Introduction

Solid-state batteries (SSBs) with a high-capacity lithium metal anode are considered as the ultimate alternative to liquid lithium-ion batteries1, which not only exhibit higher energy density but also fundamentally solve the safety problems of the liquid batteries due to the utilization of non-flammable solid-state electrolytes (SSEs). However, the solid–solid interfaces between SSEs and electrodes cause a large inherent impedance in the SSBs. At the same time, the SSEs are completely non-wetting compared with the liquid electrolyte so that the electrolyte cannot be immersed in the cathode to construct lithium-ion transport pathways, slowing the diffusion of lithium ions between particles inside the cathode. In order to address this issue, solid electrolyte powders are often added to the cathodes and the interface between solid electrolytes and active materials is designed to increase the ionic conductivity and decrease the polarization of electrode2,3,4,5. In the recent year, the SSBs employing a garnet-structured Li7La3Zr2O12 (LLZO) electrolyte have shown significant promise in practical applications because the LLZO electrolyte has high lithium-ion conductivity and is stable to lithium metal, but again, the ionic conductivity inside the cathode is low due to the use of the non-wetting LLZO solid electrolyte piece. Some efforts have been made to build lithium-ion transport channels inside cathode to preparing high performance composite cathodes in LLZO-based SSBs. For instance, Wakayama et al. reported a three-dimensional bicontinuous composite cathode which increased the surface area of the interface between the active materials and the LLZO particles6. Broek et al. embedded the electrode materials to the porous LLZO electrolyte, and it is beneficial to converting of lithium ions inside the electrode7. Besides, the interface properties of the composite cathode can also be improved by forming a coated structure in which active materials are coated with the LLZO particles8.

Unfortunately, it has been reported that LLZO is unstable in moist air and it is spontaneous to react with H2O and CO2 to generate a Li2CO3 layer on the surface9. The Li2CO3 layer is lithiophobicity and has an ultralow low lithium-ion conductivity so that it is one of the sources of the high interfacial impedance in SSEs10,11. So far, although it has been reported that Li2CO3 on the surface of LLZO can be removed by surface polishing11 or chemical reaction12, these approaches based on “eliminating” concept have just short-term effectiveness and in particular, only suitable for handling the large-sized surface of electrolyte piece. In contrast, a method for removing the Li2CO3 layer on the surface of the LLZO powder that has a larger surface area with more Li2CO3 has not been reported. This hinders the fast transport of lithium-ions inside cathode when adding LLZO powder to the cathode as an ion conductor or designing the internal interface inside the cathode. Therefore, reliable solutions to remove the Li2CO3 layer and to establish intimate physical contact between LLZO and active cathode materials are still needed.

Herein, we propose an “interface homogeneity” strategy to transform the Li2CO3 into LiCoO2 active material on the surface of Li6.4La3Zr1.4Ta0.6O12 (LLZTO) by an on-surface lithium-donor reaction. Significantly, the LLZTO coated with LiCoO2 (LLZTO@LCO) was obtained by the reaction of the Li2CO3 layer on the surface of LLZTO with Co3O4. The transformation from Li2CO3 into Li2CoO2 is complete and not reversible, indicating that the Li2CO3 layer can be fully removed. The formed LiCoO2 layer ensures direct contact between the solid electrolyte particles and the homogeneous LiCoO2 cathode material, circumventing the conventional heterogeneous solid–solid interface problem inside composite cathodes. In this work, the proposed LLZTO@LCO materials were successfully synthesized and characterized. For comparison, the LLZTO coated with naturally formed Li2CO3 (LLZTO@Li2CO3) and the LLZTO@LCO were used as an ionic conductor to prepare composite cathodes with LiCoO2 active materials, separately. And then LLZTO-based SSBs were assembled. We found that the battery with an LLZTO@LCO-containing LiCoO2 composite cathode exhibited a high Coulombic efficiency (CE) and improved cycling performance.

Results

Characterization of LLZTO and LLZTO@LCO

The LLZTO@LCO materials were prepared by a two-step solid sintering process. Figure 1a shows the transmission electron microscopy (TEM) image of LLZTO after air exposure for 4 weeks. It can be seen that a 0.1 μm thick layer formed on the surface of LLZTO. The coating layer can be indexed to the monoclinic Li2CO3 by subjecting selected area electron diffraction of the encircled region (Fig. S1). Figure 1b shows the scanning electron microscopy (SEM) image of LLZTO, the particle size of LLZTO is about 4 μm and has a smooth surface as well as irregular shape. After the reaction, instead of Li2CO3 layer, LiCoO2 is evenly distributed on the surface of LLZTO to form a coating with a thickness in the range of 0.3–0.5 μm, as shown in Fig. 1c. The inset in Fig. 1c shows the high resolution TEM (HRTEM) image of the encircled region. The lattice spacing of 0.245 nm agrees with the (101) facets of crystallized LiCoO2, and the widely exposed (101) facets (Fig. S2) exhibit higher ionic conductivity and electrochemical activity13. Moreover, as Fig. 1d shows, the LLZTO@LCO exhibits a spherical structure, which can increase the stacking density of composite cathode. The LiCoO2 exhibits a nanoplate-like, which is beneficial to the rate performance of the battery14. Figure 1e shows the energy-dispersive X-ray mapping analysis results. The Co, O elements are uniformly distributed on the surface of particles and the La, Zr elements can also be detected in that region, corresponding to the LLZTO@LCO structure. The X-ray diffraction (XRD) patterns, as shown in Fig. 1f, are in agreement with the TEM and SEM results. Compared with Li5La3Nb2O12 PDF card (45-0109), the existence of Li2CO3 on the surface of LLZTO was confirmed before reaction. Apparently, the ultimate materials only contain LLZTO and LiCoO2 after the reaction, and the LiCoO2 coating has R-3m symmetry as the traditional LiCoO2 active materials15. Above results confirm that the transformation of Li2CO3 to LiCoO2 is well achieved, but we find that an impurity is also formed during the first sintering. Figure 1g shows the XRD pattern of the lithium-donor reaction products after the first sintering. It should be noted that the impurity is La2Zr2O7. This is due to the volatilization of lithium from LLZTO during sintering16. Besides, after first sintering, the substitution of Li2CO3 by LiCoO2 indicates that the Li2CO3 coating has been fully reacted with Co3O4 to generate LiCoO2, but the LiCoO2 exhibits a block-like rather than a nanoplate-like (Fig. S3).

Fig. 1: Characterization of LLZTO@Li2CO3 and LLZTO@LCO.
figure 1

a, b TEM image and SEM image of LLZTO powder exposed to air for four weeks. c, d TEM image and SEM image of LLZTO@LCO. e EDX mapping analysis of LLZTO@LCO corresponding to (d). f XRD patterns of LLZTO@Li2CO3 and LLZTO@ LCO. g XRD pattern of materials after first sintering containing impurity.

Full size image

Reaction mechanism of transforming LLZTO@Li2CO3 to LLZTO@LCO

As shown in Fig. 2a, transforming LLZTO@Li2CO3 to LLZTO@LCO was achieved by the on-surface lithium-donor reaction. By sintering different rations of Co3O4 and LLZTO@Li2CO3, we found that there is no Co3O4 left after the excessive Co3O4 reacts with a small amount of Li2CO3 from the surface of LLZTO (Fig. S4). This is because, in addition to the reaction of Li2CO3 with Co3O4, there is also a reaction between lithium volatiles from LLZTO and the remaining Co3O4 to generate LiCoO2. Li2CO3 and LLZTO as the lithium donors together provide the lithium sources for the transformation reaction to form the complete LiCoO2 coating on the surface of LLZTO. To confirm the above process, we designed two experiments. First, Co3O4 with Li2CO3 materials were sintered under the first sintering condition and the result indicates that Co3O4 can react with Li2CO3 to generate LiCoO2 under this condition (Fig. S5). Second, we designed an experiment of sintering Li2CO3-free LLZTO and Co3O4 under the same condition. The La2Zr2O7 and LiCoO2 were still found in the sintered products, indicating that lithium volatilization exists in LLZTO during the sintering process, and the volatilized lithium can react with Co3O4 to form LiCoO2 (Fig. S6).

Fig. 2: Process of transforming LLZTO@Li2CO3 into LLZTO@LCO.
figure 2

a Schematic illustration of the lithium-donor reaction to achieve interface homogeneity. b Schematic illustration of the two-step solid state reaction process of transforming LLZTO@Li2CO3 into LLZTO@LCO.

Full size image

Based on the experimental results, Fig. 2b summarizes the process of the transformation. Initially, LLZTO exposed to air will form a Li2CO3 layer on the surface. Then, Co3O4 and LLZTO are fully mixed and sintered at 600 °C in air for 4 h. The Li2CO3 layer and Co3O4 undergo lithium-donor reaction to generate LiCoO2 on the surface of LLZTO. Meanwhile, the lithium source inside LLZTO also reacts with Co3O4 to generate LiCoO2 layer, but Li6.4-3xLa3Zr1.4Ta0.6O12 lithium-deficient phase is formed due to the loss of lithium, which contains many lithium defects and leads to the formation of La2Zr2O717. After that, in order to supply La2Zr2O7 with lithium-ion and let it return to the original LLZTO structure, Li2O salt is added and sintered again at 600 °C for 5 h in air. Excitingly, the lithiumization of La2Zr2O7 to LLZTO is realized. This process is the same as the preparation of LLZTO materials18. Finally, the pure LLZTO@LCO is obtained by washing and centrifuging. Different sintering temperatures (600, 700, 800, 900 °C) were attempted and we found that LiCoO2 can be synthesized at all of the above temperatures. However, when the temperature is higher than 700 °C, the diffusion of elements occurs (Fig. S7), which is consistent with previously reported19,20. In addition, we also tried the one-step sintering of LLZTO@Li2CO3, Co3O4, and Li2O, but the structure of LLZTO coated with LiCoO2 could not be formed and Li2CO3 still existed in store (Fig. S8).

Stability and electrochemical activity of LLZTO@LCO

LLZTO@LCO was then subjected to air-stability and activity measurements. To clarify its stability, the LLZTO@LCO was exposed to air for 4 months. The XRD results are shown in Fig. 3a. It is obvious that Li2CO3 was not formed after exposure to air for a long time. This means that the LiCoO2 coating can restrain the formation of Li2CO3 layer and improve the stability of LLZTO significantly. The Fourier transform infrared (FTIR) spectra of the LLZTO and LLZTO@LCO samples exposed to air for 4 months also verified this result (Fig. 3b). It can be seen that the strong peaks of 1438 and 863 cm−1 were formed in LLZTO, which corresponds to the Li2CO3 FTIR spectrum9. In addition, the weak peak of 3569 cm−1 agrees with the LiOH H2O FTIR spectrum, which is due to the reaction between water and LLZTO9,10, but it does not affect the formation of LiCoO2 layer (Fig. S9). Conversely, Li2CO3 and LiOH H2O were not formed in LLZTO@LCO. To investigate lithium-intercalated activity of the LLZTO@LCO, it is used as the active material to prepare a cathode with a binder and conductive carbon, without the addition of any other active materials. A Li/LiClO4-EC-DEC/LLZTO@LCO liquid cell was then assembled (the inset in Fig. 3c). It should be noted that the liquid electrolyte is used here to describe the lithium-intercalated behavior of the LLZTO@LCO in more detail, so that it can be accurately compared with the characteristic redox peaks and charging/discharging voltage platforms (~3.9 V) of commercial LiCoO2. As shown in Fig. 3c, charging and discharging voltage platforms are confirmed at about 3.9 V, which is corresponding to that of the commercial LiCoO2. Figure 3d shows the cyclic voltammetry (CV) of the battery. There are strong redox peaks at about 3.87 and 3.95 V, indicating that the LiCoO2 coating in LLZTO@LCO exhibits activity, improving the transport of ions on the LLZTO surface. Two weak peaks appear at 4.05 and 4.2 V, where LiCoO2 lattice changes from hexagonal to monoclinic21.

Fig. 3: Stability and activity test of LLZTO@LCO.
figure 3

a XRD patterns of LLZTO@LCO before and after exposure to air for four months. b FTIR spectra of the LLZTO and LLZTO@LCO samples after exposure to air for four months. c, d Charge/Discharge curves and cyclic voltammetry profile of the liquid cells. The inset in Fig. 3c shows the structural illustration of the cell, in which the cathode was prepared by mixing LLZTO@LCO, PVDF and KB.

Full size image

Characterization and ionic transport mechanism of the LCO-LLZTO@LCO composite cathode

In order to investigate the effect of LLZTO@LCO on the lithium-ion transfer kinetics of the cathode, the lithium-ion apparent diffusion coefficient was tested by performing CV measurements. Figure 4a, b shows the CV profiles of the LCO-LLZTO@LCO and LCO-LLZTO@Li2CO3 cathodes at different scan rates. The lithium-ion apparent diffusion coefficient can be calculated according to the Randles–Sevcik equation22

$$I_p = 2.68 \times 10^5n^{3/2}A\,C\,D^{1/2}v^{1/2},$$
(1)

where Ip is the peak current (A); n is the charge-transfer number of the redox reaction; A is the area of the cathode plate (cm2); C is the lithium-ion concentration in LiCoO2 cathode (0.051 mol cm−3); D is the lithium-ion diffusion coefficient (cm2 s−1); v is the scan rate (V s−1). Ip is linearly related to v1/2 and the value of Ip/v1/2 can be obtained from the linear fitting results as 0.01744 and 0.01391 for LCO-LLZTO@LCO and LCO-LLZTO@Li2CO3 cathodes (Fig. 4c). The lithium-ion apparent diffusion coefficient could be calculated to be 2.04 × 10−13 cm2 s−1 for LCO-LLZTO@LCO cathode and 1.28 × 10−13 cm2 s−1 for LCO-LLZTO@Li2CO3 cathode. Significantly, after Li2CO3 is converted to LiCoO2, the lithium-ion diffusion coefficient of the cathode is increased by about 59%. This is because the activated LLZTO@LCO promotes ionic transport between particles, decreasing the resistance inside cathode. (Fig. S10).

Fig. 4: Characterization and ionic transport mechanism of the cathodes.
figure 4

CV profiles of a LCO-LLZTO@LCO and b LCO-LLZTO@Li2CO3 cathodes at different scan rates. c Peak current as a function of the square root of the scan rate of LCO-LLZTO@LCO and LCO-LLZTO@Li2CO3 cathodes. d–f Cross-sectional SEM images of the LiCoO2, LCO-LLZTO@Li2CO3 and LCO-LLZTO@LCO cathodes, separately. g Schematic illustration of the ionic transport mechanism inside cathodes.

Full size image

To explore the mechanism by which LLZTO@LCO particles enhance the transport of lithium ions inside cathode, the cross-sectional micromorphology of the cathodes were observed. As shown in Fig. 4d, before adding LLZTO to the LiCoO2 cathode, the LiCoO2 nanoparticles are distributed on the cathode layer with a thickness of 15 μm, constructing a lithium-ion transport network. After LLZTO@Li2CO3 and LLZTO@LCO are introduced (Fig. 4e, f), they are distributed throughout the cathode and are in close contact with LiCoO2 particles around, providing composite channels for lithium-ion transport. Significantly, LLZTO@Li2CO3 and LLZTO@LCO particles with a large particle size cross the cathode layer, which can construct rapid large-span channels for the transport of the lithium ions to the interior of the cathode. However, the Li2CO3 layer on the surface of LLZTO with low conductivity will increase the interface impedance between the active material and the ion conductor particles. Conversely, LLZTO@LCO particles not only have an active surface in close contact with the active materials (Fig. S11), but also have a tight and low-impedance interface at the junction of LLZTO core and LiCoO2 shell (Figs. S12 and 13), promoting the transport of lithium ions inside the cathode.

Based on the above results, it can be concluded that the optimization of the transport channels for lithium ions by LLZTO@LCO may be the reason for the improved ionic conductivity, thus a possible mechanism is provided in Fig. 4g. In the LiCoO2 cathode without an ionic conductor, lithium ions diffuse into cathode through the ionic channels constructed by the active material LiCoO2 with low ionic conductivity, which can reduce the diffusion rate and diffusion depth of lithium ions, causing a large voltage polarization and limiting electrode reaction to occur in the shallow layer of the cathode. But, after LLZTO@Li2CO3 is added, large-span transport channels for lithium ions are formed around LLZTO@Li2CO3 particles, which can quickly transport lithium ions deeper into the LCO-LLZTO@ Li2CO3 composite cathode. But, the presence of Li2CO3 on the surface of LLZTO is like sludge in the channels. Lithium ions can only enter and exit LLZTO particles from the thin layer of Li2CO3 and can only be transported to LiCoO2 particles through the LLZTO bulk phase (Fig. S14), limiting the diffusion direction of lithium ions to the periphery, short of a crisscross lithium-ion transport network. Transforming Li2CO3 layer into active LiCoO2 layer is like dredging the channels. Lithium ions can be freely transported not only in the bulk phase but also the surface of LLZTO@LCO, allowing the rapid lithium-ion transport paths to branch in any direction (Fig. S14), which realizes uniform diffusion of lithium ions on the shallow and deep layer of the LCO-LLZTO@LCO composite cathode. The rapid lithium-ion transport channel can be compared to an irrigation canal, in which the main channel is constructed along the high ionic conductivity area where more LLZTO@LCO particles are distributed, and the main channel branches to the surroundings to deliver lithium ions to various locations of the cathode, which greatly improves the transport efficiency of lithium ions. This transport mechanism allows lithium ions to be quickly and evenly distributed throughout the cathode, so the LCO-LLZTO@LCO composite cathode exhibits higher ionic conductivity.

LLZTO-based SSBs with LCO-LLZTO@LCO composite cathode

The electrochemical properties of the LLZTO@LCO and LLZTO@Li2CO3 were also compared in SSBs consisting of a lithium anode and LLZTO solid electrolyte pellet. The commercial LiCoO2 was used as the active material, and LLZTO@Li2CO3 and LLZTO@LiCoO2 were used as ionic conductors to prepare composite cathodes, separately. Then, the prepared LLZTO@Li2CO3-containing LiCoO2 (LCO-LLZTO@Li2CO3) cathodes and LLZTO@LCO-containing LiCoO2 (LCO-LLZTO@LCO) cathodes were assembled into coin-type cells separately (Fig. 5a). Meanwhile, a thin buffer layer that is solid at room temperature was used to reduce the interface impedance between the electrolyte piece and the electrode plates (Fig. S15). All the cells were cycled at room temperature as well as 0.1 C (1 C = 140 mA g−1). The cycling performance of the battery with an LCO-LLZTO@LCO cathode is shown in Fig. 5b. The discharge capacity of the first cycle reached 131 mA h g−1 and the discharge capacity can be retained at 81% after 180 cycles with a voltage polarization of 0.08 V. Moreover, after 180 cycles the structure of LLZTO@LCO particles remained stable (Fig. S16). In contrast, the capacity retention of the battery with an LCO-LLZTO@Li2CO3 cathode reached only 60% after 180 cycles (Fig. 5c), but better than that of the battery with a pure LiCoO2 cathode without an LLZTO@LCO or LLZTO@Li2CO3 ionic conductor (Fig. S17). Furthermore, the CE of the battery with an LCO-LLZTO@LCO cathode reaches 91.1% at first cycle and stable at above 99% after the first five cycles, but the battery with an LCO-LLZTO@Li2CO3 cathode exhibited lower CE of 87.8% at first cycle (Fig. 5d). The increase in CE is owing to the reduction of side reactions by removing the Li2CO3 inside the cathode. Figure 5e, f shows the electrochemical impedance spectroscopy of the batteries. It can be seen that the impedance plot includes an incomplete semicircle in the high frequency region, a semicircle in the middle frequency region and a tail in low frequency region, in which the semicircle in the middle frequency region corresponds to the overall interface resistance (Rint) inside the battery. The interface Rint in the battery with an LCO-LLZTO@LCO cathode is 600 Ω cm2 after 180 cycles, lower than 988 Ω cm2 in the battery with an LCO-LLZTO@Li2CO3 cathode (Fig. S18). The smaller interface resistance is mainly due to the optimized ion transfer channels of the LLZTO@LCO particles. In addition, the CVs measured on the composite cathodes (Fig. 5g) show that LCO-LLZTO@LCO cathode has a lower polarization. This can be explained by the fact that the transformation of insulating Li2CO3 to active LiCoO2 with high ionic conductivity achieves interface homogeneity inside cathode and can speed up the transport of lithium-ion between the particles. The change of LiCoO2 lattice from hexagonal to monoclinic is also observed at about 4.05 and 4.2 V by testing dQ dV−1 of the SSB with an LCO-LLZTO@LCO cathode (Fig. S19), but that is not obviously shown in the CV curve in Fig. 5g. This is because the SSB has a higher impedance than the liquid battery, leading to a larger polarization, which results in a shift and widen in the main peak of the LiCoO2 so that the weak peaks at 4.05 and 4.2 V are covered. Figure 5h shows the rate performance of the batteries. The discharge capacity of the battery with an LCO-LLZTO@LCO cathode still reached 116 mA h g−1 at 0.2 C and 100 mA h g−1 at 0.5 C, but the discharge capacity of the battery with an LCO-LLZTO@Li2CO3 cathode only reached 105 and 80 mA h g−1 at 0.2 and 0.5 C, which corresponds to higher ionic conduction of the LCO-LLZTO@LCO cathode. Significantly, instead of using low-voltage active materials which were mostly used in the LLZO-based SSBs in previous reports, the high-voltage LiCoO2 active materials with LLZTO@LCO ionic conductor are used to prepare composite cathodes to assemble solid cells in this work, and show improved cycleability and rate performance (Fig. S20).

Fig. 5: Electrochemical performance of the LLZTO-based SSBs.
figure 5

a Schematic illustration of the LLZTO-based SSB. The buffer layer was formed by dissolving 10 wt% of lithium trifluoromethanesulfonyl (LiTFSI) in succinonitrile (SN), and polyacrylonitrile (PAN) was added to enhance film-forming property. b, c Discharge/charge curves of the SSBs with an LCO-LLZTO@LCO and LCO-LLZTO@Li2CO3 cathode, separately. d Cycling performance of the SSBs with an LCO-LLZTO@LCO and LCO-LLZTO@Li2CO3 cathode, separately. e, f EIS of the SSBs with a LCO-LLZTO@LCO and LCO-LLZTO@Li2CO3 cathode, separately. g CVs of the SSBs with an LCO-LLZTO@LCO and LCO-LLZTO@Li2CO3 cathode, separately. h Rate capability of the SSBs with an LCO-LLZTO@LCO and LCO-LLZTO@Li2CO3 cathode, separately. All tests were performed at room temperature.

Full size image

Discussion

Purity of electrolytes has guided the history of commercial batteries. For instance, the successful development of high-purity LiPF6 in 1994, coupled with 99.9% pure ethylene carbonate, offered a leap forward in cycling ability of commercial lithium-ion batteries. The ubiquitous Li2CO3 can be considered as an impurity of LLZO particles, leading to inferior purity (<95%), which is far away from the practical needs. Hence, when using in LiCoO2 cathodes, the purity of the LLZO electrolyte is equivalent to 100% owing to the substitution of insulating Li2CO3 impurity for active LiCoO2, and the latter provides a homogeneous contact with the LiCoO2 active material inside composite cathodes. This is an apparent advantage of the LLZTO@LCO from the view point of electrolyte purity.

To clarify the electrochemical difference of LLZTO@Li2CO3 and LLZTO@LCO inside the composite cathode, the LCO-LLZTO@Li2CO3 and LCO-LLZTO@LCO composite cathodes were analyzed by the differential electrochemical mass spectrometry analysis (DEMS) separately. The assembled liquid batteries were used to detect the release of CO2. Figure 6a shows the charge curve (top) and corresponding CO2 emission (bottom) for the LCO-LLZTO@Li2CO3 composite cathode. Notably, the intensity of CO2 began to increase when charged to about 4.0 V (vs. Li/Li+), which is considered to be due to the decomposition of Li2CO3 in LLZTO@Li2CO3, consistent with previous observations that Li2CO3 was decomposed above 4.0 V23,24. It should be noted that this is the first demonstration of the electrochemical decomposition of Li2CO3 formed on the surface of LLZTO by experiments. In stark contrast, benefiting from the transformation of Li2CO3 into LiCoO2, CO2 was not released in the homogeneous LCO-LLZTO@LCO composite cathode (Fig. 6b), exhibiting much higher electrochemical stability, which explains the high initial CE of the battery with an LCO-LLZTO@LCO composite cathode.

Fig. 6: DEMS analysis of LLZTO@Li2CO3 and LLZTO@LCO in composite cathodes.
figure 6

a, b Charge curves (top) and corresponding CO2 emission (bottom) of LCO-LLZTO@Li2CO3 and LCO-LLZTO@LCO cathode, separately.

Full size image

We further evaluated the activity and extensive applicability of LLZTO@LCO by adding it to LiFePO4 cathodes with a low ionic diffusion rate. The prepared LLZTO@LCO-containing LiFePO4 (LFP-LLZTO@LCO) composite cathode and LLZTO@Li2CO3-containing LiFePO4 (LFP-LLZTO@Li2CO3) composite cathode (Fig. 7a) were assembled into SSBs using a lithium anode and a LLZTO solid electrolyte pellet, separately. As shown in Fig. 7b, the initial discharge capacity of the battery with an LFP-LLZTO@LCO cathode reached 163.2 mA h g−1, closing to the theoretical capacity of 170 mA g−1, and can be retained at 97% after 120 cycles with a low-voltage polarization of 0.09 V. In contrast, the battery with an LFP-LLZTO@Li2CO3 cathode exhibits an initial discharge capacity of 146.4 mA h g−1 and has a large voltage polarization of 0.16 V after 120 cycles (Fig. 7c). In addition, the SSB with an LFP-LLZTO@LCO cathode exhibits longer-term cycling performance than that previously reported (Fig. S21)4,25,26,27,28,29.

Fig. 7: Extensive applicability of LLZTO@active-material and the two-step solid state reaction.
figure 7

a Schematic illustration of the LFP-LLZTO@Li2CO3 and LFP-LLZTO@LCO composite cathodes. b, c Discharge/charge curves of the SSBs with an LFP-LLZTO@LCO and LFP-LLZTO@Li2CO3 composite cathode, separately. d Schematic illustration of the two-step solid state reaction. e, f SEM image and XRD pattern of the ultimate materials after converting Li2CO3 into LMO. g, h Discharge/charge curves of the SSBs with an LMO-LLZTO@LMO and LMO-LLZTO@Li2CO3 composite cathode, separately.

Full size image

The two-step solid-state reaction process (Fig. 7d) was also successfully extended to prepare LLZTO@LiMn2O4. Figure 7e shows the XRD pattern of the converted materials. The ultimate materials contain LLZTO, LiMn2O4, and Li2Mn2O4, in which the Li2Mn2O4 is a discharged state of LiMn2O4 due to the presence of excess lithium salt during sintering, causing the insertion of lithium ions into LiMn2O4. Figure 7f shows the SEM image of LLZTO@LiMn2O4/Li2Mn2O4 (LLZTO@LMO). LiMn2O4 and Li2Mn2O4 with a nanoparticle-like morphology are evenly distributed on the surface of LLZTO to form a coating. The electrochemical properties of the LLZTO@LMO-containing LiMn2O4 (LMO-LLZTO@LMO) and LLZTO@Li2CO3-containing LiMn2O4 (LMO-LLZTO@Li2CO3) composite cathodes were compared in SSBs with the lithium anode and LLZTO electrolyte pellet. The initial discharge capacity of the battery with an LMO-LLZTO@LMO cathode reached 122.8 mA h g−1, which is higher than that of another battery (108.7 mA h g−1). In addition, compared with the battery with an LMO-LLZTO@Li2CO3 cathode, the battery with an LMO-LLZTO@LMO cathode exhibits improved cycling stability (Fig. S22) and a lower voltage polarization of 0.19 V after 120 cycles at 0.1 C (1 C = 148 mA g−1), while the initial CE is lower due to the presence of discharged Li2Mn2O4 (Fig. 7g, h).

In conclusion, by transforming the ubiquitous insulating Li2CO3 layer on the surface of LLZTO solid electrolytes into an active LiCoO2 layer, pure LLZTO@LCO particles are successfully synthesized with an on-surface lithium-donor reaction. The R-3m symmetry LiCoO2 is mainly generated by the lithium donor reaction of the Li2CO3 layer and Co3O4 in the first sintering. At the same time, it is found that lithium volatiles from LLZTO also reacts with Co3O4, resulting in part of LiCoO2, accompanying by La2Zr2O7 impurity. At the heart of our technology is offsetting the formidable impurity La2Zr2O7 by precisely supplementing extra lithium sources, thus restoring it to the pristine LLZTO structure in the second sintering step. The LLZTO@LCO particles are exposed to air for 4 months without Li2CO3 formation, indicating excellent store stability and demonstrating a radical solution of the Li2CO3 issue. We found that the decomposition of Li2CO3 formed on the surface of LLZTO occurs at voltages above 4.0 V, which is one of the reasons for the low initial Coulomb efficiency. Meanwhile, the converted LiCoO2 layer of the LLZTO@LCO particles exhibits the same lithium-intercalated electrochemical activity with commercial LiCoO2, which enables it to interact favorably with the LiCoO2 active material in the solid LCO-LLZTO@LCO composite cathode. As a consequence of the interface homogeneity inside cathode, the solid-state lithium metal battery with the LLZTO@LCO and LiCoO2 composite cathode shows 81% capacity retention after 180 cycles at 0.1 C, room temperature, superior to that with the LLZTO@Li2CO3 and LiCoO2 one, representing the highest level among LiCoO2-based solid batteries. Our results indicate that although the formation of Li2CO3 on LLZO is inevitable, it would no longer hinder. Lithium-ion transfer at the solid electrolyte/cathode interface, but provide a chance to be transformed into active materials, thus achieving an in-situ intimate contact of ion conductor and active materials inside the cathode. In addition, the solid sintering reaction, which is the most common mass production method for ceramics-type electrolytes and cathode materials, has also been successfully applied to the in-situ transformation of Li2CO3 to LiMn2O4. It is hopeful to develop a series of LLZO@LiFePO4, LLZO@layered Ni-Co-Mn or Ni-Co-Al, etc. to precisely match active materials inside the composite cathodes for solid-state lithium metal batteries.

Methods

LLZTO@LiCoO2 materials

LLZTO@LiCoO2 materials were prepared by a two-step solid-state reaction process. LLZTO powders and Co3O4 (Aladdin, 99.99%) were mixed in a mass ratio of 20:3 at an agate mortar for 10 min and sintered at 600 °C for 4 h. Then Li2O were added to the precursors in a mass ratio of 3:10, and heated to 600 °C and dwelled on for 5 h. The obtained materials were finally washed with ethyl alcohol and centrifuged giving rise to LLZTO@LiCoO2. The LLZTO powders and pellets was prepared by a method previously reported30,31.

Composite cathodes

The LCO-LLZTO@LCO and LCO-LLZTO@Li2CO3 composite cathodes were prepared in the air. LLZTO@LiCoO2 and LLZTO@Li2CO3 solid electrolytes were, respectively, mixed with LiCoO2 active materials, PVDF, KB in a mass ratio of 3:5:1:1 in N-methylpyrrolidone (NMP) solvent. After stirring for 12 h, the slurry was scraped on the carbon-containing aluminum foil, and heated at 60 °C in atmospheric pressure for 2 h, then dried at 80 °C in vacuum for 24 h to obtain cathode foil and cut it into discs of 12 mm in diameter. A total of 80 wt% of commercial LiCoO2, 10 wt% of PVDF, and 10 wt% of KB were mixed to prepare pure LiCoO2 cathodes. The LFP-LLZTO@LCO and LFP- LLZTO@Li2CO3 composite cathodes were prepared by the same method as above.

Assembly of liquid cells

CR2032-type liquid coin cells were assembled in an argon-filled glovebox to detect the air-stability and activity of LLZTO@LCO. The cathodes were prepared by mixing LLZTO@LiCoO2 active materials, PVDF and KB in a mass ratio of 8:1:1 in NMP solvent. After stirring for 12 h, the slurry was scraped on the carbon-containing aluminum foil, and heated at 60 °C in atmospheric pressure for 2 h, then dried at 80 °C in vacuum for 24 h to obtain cathode plate and cut it into discs of 12 mm in diameter. Li foil with 12 mm in diameter was used as anode and dissolving 0.1 M LiClO4 in EC-DEC (1:1, v/v) was used as a liquid electrolyte.

Assembly of solid-state cells

CR2032-type solid-state coin cells were assembled in an argon-filled glovebox. In all-solid-state cells, the LLZTO plates were used as SSEs and Li foils with 12 mm in diameter were used as anodes. In order to improve the interface between SSE and electrodes, a buffer layer that exhibits a film at room temperature was introduced, which was formed by dissolving 10 wt% of trifluoromethanesulfonyl in succinonitrile at 80 °C and adding polyacrylonitrile to enhance film-forming property. The gelatinous slurry was scraped on the electrode surface at 80 °C, and cooled down to room temperature to form a solid film.

Electrochemical measurement

The charge/discharge tests of the cells were carried out using Land machines at room temperature. The specific capacity of the batteries with an LCO-LLZTO@LCO cathode was calculated based on the weight of the cathode active materials including both the LiCoO2 on the surface of LLZTO@LiCoO2 and the commercial LiCoO2. The proportion of Co element is 11.58 wt% in the LLZTO@LiCoO2, which is provided by the inductively coupled plasma spectrum test. The loading of cathodes is about 2 mg cm−2, corresponding to the active materials of around 1.12 mg cm−2.

TEM, selected area electron diffractions, and high resolution transmission electron microscopy observation

The coating structure of LLZTO@Li2CO3 and LLZTO@LiCoO2 were observed using Field Emission JEM-2100F TEM. The diffraction fringes and lattice fringes of LLZTO@Li2CO3 and LLZTO@LiCoO2 were observed using selected area electron diffractions and high resolution transmission electron microscopy of field emission JEM-2100F TEM, separately.

SEM observation and energy-dispersive X-ray spectroscopy analysis

The microstructures and X-ray (energy-dispersive X-ray spectroscopy) mapping of LLZTO and LLZTO@LiCoO2 were observed using SU-8220 field emission SEM.

X-ray powder diffraction and FTIR analysis

The phases and crystalline structure of the materials before and after transformation were analyzed using X-ray powder diffraction with Cu Kα radiation. LLZTO and LLZTO@LiCoO2 were exposed to air for four months for FTIR measurements.

Differential electrochemical mass spectrometry analysis

The LCO-LLZTO@LCO and LCO-LLZTO@Li2CO3 slurry were dripped on stainless steel with 12 mm in diameter, separately. Then, the dried composite cathodes were assembled into customized Swagelok cells. Dissolving 1 M LiPF6 in EC/DMC (1:1 v:v) was used as the liquid electrolyte and Li foil with 12 mm in diameter was used as the anode. The liquid cells were charged to 4.5 V at 0.25 C.