Reinforced cathode-garnet interface for high-capacity all-solid-state batteries

Garnet-type solid-state electrolytes (SSEs) are particularly attractive in the construction of all-solid-state lithium (Li) batteries due to their high ionic conductivity, wide electrochemical window and remarkable (electro)chemical stability. However, the intractable issues of poor cathode/garnet interface and general low cathode loading hinder their practical application. Herein, we demonstrate the construction of a reinforced cathode/garnet interface by spark plasma sintering, via co-sintering Li6.5La3Zr1.5Ta0.5O12 (LLZTO) electrolyte powder and LiCoO2/LLZTO composite cathode powder directly into a dense dual-layer with 5 wt% Li3BO3 as sintering additive. The bulk composite cathode with LiCoO2/LLZTO cross-linked structure is firmly welded to the LLZTO layer, which optimizes both Li-ion and electron transport. Therefore, the one-step integrated sintering process implements an ultra-low cathode/garnet interfacial resistance of 3.9 Ω cm2 (100 °C) and a high cathode loading up to 2.02 mAh cm−2. Moreover, the Li3BO3 reinforced LiCoO2/LLZTO interface also effectively mitigates the strain/stress of LiCoO2, which facilitates the achieving of superior cycling stability. The bulk-type Li|LLZTO|LiCoO2-LLZTO full cell with areal capacity of 0.73 mAh cm−2 delivers capacity retention of 81.7% after 50 cycles at 100 μA cm−2. Furthermore, we reveal that non-uniform Li plating/stripping leads to the formation of gaps and finally results in the separation of Li and LLZTO electrolyte during long-term cycling, which becomes the dominant capacity decay mechanism in high-capacity full cells. This work provides insight into the degradation of Li/SSE interface and a strategy to radically improve the electrochemical performance of garnet-based all-solid-state Li batteries.


Future perspectives
Garnet-type solid-state electrolytes demonstrate great potential to improve cell safety without employing organic liquid electrolytes and increase battery energy density with high-voltage cathodes and lithium (Li) metal anodes. However, the high interfacial impedance aroused by the poor interfacial contact and parasitic reactions remains a major challenge for high-performance bulk-type all-solid-state batteries (ASSBs). Future research into the interface engineering between garnet electrolytes and the electrodes will play a key role in the development of practical ASSBs. For Li/garnet interface, designing multifunctional flexible interlayers with excellent lithiophilicity and electronic insulativity is the key to enhancing the cycling stability and achieving high areal capacity. For cathode/garnet interface, advanced sintering methods such as spark plasma sintering have successfully attained intimate interface bonding. Further efforts can be made in developing bulk composite cathode with superior kinetics and stability to realize high rate performance and energy density.

Introduction
Garnet-type solid-state electrolyte (SSE) is considered as a prospective candidate for building all-solid-state batteries (ASSBs) owing to its high ionic conductivity, wide electrochemical window and high stability against lithium (Li) metal and air [1][2][3], which enable it to fit high-voltage cathode materials and Li metal anode (3860 mAh g −1 and −3.040 V versus the standard hydrogen electrode) for higher energy density [4,5]. However, the practical application of garnetbased ASSBs is restricted by the large interfacial resistance between the electrolytes and electrodes. Many attempts have been made to ameliorate the interfacial contact between the garnet SSE and Li anode, including introducing interface coatings (e.g. C, Mg, Al 2 O 3 , g-C 3 N 4 , LiF, Cu 3 N) [6][7][8][9][10][11][12], and removing the Li 2 CO 3 from the garnet by surface treatment (e.g. HCI, H 3 PO 4 , NH 4 F, H 3 BO 3 -HF) [13][14][15][16], which have effectively reduced the interfacial resistance and improved the performance of Li symmetric batteries. Nevertheless, building an intimate cathode/garnet SSE interface is still limited by the following challenges. (a) High-temperature sintering may give rise to unexpected side reactions between the cathode active material (CAM) and the electrolyte in the composite cathode, while low-temperature sintering leads to poor contact between CAM and the electrolyte, and between the composite cathode layer and the electrolyte layer [17][18][19]; (b) non-negligible volume change in CAM during cycling may cause particle fracture and interface delamination [20,21]; (c) electrochemical reactions may occur between CAM and the electrolyte [22,23]. All the above ultimately results in low cathode loading and poor cyclic performance of garnet-based ASSBs.
Recent studies have made extensive efforts to reduce the resistance of the cathode/garnet SSE interface, where LiCoO 2 (LCO) is extensively studied as CAM due to its high electronic conductivity, high thermal stability and similar thermal expansion coefficient to garnet SSEs [24][25][26][27]. Among them, three main approaches have been highlighted. (a) Depositing LCO on the garnet pellet directly to form a thin film [28][29][30]. However, the loading of the cathode film, which is typically hundreds of nanometers thick, is extremely low. (b) Casting the composite cathode slurry on the top surface of garnet pellets followed by sintering [31][32][33][34][35][36][37][38]. To improve the contact of LCO and garnet, sintering additives such as Li 3 BO 3 (LBO) and its derivatives are added to the composite cathode slurry. Despite all this, most cells fabricated by this method still suffer from low operating current density and areal loading, which cannot meet the requirements of high-energy bulk-type Li batteries. (c) Co-sintering the electrolyte and the composite cathode into a dual-layer by simultaneous pressurization and heating, such as spark plasma sintering (SPS) [22,[39][40][41], and hot-press sintering [42]. The significant merit of the cosintering process is to construct a cathode/SSE interface directly from each specific powder rather than rigid pellets under applied pressure, which can easily attain tight interface bonding and high-capacity cathodes (>1 mAh cm −2 ). Generally, a lower sintering temperature (∼700 • C) is usually chosen to avert side reactions between LCO and garnet, yet leading to poor interfacial contact and high interfacial impedance. Moreover, degradation may occur at the LCO/garnet interface during electrochemical cycling [22], both resulting in poor cycling stability. Therefore, in terms of the state-of-the-art garnet-based ASSBs, there is still much scope for improvement to reach the desired target, and it remains a major challenge to enhance the stability of cathode/garnet interface.
Herein, we report a reinforced cathode/garnet interface constructed by the SPS method, which achieves an intimate contact between high-capacity cathode and garnet SSE. The electrolyte is composed of Li 6.5 La 3 Zr 1.5 Ta 0.5 O 12 (LLZTO) and LBO, and the composite cathode is composed of LCO, LLZTO and LBO (figure 1). LBO is served as a sintering additive with a room-temperature Li-ion conductivity of 2 × 10 −6 S cm −1 and melts at ∼700 • C, which can wet the particles and accelerates the interdiffusion of atoms in liquid phase, beneficial to the enhancement of densification and Li + conductivity [43,44]. Profiting from the SPS and LBO strategies, the composite cathode with conductive network structure is firmly bonded to the LLZTO layer, thereby reducing its interfacial impedance to as low as 3.9 Ω cm 2 at 100 • C. Thus, the bulk-type Li|LLZTO|LCO-LLZTO ASSB with an areal loading of 0.73 mAh cm −2 exhibits superior cycling stability with a capacity retention of 81.7% after 50 cycles at 100 µA cm −2 . In addition, we reveal the evolution of LLZTO/electrode interfaces of full cells during cycling from multiple aspects and propose research directions to further improve the electrochemical performance of LLZTO-based ASSBs, laying an essential foundation for its future application.  stoichiometric amounts with 2-propanol as dispersant and 10 wt% excess LiOH·H 2 O was added to compensate for Li loss during subsequent sintering. The mixed powder was then dried and transferred to a muffle furnace for calcining at 900 • C for 12 h. The sintered powder was finally sieved into a uniform fine powder with a 200-mesh sieve. For LBO synthesis, LiOH·H 2 O (99%, Aladdin) and B 2 O 3 (98%, Aladdin) were mixed in a mole ratio of 6:1 and then calcined at 600 • C for 4 h.

Fabrication of ASSBs
LCO-LLZTO/LLZTO half cells were sintered using a LABOX-212 SPS system. LCO (Canrd, China), LLZTO, LBO were mixed in a mass ratio of 50:50-x:x (x = 0, 2, 5, 10) as the composite cathode. LLZTO, LBO were mixed in a mass ratio of 100-x:x (x = 0, 2, 5, 10) as the electrolyte. The two powders were layered into a graphite mold with an inner diameter of 10 mm, and sintered at 80 MPa and 690 • C for 1 h at high vacuum. Then the LCO-LLZTO/LLZTO half cells were transferred to a tube furnace and annealed at 875 • C for 2 h in O 2 flow. The as-sintered pellets were firstly polished, where the LLZTO layers were about 800 µm thick and the composite cathode layers were about 15-50 µm thick according to different cathode areal loadings. Then Au thin film was sputtered on the composite cathode side served as current collector, and a thin graphite-based layer was coated on the electrolyte side to enhance the adhesion of Li foil anode [6]. The assembled full cells were sealed in a Swagelok cell and applied pressure of ∼10 kPa for further tests.

Characterizations
The phase structure was analyzed by x-ray diffraction (XRD), using a Rigaku Ultima IV x-ray Diffractometer, which operated at 40 kV and 30 mA over the range of 10 ∼ 90 • (2θ) with Cu Ka radiation (λ = 1.5405 Å). The microstructure and elemental distribution of samples were characterized using the SUPRA 55 scanning electron microscope (SEM) combined with energy dispersive spectrometer (EDS). Samples for EDS mapping were preprocessed by ion beam cutting. Raman spectra were recorded with a Horiba Scientific XploRA PLUS unit with a 532 nm excitation solid-state laser over a Raman Shift of 200 cm −1 -1500 cm −1 . The surface constitution of samples was examined using the Thermo Scientific K-Alpha x-ray Photoelectron Spectrometer (XPS) system. The Young's modulus of samples was measured using a Cypher ES Polymer Edition Atomic Force Microscope by scanning 256 points on an area of 5 × 5 µm 2 .

Electrochemical measurements
Galvanostatic charge and discharge tests of Li|LLZTO|LCO-LLZTO ASSBs were conducted between 3.0 and 4.05 V at different current density, and collected with a Land CT-2001A (Wuhan, China). The electrochemical impedance spectroscopy (EIS) tests were performed in an Auto Lab workstation, in the frequency range from 1 MHz to 0.01 Hz and with an amplitude of 10 mV. The full cells were charged to 4.05 V and held at 4.05 V for 12 h, then rest at open circuit voltage for 2 h before EIS testing.

Results
As shown in figure S1, no discernible impurities are detected in the as-synthesized LLZTO and LBO by XRD, and the diffraction peaks of LLZTO are well indexed to cubic phase. Previous studies have reported that LCO and LLZTO would gradually set off a heat-induced reaction from 700 • C [18,45], thus we selected 690 • C as the heating temperature. Owing to the relatively low temperature, the density of pellets can be improved by prolonging the dwell time, in contrast to that of other previous reports (∼10 min) [39,41]. The Li-ion conductivity of the electrolyte pellets with various dwell time of 15 min, 30 min, 1 h, and 1.5 h were tested by EIS at room temperature (figure S2, table S1). The results indicate that the Li-ion conductivity increases with the prolongation of the dwell time within 1 h, and when the dwell time reaches 1.5 h, the excessive volatilization of Li leads to the formation of La 2 Zr 2 O 7 impurity phase (figure S3), resulting in a slight decrease in ionic conductivity, and thus 1 h is selected as the dwell time.
During sintering above 600 • C, LCO in the composite cathode is reduced to CoO due to direct contact with graphite molds [25,41,46], but interestingly, this change can be reversed by annealing in O 2 flow, confirmed by both XRD, Raman spectroscopy, and XPS, as shown in figure 2. From XRD and Raman results (figures 2(a) and (b)), the patterns show explicitly that the CoO in the composite cathode is oxidized back to LCO after annealing. From XPS results (figure 2(c)), Co2p spectrum shows spin orbit splitting into 2p 1/2 (at the high binding energy) and 2p 3/2 (at the low binding energy) intensity peaks which contain the same chemical information. For the sample of composite cathode before annealing, the 2p 3/2 peak locates at 781.1 eV can be assigned to Co 2+ , with a corresponding satellite peak at 786.8 eV, consistent with the previous report [47]. For the sample of composite cathode after annealing, the 2p 3/2 component can be well fitted into an intensive main peak at 779.4 eV assigned to Co 3+ and a weak peak at 781.3 eV assigned to Co 2+ , accompanied by a satellite peak at 788.4 eV [48], which indicates that most Co 2+ is converted to Co 3+ . By contrast, the cubic phase structure of LLZTO is well maintained during the sintering process, implying that LLZTO does not occur any chemical reaction with LCO or graphite at the current experimental conditions ( figure S4).
As mentioned above, LBO was added as a sintering additive for SPS process, which has a great influence on the microstructure of sintered products. Figures S5(a) and (d) show the SEM images of the cross section of the sintered LCO-LLZTO/LLZTO half cells with 0 wt%, 2 wt%, 5 wt% and 10 wt% LBO, respectively. Figures S5(e)-(h) show the corresponding contact state of particles inside the composite cathode at high magnification, respectively. With the increase of LBO content, the voids between the composite cathode layer and the electrolyte layer and between the particles reduce accordingly, and the sintered density increases. When the LBO content reaches 5%, the LCO-LLZTO/LLZTO half cell achieves a fairly dense state, as shown in figure 3(a). An enlarged SEM image of the interface between the LCO composite cathode and the LLZTO electrolyte (red boxed area in figure 3(a)) is shown in figure 3(b), which shows a densified sintered structure and a well-welded interface. The EDS mappings of La, Zr, Co and B (figures 3(c)-(f)) demonstrate that the LLZTO and LCO are each connected into a conductive network, which are tightly cross-linked throughout the cathode, providing composite channels to facilitate Li-ion and electron transport. LBO is evenly scattered in the LCO-LLZTO/LLZTO half cell which boosts the cementing among particles to form a reinforced dual-layer structure and mitigate the strain/stress of LCO. The EDS line scan (figure 3(g)) shows the variation trends of elements in the direction perpendicular to the LLZTO/LCO interface (the inset in figure 3(g)), where a thin interfacial layer of about 320 nm think is formed, and no additional elements are detected in the bulk of LLZTO and LCO particles, implying that there is no obvious element diffusion or chemical reactions between the bulk of two particles. Furthermore, the electrolyte pellet with LBO exhibits a higher mechanical strength, as shown in the figure 3(h) and S6. The Young's modulus of the electrolyte pellet with 5% LBO is 6.5 GPa, while that without the additive is only 3.5 GPa. This also indicates that LBO enhances the compactness of the pellets, which conduces to improve Li-ion conductivity and restrains the growth of Li dendrite.
The optimal LBO content was explored by evaluating the impedance of the full cell. As shown in figure S7, the EIS spectra of Li|LLZTO|LCO-LLZTO ASSBs with different LBO contents at 100 • C all comprise four parts. The impedance value of the real axis intercept in the high frequency can be assigned to the resistance of LLZTO electrolyte (R LLZTO ), the semicircle in the high-frequency region (1 MHz-150 Hz) represents the interfacial resistance between Li and LLZTO (R Li/LLZTO ), the semicircle in the mid-frequency region (150 Hz-1 Hz) represents the interfacial resistance between composite cathode and LLZTO (R LCO-LLZTO/LLZTO ), and the diagonal line in the low frequency is related to the Warburg impedance attributed to the Li-ion diffusion process [31,49]. The inserted equivalent circuit is used for simulating the electrical response of the full cells, wherein R is the resistance element, CPE is the constant phase element, W is the Warburg element. The fitting results of R LLZTO , R Li/LLZTO , and R LCO-LLZTO/LLZTO for the full cells with different LBO content are shown in table S2. It shows that each resistance component of the full cell with 5% LBO is the smallest. This can be attributed to the influences of sintering additive LBO. Wherein, moderate LBO can promote the densification, thus facilitate rapid ionic/electronic conduction. While excess addition is adverse to Li-ion transport due to the low Li-ion conductivity of LBO. Furthermore, the interfacial resistance between the cathode and LLZTO is as low as 3.9 Ω cm 2 (at 100 • C, 5% LBO), confirming that the SPS and sintering additive strategies are highly effective at improving the poor solid-solid contact between the garnet and the cathode.
We then selected Li|LLZTO|LCO-LLZTO full cells with 5% LBO for electrochemical performance measurements due to their lower impedance. The full cells with different areal capacity are assembled and subjected to the galvanostatic charge-discharge test at 100 • C with a current density of 100 µA cm −2 . As shown in figure S8, the full cell with a high initial discharge areal capacity of 2.02 mAh cm −2 shows poor cycling stability with a capacity retention of only 59.8% after 30 cycles. The charge-discharge profiles exhibit a large increase of the charge/discharge potential gap. Nevertheless, when the areal capacities of the full cells are decreased to 1.40 mAh cm −2 and 0.73 mAh cm −2 , the electrochemical performance is significantly improved (figures 4(a), (b) and S9). Both cells show initial coulombic efficiencies of about 80%, and the coulombic efficiencies rapidly increase to ∼97.5% in the third cycle, then slightly increase to ∼98.5% during the following cycles. It also shows that the cell with lower areal capacity exhibits superior cycling stability. As for the cell with areal capacity of 1.40 mAh cm −2 , fast capacity decay is observed with a low capacity retention of 66.7% after 50 cycles. While for the cell with areal capacity of 0.73 mAh cm −2 , the cycling stability is greatly improved. It still delivers a high capacity of 0.60 mAh cm −2 with a capacity retention of 81.7% after 50 cycles. Operando EIS was employed to investigate the fading mechanism of the Li|LLZTO|LCO-LLZTO full cell. Figure 4(c) shows the EIS spectra of the battery with a capacity of 1.40 mAh cm −2 underwent different cycle numbers. It shows that the impedance steadily increasing throughout the cycling process, corresponding to the continuous increase in polarization. Figure 4(d) displays the value of R LLZTO , R Li/LLZTO , and R LCO-LLZTO/LLZTO by fitting the impedance data. It shows that only R Li/LLZTO of the three resistance components apparently increases as the galvanostatic cycle progresses, which is related to the acute deterioration of the Li/LLZTO interface. Conversely, the largely stable R LCO-LLZTO/LLZTO is because, firstly, the interface formed by integrated sintering is inherently more stable than that formed by attaching, and secondly, the charge cut-off voltage of 4.05 V sufficiently limits the volume change of LCO to prevent structural collapse. Therefore, the main reason for the performance decay of full cells is presumably the incremental impedance of the Li/LLZTO interface. Moreover, in practice, the batteries usually need to operate at a lower temperature, so the electrochemical performances of the Li|LLZTO|LCO-LLZTO full cells at 60 • C were also investigated. As shown in figure 4(e), the cell still exhibits stable cycling for more than 100 cycles at a current density of 50 µA cm −2 with an initial discharge capacity of 1.1 mAh cm −2 . The coulombic efficiency of the cell gradually increases from 71% to 99% in the first 15 cycles and then remains stable. After 50 cycles, the cell reveals a substantial capacity loss, with a retention rate of 60.2%. The reason is that lower temperature results in worse mass transfer kinetics of Li metal, making it less prone to creep to fill the voids at the Li/LLZTO interface. Meanwhile, the conduction of Li ions inside the cell is also retarded, eventually leading to worse electrochemical performance than that at 100 • C.
To further explore the evolution of electrode/electrolyte interfaces of the cell during cycling, we performed exsitu SEM tests on the same batch of high-capacity cells (∼1 mAh cm −2 ) with different cycle numbers at 60 • C. For LLZTO/anode interface, the Li metal exhibits intimate contact with the LLZTO electrolyte prior to cycling ( figure 5(a)). After 30 cycles, as shown in figure 5(b), there are several gaps between Li and LLZTO electrolyte, triggered by the accumulated stress due to the persistent uneven Li plating/stripping at the Li/LLZTO interface. The loss in the interface contact would accelerate the generation of an uneven electric field, causing the continuous deterioration of the Li/LLZTO interface. After 130 cycles, the Li/LLZTO interface shows severe degradation with the formation of large voids and the cell fails ( figure 5(c)). In comparison, there is no obvious change observed for the LCO/LLZTO interface, which is still in intimate contact even after 130 cycles ( figure 5(d)). The composite cathode layer maintains a dense state as well, where most particles show no obvious fractures (the inset of figure 5(d)). Besides, no impurity is generated in the composite cathode after 130 cycles, validated by the XRD pattern (figure S10).
These results indicate that the electrochemical reaction and the resultant volume change are uniform throughout the composite cathode, and the SPS process builds a durable SSE/cathode interface that keep unchanged and intact in structure during long-term cycling. Hence, it can be concluded that the performance degradation of the bulk-type Li|LLZTO|LCO-LLZTO ASSBs is primarily due to the continuous deterioration of the Li/LLZTO interface. Based on the previous electrochemical results (figures S8, 4(a)), generally, lowering the areal Li plating/stripping capacity could alleviate the deterioration of the Li/LLZTO interface, thus improving the cycling stability the ASSBs. To investigate the effects of cathode areal capacity on the electrochemical performance of Li|LLZTO|LCO-LLZTO ASSBs, we decreased the areal capacity to 0.49 mAh cm −2 . As expected, the cycling stability is significantly enhanced. It shows that the capacity retention is high up to 92.3% after 50 cycles ( figure S11(a)). And the Li metal anode still maintains intimate contact with LLZTO electrolyte ( figure S11(b)). The difference in behavior of the Li/LLZTO interface demonstrates the effectiveness of the low cathode capacity in lessening the contact loss between Li anode and LLZTO electrolyte, which forcefully proves that the stability of the Li/LLZTO interface plays a vital role in the cycling stability of the bulk-type Li|LLZTO|LCO-LLZTO ASSBs with high areal capacities. To date, various works have been reported to construct LLZO-based ASSBs, as shown in table 1. It is obvious that the Li|LLZTO|LCO-LLZTO full cell in this work demonstrates outstanding operating current density, areal capacity, and cycling stability. For pursuing better electrochemical performance, subsequent research endeavors can be made in improving the mechanical stability of a Li/LLZO interface during repeated high-capacity (>3 mAh cm −2 ) Li plating/stripping process, for example, combining 3D interface construction and multifunctional interlayers, and then adapting it to ASSB systems, thereby achieving high capacity and extended cycling.

Conclusion
In summary, a solid and durable cathode/garnet interface is successfully constructed by low temperature densification strategy using the SPS method with LBO as sintering additive. The intimate LCO-LLZTO/LLZTO interface and the cross-linked structure of the composite cathode are efficiently constructed and provide unimpeded ionic/electronic conducting pathways, ensuring high electrochemical activity under high areal capacity. As a consequence, the initial LCO-LLZTO/LLZTO interfacial impedance R LCO-LLZTO/LLZTO is remarkably reduced to 3.9 Ω cm 2 (100 • C), and the areal capacity is high up to 2.02 mAh cm −2 . EIS and SEM analyses reveal that the capacity decay of the bulk-type Li|LLZTO|LCO-LLZTO ASSBs is mainly attributed to the degradation of the Li/LLZTO interface induced by the uneven Li plating/stripping, which results in interfacial contact loss and continuous increase in interfacial impedance, while the LCO/LLZTO interface possesses high stability, and maintains intimate contact and extremely low interfacial impedance even after long-term operation. The deterioration of the Li/LLZTO interface can be mitigated via decreasing the areal capacity of Li plating/stripping. The Li|LLZTO|LCO-LLZTO full cells with areal capacities of 0.73 mAh cm −2 and 0.49 mAh cm −2 exhibit superior cycling stability with capacity retention of 81.7% and 92.3% after 50 cycles, respectively. Overall, our investigation proposes a promising methodology to address the key challenge of a cathode/LLZTO interface and provides guidance for future development of high-capacity garnetbased ASSBs.