Degradation Mechanism of All-Solid-State Li-Metal Batteries Studied by Electrochemical Impedance Spectroscopy.

Solid-state Li-metal batteries have the potential to achieve both high safety and high energy densities. Among various solid-state fast-ion conductors, the garnet-type Li7La3Zr2O12 (LLZO) is one of the few that are stable to Li metal. However, the large interfacial resistance between LLZO and cathode materials severely limits the practical application of LLZO. Here a LiCoO2 (LCO) film was deposited onto an Al-doped LLZO substrate at room temperature by aerosol deposition, and a low interfacial resistance was achieved. The LCO particles were precoated by Li3BO3 (LBO), which melted to join the LCO particles to the LLZO substrate at heating. All-solid-state Li/LLZO/LBO-LCO cells could deliver an initial discharge capacity of 128 mAh g-1 at 0.2 C and 60 °C and demonstrated relatively high capacity retention of 87% after 30 cycles. The cell degradation mechanism was studied by electrochemical impedance spectroscopy (EIS) and was found to be mainly related to the increase of the interfacial resistance between LBO and LCO. In-situ SEM analysis verified the hypothesis that the increase of the interfacial resistance was caused primarily by interfacial cracking upon cycling. This study demonstrated the capability of EIS as a powerful nondestructive in-situ technique to investigate the failure mechanisms of all-solid-state batteries.

Introduction

Increasing concerns about environmental pollution accelerate the development of electric vehicles (EVs), which are quickly taking over the automobile industry despite the pandemic.[1] In 2020, the global transportation sector was responsible for 24% of direct CO2 emissions from fuel combustion.[2] As a result, reducing transportation emissions is crucial in tackling the climate emergency. Replacing gasoline cars with EVs is expected to be indispensable in reaching net-zero emissions worldwide.

However, the EV industry is concerned with maximizing the driving range and improving vehicle safety. Buyers are often deterred by range anxiety because public charging stations are lacking. In addition, Li-ion batteries with liquid electrolytes are potentially dangerous as the liquid electrolytes are generally highly flammable. Recently all-solid-state Li-metal batteries have regained tremendous research interest because of the discovery of some new solid-state fast-ion conductors, such as LLZO and Li10GeP2S12 (LGPS).[3] For example, the ionic conductivity of cubic LLZO at room temperature is about 10–3 S cm–1.[4] Moreover, LLZO is reported to be chemically and electrochemically stable against Li metal,[5] which has a very high theoretical specific capacity of 3860 mAh g–1 and the lowest redox potential of −3.04 V versus standard hydrogen electrode.

Nevertheless, it is difficult to form good interfacial contact between LLZO and electrode materials because LLZO has a high elastic modulus.[6] This results in high interfacial resistance, preventing the practical application of LLZO as a promising solid electrolyte.[7] Recently, a great effort has been made to decrease the interfacial resistance between LLZO and Li metal anode.[8,9] For instance, the LLZO/Li interfacial resistance can be effectively reduced by introducing a thin Au or Al2O3 interlayer.[10,11] Good contact between LLZO and Li metal can also be achieved by high pressure and heating because Li metal is ductile, and its melting point is relatively low (180.5 °C).[12] In contrast, much less effort has been devoted to reducing the interfacial resistance between LLZO and cathode materials.[13] Although good interfacial contact between LLZO and LCO can be achieved by cosintering,[14] LCO decomposes at temperatures greater than 900 °C.[15] A convenient way to decrease the interfacial resistance is to introduce a small amount of liquid electrolyte to wet the solid electrolyte/cathode interface. Moreover, an ionic-liquid-containing cathode concept has been explored to tackle the interfacial resistance issue.[16,17] Polymer-in-LLZO flexible sheet electrolytes have also been proposed to improve the physical contact between solid electrolytes and cathodes.[18,19] However, introducing a liquid electrolyte or applying a polymer-based electrolyte to reduce the interfacial resistance will compromise battery safety. As a result, Han et al. suggested a new approach to lower the interfacial resistance between LLZO and LCO: thermally soldering LCO and LLZO together with an in-situ-formed ceramic interlayer.[20]

Aerosol deposition (AD) is a ceramic coating technology that can deposit cathode films on ceramic solid electrolytes at room temperature.[21,22] An illustration of the AD system used for this study is shown in Figure a. Accelerated cathode particles are cracked and deformed after hitting the ceramic substrate (Figure b,c). The high surface energy generated by particle fracture or plastic deformation is likely to be the driving force for the densification of the fine cathode particles at room temperature.[23] However, ceramic-based all-solid-state cells prepared by AD generally degrade rapidly with cycling, which is very likely due to the mechanical degradation (e.g., cracking) of the solid electrolyte/electrode interface.[24] Since both the cathode and solid electrolyte materials are brittle ceramics, the stress generated by the expansion and contraction of the cathode material during cycling will likely lead to cracking of the interface. Thus, the solid electrolyte/cathode interface prepared by AD requires further engineering. One practical way is to thermally solder LCO and LLZO together with Li3BO3 (LBO), a mechanically soft, low melting point (about 700 °C), and Li-ion-conducting solid electrolyte.[25,26]

Figure 1

Illustration of the AD process. (a) AD system. (b) Solder-coated cathode particles bombarding an LLZO substrate. (c) As-deposited cathode layer, where the particles are cracked and deformed. (d) Heat-treated cathode layer/LLZO substrate interface, where the void spaces are filled by the LBO solder upon heating.

Here we deposited an LBO-LCO cathode film on an Al-doped LLZO substrate at room temperature by AD. The LCO particles were precoated by LBO, which melted at heating and functioned as a binder to connect the LCO particles with the LLZO electrolyte. The all-solid-state Li/LLZO/LBO-LCO cell showed relatively high capacity retention, possibly better than any reported results for aerosol-deposited all-solid-state Li-metal batteries. The cell degradation mechanism was studied by EIS and in-situ scanning electron microscopy (SEM).

Experimental Section

Synthesis of LCO

The LCO powder was synthesized through a sol–gel method using lithium acetate (C2H3LiO2, 98.0%, FUJIFILM Wako Pure Chemical Corporation, Japan) and cobalt acetate tetrahydrate (C4H6CoO4·4H2O, 99.0%, FUJIFILM Wako Pure Chemical Corporation, Japan) as starting materials. The starting materials were dissolved in water to form a solution, in which citric acid (C6H8O7, 99.0%, Kanto Chemical Co., Inc., Japan) was added as a chelating agent for gelation. The solution was stirred at 80 °C for 1 h and dried at 120 °C. The obtained solid mixture was calcined at 400 °C for 9 h in air, followed by sintering at 800 °C for 8 h in air to produce a crystalline LCO powder. The XRD pattern of the synthesized LCO powder is shown in Figure S1.

Synthesis of LBO

The LBO powder was synthesized by heating mixtures of Li2CO3 (99.0%, FUJIFILM Wako Pure Chemical Corporation, Japan) and B2O3 (95%, Kanto Chemical Co., Inc., Japan) in a molar ratio of 3:1 at 600 °C for 10 h in air. The LBO powder was further pulverized (to reduce the average particle size to less than 1 μm) by ball milling (400 rpm, 10 min milling followed by a 20 min pause to avoid overheating, repeated for 99 times). Figure S2 shows the XRD pattern of the synthesized LBO powder.

Synthesis of Al-Doped LLZO

The Al-doped LLZO powder was prepared by a solid-state reaction method. Starting materials of LiOH·H2O, La(OH)3, and ZrO2 powders were mixed by ball milling and calcined at 900 °C for 15 h. After mixing with a γ-Al2O3 powder, the calcined powder was compacted and sintered at 900 °C for 3 h and then at 1200 °C for 24 h. The molar ratio of LiOH·H2O, La(OH)3, ZrO2, and γ-Al2O3 was 6.9:3.0:2.0:0.125. The XRD pattern of the synthesized Al-doped LLZO pellet is shown in (Figure S3). An optical image of the Al-doped LLZO pellet is shown in Figure S4a, and an SEM image of a fracture surface is shown in Figure S4b. The relative density of the LLZO pellet was about 95%, determined by measuring its weight and dimension.

Aerosol Deposition

The LBO and LCO powders in a mass ratio of 3:7 were ball milled at 400 rpm for 30 min, following calcination at 800 °C for 2 h in air. The LBO-coated LCO was transferred into the deposition chamber, which was evacuated by a rotary pump to 80 Pa. The diameter of the nozzle hole of the AD system was 1 mm, and the deposition area on the LLZO substrate was 0.785 cm2. The distance between the LLZO substrate and the nozzle was 40 mm. The Ar gas pressure was set to be 0.6 MPa. The optimal particle size for AD was found to be about 0.1–10 μm. The amount of LBO-LCO deposited on the LLZO substrate was determined from the weight increase of the LLZO substrate after after deposition. The LLZO substrate coated with the LBO-LCO composite cathode layer was further heated at 750 °C for 2 h in air to increase the interfacial contact between the LBO-LCO cathode and the LLZO substrate.

Materials Characterization

The crystal structure was analyzed by X-ray diffraction (XRD, Rigaku SmartLab). The microstructure was analyzed by a scanning electron microscope (SEM, JSM-6490A). The element distribution at the LBO-LCO/LLZO interface was analyzed by energy dispersive X-ray spectroscopy (EDS). An in-situ electrochemical cell was designed for continuously monitoring the microstructure evolution of the cross section of the LBO-LCO/LLZO interface region during cycling.

Electrochemical Measurements

AC electrochemical impedance spectroscopy (EIS) was performed in the frequency range from 7 MHz to 0.1 Hz at 25–60 °C (Biologic SP-300, France). The amplitude of the perturbation was 10–50 mV. The EIS spectrum of the Al-doped LLZO pellet is shown in Figure S5a, and the Arrhenius plot is shown in Figure S5b. Galvanostatic cycling of the all-solid-state Li/Al-doped LLZO/LBO-LCO cells was performed at 0.2 C and 60 °C with a cutoff voltage range of 2.8–4.3 V. The configuration of the all-solid-state cell is illustrated in Figure S6a. A UFO-shaped battery testing apparatus was designed and built for the cycling test, as shown in Figure S6b. The EIS data were fitted to equivalent electrical circuit models with the EC-Lab software.

Results and Discussion

An SEM image of the pristine LCO powder synthesized by a sol–gel method is shown in Figure a. The particle size is in the range of 0.5–3.0 μm, and the particle surface is clean. The LCO powder has a trigonal structure with R3m symmetry, as confirmed by X-ray diffraction (Figure S1). On the other hand, the LBO powder synthesized by a solid-state reaction shows a monoclinic structure with P21/c symmetry (Figure S2). The LCO particle was coated by LBO (Figure b,c) through ball milling and heat treatment (800 °C). The mass ratio between LCO and LBO in the LCO-LBO composite cathode is 7:3. It should be noted that LBO became amorphous-like after ball milling (Figures c and S7).

Figure 2

SEM micrographs of the LBO-LCO/LLZO interface. (a) As-synthesized LCO powder, (b) LBO-coated LCO powder, (c) High-resolution SEM image of the circled area in (b), (d) LBO-LCO cathode layer deposited on LLZO by AD at room temperature, following heat treatment at 750 °C for 1 h, (e) Element mapping of oxygen at the LBO-LCO/LLZO interface area, and (f) Element mapping of cobalt at the LBO-LCO/LLZO interface area.

The LBO-LCO powder was deposited onto a cubic LLZO (Figure S3) substrate by AD at room temperature. The interfacial contact between LBO-LCO and LLZO was further improved following heat treatment at 750 °C for 1 h (Figure d). The cubic LLZO with a relative density of about 95% was prepared by a solid-state reaction and pressureless sintering in air (Figure S4). The thickness of the deposited composite cathode layer is generally in the range 4–8 μm, while it is also possible to deposit much thicker cathode films (e.g., a 25 μm thick film, Figure S8) on LLZO by AD. Figure e,f shows the EDS mappings of oxygen and cobalt at the LBO-LCO/LLZO interface region, indicating good interfacial contact between the LBO-LCO cathode and the LLZO electrolyte.

Figure shows the galvanostatic cycling profiles of the Li/LLZO/LBO-LCO cells at 0.2 C and 60 °C. The cutoff voltage range was 2.8–4.3 V. As shown in Figure a, the ceramic-based all-solid-state cell delivered a high initial discharge capacity of 123 mAh g–1. The LBO-LCO cathode layer was 5.6 μm in thickness, and the corresponding loading of the LCO active material was about 1.1 mg cm–2. The cell also showed relatively high-capacity retention of above 84% after 30 cycles (Figure b), better than any previously reported results for aerosol-deposited all-solid-state Li metal batteries with a ceramic electrolyte.[21,24,27] The cell performance was further improved when the thickness of the LBO-LCO cathode layer was decreased from 5.6 to 4.3 μm because of reduced cell resistance. Specifically, the initial discharge capacity increased to 128 mAh g–1 (Figure c), and the capacity retention after 30 cycles increased to 87% (Figure d). The voltage profile of the all-solid-state cell at different current densities is shown in Figure e. The cell cycling performance was sensitive to current density, and well-defined plateaus were only observed at current densities less than 0.5 C. The discharge capacities were 126.3, 114.2, 95.1, 73.8, and 51.6 mA h g–1 when the current densities were 0.1, 0.2, 0.5, 1, and 2 C, respectively. However, the discharge capacity recovered promptly to 110.1 mAh g–1 when the C rate returned to 0.1 C, indicating a relatively high cycling capability at various C rates. It is worth mentioning that the Li/LLZO/LBO-LCO cell cannot be reversibly cycled without LBO because the interfacial resistance between the LCO cathode and the LLZO electrolyte is too high. The LBO coating is critical for the functioning of the all-solid-state cell.

Figure 3

Galvanostatic cycling performance and rate capability of the Li/LLZO/LBO-LCO cell at 60 °C. (a, b) Cycling performance. The LBO-LCO cathode layer is 5.6 μm thick, and the LCO active material loading is 1.1 mg cm–2. (c, d) Cycling performance. The LBO-LCO cathode layer is 4.3 μm thick, and the LCO active material loading is 0.84 mg cm–2. (e, f) Rate capability. The LBO-LCO cathode layer is 4.3 μm thick, and the LCO active material loading is 1.0 mg cm–2.

To understand the cell degradation mechanism, EIS analysis was performed at 10% state-of-charge (SOC) and in the frequency range from 3 MHz to 0.1 Hz at 60 °C (Figure ). The perturbation amplitude was chosen to be 50 mV to reduce noise and increase signal intensity. Typically the perturbation in potential is very small, about 5–10 mV at a controlled frequency, to ensure linear behavior of the current according to the Butler–Volmer model and to minimize irreversible changes to the electrochemical system.[28,29] The physical configuration of the all-solid-state Li/LLZO/LBO-LCO cell is illustrated in Figure a, where an Au interlayer is introduced to improve the interfacial contact between cell components.

Figure 4

EIS analysis of the degradation mechanism of the Li/LLZO/LBO-LCO cell. (a) Illustration of the all-solid-state cell configuration. (b) An ideal equivalent electrical circuit of the cell. (c) EIS spectrum of the cell before cycling. (d–g) EIS spectra at 10% of SOC during the 1st, 10th, 20th, and 30th charging processes, respectively. (h) Logarithm of the total impedance vs the logarithm of the frequency (the Bode plot is shown in Figure S9). (i) Evolution of the resistance components with cycling. In particular, the interfacial resistance between LBO and LCO (RLBO/LCO) increased significantly with cycling.

EIS is an electrochemical technique to measure a system’s static or dynamic impedance by monitoring the current response as a function of the applied AC potential.[30,31] Regardless of galvanostatic or potentiostatic conditions, the AC signal is sinusoidal. The impedance Z(ω) is a frequency-dependent complex number:[32,33]where j = is the imaginary unit, ω = 2πf the angular frequency of the applied AC voltage, f the AC frequency, ϕ the phase angle shift between the voltage and current, t the time, V the maximum value of V, and I the maximum value of I. Typical impedance spectra consist of both resistive (R) and capacitive (C) components, and the resistance consists of both electronic and ionic resistance. The impedance data are commonly plotted in two ways: (i) a Nyquist plot that shows the resistance along the real or x-axis and negative reactance (such as capacitance) along the imaginary or y-axis; (ii) a Bode plot that shows the logarithm of the frequency along the x-axis, and the logarithm of the impedance Z along one y-axis and the phase shift along the second y-axis. The Nyquist plot is very sensitive to changes and useful for analyzing the possible reaction mechanisms in an electrochemical system. In contrast, the Bode plot is advantageous for observing phase margins when the system becomes unstable because all information can be directly visible.[32]

An ideal equivalent electrical circuit of the Li/LLZO/LBO-LCO cell is shown in Figure b.[34] Because the resistance of the current collectors, the Au interlayers, and the LLZO electrolyte bulk are free of correlation with capacitance or inductance, they can be combined into an overall bulk resistance of Rbulk. In addition, there is only one inclined tail about 45° to the x-axis at the low-frequency region, which represents a Warburg element, describing a diffusion-controlled process. Similarly, all the Warburg resistances can be combined into an overall Warburg resistance of WWarburg to minimize the equivalent circuit elements for practical impedance analysis. A simplified equivalent circuit is shown in Figure a. The Nyquist plot before cycling is shown in Figure c, which shows only a high-frequency distorted semicircle and a low-frequency tail characteristic. The total resistance of the cell was about 115 Ω, corresponding to an area-specific resistance (ASR) of 90 Ω·cm2. After charging, the ASR decreased significantly to 65 Ω·cm2 because of reduced overall charge transfer resistance Rct (Figure d). This result suggested that the interfacial contact between LBO-LCO and LLZO was improved after the initial charging process. In addition, three poorly resolved arcs are observed in the frequency domain from 3 MHz to 8 Hz, which can be assigned to the interfacial resistance between Li–Au and LLZO (RLi–Au/LLZO), that between LLZO and LBO (RLLZO/LBO), and that between LBO and LCO (RLBO/LCO). This is because cubic LLZO has much higher ionic conductivity (about 10–4 S cm–1 at 25 °C) than LBO (about 10–6 S cm–1 at 25 °C). As a result, Li ions are supposed to transport much faster through the Li–Au/LLZO interface than through the LBO/LCO interface. The arc at the high-frequency region, which corresponds to a simplified RC Randles equivalent circuit (a resistor and a capacitor in parallel), is the response of the Li–Au/LLZO interface. On the other hand, the arc at the low-frequency region is the response of the LBO/LCO interface. The simulated spectrum (Figure d) based on the simplified equivalent electrical circuit (Figure a) agrees well with the measured spectrum. Similarly, the EIS spectra after 10, 20, and 30 cycles (as well as the corresponding simulated spectra) are shown in parts e, f, and g of Figure , respectively. The overall ASR of the cell increased rapidly with cycling, which is also reflected in Figure h. In particular, the RLBO/LCO increased significantly with cycling, while the Rbulk and RLi–Au/LLZO (Figure S10) were almost unchanged after 30 cycles (Figure i). Besides, the RLLZO/LBO increased gradually with cycling. The increase of RLBO/LCO is very likely related to the electromechanical degradation of the LBO/LCO interface, caused by the insertion and extraction of Li ions into and from LCO:Similarly, the increase of RLLZO/LBO is also likely caused by electromechanical degradation of the LLZO/LBO interface. All the impedance parameters used for the equivalent electrical circuit modeling are listed in Table .

Figure 5

In-situ SEM analysis of the degradation mechanism of the Li/LBO-LCO/LLZO cell. (a) A simplified equivalent electrical circuit of the one shown in Figure b. (b–d) Cross-sectional SEM images showing the crack evolution in the LBO-LCO/LLZO interface region before charging, after charging to 4. 2 V, and after discharging to 2.8 V, respectively. (e–g) Cross-sectional SEM images showing the crack evolution in the LBO-LCO/LLZO interface region after charging to 3.5, 4.05, and 4.2 V, respectively.

Table 1

Impedance Parameters Obtained by Using the Impedance Fitting Tool of ZFit Available in EC-Laba

Cycle	Rbulk (Ω·cm2)	RLi–Au/LLZO (Ω·cm2)	RLLZO/LBO (Ω·cm2)	RLBO/LCO (Ω·cm2)	QLi–Au/LLZO (F sn–1)	QLLZO/LBO (F sn–1)	QLBO/LCO (F sn–1)	WWarbug (Ω·s–0.5)
1st	30.0	19.2	10.8	7.2	4 × 10–6	4 × 10–5	9 × 10–4	14.34
					n = 0.7856	n = 0.8302	n = 0.8142
10th	30.0	19.2	11.2	13.4	3 × 10–6	4 × 10–5	8 × 10–4	17.08
					n = 0.7911	n = 0.8219	n = 0.7494
20th	31.7	21.8	13.0	19.0	1 × 10–3	5 × 10–5	2 × 10–6	19.41
					n = 0.6144	n = 0.7587	n = 0.8021
30th	30.0	19.2	13.1	21.0	1 × 10–3	2 × 10–6	1 × 10–4	21.27
					n = 0.6692	n = 0.8611	n = 0.6160

The value of the χ2 distribution for the fitting is in the range 0.9–3.0. Q stands for constant phase element or CPE, used for modeling the behavior of a double layer that is an imperfect capacitor. For practical impedance analysis, RCu, Rbulk LLZO, RAu, and RAl are combined into Rbulk. The Warburg resistances of WLi–Au/LLZO, WLLZO/LBO, and WLBO/LCO are combined into WWarburg, as shown in Figure a.

In-situ SEM was performed to verify the proposed crack evolution mechanism discussed above. The pre-existing cracks (Figure b) become larger and deeper after the initial charging (Figure c) and discharging (Figure d), as indicated by the arrowheads in Figure b–d. In addition, as expected, crack growth is more likely to occur at higher charging voltages. The crack in Figure e becomes slightly larger as the voltage increases from 3.5 to 4.2 V (Figure e–g). Thus, interfacial cracking between LBO and LCO was the primary degradation mechanism of the cell. On the other hand, no visible cracking was observed between the LBO/LLZO interface, probably because the degree of the interfacial cracking was less severe. Another possibility accounting for the gradual increase of RLLZO/LBO might be due to the potential reaction between LBO and LLZO. Thus, identifying a more suitable solder material than LBO is critical for developing all-solid-state Li/LLZO/LCO batteries.

Conclusions

All-solid-state Li/LLZO/LBO-LCO cells were prepared by aerosol deposition, and the cell degradation mechanism was investigated by EIS. LBO-LCO cathode films were deposited onto an Al-doped LLZO ceramic substrate at room temperature. LCO, the active material, was 70 wt % in the LBO-LCO cathode layer, while LBO functioned as a Li+-conducting solder that joined the LCO particles to the LLZO substrate at heating. The LBO coating was critical for the functioning of the all-solid-state Li/LLZO/LBO-LCO cell. The cell delivered a high initial discharge capacity of 128 mAh g–1 at 0.2 C and 60 °C. Moreover, it demonstrated high capacity retention of 87% after 30 cycles, possibly representing the best results available for all-solid-state Li metal batteries with ceramic oxide electrolytes.

On the basis of the EIS analysis results, the cell degradation was mainly due to the increased LBO/LCO interfacial resistance. In-situ SEM analysis further confirmed the hypothesis that the interfacial cracking between LBO and LCO evolved with cycling, resulting in the significant increase of the interfacial resistance. Electrochemical impedance spectroscopy has been demonstrated here to be a powerful nondestructive in-situ technique that can monitor a cell’s impedance evolution at different stages and provide insights into the cell failure mechanism.