2022 年 90 巻 4 号 p. 047003
All-solid-state batteries (ASSBs) using sulfide-based solid electrolytes (SEs) are promising energy storage devices beyond the present liquid-type lithium-ion batteries (LIBs) using organic solvents, which are expected to realize the adaptation of new systems such as higher-voltage cathodes. However, in recent years, undesirable side reactions are being reported in the nanometer-order region at the interface of cathode active materials/SEs. Therefore, we evaluated the cycle durability of the all-solid-state cathode half-cells using an argyrodite-structured sulfide-based solid electrolyte at various potentials at 60 °C and then measured the coulombic behavior by high precision coulometry. Furthermore, the interfaces of the cathode active material/SE were observed using secondary electron microscopy (SEM), transmission electron spectroscopy (TEM), and electron diffraction (ED). In conclusion, a strong correlation was found between the coulombic behavior and material decomposition at the interface.
Recently, all-solid-state batteries (ASSBs) using solid electrolytes (SEs) have been studied because of their potential for higher energy densities1,2 and safety3 compared to general liquid-type lithium-ion batteries (LIBs). There are various types of SEs, polymers,4 and molecular crystals,5 organics, such as halides, and oxides, and inorganics such as sulfides.1 Among them, sulfide-based SEs have several advantages, including high ionic conductivity (>10 mS cm−1)6,7 and plasticity.8,9 These excellent properties are expected to lead to their rapid commercialization, which will enhance battery performance and fabrication process suitability. In addition, the solidification of batteries using sulfide-based SEs has expanded the operating temperature range.2,6,7,10 We have previously reported on good performances by an argyrodite-structured sulfide SE/LiNbO3-coated LiNi0.5Co0.2Mn0.3O2 (NCM523) at a low temperature of −60 °C that partially freezes ordinary organic solvent electrolyte10 and at a high temperature of 60 °C that pose major problems such as battery-package swelling because of the gas generation by decomposition side reactions of the organic solvent electrolyte.2 However, the decomposition reactions gradually proceed at the interface of sulfide SE/NCM in conditions such as a high temperature environment, high cathode potential, and after a long-term charge/discharge cycle testing.2,11,12 Elucidating the not-fully-understood decomposition side reactions in more detail will be an important finding for further material development and improvement of battery performance of ASSBs using sulfide SE in the future.
In this study, we aimed to evaluate the charge/discharge cycle durability of the all-solid-state half-cells using an argyrodite-structured sulfide-based SE and LiNbO3-coated NCM cathode active material at various potentials at 60 °C and to measure the coulombic behavior using high precision coulometry (HPC). HPC measurement is one of the electrochemical and non-destructive evaluations first reported by Dahn et al., which is a method to analyze undesirable side reactions inside the cell by accurately measuring coulombic efficiency (CE) and the charge and discharge capacity endpoint slippages per cycle (Δc and Δd, described in detail later in the chapter).13–16 Later, Yamazaki et al. also reported on a simple equipment configuration for HPC measurement.17 Furthermore, we proposed to observe the cycled interfaces of sulfide SE/LiNbO3-coated NCM using secondary electron microscopy (SEM), transmission electron spectroscopy (TEM), and electron diffraction (ED) transfer without exposure to air.
Half-cells were fabricated as described in our previous report.2,10 Briefly, LiNi0.5Co0.2Mn0.3O2 (NCM523, Sumitomo Metal Mining) was used as the cathode active material to form agglomerated particles.2 The LiNbO3 coating on the NCM523 particles was prepared with Li–Nb double ethoxide in anhydrous ethanol solution using a rolling fluidized coating machine (MP-01, Powrex).2,10,18,19 The average thickness of the coating was approximately 2–10 nm (observed under secondary electron microscope2 and transmission electron microscope10). An argyrodite-structured Li(7−x)PS(6−x)Clx (x ≈ 1, 2 × 10−3 S cm−1, Mitsui Mining & Smelting) was used as the sulfide-based SE. The pellet-type half-cell consisted of three layers with confining pressure of 1 t cm−2: (1) cathode mixture layer with LiNbO3-coated NCM523 and SE at 50 : 50 (v/v) as a working electrode with press pressure of 6 t cm−2; (2) an SE separator layer of ∼600 µm with press pressure of 6 t cm−2; and (3) an ∼600 µm In–Li alloy as the counter electrode with press pressure of 1 t cm−2. The In–Li alloy counter electrode exhibited a flat potential plateau (0.62 V vs. Li/Li+). The cathode mixture layer was designed with a cell capacity of 2 mAh (defined as 1 C-rate) based on the theoretical capacity of 160 mAh g(NMC532)−1 at an upper charging potential limit of 4.25 V vs. Li/Li+. The capacity of the In/In–Li phase was designed to be ∼30 times larger than the cathode capacity so that the lithium supply source would not be depleted due to side reactions at the cathode. All fabrication processes were performed in an Ar-filled glove box (dew point less than −80 °C, oxygen concentration <1 ppm).
HPC cycle evaluation was performed using an ultrahigh precision coulometry system (UHPC-2A, Novonix) with a thermostatic chamber (SU-242, Espec). The half-cells were placed for 3 hours in a thermostatic chamber at 60 °C to make the temperature inside the cells uniform. Subsequently, the charge/discharge cycle tests were started. In this study, the upper charging potential limits were set to 4.25, 4.35, 4.45, 4.55, and 4.65 V vs. Li/Li+. The cells were charged to each limit vs. Li/Li+ at a constant current (CC) of 600 µA (0.3 C-rate). Then, the cells were discharged to 3.0 V vs. Li/Li+ at a 0.3 C-rate CC to maintain a constant potential (CP) of 3.0 V vs. Li/Li+ with a 0.1 C-rate cutoff. The rest period for 30 min was set between each charge and discharge process. The charge/discharge processes for each upper charging potential limit were cycled over 25 cycles. Figure 1 shows a schematic of each parameter of coulombic efficiency (CE) and the charge and discharge capacity endpoint slippages per cycle (Δc and Δd, respectively). We evaluated the HPC parameters as coulombic behavior. Each parameter refers to the following equations based on the reports by Dahn et al. for liquid-type LIBs:14
| \begin{equation} \mathrm{CE} = (\text{discharge capacity})/(\text{charge capacity}) \end{equation} | (1) |
| \begin{equation} \varDelta_{\text{c}} = Q_{\text{ox}} - 2Q_{\text{c}} \end{equation} | (2) |
| \begin{equation} \varDelta_{\text{d}} = 2Q_{\text{Li}} - Q_{\text{ox}} = [1 - (\mathrm{CE})] Q_{0} \end{equation} | (3) |
| \begin{equation} \text{Capacity fade per cycle} = \varDelta_{\text{d}} - \varDelta_{\text{c}} = 2Q_{\text{Li}} - 2Q_{\text{ox}} + 2Q_{\text{c}}, \end{equation} | (4) |
Schematic image of each parameter of coulombic efficiency (CE) and the charge and discharge capacity endpoint slippages per cycle (Δc and Δd, respectively).
After all cycle tests, we collected the composites of SE and LiNbO3-coated NCM523 from the cells at 3.95 V vs. Li/Li+. Bright-field (BF) images and electron diffraction (ED) patterns at the interface regions were observed using a field-emission transmission electron microscope (JEM-2100F, JEOL) equipped with a 14-bit CCD camera. To prevent the exposure of the samples to air, a vacuum transfer holder (Gatan, model 648) was used. The microstructure of the NCM particle surface was examined using high-resolution TEM imaging. The ED patterns were analyzed with a computer program, “ProcessDiffraction”20 and the simulated X-ray diffraction patterns from inorganic crystal structure database (ICSD). By this observation method, the changes in the crystal structure of the surface and interface can be followed on a nanometer scale.21–23 Additionally, cross-sectioned SEM images (JSM-7200F and IB-19520CCP, JEOL) were obtained for some samples.
The charge/discharge curves and discharge capacity retentions for 10 cycles at 4.25–4.65 V vs. Li/Li+ are shown in Fig. 2. All charge/discharge curves show that the capacity shifts as the number of cycles increases and the shift directions depend on the cathode potential. For example, in the 4.25 V vs. Li/Li+ cycles, the charge and discharge endpoints gradually shifted in the positive direction. Moreover, the shifts of the charge and discharge endpoints were in the same direction up to 4.45 V vs. Li/Li+ cycles. In contrast, in the 4.55 V and 4.65 V vs. Li/Li+ cycles, the discharge endpoints shifted in the positive direction like that in the 4.25 V vs. Li/Li+ cycles, but the charge endpoints shifted in the negative direction. This shift direction reversal of the charge endpoint occurred at a threshold potential between 4.45 V and 4.55 V vs. Li/Li+. Next, we analyzed the discharge capacity at cathode potential of 4.25–4.65 V vs. Li/Li+ (Fig. 3). In this study, we performed CC and CP discharge to separate the apparent capacity attenuation due to the increase in resistance and the attenuation due to degradation of the active material itself from the capacity retention at each potential. That is, the CC + CP total discharge capacity decreases if the active material itself degrades, and the decrease in the discharge capacity of CC concerning CC + CP is assumed to indicate an increase in resistance. We previously reported that the interfacial resistance increased due to the oxidative decomposition of the SE and the CC discharge capacity decreased even when the degradation of the NCM cathode active material itself was not observed.2 Here, we evaluated the resistance transition by this simple method, although the analyses of the current density dependency of the capacity ratio of CC/(CC + CP), and electrochemical impedance spectroscopy (EIS) could provide deeper insights. In the 4.25 V vs. Li/Li+ cycles, the CC and CC + CP discharge capacity retentions were extremely flat and nearly 100 % at 10 cycles, which means that the cell performance had hardly deteriorated, that is, neither the resistance increase nor the degradation of the NCM active material had occurred, even in a high temperature environment of 60 °C. In the 4.35 V vs. Li/Li+ cycles, the capacity retention at 10 cycles decreased slightly to 99.01 %, but no difference between the CC and CC + CP capacity retentions was observed like the 4.25 V vs. Li/Li+ cycles. In the cycles at potentials higher than 4.45 V vs. Li/Li+, the higher the cathode potential, the lower the capacity retention, and the wider the difference between the CC and CC + CP capacity retention for each potential. In particular, at 4.55 V and 4.65 V vs. Li/Li+, the difference was significant and was assumed to differ from the slower degradation modes observed from 4.25 V to 4.45 V vs. Li/Li+. Figure 3c shows the capacity ratio of CC and CC + CP, whose decrease reflects the increase in resistance for each cathode potential.
The charge/discharge curves for 10 cycles at (a) 4.25 V, (b) 4.35 V, (c) 4.45 V, (d) 4.55 V, and (e) 4.65 V vs. Li/Li+, and the arrows at charge and discharge endpoints indicate the shift direction.
The retention of (a) CC and (b) CC + CP discharge capacity in 10 cycles at each cathode potential. The dotted line indicates the retention of 100 %.
Then, the HPC parameters CE, Δc, and Δd were analyzed to elucidate the degradation mechanism from an electrochemical viewpoint. Figure 4 shows the parameters and capacity fade per cycle obtained from Eq. 4. Some of them are defined by the above equations and thus were plotted from the 2nd cycle. Although the differences of CE were extremely small, the slight decreases with increasing cathode potential, e.g., 0.0003 pts (0.03 %) between 4.25 V and 4.35 V vs. Li/Li+, were certainly observed. The CE was associated with capacity retention at each potential, but its value was not sufficient to understand the factors of capacity deterioration in detail. Therefore, it is important to consider other HPC parameters. The discharge endpoint slippage Δd is defined by Eq. 3, and the changes in the values were naturally similar to the inversion. In contrast, the charge endpoint slippage Δc showed a clear dependency on the cathode potential. The slippage is the quantified value of the amount and direction of the abovementioned endpoint shift; therefore, we studied the cathode potential dependency of Δc. The slippage of Δc comprises Qox and Qc, as shown in Eq. 2. Although the coulombic amount is derived from the loss of Li+ by electrolyte oxidation, Qox is expected to increase as the cathode potential increases, as in our previous reports.2 However, the potential dependency of Δc showed the opposite trend. Considering it is unlikely that the oxidative decomposition of the sulfide SE was suppressed for the increase in the cathode potential, the value of Qc must have increased, which means that the amount of trapped Li+ in the cathode increased at higher cathode potentials. TEM/ED measurements were performed to analyze these mechanisms from chemical and physical perspectives.
The high precision coulometry parameters of the coulombic efficiency, the charge endpoint slippage Δc, the discharge endpoint slippage Δd, and the capacity fade per cycle in 10 cycles at each cathode potential.
Figures 5 and 6 show TEM/ED results for SE in the interfacial regions after the cycle testing in the potential ranges of slow degradation mode from 4.25 V to 4.45 V vs. Li/Li+ and rapid degradation mode from 4.55 V to 4.65 V vs. Li/Li+, respectively. In the former potential range (Fig. 5), the ED profiles of SE at 4.25 V and 4.35 V vs. Li/Li+ did not change but partially changed at 4.45 V vs. Li/Li+, where the phases of LiCl and Li2S generated with the argyrodite-structured pattern remained. The potential of 4.45 V vs. Li/Li+, at which the degradation of argyrodite SE was observed, completely matched the first threshold potential at which cell performance, such as capacity retention and CE, began to deteriorate. In the latter potential range (Fig. 6), the partially remaining argyrodite-structured pattern disappeared steadily as the cathode potential increased. These changes in the ED profiles of argyrodite-structured SE indicate that the SE exhaled the products, such as LiCl and Li2S, and decomposed into an amorphous phase. These results are similar to those previously reported,2,12 and it is significant that the HPC parameters further support them in this study. Conversely, when looking at the active material of NCM523, a significant change in particle size was observed in the TEM image at the highest potential of 4.65 V vs. Li/Li+. Most of the NCM particles were subdivided to the submicron order despite maintaining a particle size of several microns at lower potentials. The SEM images strongly supported this hypothesis (Fig. 7). Furthermore, the cracks are caused by shrinkage due to excessive delithiation from NCM523 in liquid-type LIB.24–29 In particular, the contraction of the particles is already known to be prominent above the potential of ∼4.5 V (x > 0.8, in Li(1−x)Ni0.5Co0.2Mn0.3O2).28,29 In contrast to the drastic physical changes in particle size, a lattice fringe with a dark line distance of ∼4.7 Å was observed on the NCM surface at 4.25 V, 4.45 V, and 4.65 V vs. Li/Li+ (Fig. 7). The lattice fringe is assigned to the distance of the (003) plane in the NCM crystal structure,25,26 which means that chemical degradation, for example, a rock-salt structure due to transition metal ion migration into the lithium layer,25,26 did not occur significantly in this study. The larger the contribution of the physical, not chemical, degradation of NCM523, the larger value of Qc in Eq. 2 and Δc is negative. That is, the irreversible Li+ loss due to the disconnection of the conduction path derived from the cracks in the NCM particles led to rapid decrease of the discharge capacity, rapid resistance increase (Fig. 3), and Δc inversion to a negative value (Fig. 4). This deterioration in cycle performance due to the degradation of NCM active material itself was partially consistent with our previous report of the potential dependency for the EIS.2 The physical factors of crack and subdivision were considered more dominant in this study, although it cannot be said that chemical degradation had not occurred at all. Janek et al. have reported that the NCM811 cathode of an ASSB, which is more susceptible to physical factors due to its larger contraction/expansion ability compared to NCM523, might be more affected by physical factors such as cracks and voids than that of liquid-type LIBs.27
(a–c) BF images of SE/NCM composites, (d–f) ED patterns at green circles in BF images of (a–c), and (g–i) intensity profiles of each ED pattern at the interfacial SE after the cycle testing in the potential ranges from 4.25 V to 4.45 V vs. Li/Li+. The simulated X-ray diffraction patterns of argyrodite-typed Li6PS5Cl (ICSD Collection code: 259200), LiCl (ICSD Collection code: 26909), and Li2S (ICSD Collection code: 56023) are also shown.
(a–c) TEM BF images of SE/NCM composites, (d, e) ED patterns at green circles in BF images of (a, b), and (f, g) intensity profiles of each ED pattern at the interfacial SE after the cycle testing in the potential ranges from 4.25 V to 4.45 V vs. Li/Li+. The simulated X-ray diffraction patterns of argyrodite-typed Li6PS5Cl (ICSD Collection code: 259200), LiCl (ICSD Collection code: 26909), and Li2S (ICSD Collection code: 56023) are also shown.
From these electrochemical, chemical, and physical analyses by using HPC and TEM/ED, we observed a potential-dependent model of cycle performance in the system of argyrodite-structured sulfide Li6PS5Cl/LiNbO3-coated NCM523 (Fig. 8). Initially, in charge/discharge cycles up to 4.35 V vs. Li/Li+, both argyrodite SE and NCM cathode active material showed minimal degradation. Subsequently, at potentials higher than 4.35 V, the oxidative decompositions of SE proceeded, which led to the decomposition of SE into amorphous phase and generation of LiCl and Li2S. Finally, at potentials higher than 4.55 V, the physical degradation (such as nano/microcrack and subdivision) of NCM active material particles began together with the oxidative decomposition of SE. To circumvent the chemical and physical degradations, it may be effective to develop higher oxidation-tolerant SE materials30 and cathode active materials, such as single crystal particles,31 that are less likely to crack.
Schematic images of degradation mechanism for cathode potential.
In this study, we investigated the potential dependency of the cycle performance of an all-solid-state cathode in an argyrodite-structured sulfide, Li6PS5Cl/LiNbO3-coated NCM523, using HPC measurements and SEM and TEM/ED analyses. HPC parameters, such as coulombic efficiency and charge and discharge endpoint slippages, showed two threshold potentials for the deterioration of the cycle performance. The material analysis of TEM/ED after the cycle test at potentials from 4.25 V to 4.65 V vs. Li/Li+ at 60 °C showed that threshold values were related to the subsequent degradation modes. The first threshold was at 4.45 V vs. Li/Li+, where the oxidative decomposition of the solid electrolyte began. The second threshold was at 4.55 V vs. Li/Li+, where the physical degradation (e.g., nano/microcracks and subdivision) of cathode active material particles began together with the oxidative decomposition of the solid electrolyte. These results revealed the essential tasks leading to their rapid commercialization of ASSBs.
This article is based on results obtained from the “Development of Fundamental Technologies for All Solid State Battery applied to Electric Vehicles” project (SOLiD-EV, JPNP18003) commissioned by the New Energy and Industrial Technology Development Organization (NEDO).
Yusuke Morino: Conceptualization (Lead), Data curation (Equal), Investigation (Equal), Methodology (Equal), Visualization (Lead), Writing – original draft (Lead), Writing – review & editing (Lead)
Hirofumi Tsukasaki: Data curation (Equal), Investigation (Equal), Methodology (Equal)
Shigeo Mori: Data curation (Equal), Investigation (Equal), Methodology (Equal)
There are no conflicts of interest to declare.
New Energy and Industrial Technology Development Organization: JPNP18003
Y. Morino and H. Tsukasaki: ECSJ Active Members
