2024 Volume 92 Issue 4 Pages 047002
The sulfide solid electrolyte Li4SnS4 has gained attention owing to its high moisture durability. In this study, we quantitatively investigated the changes in the electrochemical properties and chemical/physical states of Li4SnS4 resulted from moisture exposure using the XRD, Raman spectroscopy, and high-frequency electrochemical impedance spectroscopy (HF-EIS). Li4SnS4 was subjected to Ar gas flow at a dew point ranging from −20 °C to 0 °C for 1 h, and sulfide hydrolysis generated only a minute amount of H2S. The XRD patterns and Raman spectra revealed the formation of Li4SnS4·4H2O with increasing dew point. The HF-EIS analysis, which was conducted to clarify the spatial distribution of the hydrate within the particle, revealed a significant decrease in the ionic conductivity of Li4SnS4; this result can be attributed to the increased grain-boundary (SE/SE particle contact) resistance due to the formation of Li4SnS4·4H2O at the particle surface, despite the generation of a minute amount of H2S. By combining these multifaceted analytical methods, we demonstrated that the thermodynamically stable surface hydrate Li4SnS4·4H2O reduced the lithium-ion conductivity without H2S generation owing to the hydrolysis of sulfide. Thus, we chemically, spatially, and quantitatively verified the mechanism underlying the observed decrease in the ionic conductivity.
A sulfide solid electrolyte (SE), as a lithium ionic conductor, exhibits excellent properties, including a high lithium-ion conductivity, thermal stability, and plasticity,1–4 which are suitable for implementing all-solid-state batteries (ASSBs). ASSBs are the next-generation energy storage alternatives to conventional lithium-ion batteries, which use organic liquid electrolytes. Recently, Kanno et al. investigated the operation of an ASSB with an extremely high capacity, utilizing millimeter-thick electrodes and a sulfide SE composed of Li–Si–P–S–Br–O.5 This result further heightens expectations for practical applications of ASSBs using sulfide SEs.
However, sulfide SEs exhibit an inherent disadvantage—they produce toxic H2S gas when exposed to humidity,6–10 and this H2S gas production hinders their application in ASSBs. Therefore, numerous research groups have focused on developing more stable materials.11–15 The argyrodite-structured Li6PS5Cl is known to produce a significant amount of H2S gas,9,10 and thus, some strategies, such as adjusting the atomic composition,7,11 replacing S with other elements,12 and applying nano-coatings to the particle surface,15,16 have been proposed to suppress the formation of H2S gas in Li6PS5Cl. In addition to these material modifications, some recent studies introduced Li4SnS4 (LSS), which is known for its inherently high moisture durability.17–23 Hexagonal LSS synthesized via mechanical milling produces an exceptionally small quantity of H2S gas (∼0.2 cm3 g−1) when exposed to air with temperatures in the range of 20–22 °C and relative humidity of 70 % for 40 min.17 Additionally, Kimura et al.18 reported the overall picture that the hydration and dehydration of LSS, which suggested that thermodynamically stable hydrated state of Li4SnS4·4H2O suppresses the generation of H2S gas. Therefore, taking advantage of its stability against H2O, LSS can be synthesized synthetically through an aqueous-solution processes.21–23 Matsuda et al. also proposed an aqueous-phase synthesis method using an ion-exchange membrane,23 with expectations for its industrial development.
In the present study, we focused on the quantitative analysis of the variations in the electrochemical properties and chemical/physical states of LSS exposed to an inert Ar gas mixed with H2O as various dew points (indicating moisture contents). To verify the changes in the electrochemical properties and chemical/physical states associated with moisture-exposure, we conducted various measurements using XRD, Raman spectroscopy, and high-frequency electrochemical impedance spectroscopy (HF-EIS). We examined the mechanisms underlying the changes observed in LSS exposed to a small amount of moisture, dry environment close to simulating actual battery manufacturing, based on the results of the above analyses.
LSS powder was synthesized using a previously reported procedure.21,22 Li2S, S, and Sn were introduced into ultrapure H2O at a molar ratio of 2 : 2 : 1 without exposure to air. The mixture was dissolved while stirring at 80 °C for more than 12 h and then dried at 120 °C in a vacuum for 3 h to obtain a hexagonal LSS powder with an average particle size of approximately 1 µm (Fig. 1). Controlled moisture exposure of LSS was performed using an apparatus constructed based on the method reported by Yamada et al.24 An Ar gas cylinder (<0.5 ppm H2O, Grade 1, Taiyo Nippon Sanso Co., Japan) was used as the upstream gas source. The dew points of the gas flow during the moisture exposure test were −20 and 0 °C, and the corresponding H2O contents were approximately 1000 and 6000 ppm, respectively. In the experiment, 200 mg of the LSS powder was exposed to the moisture-controlled Ar gas at a flow rate of 0.8 L min−1 for 1 h. An H2S sensor (Model RS3000, Advanced Micro Instruments, USA) was connected to the gas line after moisture exposure of the LSS sample to monitor the amount of generated H2S gas. The total amount of generated H2S was calculated per 1 g of SE based on the recorded concentration values. The data for a dew point of 0 °C were obtained from a previous report.
SEM images were obtained to investigate the particle size of synthesized LSS with an electron gun at an accelerating voltage of 5 keV in secondary electron collecting mode.
The particle size of the synthesized LSS was determined using a SEM apparatus (S-4800, Hitachi Hi-Tech, Japan) with an air-tight sample holder. The particles were dispersed and fixed onto a conductive carbon tape. SEM images were obtained using an electron gun at an accelerating voltage of 5 keV in secondary electron collecting mode.
2.3 XRDXRD measurements were conducted using a reflective configuration system (Empyrean, Malvern PANalytical, UK) with an air-tight holder. The diffraction patterns were obtained for a 2θ range of 10°–80° with a step width of 0.1° using CuKα X-rays. Rietveld analysis of the acquired diffraction patterns was conducted based on the reported crystallographic information files of hexagonal LSS17 and Li4SnS4·4H2O.18 General Structure and Analysis Software II (GSAS-II) software package25 was used for data processing.
2.4 Raman spectroscopyRaman spectroscopy was carried out using the RAMANforce (Nanophoton, Japan) system, equipped with an incident green laser with a wavelength of 532 nm. The laser was directed through a quartz glass window in an air-tight sample holder. The laser output was reduced to 10 µW µm−2, and the laser exposure time was set to 10 s to prevent damage. The Raman spectra were calibrated using the fixed Si wafer peak at 520.6 cm−1 and normalized to the peak corresponding to the Sn–S vibration of the SnS4 unit (located at ∼350 cm−1).17
2.5 Impedance analysisEIS measurements were conducted within a high-frequency range of up to 120 MHz using an integrated commercial system with a temperature control unit (4990EDMS-120K, LakeShore 33x, TOYOTech, Japan). First, 80 mg of the LSS powder samples (before and after exposure) were pelletized at 360 MPa in a zirconia cylinder (diameter: 10 mm) between two stainless steel (SUS) pistons. The pellet was subsequently extracted and loaded into the measurement cell. The EIS measurements were conducted over a frequency range from 100 MHz to 20 Hz at a setting temperature of 233.15–298.15 K. The normalized impedance value Z (unit = Ω cm) was calculated as Z = (Zmeasure × S)/d, where S is the area of the pellet (10 mmφ diameter), and d is the thickness (usually ∼0.55 ± 0.02 mm). For convenience, we conducted a simple fitting analysis using the equivalent circuit depicted in Fig. 2, in full accordance with previously reported methodologies.26,27 All the sampling procedures for the EIS measurements were performed in an Ar-filled glove box (dew point < −80 °C).
Equivalent circuit used for the EIS spectral fitting analysis.
Figure 3 shows the amount of H2S gas generated in cm3 volume per 1 g of LSS at different dew points. A low amount of H2S gas, approximately 0.01 cm3 g−1, was detected at a dew point of −20 °C, and a maximum amount of 0.17 cm3 g−1 of H2S gas was generated at a dew point of 0 °C. The maximum H2S gas amount of 0.17 cm3 g−1 at a dew point of 0 °C accounts for only 0.05 mol% of the S content in LSS, indicating that the stoichiometric composition remains nearly unchanged. This result indicates that LSS exhibits extremely high moisture durability. Moreover, under similar moisture conditions, the amount of H2S gas released by the argyrodite-structured sulfide SE Li6PS5Cl, which is a typical sulfide SE, is more than 100 times that released by LSS in this study.8,9,28,29 Recently, Scharmann et al.28 systematically investigated the H2S generation amount of argyrodite-structured Li6PS5Cl for various dew points, whose results implies challenging issue for implementation of all-solid-state batteries using sulfide SEs.
Amount of H2S gas generated converted to per 1 g of LSS at dew points of −20 °C (dashed line, blue) and 0 °C (solid line, red) in Ar flow.
To clarify the chemical state of LSS after moisture exposure, we analyzed the exposed samples using XRD and Raman spectroscopy. Figure 4 shows the XRD patterns of LSS before (pristine) and after moisture exposure. The pattern of the pristine LSS sample indicates that it has a hexagonal crystal structure.17 No changes are observed in the XRD pattern after exposure up to a dew point of −20 °C. However, a noticeable change is evident after exposure at a dew point of 0 °C. The prominent peak at 14.9° with several subpeaks align with the pattern of the hydrate Li4SnS4·4H2O.18 The quantitative XRD analysis results22 show that the molar ratio of Li4SnS4·4H2O in the pristine sample is <0.2 % and those in the moisture-exposed samples at dew points of −20 and 0 °C are 0.4 %, and 14.1 %, respectively. Therefore, exposure to moisture ultimately leads to hydration of LSS. Nevertheless, in the XRD measurement, which is a bulk analysis, no distinct hydrates could be discerned in the samples exposed to moisture at low dew points. It is generally known that the analysis depth of Raman spectroscopy measurements using a green laser of 532 nm is approximately several hundred nanometers,30 although strictly speaking it depends on the reflectance and absorption characteristics of samples. Yamamoto et al.31 conducted a depth analysis of the reaction process in SE synthesis using XRD and Raman spectroscopy. Figure 5a shows the spectra of the Raman shift region attributed to Sn–S vibrations. Referring to the previously reported,18 Li4SnS4·4H2O has a sharper and lower Raman-shift peak than SnS44− of pristine LSS. Therefore, the peak at ∼350 cm−1 corresponds to the SnS44− unit of pristine LSS and that at ∼340 cm−1 is ascribed to the hydrate Li4SnS4·4H2O. Supplementarily, a sub-peak at ∼240 cm−1 derived from Li4SnS4·4H2O18 was also observed in the Raman spectra of the samples exposed to moisture at a dew point of 0 °C. The peaks clearly show the presence of hydrate, which is consistent with the XRD result. The subtraction spectra of the pristine sample are shown in Fig. 5b, which show the variations in the samples exposed to different dew points. Notably, a small amount of hydrate is formed even at a dew point of −20 °C. The XRD and Raman spectroscopy results suggest that the hydration reaction of LSS may commence at the particle surface. Furthermore, these findings indicate that elevating the dew point (corresponding to an increase in moisture content) promotes the hydration reaction.
Diffraction pattern analysis; (a) pristine (black), moisture-exposed LSS at dew points of (b) −20 °C (blue), and (c) 0 °C (red). According to the results of the quantitative XRD analysis, the molar ratio of Li4SnS4·4H2O in the pristine sample is <0.3 % and those in the moisture-exposed samples at dew points of −20 and 0 °C are 0.4 %, and 14.1 %, respectively. The vertical bars at the bottom correspond to hexagonal LSS and Li4SnS4·4H2O crystal structures.17,18
(a) Raman spectra in the Sn–S vibration region of pristine and moisture-exposed samples at dew points of −20 and 0 °C.18 (b) Subtraction spectra for comparing the moisture-exposed samples at dew points of −20 and 0 °C with the pristine samples. The dotted lines correspond to the SnS44− unit and Li4SnS4·4H2O.18
To verify that the hydration reaction is initiated from the particle surface and to assess its impact on lithium-ion conductivity, depth-resolved methods that can effectively distinguish and characterize both the surface and internal bulk of the particles are employed. In this study, we adopted the HF-EIS26,27,32–34 method for this analysis. The HF-EIS method can separate the lithium-ion conduction between the particles, that is, at the grain boundaries (SE/SE particle contact), from that within the particle bulk in the macroscopic overall conduction of the SE based on the differences in the time constants. In other words, it has an indirect depth resolution, which enables tracking of the impedance changes both at the particle surface and within the particle. Because of the significantly small time constant of the internal bulk impedance component, impedance measurements at frequencies higher than those typically employed in general-purpose measurements are required.8,26,32–34 Previous studies also successfully separated and analyzed the grain boundary and internal bulk impedances of exposed sulfide SE particles.26,32 For example, Tian et al. reported the separation of the physical structure (grain boundary and internal bulk) and impedance factors of sulfide SEs by combining HF-EIS measurements and the distribution of relaxation times (DRT) analysis method, providing valuable insights for industrial production.32 In addition, Kuwata et al. precisely investigated lithium-ion diffusion in Li0.29La0.57TiO3 and visualized that a remarkable local concentration gradient of lithium-ion was formed at grain boundary where primary particles contacted,35 which would support that the equivalent circuit has a RC component corresponding to the grain boundary as shown in Fig. 2. Figures 6a and 6b show the EIS data measured at 298 and 253 K in the form of Nyquist and Bode plots. Evidently, the total impedance at a high dew point increases, and the rate of increase is high at low temperatures. The spectra reveal that the two overlapped components of the pristine sample (component 1: a high-frequency semicircular arc; component 2: a low-frequency semicircular arc) separate as the dew point increases during moisture exposure. While component 2 shifts toward the lower frequency side owing to the high dew point during moisture exposure, the response frequency of component 1, as indicated by the dotted lines in Figs. 6a and 6b, remains unchanged for all the samples. The two components are then fitted using the equivalent circuit depicted in Fig. 2, and the fitted capacitance and resistance values are shown in Figs. 6c and 6d. Interestingly, the resistance of component 1 changes only negligibly, whereas that of component 2 significantly increases. Based on a comparison of the capacitance values,26,34 component 1 in the high-frequency region of ∼10−11 F can be assigned to the lithium-ion conduction of the internal bulk, whereas component 2 in the low-frequency region of ∼10−10 F is ascribed to the grain boundary (SE/SE particle contact). The spectroscopic impedance analysis reveals that the observed increase in the resistance, which is equal to a decrease in the lithium-ion conductivity, with an increase in the dew point can be attributed to the grain boundaries. Table 1 presents a summary of the relationship between the amount of H2S gas generated and total lithium-ion conductivity. The impedance spectroscopic analysis reveals a significant decrease in the lithium-ion conductivity, which can be attributed to the increase in the grain-boundary (SE/SE particle contact) resistance, despite the minute amount of H2S gas generated, as described before.
Nyquist and Bode plots of the EIS data at (a) 298 K and (b) 253 K; pristine (black circle), moisture-exposed at dew points of −20 °C (blue square) and 0 °C (red triangle). The solid lines are the fitting results for the EIS data. (c) Capacitance value at 298 K for each sample obtained by fitting. (d) Arrhenius plot of the impedance components 1 (opened marks) and 2 (closed marks), along with the activation energy derived from linear approximation (dotted and dashed lines).
| Pristine | Dew point −20 °C | Dew point 0 °C | |
|---|---|---|---|
| Total amount of H2S gas (cm3 g−1) | — | 0.01 | 0.17 |
| Lithium-ion conductivity (mS cm−1) | 0.19 | 0.075 | 0.026 |
| [Retention rate] | [100 %] | [39 %] | [14 %] |
Furthermore, the activation energy for lithium-ion conduction is calculated based on the temperature dependence of the impedance components 1 and 2, and the results are shown in Fig. 7. The activation energies for lithium-ion conduction in the inner bulk and grain boundary of the pristine sample are 42.8 and 44.6 kJ mol−1, respectively. However, in the moisture-exposed samples, the activation energy for lithium-ion conduction in the grain boundary significantly increases to ∼53 kJ mol−1, whereas that of the inner bulk remains unchanged. This observed trend is consistent with the observed increase only in the grain-boundary resistance. The XRD and Raman spectroscopy results collectively indicate that remarkable increase in the resistance and activation energy at the grain boundary is attributed to the formation of the hydrate Li4SnS4·4H2O only on the surface of the SE particle. Indeed, a previous study reported that this hydrate exhibits a lower lithium-ion conductivity and higher activation energy compared with those of LSS.18 Moreover, the increase in the dew point from −20 °C to 0 °C does not affect the activation energy, which remains constant (Fig. 7); only the grain-boundary resistance increases with the increasing dew point (Fig. 6d), suggesting that the thickness of Li4SnS4·4H2O on the surface increases with increasing dew point of exposure. This result is also consistent with the XRD and Raman spectroscopy results. We mention here that various technical limitations exist the measurable resistance range of the measurement device, the thickness of the sample pellet, and pellet shrinkage at extremely low temperatures, etc.
Activation energy of 1/Rx for components 1 (opened bar) and 2 (closed bar) calculated from the linear-approximation slope.
The correlation between the hydration process of moisture-exposed LSS and lithium-ion conductivity was summarized as follows (Fig. 8), where the thickness of the surface hydrate layer is geometrically calculated based on the results of qualitative XRD analysis, assuming LSS sphere particle with diameter of 1 µm:
Schematic illustrations of the LSS hydration processes observed in this study; (i) initial state, (ii) lower dew point (e.g. dew point of −20 °C), and (iii) higher dew point (e.g. dew point of 0 °C).
The lithium-ion conductivity of the hydrate Li4SnS4·4H2O is low (10−9 S cm−1).18 Therefore, the formation of a surface layer of the thermodynamically stable hydrate Li4SnS4·4H2O results in a significant decrease in the lithium-ion conductivity, without H2S gas generation owing to sulfide hydrolysis. Further investigations are required to elucidate the mechanism by which the hydration reaction erodes into the bulk as well as its erosion rate. In this study, we succeeded in indirectly revealing the chemical and spatial states of moisture-exposed LSS by combining XRD, Raman spectroscopy, and HF-EIS measurement results. These facts cannot be easily and accurately analyzed using general vacuum-based surface evaluation methods owing to the occurrence of dehydration. The proposed methodology and insights obtained from the findings of this study will promote the improvement and enhancement of sulfide SEs in the future.
In this study, we investigated the relationship between lithium-ion conductivity and the chemical state of LSS, which is a sulfide SE, with minimal H2S gas generation from side reactions with H2O (compared to other typical materials) upon moisture exposure at dew points of −20 and 0 °C in Ar gas flow. The moisture-exposed powder was analyzed using XRD, Raman spectroscopy, and HF-EIS. The major conclusions drawn from the results of these analyses are as follows:
We succeeded in revealing the chemical and spatial states of moisture-exposed LSS, which is difficult to analyze using general surface analysis methods in vacuum owing to dehydration. We also verified the mechanism corresponding to the decrease in the lithium-ion conductivity. More direct measurements will be required in the future to obtain deeper insights into these phenomena.
Yusuke Morino: Conceptualization (Lead), Data curation (Lead), Formal analysis (Lead), Investigation (Lead), Methodology (Lead), Validation (Lead), Visualization (Lead), Writing – original draft (Lead), Writing – review & editing (Lead)
Misae Otoyama: Formal analysis (Supporting), Supervision (Supporting), Writing – review & editing (Supporting)
Toyoki Okumura: Formal analysis (Supporting), Supervision (Supporting), Writing – review & editing (Supporting)
Kentaro Kuratani: Formal analysis (Supporting), Supervision (Supporting), Writing – review & editing (Supporting)
Naoya Shibata: Data curation (Supporting), Formal analysis (Supporting), Investigation (Supporting), Writing – review & editing (Supporting)
Daisuke Ito: Formal analysis (Supporting), Investigation (Supporting), Supervision (Supporting), Writing – review & editing (Supporting)
Hikaru Sano: Conceptualization (Lead), Data curation (Lead), Formal analysis (Lead), Investigation (Lead), Methodology (Lead), Validation (Lead), Visualization (Lead), Writing – original draft (Lead), Writing – review & editing (Lead)
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Y. Morino, M. Otoyama, T. Okumura, K. Kuratani, N. Shibata, D. Ito, and H. Sano: ECSJ Active Members
