corresponding to the degree of overpotential), while the horizonal axis represents the discharge capacity normalized by electrode area (including the information about the loading of the cathode active material). The performance indices plotted on the figure are averaged values calculated from the results obtained over five cycles. It can be seen that the efficiencies of the Ge-, Si-, and Sn-containing cells (plots G, H, and I) are low, even when their capacities are extremely small, thereby indicating that the interfaces between Li metal and the LGPS-type solid electrolytes containing group 14 elements are prone to reduction by Li metal. In contrast, the cells prepared using Li–P–S- or Li–P–S–O-based electrolytes exhibited higher energy efficiencies (plots D, E, and F) when the loading of the cathode active material was small.^3,39 However, for these solid electrolytes, short circuits caused by Li dendrite formation induced cell malfunction when the mass loading was increased (plot C).³⁸ In contrast, stable cycling with comparably large capacities was observed at the interface between the Li metal and the LGPS-type phase for the Li–P–S–Br and Li–P–S–I systems (plots A and B).³⁸ These phases exhibit relatively high conductivities exceeding 5 mS cm⁻¹,³⁸ indicating that the reduction of LGPS could be prevented by selecting the appropriate component atoms, whilst also maintaining a crystal structure that favors ionic conduction.

4. Application in All-solid-state Batteries

4.1 Thick cathode materials

The energy density of a battery can be increased by reducing the volumetric ratio of all components except the active materials, provided that thick cathodes are applied for all-solid-state batteries which allow for bi-polar stacking. The application of a thick cathode becomes further attractive when combined with metal anodes with a high volumetric energy density. However, a previous simulation study on thick cathode composites demonstrated that an extremely high ionic conductivity was required for the cathode-side solid electrolyte to achieve both the theorical capacity and fast charging.⁴⁰

To investigate the feasibility of all-solid-state Li-metal batteries with thick cathodes, the Li cells were cycled using the novel LSiGePSBrO solid electrolyte with a cathode thickness up to 800 µm. More specifically, for a cell containing a cathode composite that comprises LSiGePSBrO and LiNbO₃-coated LiCoO₂ in a weight ratio of 30 : 70, along with a separator composed of an LGPS-type electrolyte in a Li–P–S–Br–I system for stable cycling of the Li-metal anode, a charge-discharge test was conducted. Figure 6 shows the charge–discharge curves recorded for this cell, wherein it can be seen that at 60 °C, the cell delivered a discharge capacity of 23.5 mAh cm⁻² under a current density of 1.1 mA cm⁻², which corresponds to a 0.05 C rate. Upon increasing the current density to 11.5 mA cm⁻² (0.5 C rate), a discharge capacity of >20 mAh cm⁻² was retained. Notably, these values are superior to those reported for all-solid-state Li-metal battery cells with non-modified and simple cell designs. It is expected that the cell performance could be further improved through the use of interfacial modifications between the Li metal and the separator.

Figure 6.

Dimensions of the Li metal cell prepared using LSiGePSBrO (right), and the corresponding charge–discharge curves recorded at 60 °C (left). The upper right image is adapted from Ref. 12 with minor modifications.

4.2 Electrochemical impedance spectroscopy for an improved understanding of LGPS all-solid-state batteries

LGPS-type solid electrolytes that enable the preparation of high-performance all-solid-state battery cells serve as model materials for understanding the physico(electro)chemical phenomena occurring in batteries. Such an understanding can be achieved by studying the kinetics of the side-reactions between LGPS and the cathode active materials in all-solid-state batteries,⁴¹ the effects of post-annealing treatment of the buffer layers coated on cathode active materials to prevent side-reactions,⁴² and the mechanism of capacity decay during charge–discharge cycling.⁴³ These investigations are indispensable for achieving batteries with longer lifetimes. Since EIS measurements can be used to support such examinations, the author aimed to establish a reproducible and consistent approach for the analysis for EIS data obtained from all-solid-state battery cells based on LGPS-type superionic conductors,⁴⁴ by applying the distribution of relaxation times (DRT) analysis.⁴⁵

The cells were prepared using an In–Li anode, a cathode mixture composed of LiNbO₃-coated-LiCoO₂ and LGPS powders, and a separator comprising either of LGPS or LPSBr.³⁸ EIS was performed over a frequency range of 10 mHz to 7 MHz at a 100 % state of charge (SOC). Figure 7 shows the corresponding Nyquist plots (Figs. 7a1 and 7b1) and their transformations (Figs. 7a2 and 7b2) obtained by DRT analysis at 298 K and SOC 100 % (3.6 V vs. Li⁺/Li–In) for the cells containing the LGPS (Figs. 7a1 and 7a2) and LPSBr separators (Figs. 7b1 and 7b2). Two distorted semicircles can be seen for each cell at the low and high frequency areas of their respective Nyquist plots, indicating that at least two electrochemical processes exist in each cell. The sizes of both semicircles at 1 M > f/Hz > 1 k and at 1 k > f/Hz > 0.1 are larger for the LGPS cell than for the LPSBr cell, indicating that the former suffers from resistive physical/electrochemical processes. Subsequently, the DRT transformation (Figs. 7a2 and 7b2) was used to separate the impedance components and construct an equivalent circuit model (ECM), wherein the horizonal and vertical axes represent the measurement frequency and the polarization contribution at each frequency, respectively. In addition, the center frequency of each distinct peak corresponds to the time constant [τ = 1/(2πf)] for the respective electrochemical process, and thus in simple ECMs the number of peaks indicates the number of electrochemical processes taking place in the cell. The peak area represents the polarization contribution of each process. Thus, DRT analysis of the LGPS full cell (a2) indicates that the semicircle at the high frequency side can be divided into two impedance contributions, while that at the low frequency side cannot be separated. In contrast, only two DRT peaks are observed for the LPSBr cell (b2), indicating that the number of resistive processes is different between these two cells. Since LGPS powder is commonly used as a solid electrolyte in cathode composites, it was considered that this difference originates from the interface between the In–Li anode and the separator solid electrolyte.

Figure 7.

Nyquist plots (left side) and their transformation by distribution of relaxation time analysis (right side) for all-solid-state battery cells using LGPS (a, c, e) or LPSBr as separator (b, d, f). (a1–b2) Full cells. (c1–d2) Cathode symmetric cells. (e1–f2) Anode In–Li symmetric cells. Data are reproduced from Ref. 44.

This assumption was validated using the EIS data recorded for the cathode-composite symmetric cells (Figs. 7c1–7d2) and for the In–Li symmetric cells (Figs. 7e1–7f2). The corresponding Nyquist diagrams and the DRT spectra are shown in Figs. 7c1, 7d1, 7e1, and 7f1 and Figs. 7c2, 7d2, 7e2, and 7f2, respectively. It can be seen that the DRT spectrum for the cathode symmetric cell at 100 % SOC (Figs. 7c2 and 7d2) shows a single peak at ∼1 kHz, while the In–Li anode symmetric cell exhibits two peaks at ∼0.1 MHz and 1 Hz for LGPS separator (Fig. 7e2), but one peak at ∼10 Hz for the LPSBr separator (Fig. 7f2). These results indicate that the number and position of the DRT peaks for the symmetric cells are consistent with those of the full cells.

The above results allow interpretation of the physical/electrochemical processes detected by EIS for the LGPS and LPSBr full cells, as indicated by arrows in Figs. 7a2 and 7b2. The interphase resistance at the In–Li/separator SE was observed only for the LGPS full cell, indicating that LGPS was more prone to reduction than LPSBr. Consequently, the LGPS full cell suffers from a large impedance from the In–Li side (151 Ω for the sum of the interface and interphase resistance values evaluated by ECM analysis), which is higher than that of its LPSBr counterpart (18 Ω). Thus, application of the DRT method provides guidance for the physically meaningful and consistent quantification of impedance components for all-solid-state batteries.

5. Conclusion

This article reviews the author’s studies into the lithium superionic conductor Li₁₀GeP₂S₁₂ (LGPS) and its structural derivatives to describe the crystal structure in terms of its ion dynamics and elemental substitution, with the aim of enhancing the electrochemical properties of this material. In addition, the current understanding of electrochemical phenomena in all-solid-state batteries with LGPS-type materials is discussed based on the use of electrochemical impedance spectroscopy (EIS). The discrepancy between the low potential barrier for site-to-site jumping and the activation barrier evaluated by EIS is discussed, wherein the former was implied by the static structural information and the observation of microscopic ionic motion, while the latter was measured for a single crystal sample that was grown based on the phase diagram for the Li₄GeS₄–Li₃PS₄ pseudo-binary system. The observed discrepancy would imply that further enhancements in the ionic conduction could be possible via elemental substitution on LGPS. Indeed, the LSiGePSBrO (Li_9.54[Si_0.6Ge_0.4]_1.74P_1.44S_11.1Br_0.3O_0.6) phase obtained by multiple substitutions exhibited an ionic conductivity of 32 mS cm⁻¹ at room temperature which is approximately three-times higher than that achieved for LGPS itself. Moreover, a Li-metal battery cell prepared using a thick cathode composite composed of LSiGePSBrO and LiNbO₃-coated LiCoO₂ demonstrated a large areal capacity of 20 mAh cm⁻² and a current density of 10 mA cm⁻² at 60 °C, implying that thick-type all-solid-state Li metal batteries constitute a promising future battery configuration. The presented EIS results and distribution of relaxation times analyses aimed at separating the resistive contributions from the various electrochemical phenomena will be expected to support the development of such future batteries. The discovery of novel solid electrolytes will be important for further clarification of the superionic conduction mechanism and the improvement of battery performances.

Acknowledgments

This article encompasses a number of collaborations. In particular, the author would like to thank T. Yajima (Nagoya University) and Z. Hiroi (University of Tokyo), who carried out investigations into the single crystals. In addition, the studies using neutron beams were performed at J-PARC (Ibaraki, Japan).

The APC for the publication of this paper was supported by The Electrochemical Society of Japan.

CRediT Authorship Contribution Statement

Satoshi Hori: Investigation (Lead), Visualization (Lead), Writing – original draft (Lead), Writing – review & editing (Lead)

Conflict of Interest

The authors declare no conflict of interest in the manuscript.

Footnotes

A part of this paper has been presented in the 90th ECSJ Meeting in 2023 (Presentation #S8-1_2_04).

S. Hori: ECSJ Active Member

References

• 1)  K. Takada, Acta Mater., 61, 759 (2013).
• 2)  J. Janek and W. G. Zeier, Nat. Energy, 1, 16141 (2016).
• 3)  Y. Kato, S. Hori, T. Saito, K. Suzuki, M. Hirayama, A. Mitsui, M. Yonemura, H. Iba, and R. Kanno, Nat. Energy, 1, 16030 (2016).
• 4)  Y. G. Lee, S. Fujiki, C. Jung, N. Suzuki, N. Yashiro, R. Omoda, D.-S. Ko, T. Shiratsuchi, T. Sugimoto, S. Ryu, J. H. Ku, T. Watanabe, Y. Park, Y. Aihara, D. Im, and I. T. Han, Nat. Energy, 5, 299 (2020).
• 5)  D. H. S. Tan, Y.-T. Chen, H. Yang, W. Bao, B. Sreenarayanan, J.-M. Doux, W. Li, B. Lu, S.-Y. Ham, B. Sayahpour, J. Scharf, E. A. Wu, G. Deysher, H. E. Han, H. J. Hah, H. Jeong, J. B. Lee, Z. Chen, and Y. S. Meng, Science, 373, 1494 (2021).
• 6)  L. Ye and X. Li, Nature, 593, 218 (2021).
• 7)  A. Manthiram, X. Yu, and S. Wang, Nat. Rev. Mater., 2, 16103 (2017).
• 8)  N. Kamaya, K. Homma, Y. Yamakawa, M. Hirayama, R. Kanno, M. Yonemura, T. Kamiyama, Y. Kato, S. Hama, K. Kawamoto, and A. Mitsui, Nat. Mater., 10, 682 (2011).
• 9)  Y. Kato, S. Hori, and R. Kanno, Adv. Energy Mater., 10, 2002153 (2020).
• 10)  T. Yajima, Y. Hinuma, S. Hori, R. Iwasaki, R. Kanno, T. Ohhara, A. Nakao, K. Munakata, and Z. Hiroi, J. Mater. Chem. A, 9, 11278 (2021).
• 11)  D. Zagorac, H. Müller, S. Ruehl, J. Zagorac, and S. Rehme, J. Appl. Crystallogr., 52, 918 (2019).
• 12)  Y. Li, S. Song, H. Kim, K. Nomoto, H. Kim, X. Sun, S. Hori, K. Suzuki, N. Matsui, M. Hirayama, T. Mizoguchi, T. Saito, T. Kamiyama, and R. Kanno, Science, 381, 50 (2023).
• 13)  Y. Zeng, B. Ouyang, J. Liu, Y.-W. Byeon, Z. Cai, L. J. Miara, Y. Wang, and G. Ceder, Science, 378, 1320 (2022).
• 14)  O. Kwon, M. Hirayama, K. Suzuki, Y. Kato, T. Saito, M. Yonemura, T. Kamiyama, and R. Kanno, J. Mater. Chem. A, 3, 438 (2015).
• 15)  D. A. Weber, A. Senyshyn, K. S. Weldert, S. Wenzel, W. Zhang, R. Kaiser, S. Berendts, J. Janek, and W. G. Zeier, Chem. Mater., 28, 5905 (2016).
• 16)  S. Hori, R. Kanno, O. Kwon, Y. Kato, T. Yamada, M. Matsuura, M. Yonemura, T. Kamiyama, K. Shibata, and Y. Kawakita, J. Phys. Chem. C, 126, 9518 (2022).
• 17)  P. Heitjans and S. Indris, J. Phys.: Condens. Matter, 15, R1257 (2003).
• 18)  C. T. Chudley and R. J. Elliott, Proc. Phys. Soc., 77, 353 (1961).
• 19)  J. Sugiyama, H. Nozaki, I. Umegaki, Y. Higuchi, S. P. Cottrell, S. Hori, R. Kanno, and M. Månsson, JPS Conf. Proc., 21, 011016 (2018).
• 20)  R. Iwasaki, S. Hori, R. Kanno, T. Yajima, D. Hirai, Y. Kato, and Z. Hiroi, Chem. Mater., 31, 3694 (2019).
• 21)  T. Lapp, S. Skaarup, and A. Hooper, Solid State Ionics, 11, 97 (1983).
• 22)  S. Hori, M. Kato, K. Suzuki, M. Hirayama, Y. Kato, and R. Kanno, J. Am. Ceram. Soc., 98, 3352 (2015).
• 23)  R. Kanno and M. Murayama, J. Electrochem. Soc., 148, A742 (2001).
• 24)  K. Homma, M. Yonemura, T. Kobayashi, M. Nagao, M. Hirayama, and R. Kanno, Solid State Ionics, 182, 53 (2011).
• 25)  K. Homma, M. Yonemura, M. Nagao, M. Hirayama, and R. Kanno, J. Phys. Soc. Jpn., 79, 90 (2010).
• 26)  S. Hori, K. Suzuki, M. Hirayama, Y. Kato, T. Saito, M. Yonemura, and R. Kanno, Faraday Discuss., 176, 83 (2014).
• 27)  P. Bron, S. Johansson, K. Zick, J. Schmedt auf der Günne, S. Dehnen, and B. Roling, J. Am. Chem. Soc., 135, 15694 (2013).
• 28)  J. M. Whiteley, J. H. Woo, E. Hu, K.-W. Nam, and S.-H. Lee, J. Electrochem. Soc., 161, A1812 (2014).
• 29)  R. D. Shannon, Acta Crystallogr., Sect. A, 32, 751 (1976).
• 30)  Y. Sun, K. Suzuki, S. Hori, M. Hirayama, and R. Kanno, Chem. Mater., 29, 5858 (2017).
• 31)  M. Tatsumisago, N. Machida, and T. Minami, J. Ceram. Soc. Jpn., 95, 197 (1987).
• 32)  D. Bérardan, S. Franger, A. K. Meena, and N. Dragoe, J. Mater. Chem. A, 4, 9536 (2016).
• 33)  Y. Deng, C. Eames, B. Fleutot, R. David, J.-N. Chotard, E. Suard, C. Masquelier, and M. S. Islam, ACS Appl. Mater. Interfaces, 9, 7050 (2017).
• 34)  Y. Wang, D. Lu, J. Xiao, Y. He, G. J. Harvey, C. Wang, J.-G. Zhang, and J. Liu, Energy Storage Mater., 19, 80 (2019).
• 35)  A. D. Bui, S.-H. Choi, H. Choi, Y.-J. Lee, C.-H. Doh, J.-W. Park, B. G. Kim, W.-J. Lee, S.-M. Lee, and Y.-C. Ha, ACS Appl. Energy Mater., 4, 1 (2021).
• 36)  S. Yang, M. Takahashi, K. Yamamoto, K. Ohara, T. Watanabe, T. Uchiyama, T. Takami, A. Sakuda, A. Hayashi, M. Tatsumisago, and Y. Uchimoto, Solid State Ionics, 377, 115869 (2022).
• 37)  M. Takahashi, T. Watanabe, K. Yamamoto, K. Ohara, A. Sakuda, T. Kimura, S. Yang, K. Nakanishi, T. Uchiyama, M. Kimura, A. Hayashi, M. Tatsumisago, and Y. Uchimoto, Chem. Mater., 33, 4907 (2021).
• 38)  S. Song, S. Hori, Y. Li, K. Suzuki, N. Matsui, M. Hirayama, T. Saito, T. Kamiyama, and R. Kanno, Chem. Mater., 34, 8237 (2022).
• 39)  M. Xu, S. Song, S. Daikuhara, N. Matsui, S. Hori, K. Suzuki, M. Hirayama, S. Shiotani, S. Nakanishi, M. Yonemura, T. Saito, T. Kamiyama, and R. Kanno, Inorg. Chem., 61, 52 (2022).
• 40)  A. Bielefeld, D. A. Weber, and J. Janek, ACS Appl. Mater. Interfaces, 12, 12821 (2020).
• 41)  T. T. Zuo, R. Rueß, R. Pan, F. Walther, M. Rohnke, S. Hori, R. Kanno, D. Schröder, and J. Janek, Nat. Commun., 12, 6669 (2021).
• 42)  X. Sun, S. Hori, Y. Li, Y. Yamada, K. Suzuki, M. Hirayama, and R. Kanno, J. Mater. Chem. A, 9, 4117 (2021).
• 43)  X. Sun, Y. Yamada, S. Hori, Y. Li, K. Suzuki, M. Hirayama, and R. Kanno, J. Mater. Chem. A, 9, 17905 (2021).
• 44)  S. Hori, R. Kanno, X. Sun, S. Song, M. Hirayama, B. Hauck, M. Dippon, S. Dierickx, and E. Ivers-Tiffée, J. Power Sources, 556, 232450 (2023).
• 45)  S. Dierickx, A. Weber, and E. Ivers-Tiffée, Electrochim. Acta, 355, 136764 (2020).

Biographies

Satoshi Hori (Specially Appointed Associate Professor, Institute of Innovative Research, Tokyo Institute of Technology)

Satoshi Hori received his B.S. degree from the University of Tokyo and Ph.D. degree from Tokyo Institute of Technology (2016). He continued his study on synthesis and crystal structure analyzes of lithium conductive sulfides as a JSPS research fellow and then as a project assistant professor. His research topic is synthesis of new materials for all-solid-state batteries.