Insights Into the Interfacial Degradation of High-Voltage All-Solid-State Lithium Batteries

Highlights The cycle performance of poly(ethylene oxide) (PEO)-based all-solid-state lithium batteries with LiCoO2 cathode was greatly improved via coating LiCoO2 with high-voltage stable Li3AlF6. At the upper cutoff voltage of 4.2 V, the poor electrochemical performance is mainly originated from the structure collapse of LiCoO2 at the surface instead of the decomposition of PEO. When the voltage reaches 4.5 V or even higher potentials, the intensive electrochemical decomposition of PEO-based solid polymer electrolyte accelerated interfacial degradation. Supplementary Information The online version contains supplementary material available at 10.1007/s40820-022-00936-z.

Recently, numerous efforts have been devoted to improving cycling performance of ASSLBs with PEO-based SPE and high-voltage cathodes [24][25][26]. Zhou et al. [27] introduced high-voltage stable poly(N-methyl-malonic amide) middle layer to protect PEO-based SPE from high-voltage oxidation, which stabilized the PEO-based ASSLBs operating in the voltage range of 3.0-4.25 V. Besides, Liang et al. [28] adopted lithium niobium oxide (LNO) thin film on the surface of NMC811 particles to stabilize the interface of NCM811 and electrolyte and thus suppress the oxidation of the PEO-based SPE.
Nevertheless, the failure mechanism of PEO-based SPE with high-voltage cathodes remains unclear. Yang et al. [29] tended to attribute the poor cycling performance to the structure instability of PEO-based SPE at high voltage. Meanwhile, it is supposed that the reactive terminal -OH group may be the root cause of the narrow electrochemical window of PEO-based SPE. By replacing terminal -OH group with more stable -OCH 3 , the electrochemical window could be extended to 4.3 V, which boosted the cycling stability of solid-state Li-NCM532 batteries. In another study, Chen's group [30] concluded that electrochemical decomposition of PEO-based SPE is not the only cause for the poor performance of LiCoO 2 /PEO-LiTFSI/Li batteries. When paired with LiMn 0.7 Fe 0.3 PO 4 cathode, the PEO-based solid-state battery shows much higher capacity retention than that with LiCoO 2 under the same cutoff voltage (4.2 V vs Li/ Li + ). To better understand such inconsistency, a systematic study regarding to the interfacial decomposition of PEO is required.
Herein, Li 3 AlF 6 (LAF) solid-state electrolyte is chosen as coating material on the surface of LiCoO 2 particles to systematically analyze the failure mechanism of PEO-based solid-state batteries with LiCoO 2 electrode. LAF was predicted to have good electrochemical stability and high ionic conductivity through theoretical calculation [31][32][33]. The results reveal that, at charging voltage of 4.2 V, the serious capacity decay mainly derives from structure collapse of LiCoO 2 , resulting in phase transition of the LiCoO 2 . When the charging cutoff voltage reaches up to 4.5 V or even higher, PEO-based SPE decomposed severely, leading to the constant increase of cell impedance and the degradation of battery. With the protection of LAF coating layer, structure failure of LiCoO 2 is inhibited and the decomposition of PEO-based SPE is reduced, exhibiting improved capacity retention.

Preparation of PEO-Based SPE
PEO (Mw = 600,000, ACROS ORGANICS), lithium bis(trifluoromethanesulfonyl)imide (LiTFSI, Aladdin, 99%) and Al 2 O 3 were dispersed in anhydrous acetonitrile (Innochem, 99.9%) and stirred continuously for 10 h at 80 °C. The mole ratio is EO:Li = 25:1. The obtained homogeneous gel-like solution was then casted onto a Teflon substrate, and the solvent was slowly evaporated at room temperature first, and SPE was then further dried for 24 h at 80 °C in an Ar-filled glovebox to obtain PEO-based SPE.

Preparation of LAF-coated LiCoO 2
Firstly, LiCoO 2 and NaF were dispersed and stirred for 4 h in deionized water. LiCl was added into the mixture solution and stirred for 6 h to obtain in situ LiF-coated LiCoO 2 . The 1 3 obtained LiF-coated LiCoO 2 and AlF 3 were dispersed and stirred in deionized water for 8 h and then dried at 80 °C. Finally, the composite was sintered at 400 °C for 10 h to obtain LAF-coated LiCoO 2 .

Electrochemical Measurements
LAF-coated LiCoO 2 or bare LiCoO 2 cathode electrodes were prepared by mixing LAF-coated LiCoO 2 or pristine LiCoO 2 powder, PEO, LiTFSI and SP in anhydrous acetonitrile to form a homogeneous slurry. The mass ratio of LiCoO 2 : PEO: SP was 7: 2: 1 by weight, and the amount of LiTFSI was based on the content of PEO (EO: Li = 10: 1). Then, the slurry was casted on a carbon-coated aluminum foil with doctor blade, followed by drying overnight at 110 °C in Ar-filled glove box. 2032 coin-type cells were assembled in Ar-filled glove box and constructed using bare LiCoO 2 and LAF-coated LiCoO 2 electrode, lithium metal anode and PEO-based SPE. The assembled cells were kept at 60 °C for 30 h before test. The cycling test was performed in the voltage range of 3.0-4.2 V (vs. Li/Li + ) at current density of 0.2C (1C = 150 mAh g −1 ) at 60 °C. Electrochemical impedance spectroscopy (EIS) measurements were taken with an electrochemical workstation (1400 cell test system, Solartron) in the frequency range from 1 MHz to 0.1 Hz with 10 mV amplitude at 60 °C.

Materials Characterization
The X-ray diffraction (XRD) patterns of bare LiCoO 2 and LAF-coated LiCoO 2 were collected by Bruker D8 Discover diffractometer with Cu Kα radiation within the 2θ range of 10-80°. The morphology and elements distribution of samples were characterized by scanning electron microscope (SEM, ZEISS SUPRA55) with an energy-dispersive spectroscopy (EDS, OXFORD, X-MaxN TSR). High-resolution transmission electron microscopy (HRTEM) images were collected using JEOL3200FS field-emission transmission electron microscopy (FETEM). All these cross-section samples were prepared by focused ion beam (FIB, FEI, Scios).

Results and Discussion
The thermodynamic electrochemically stable window of LAF was calculated by density functional theory (DFT) method. The equilibrium voltage profile and corresponding phase equilibria are shown in Fig. 1a. The LAF exhibits a wide electrochemical window with a cathodic limit of 1.03 V (vs Li + /Li) and an anodic limit of 6.51 V (vs Li + / Li). Meanwhile, LAF was proved to possess acceptable ionic conductivity in previous study [32]. In view of that PEObased SPE is likely to undergo electrochemical decomposition charging voltage of above 3.9 V (vs. Li/Li + ) and the decomposition may cause battery performance degradation [15][16][17][18][19]. The introduction of a modification layer with a wider electrochemical window is a common method to improve battery performance [34,35]. Therefore, the highvoltage stable LAF thin film with acceptable ionic conductivity is suitable to modify the interface between LiCoO 2 cathode and PEO-based SPE to demonstrate the failure mechanism of LCO/PEO solid-state battery.
The as-prepared LAF-coated LiCoO 2 (LAF@LCO) samples show the same XRD pattern as the bare LiCoO 2 (LCO) (Fig. 1b, JCPDS No. 75-0532), indicating the well-preserved structure of LiCoO 2 after surface modification. The SEM images show that the relatively smooth surface of LCO becomes rough after coated with a layer of Li 3 AlF 6 (Fig. 1c,  d). In Fig. 1e, EDS mapping of LAF@LCO particle shows clear signals of Al and F element. But the bare LCO with smooth surface only exhibits signals of Co and O element (Fig. S1). HRTEM is employed to further characterize the structure of LAF layer, which was uniformly coated over the whole surface of LCO. As shown in Fig. 1f, the clear lattice fringes of bulk area in HRTEM exhibit interplanar distance of 0.4728 nm belonging to the (003) planes of the layered phase in LAF@LCO particle [36][37][38], which is the same as bare LCO (Fig. S2). A lattice stripe in surface area shows different interplanar distance of 0.2135 nm, close to the interplanar distance of the (332) plane of Li 3 AlF 6 (JCPDS No. 88-0860) [39][40][41].
To evaluate the effectiveness of the modification strategy, the galvanostatic charge/discharge performance of LCO and LAF@LCO with PEO was tested using 2032 coin-type cells in the voltage range of 3.0-4.2 V (vs Li + /Li) at 0.2C at 60 °C. The cycling performance of LCO and LAF@LCO at 0.2C (Fig. 2a) shows that despite the slightly lower initial capacity, LAF@LCO delivered much-improved cycling stability as well as higher Coulombic efficiency compared with pristine LCO, which completely failed after 50 cycles. Figure 2b, c shows the charge/discharge curves of LCO and LAF@LCO at different cycles, respectively. The cycle performance of LAF@LCO is much better than that of LCO. The LAF@LCO cell shows an initial specific capacity of 122 mAh g −1 at 0.2C and increases to 125 mAh g −1 , followed by slightly decrease to 98 mAh g −1 over 100 cycles with capacity retention of 75.1%. The initial increase of capacity could result from gradual wetting of LiCoO 2 electrodes and PEO-based SPE [42,43]. In contrast, the LCO cell decays rapidly from 127 to 10 mAh g −1 after 50 cycles. The LAF@LCO also exhibits superior cycle performance to LCO at 0.5C, with 80.2% of discharge capacity retention (Fig. 2d). The charge/discharge curves of LCO and LAF@ LCO at 0.5C are presented in Fig. S3. The above results suggest that the surface modification promotes the interfacial stability between LCO and PEO-based SPE.
To investigate the role of LAF coating layer in promoting electrochemical performance, EIS, TEM and X-ray photoelectron spectroscopy (XPS) were conducted. Figure 3a shows the C 1s spectra from pristine PEO-based SPE. Three peaks can be observed at 292.1, 285.7 and 284.4 eV, which can be assigned to -CF 3 , C-O, C-C, respectively [44][45][46][47]. The -CF 3 signal in C 1s spectra comes from LiTFSI [48,49]. The C-O signal in C 1s and O 1s (532.5 eV) spectra (Fig. S4a) can be attributed to the ether chain (-C-O-C-) in PEO-based SPE. The initial areal ratio of C-O to C-C peak is 1: 0.21. Figure 3b, c shows the C 1s spectra from PEO-based SPE surface after cycling with LCO and LAF@ LCO at 3.0-4.2 V, respectively. By comparison, the areal ratio of C-O to C-C peak in C 1s XPS spectra decreases to 1: 0.55 after cycling with LCO electrode, which indicates the loss of ether chain. O-C=O (288.5 eV) peak can be originated from the decomposition products of PEObased SPE, because of the redox reaction with LCO during charge/discharge process. As PEO consists of repeating -(O-CH 2 -CH 2 )-units, the emerging peak corresponding to O-C=O group (288.5 eV) group is attributed to the oxidative decomposition of PEO [50]. Previous studies have shown that the hydroxyl group (-OH) at the end of PEO molecular chain will lose electrons and form monatomic free radicals (-O), while protons are trapped by lithium salts (as shown in Eq. 1) [54]. In view of the low proportion of hydroxyl groups in the molecular chain, we have made the following assumptions regarding to the decomposition process of PEO at 4.2 V voltage: C-O bond in PEO molecular chain breaks, forming long-chain polymers with -O · and -CH 2 · groups, respectively (Eq. 2). The unstable -CH 2 · group further converts to polymer with -CH · -CH 3 (Eq. 3). -O · groups will then react with -CH · -CH 3 groups, forming O-C-O group (Eq. 4), which tends to be oxidized into O-C=O group under oxidative potential (Eq. 5), and protons are also believed to be trapped by lithium salts. Delithiated Li x CoO 2 containing highly oxidative Co 4+ will accelerate this reaction process by attracting electrons and protons.
Herein, we use the areal ratio of O-C=O to C-O peaks in C 1s spectra as the indicator of the oxidation degree of PEO. With an upper cutoff voltage of 4.2 V, the ratios obtained from LCO and LCO@LAF after cycling are 0.08 and 0.05, respectively. indicating that the decomposition of PEO is effectively inhibited by LAF layer.
EIS was carried out to evaluate the cell impedance (Fig. 3d). However, the bulk resistance (R b ) of PEO-based SPE and total impedance of cells with LCO and LAF@LCO electrode are unchanged and almost equal after 50 cycles (Fig. S5), which demonstrates that the decomposition products still possess favorable ionic conductivity during longterm cycles. It could be concluded that the decomposition of PEO-based SPE may not be the primary cause for capacity fading.
The surficial structure change of cycled LiCoO 2 is also analyzed by HRTEM. As seen in Fig. 3e, different lattice stripes appear at the surface of LiCoO 2 particle. The interplanar distance of 0.2472 and 0.2137 nm belongs to the (111) and (200) plane of CoO [51,52], respectively, which indicates the structure of LiCoO 2 is transformed into CoO phases during cycling. The structure transition of LiCoO 2 is consistent with previous report, and CoO has been found to suppress the Li + transport [30]. The phase transition unveils that chemical redox reaction occurs at the interface between the highly reactive LiCoO 2 surface and PEO-based SPE, resulting in structure collapse of LiCoO 2 , hence the poor cycle performance of LCO at voltage of 4.2 V.
By sharp contrast, no obvious structure change can be found at the surface of LAF@LCO, and the original layered phase was well retained (Fig. 3f). It can be concluded that under this testing condition, LAF serves as a protective layer to shield LiCoO 2 from chemical degradation and promotes the cycling stability. Compared with PEO in contact with LCO@LAF, the direct contact between PEO and LCO accelerates the oxidation of PEO. It is believed that delithiated Li x CoO 2 containing highly oxidative Co 4+ tends to react with PEO, leading to oxygen loss in LCO lattice and degradation of PEO chains that forms decomposition products. Such chemical redox reaction will be inhibited by the electronic insulating LAF coating.
Differential electrochemical mass spectrometry (DEMS) was adopted to study the gas release behavior in PEO-based ASSLBs. LiCoO 2 /PEO-LiTFSI/Li was cycled under different upper cutoff voltages of 4.2, 4.3, 4.4, 4.5, 4.6, 4.7 and 4.8 V at 0.2C (Fig. 4a). Only trace amount of gas can be detected before 4.4 V, indicating negligible PEO decomposition. This result agrees with the conclusion drawn above that PEO decomposition might not be the reason for capacity fading of battery in the voltage range of 3.0-4.2 V. A small amount of C 2 H 4 O comes from the fracture of adjacent C-O bond in PEO as shown in Eq. 2. As the cutoff voltage gradually increases to 4.5 V, significant amounts of O 2 are released by the cell. Before the voltage could reach In contrast, gas release started to occur at 4.6 V for LAF@ LCO cells and intensive decomposition of PEO cannot be observed before 4.8 V (Fig. 4b). The improved performance may be ascribed to more stable surface of LAF@LCO particles, where PEO and LiCoO 2 are physically separated by the LAF layer, inhibiting the chemical decomposition of PEO at the PEO-LCO interface. Moreover, in the voltage range of 3.0-4.5 V, the capacity fading of LCO cell became even faster (Fig. 4c), while the LAF@LCO cell remained relatively stable (Fig. 4d) despite a mild attenuation. The contrast of cycle performance can be observed clearly in Fig. S6. In general, the improvement of LCO/PEO performance by LAF coating in our work is satisfactory (as shown in Table S1). The DEMS data indicate that compared with 4.2 V, the decomposition reaction of PEO becomes more drastic above 4.5 V, where a large amount of gas (e.g., C 2 H 2 , H 2 , C 2 H 4 , and O 2 ) are released.
The electrochemical decomposition of PEO-based SPE would lead to the increase of cell impedance. EIS is conducted to evaluate the cell impedance (Fig. 5a, b). The impedance of cell with LCO electrode increased rapidly Scheme 1 Proposed mechanism of LAF coating layer for enabling stable and high-voltage ASSLBs with the increase of the cycle number and reached over 900 Ω after 20 cycles. The increase of the cell impedance reveals that continue decomposition of PEO-based SPE occurs at the interface. In contrast, the impedance of the cell with LAF@LCO electrode was retained at around 200 Ω after 20 cycles, indicating more stable interface between PEO and LAF@LCO electrode due to the presence of LAF coating layer.
XPS is carried out to characterize the electrochemical decomposition of PEO-based SPE. Figure 5c shows the C 1s spectra from PEO-based SPE surface after cycling with LCO. Compared with XPS data obtained from battery cycled in the voltage range of 3.0-4.2 V, a new peak corresponding to R-C=O (287.2 eV) emerged markedly after cycling with LCO electrode in the voltage range of 3.0-4.5 V, accompanied with the apparent decrease of C-O peak, which can be attributed to further oxidation of PEO segments at 4.5 V with the areal ratio of R-C = O to O-C of 0.24. This process is also accompanied with the apparent decrease of C-O peak, which can be attributed to the oxidation of PEO segments at 4.5 V. It is speculated that (shown in Eq. 6), monatomic free radicals (-O · ) obtained previously will be further oxidized into carbonyl group [54]. In the process, delithiated Li x CoO 2 containing highly oxidative Co 4+ will also accelerate this reaction process by attracting electrons and protons. As for PEO cycled with LAF@LCO, the areal ratio of O-C = O to O-C peaks significantly decreased from 0.24 to 0.06 (Fig. 5d), suggesting a well-protected electrode-electrolyte interface at a voltage as high as 4.5 V. The same conclusion can be drawn from the O 1 s results for both samples (Fig. 5e, f). The XPS results are in good accordance with the constant increase of impedance and the results of DEMS. Different from the capacity fading at 4.2 V, the even poorer performance at 4.5 V could be attributed to both the degradation of LCO and the decomposition of PEO. (6) For LAF@LCO, the mild decomposition of PEO-based SPE could be ascribed to the deteriorated conductive carbon/ SPE interface at elevated voltages. The linear sweep voltammetry results of carbon, LCO and LAF@LCO electrodes confirm the deterioration of carbon/SPE interface at high voltage. As shown in Fig. S7, the current increases abruptly at the voltage beyond 4.2 V for LCO electrode, indicating the intense decomposition of PEO-based SPE, whereas similar current increase is observed after 4.5 V for the LAF@ LCO and carbon electrodes, which further demonstrates the protective effect of LAF coating layer. Since the bare carbon particles may cause the decomposition of PEO-based SPE at high voltages (≥ 4.5 V), future optimization should focus on interfacial protection at the electrode level.

Conclusions
In conclusion, the LAF is applied to coat on LiCoO 2 particles for investigating the deterioration mechanism of PEO-based ASSLBs paired with high-voltage cathode. As shown in Scheme 1, at charging voltage of 4.2 V, the poor electrochemical performance is mainly originated from the structure collapse of LiCoO 2 at the surface induced by chemical redox reaction between the highly reactive LiCoO 2 surface and PEO, indicating that structure stability of LiCoO 2 surface and interface is critical for high-voltage application [53]. When the voltage reaches 4.5 V or even higher potentials, both the degradation of LiCoO 2 and the intensively decomposition of PEO could be the reasons for the capacity further fading. The results show that LAF coating layer can stabilize LiCoO 2 by protecting them from chemical degradation during cycling and minimize the electrochemical decomposition of PEO at high voltage. This study provides an original view of the batteries failure and presents a facile strategy to extend the compatibility of PEO-based SPE with high-voltage cathode materials.