Design of a lithiophilic and electron-blocking interlayer for dendrite-free lithium-metal solid-state batteries

Abstract

All-solid-state batteries are a potential game changer in the energy storage market; however, their practical employment has been hampered by premature short circuits caused by the lithium dendritic growth through the solid electrolyte. Here, we demonstrate that a rational layer-by-layer strategy using a lithiophilic and electron-blocking multilayer can substantially enhance the performance/stability of the system by effectively blocking the electron leakage and maintaining low electronic conductivity even at high temperature (60°C) or under high electric field (3 V) while sustaining low interfacial resistance (13.4 ohm cm²). It subsequently results in a homogeneous lithium plating/stripping, thereby aiding in achieving one of the highest critical current densities (~3.1 mA cm⁻²) at 60°C in a symmetric cell. A full cell paired with a commercial-level cathode exhibits exceptionally long durability (>3000 cycles) and coulombic efficiency (99.96%) at a high current density (2 C; ~1.0 mA cm⁻²), which records the highest performance among all-solid-state lithium metal batteries reported to date.

A layer-by-layer strategy in interfacial engineering can help to mitigate lithium penetration issues in all-solid-state batteries.

INTRODUCTION

With the advent of widespread electrified transportation, performance breakthroughs of batteries with an emphasis on high safety and high energy density are becoming increasingly important. In this respect, all-solid-state batteries (ASSB) are a promising alternative with exceptional safety features, which can benefit from the use of nonflammable solid-state electrolytes (1, 2). Moreover, a high energy density can be potentially achieved for ASSB by allowing the use of lithium-metal anodes, the ideal electrode material with the highest theoretical capacity (3860 mAh g⁻¹) and the lowest redox potential (0 V versus Li/Li⁺) for lithium batteries (2, 3). The practical deployment of ASSB has become closer to reality with the recent discoveries of solid-state electrolytes that could exhibit sufficiently high ionic conductivities comparable to those of commercial liquid electrolytes in lithium-ion batteries (1, 4, 5). In particular, garnet-type Li₇La₃Zr₂O₁₂ electrolytes showed a great promise as a near-future solid electrolyte owing to their high ionic conductivity [~1 mS cm⁻¹ with tantalum (Ta) doping at room temperature] and physical properties (e.g., air stability and mechanical robustness), offering processing feasibility (5–7). Moreover, the electrochemical and chemical compatibility with lithium metal further makes the Li₇La₃Zr₂O₁₂ electrolyte system suitable for high-energy lithium metal ASSB (5, 8).

Nonetheless, practical challenges remain in deploying the Li₇La₃Zr₂O₁₂ electrolyte in lithium-metal batteries. One of the most critical difficulties is that it is vulnerable to the short circuits caused by the penetration of lithium metal dendrites upon repeated deposition/stripping. It has been observed that both polycrystalline and single-crystalline Li₇La₃Zr₂O₁₂ electrolytes are penetrated by lithium dendrites even at low current densities of 0.1 to 1.0 mA cm⁻² at room temperature, which falls far behind the actual battery operation criteria for commercialization (9–11). It is an unexpected contrast to the common expectation that solid-state electrolytes would be mechanically robust and thus suppress the issue of the dendritic lithium short-circuiting, considering their higher elastic moduli (e.g., ~55 GPa for Li₇La₃Zr₂O₁₂) than that of ductile lithium metal (~5 GPa) (7, 10).

In recent years, many researchers have witnessed that the lithium dendrite issue is persistent regardless of the type of solid electrolytes or their processing conditions, and it can occur due to several potential reasons (10–13). One of the most commonly attributed factors is the interfacial inhomogeneity that originates from microstructural imperfections at the physical boundary between the lithium metal and the solid electrolyte, where the current density can be localized to be excessively high due to uneven electrical constriction resistance (12). The high local current density at the interface is likely to trigger the preferred growth of lithium metal at the hot spot, as schematically illustrated in Fig. 1A. As a consequence of inhomogeneous lithium deposition, pressure builds up locally at the contact point, and the current-induced intragranular fractures can develop and propagate with the lithium metal filling. Previous studies using ex situ and in situ macroscopic measurements demonstrated that intragranular fractures are involved in cell short circuits (11, 14). In addition, the grain boundaries of the solid electrolyte are microstructural imperfections susceptible to lithium dendrites due to the variations in atomic structure and density at the grain boundaries. It has been theoretically and experimentally demonstrated that the elastic modulus at the grain boundaries in the solid electrolyte is ~50% less than that of the grain, leading to the intergranular crack and lithium dendritic growth along the grain boundaries/interconnected pores in the pellets (15, 16). In this regard, previous efforts have been placed in enhancing the interfacial homogeneity by removing impurities at the interface (17, 18) or improving the wettability of lithium metal by inserting lithiophilic interlayers (19–25). Various lithiophilic interlayers were thus explored such as materials that can alloy with lithium (e.g., Ge or Ga) (20, 24) or have conversion reactions with lithium metal [e.g., Zn(NO₃)₂ or SnF₂] (19, 21), which could notably improve the performance. Nevertheless, the short circuits caused by lithium dendrites still happen after prolonged cycling or when operated at a moderately high current density. Another important factor that has been more recently attributed to is the nonnegligible electronic conductivity of the solid electrolyte itself. It has been theoretically and experimentally verified that the common solid electrolytes also exhibit nonnegligible electronic conductivities, and, more importantly, the electronic conductivity near grain boundaries or defects can be significantly higher; therefore, these imperfections can serve as electron-conducting pathways (26–29). It was further demonstrated that the electronic conductivity of garnet-type electrolytes can surge when exposed to high applied potentials or temperature, showing the electric breakdown properties at certain threshold voltages (30) and temperature, implying the risk of electrical leakage through the solid electrolyte at fast charging/discharging or high-temperature operating conditions. The presence of the electron-conducting pathways inside the solid electrolyte suggests the feasibility of the lithium metal nucleation and growth preferentially along these paths, consequently incurring a risk of short circuit (31, 32).

Fig. 1. Schematic illustrations of the interface between lithium metal and garnet-type solid electrolyte.

(A) Lithium dendrite formation at the conventional interface between lithium metal and the sintered solid electrolyte. (B) Interfacial design coupling with lithiophilic and electron-blocking interlayers. The schemes at the bottom show enlarged views of the interface in each case.

These previous findings indicate that the attributing factors of the lithium dendrite issue in ASSBs should be taken as orthogonal problems, and each issue should be dealt with a separate strategy. While various lithiophilic interlayers could improve the contact issue to some extent, it could not mitigate the electronic leakage from the electrode to the electrolyte, as many of the previous lithiophilic interlayers are either metallic or contain mixed ionic and electronic conductors after the reaction with lithium. Here, we attempt to resolve these orthogonal problems by designing a layer-by-layer assembly with a lithiophilic layer and an electron-blocking layer, simultaneously overcoming the poor interfacial contact and the electron leakage problems, as shown in Fig. 1B. The presence of the lithiophilic layer would aid in homogenizing the lithium-ion flux and maintaining good wettability, whereas the electron-blocking layer would fundamentally passivate the electron transport pathway at the interface, thereby prohibiting the electron leakage–induced lithium nucleation inside the solid electrolyte. We show that this multi-interlayer strategy is remarkably effective; the critical current density (CCD) of the lithium symmetric electrodes increases to a record-high 3.1 mA cm⁻² at 60°C, and the lithium full cell paired with the commercial LFP cathode exhibits a stable cycle performance over 3000 cycles at a high current density of 1.0 mA cm⁻², which is the highest value reported to date. Our findings indicate that the combined strategy to regulate both lithium and electron transport is essential to address the multi-origin lithium dendrite issues. Moreover, this approach of layer-by-layer assemblies is extendable to various combinations of lithiophilic and electron-blocking layers, which would offer avenues to overcome the major bottlenecks toward the practical deployment of all-solid-state lithium-metal batteries.

RESULTS

Physical/electrical properties of the layer-by-layer deposits at the interface

In our layer assembly, metallic silver (Ag) was chosen as a material for a lithiophilic layer to ensure homogeneous interfacial contact with good wettability. It can readily alloy with lithium (33) and offer the high lithium diffusivity in the Li-Ag alloy, which is reported to be significantly higher (~10⁻⁶ cm² s⁻¹) than that in the pristine lithium metal (5 × 10⁻¹¹ cm² s⁻¹) (34–36). Considering that the low self-diffusion rate of lithium has been ascribed to cause the void formation/inhomogeneity at the interface (13), the enhanced lithium mobility at the interlayer would aid in stably maintaining the wetted/uniform interface between lithium and the solid electrolyte. Lithium fluoride (LiF) layer was selected as an electron-blocking interlayer because of its low electronic conductivity (~10⁻¹⁰ S cm⁻¹) (37, 38) and the proven electrochemical compatibility with lithium metal as a common constituent of solid electrolyte interphase in conventional lithium-ion batteries (39, 40). We prepared Ta-doped Li_6.4La₃Zr_1.4Ta_0.6O₁₂ (LLZTO) solid electrolyte via conventional solid-state synthesis. In addition, the electron-blocking LiF and the lithiophilic Ag interlayers were deposited, respectively, on the surface of the LLZTO pellet using a thermal evaporator (see Materials and Methods for details). The LiF layer was first coated on the polished surface of the LLZTO pellet, followed by the deposition of Ag to be contacted with lithium metal anode as a lithiophilic layer. According to the x-ray diffraction (XRD) analysis in fig. S1, the as-prepared LLZTO was successfully synthesized as a cubic phase garnet (31) without any noticeable impurities such as Li₂CO₃. For reference, the ionic and electronic conductivities of the pristine LLZTO were measured by the electrochemical impedance spectroscopy (EIS) spectra and direct-current (DC) polarization curves and were observed to be 0.45 mS cm⁻¹ and 5.00 × 10⁻⁹ S cm⁻¹ at room temperature, respectively, consistent with previously reported values for garnet electrolytes (fig. S1) (21, 26).

Figure 2 (A to D) displays the cross-sectional image of the LLZTO surface coated with the Ag/LiF interlayers scrutinized by transmission electron microscopy (TEM) coupled with energy-dispersive spectroscopy (EDS). It illustrates that two separate layers of LiF and Ag were subsequently deposited on the LLZTO with a thickness of ~300 nm. The feature that needs to be noted is that the fluorine (Fig. 2C) is distributed over a wider region than the apparent LiF layer, particularly toward the LLZTO side. We suppose that, considering the principle of the thermal evaporator—the source is vaporized and condensed back to the solid state on the target surface—the vaporized LiF could partially fill the micropore or imperfections on the near-surface region of the LLZTO pellet. In this regard, we further investigated the interface between LiF and LLZTO using high-resolution TEM (HRTEM), fast Fourier transform (FFT) patterns, and the corresponding inverse FFT (IFFT) image.

Fig. 2. Morphological and elemental characterization of the layer-by-layer deposition at the interface.

(A) High-angle annular dark-field scanning TEM image of the interfacial morphology of Ag/LiF-coated LLZTO at low magnification. Corresponding elemental mappings of (B) La, (C) F, and (D) Ag of (A). (E) HRTEM image of the interface between LLZTO and LiF. Top right region corresponds to LiF deposits, while the bottom left region is the LLZTO. (F) Corresponding FFT patterns of LiF, LLZTO 1, and LLZTO 2 regions. (G) Corresponding IFFT image of the LiF region. (H) XPS spectra of Ag/LiF-coated LLZTO pellet at F 1s and Ag 3d. a.u., arbitrary units.

The HRTEM image in Fig. 2E shows that even the rough LLZTO surface could be well infused by the LiF. The FFT images generated for the LLZTO 1 and 2 regions (yellow boxes) confirm the appearance of the major (211), (400), (420), and (422) crystal planes of cubic garnet structure in Fig. 2F (41), corresponding to interplanar spacings of 0.528, 0.325, 0.289, and 0.264 nm, respectively. In the LLZTO 1 region, the (200) crystal planes of LiF grains were also observed, which supports the infiltration of LiF into the rough surface of the LLZTO. The interplanar spacing of 0.206 nm in the FFT pattern is well matched with the (200) lattice of cubic LiF (Fig. 2G and fig. S2) (39). The LiF region was found to consist of (i) nanoparticles with a size of ~10 nm (indicated by green boundaries) and (ii) a surrounding amorphous matrix. The effect of this unique structure of the LiF layer on the electron-blocking and the lithium-transporting function will be discussed in later sections. The x-ray photoelectron spectroscopy (XPS) analysis on the pellet surface also verifies the characteristic peak of the Li─F bond at a binding energy of 684.5 eV in the F 1s spectrum, as provided in Fig. 2H (left) (40). The two strong peaks at 367.9 and 373.9 eV are attributed to Ag 3d_5/2 and Ag 3d_3/2, respectively (33), confirming the coverage of the Ag/LiF multilayer on the LLZTO solid electrolyte.

To investigate the electron-blocking effect of the LiF layer, we comparatively measured the electronic conductivities of bare and LiF-coated LLZTO at various conditions as shown in Fig. 3 (A to C). Figure 3A depicts the current-time profiles obtained by applying DC polarization for the two samples with lithium-ion–blocking Ag/In electrodes at room temperature (additional tests conducted at other conditions are also provided in fig. S3) (26, 42). It shows that the current decreases and reaches the steady state over the time when a constant voltage (1 V) is applied in the experiment. Considering that there is no ionic transport due to the ion-blocking electrode, the measured current is mainly attributable to the electronic conduction. The analysis reveals that the electronic conductivity of LiF-coated LLZTO (5.48 × 10⁻¹⁰ S cm⁻¹) is approximately one order of magnitude lower than that of the bare LLZTO (5.00 × 10⁻⁹ S cm⁻¹) at room temperature due to the insulative nature of LiF layer. The low electronic conductivity of LiF-coated LLZTO could be maintained at higher applied voltages and/or temperature conditions that represent the actual cell operation circumstances. Figure 3B shows that the electronic conductivity rapidly increases when higher DC voltages are applied to the bare LLZTO pellet, surging from 1.8 × 10⁻¹⁰ S cm⁻¹ at 0.5 V to 6.9 × 10⁻⁹ S cm⁻¹ at 3.0 V, which agrees with the previous reports (43). In contrast, that of the LiF-coated LLZTO remained relatively constant up to a DC of 3.0 V, exhibiting a ninefold lower value than that of the bare LLZTO at the same condition. Figure 3C illustrates that the bare LLZTO also suffers from the electrical breakdown at a high temperature, displaying a sharp upsurge in the electronic conductivity at 60°C, which becomes 12 times higher than the value at room temperature. It is contrary to the case of the LiF-coated LLZTO that exhibits an only marginal increase of the electronic conductivity with higher temperatures (e.g., 3.0 × 10⁻⁹ S cm⁻¹ at 60°C), demonstrating the efficacy of the LiF insulating layer in hindering the electronic conduction through the LLZTO solid electrolyte.

Fig. 3. Physical and electrical properties of the LLZTO electrolyte with/without interlayer.

(A) Current-time profiles of LLZTO and LiF-coated LLZTO using Ag/In blocking electrodes under 1 V at room temperature. (B) Electronic conductivity of LLZTO and LiF-coated LLZTO using Ag/In blocking electrodes as a function of the external voltage and (C) temperatures. (D) Digital image of the half-area coated by the LiF layer and the other half-area coated by the Ag/LiF layer on the surface of the LLZTO electrolyte (left). Photograph of the surface of LLZTO after dipping in molten lithium for a few seconds, as depicted in the inset (right). (E) Cross-sectional SEM image of the interface between LLZTO and Li metal. (F) Comparison of the EIS spectra of bare, Ag-coated, and Ag/LiF-coated LLZTO with lithium symmetric cells.

These quantitative measurements of electrical conductivities confirm the risk of the nonnegligible leakage current in the conventional LLZTO electrolyte at the practical operational window of voltages and temperatures (28, 30). In addition, note that the presence of the thin LiF layer could effectively mitigate the increase of the electronic conductivity at these conditions simply by regulating the electron transport to the solid electrolyte at the interface. We also found that, despite the effective electron-blocking role of the LiF layer, the reduction in the ionic conductivity was not significant. The EIS measurement revealed that the total ionic conductivity decreases from 0.45 to 0.35 mS cm⁻¹ at room temperature with the LiF coating on the LLZTO (fig. S4). The slight reduction is ascribed to a lower ionic conductivity of the LiF phase compared with the LLZTO electrolyte; however, considering the intrinsic ionic conductivity reported for the crystalline LiF (~10⁻³¹ S cm⁻¹) (44), it is regarded as a relatively small decrease. We suppose that it is partly due to (i) the unique LiF film structure that is primarily composed of the low-crystallinity LiF amorphous matrix with LiF nanoparticles embedded, as evidenced by the TEM observations in Fig. 2E, and (ii) the thickness of the LiF film (~200 nm), which requires only short-length lithium ionic diffusion. According to previous literatures (45, 46), the ionic conductivity can be enhanced by partly amorphizing ionically insulating crystalline materials, inferring that the LiF amorphous matrix could allow a more facile lithium ionic transport than that expected for the crystalline LiF at the interface.

The Ag layer was subsequently deposited on the LiF-coated LLZTO, and its lithiophilic nature was examined in Fig. 3D. The figure depicts a model LLZTO system, where half of the LiF-coated LLZTO pellet was selectively coated with the Ag layer, to comparatively visualize the enhancement of the wettability. It shows that when the pellet is immersed in molten lithium metal for several seconds and is retrieved, the molten lithium completely wets on the Ag/LiF-coated side, whereas almost no lithium is observable on the LiF-coated side, indicating the improved surface-wetting properties by the Ag layer. The cross-sectional scanning electron microscopy (SEM) image of the Ag/LiF-coated pellet wetted with lithium (Fig. 3E) also reveals that the lithium and the LLZTO surface have formed an intimate contact without any microscale gaps at the interface, as indicated with yellow dashed lines. Accordingly, the interfacial resistance could be significantly reduced in the Li symmetric cell as comparatively shown in the Nyquist plots for the bare, Ag-coated, and Ag/LiF-coated LLZTO cases in Fig. 3F. It illustrates that the semicircles at the low-frequency region become substantially smaller with the Ag coating of the pellet regardless of the presence of LiF coating than that of the bare LLZTO pellet. The interfacial area-specific resistance (ASR) markedly decreases when the silver is coated—from 254.5 ohm cm² for the bare LLZTO cell to 28.6 and 5.8 ohm cm² for Ag-coated and Ag/LiF-coated LLZTO cells, respectively, indicating the enhanced interfacial properties, while the bulk and grain boundary ASR of all the cells were similar in the high-frequency region with 22.0, 48.5, and 39.1 ohm cm², respectively (47).

Enhanced electrochemical performance from the layer-by-layer strategy

The efficacy of the interlayer was scrutinized regarding the lithium dendritic formation in the lithium symmetric cell using LLZTO electrolytes under various testing conditions. Figure 4 (A and B) and fig. S5 display the results of the CCD tests measured at 60°C and room temperature, which evaluate the maximum current density before the voltage drop occurs by short circuits due to the dendritic lithium growth, for Ag-coated and Ag/LiF-coated LLZTO cells, respectively. For reference, the result for the bare LLZTO cell is also provided in the Supplementary Materials (fig. S5), which recorded the CCD value of 0.3 mA cm⁻², consistent with previous reports (48). Our results show that the short circuit of the Ag-coated LLZTO cell occurs at a current density of 1.5 mA cm⁻² upon increasing the current up to 3.2 mA cm⁻² with a step of 0.1 mA cm⁻² (Fig. 4A). It is a significantly greater CCD value over the bare LLZTO cell, which is probably due to the enhanced interfacial property by the Ag interlayer in the cell, as previously reported for the similar lithium-alloying metallic interlayer such as Al (49). The lithium symmetric cell with Ag/LiF-coated LLZTO electrolyte could retain the cell stability maintaining a low polarization even at high current densities, delivering a remarkably high CCD value of 3.1 mA cm⁻², as shown in Fig. 4B. To the best of our knowledge, the CCD of 3.1 mA cm⁻² at 60°C is the highest value obtained for garnet-based solid electrolytes that have been reported thus far (22, 24). Similarly, the CCD of the bare and Ag-coated LLZTO case was 0.15 and 0.25 mA cm⁻², whereas the CCD of the Ag/LiF-coated LLZTO case was significantly improved to 0.75 mA cm⁻² at room temperature (fig. S5).

Fig. 4. Electrochemical performance of the solid-state lithium symmetric cell and ex situ analysis of lithium deposits.

CCD of (A) Ag-coated and (B) Ag/LiF-coated LLZTO in lithium symmetric cells at 60°C with increasing current densities ranging from 0.1 to 3.2 mA cm⁻² at a step size of 0.1 mA cm⁻². (C) Ex situ cross-sectional SEM image of the fractured Ag-coated LLZTO observed by BSD before the electrical short-circuiting. (D) Enlarged image of Ag-coated LLZTO with dark spots highlighted with yellow arrows at higher magnification. (E) Ex situ cross-sectional BSD image of a fractured Ag/LiF-coated LLZTO before the electrical short-circuiting. (F) Enlarged image of Ag/LiF-coated LLZTO with pores highlighted with green arrows at higher magnification.

Although the Ag-coated LLZTO exhibited a low interfacial resistance and thus could maintain small polarization initially in the CCD test, the short circuit was inevitable later at an elevated current density (>1.5 mA cm⁻²). We suppose that it is attributed to the nonnegligible electronic conduction occurring in the Ag-coated LLZTO when applied with higher current rates at this temperature (60°C), as previously demonstrated in Fig. 3B. It hints at the importance of the LiF coating layer in keeping the low electronic conduction in the LLZTO electrolyte even at high current density/temperature conditions, and thereby delaying the lithium dendritic growth. Our speculation could be supported by ex situ backscattered electron detector (BSD) SEM analysis on the cross section of the LLZTO samples that were retrieved after the CCD tests. Figure 4 (C to F) depicts the cross-sectional images of Ag-coated (Fig. 4, C and D) and Ag/LiF-coated (Fig. 4, E and F) LLZTO pellets after the CCD test at 0.2 mA cm⁻². The bright and dark BSD images correspond mainly to the region with high–atomic number elements (e.g., Zr and La) and that of low–atomic number elements (e.g., Li), respectively, which enables probing the potential nucleation of metallic lithium inside the sample (29). Figure 4 (C and D) evidently illustrates that several black spots are extensively distributed inside the Ag-coated LLZTO electrolyte. On a closer look at these spots (yellow arrows in Fig. 4D), it is revealed that they are clearly distinguishable from the micropores observed in Fig. 4F (indicated with green arrows) for Ag/LiF-coated LLZTO or the pristine LLZTO in fig. S6, signifying the growth of lithium. The blurring boundary of the spots also strongly indicates that they are the defocused images of the lithium protrusion, which has been frequently observed in similar previous works (26, 29). It is notable that, even before the actual short circuit takes place at the current density of 0.2 mA cm⁻², several micrometer-sized lithium islands have already grown inside the Ag-coated LLZTO solid electrolyte. The absence of such signature in the Ag/LiF-coated LLZTO proposes that the electron-blocking LiF interlayer could suppress the electron transport toward the LLZTO electrolyte, which could ultimately delay the nucleation of metallic lithium in the cell.

Inspired by the significant enhancement of the CCD performance, we comparatively evaluated the galvanostatic cycling stabilities of lithium symmetric cells using LLZTO electrolytes with interlayers. Figure 5A presents the cycling curves of plating/stripping processes for lithium cells with Ag-coated and Ag/LiF-coated LLZTOs, respectively, at 0.2 mA cm⁻² at room temperature. The cell using the Ag/LiF-coated LLZTO electrolyte shows markedly robust stability for more than 600 hours without a noticeable increase in the overpotentials throughout the plating/stripping cycles, which contrasts to the Ag-coated LLZTO-based cell that displayed a cell failure after tens of cycles. In addition, to scrutinize the stability of the interlayer after cycling, we disassembled the cells before the short circuits occurred (∼10 hours of galvanostatic cycling at 0.2 mA cm⁻² at room temperature) and examined the interfaces, as shown in fig. S7. The cross-sectional SEM image of the Ag/LiF-coated LLZTO shows that the interface between lithium and LLZTO remained stable even after repeated lithium plating/stripping processes (fig. S7A). The high-resolution EDS line scan in fig. S7B indicates that the F-rich layer is stably located at the interface between the lithium electrode and LLZTO electrolyte, whereas the Ag-rich layer has been slightly diffused into the bulk lithium metal to some extent, indicating the presence of the Li-Ag alloy. According to our previous study on the interface of a bare LLZTO cell after cycling (21), significant interface degradation was observed in the uncoated LLZTO cell, where pores and voids were found at the interface and within the solid electrolyte. In contrast, the Ag/LiF-coated LLZTO maintained the initial interface structure, showing the good wetting with lithium and a stable LiF interlayer without delamination or the apparent defect formations. These results indicate that the Ag and LiF interlayers contributed to the stability of the interface during repeated stripping and plating of lithium. The symmetric lithium cell with Ag/LiF-coated LLZTO could also exhibit a steady cycle performance at a higher current density of 0.5 mA cm⁻² for more than 130 hours with the overpotential remaining below 0.1 V (Fig. 5B). We also examined the cycling performance of Ag/LiF-coated LLZTO symmetric cell at a higher current density of 1.0 mA cm⁻² at 60°C, considering the high CCD value at 60°C (fig. S8). It revealed that the symmetric lithium cell with Ag/LiF-coated LLZTO maintained steady cycling performance at a higher current density of 1.0 mA cm⁻² while keeping the overpotential below 22 mV. On the basis of these results, we fabricated hybrid full cells with a commercial LFP cathode (a mixture of LFP powder, carbon, and polyvinylidene fluoride binder in an 8:1:1 ratio; see the Materials and Methods for details) to verify the performance of the lithium metal anode–LLZTO electrolyte couple in the full cell geometry (21, 49). Note that a small amount of ionic liquid (2 M LiFSI in Pyr13FSI) was applied to the cathode mixture to ensure the lithium-ion transport among LFP powders and the contact with the LLZTO electrolyte, but the lithium anode side is still free of the liquid phase. Figure 5C demonstrates that the hybrid full cell can deliver a decent power capability; a discharge capacity of 124.7 mAh g_cathode⁻¹ can be still deliverable, when the current density increases to a 5C_cathode rate (2.5 mA cm⁻² for the lithium anode) at 60°C. This high-rate capability is a clear contrast to the bare and Ag-coated LLZTO hybrid full cells as shown in Fig. 5D, which exhibited commonly lower discharge capacities of the LFP cathode, and could not operate at current rates higher than 0.5C_cathode (bare) and 2C_cathode (Ag-coated LLZTO), respectively. The premature failures of these cells at 0.5C_cathode (bare) and 2C_cathode (Ag-coated LLZTO) are attributable to the high current densities applied to the lithium anode sides (e.g., 0.25 and 1 mA cm⁻², respectively), which exceed the CCD values of the respective LLZTO electrolytes. It manifestly emphasizes the importance of the high CCD value in achieving the decent power capability of a solid-state full cell, which is only attainable with the low interfacial resistance at the lithium/LLZTO and the low electronic conductivity of LLZTO under high currents/temperatures.

Fig. 5. Electrochemical performances of lithium symmetric cells and hybrid solid-state full cells using commercial cathodes.

(A) Galvanostatic cycling stability of Ag-coated and Ag/LiF-coated LLZTO at a current density of 0.2 mA cm⁻² for 0.5 hours at room temperature in lithium symmetric cells. (B) Cycling performance of a Ag/LiF-coated LLZTO symmetric cell at a current density of 0.5 mA cm⁻² for 0.25 and 0.5 hours at room temperature. (C) Voltage profiles of Li|Ag/LiF-coated|LFP hybrid solid-state full cells at 60°C at current densities of 0.2, 0.5, 1.0, 2.0, and 5.0 C (C rate here refers to the current rates applied to a cathode, i.e., C_cathode). (D) Rate capability tests for bare, Ag-coated, and Ag/LiF-coated LLZTO hybrid cells at 60°C. The bare and Ag-coated LLZTO hybrid full cells could not operate at current rates higher than 0.5C_cathode (bare) and 2C_cathode (Ag-coated LLZTO), respectively, due to the premature cell failures. (E) Cycle performance of Li|Ag/LiF-coated LLZTO|LFP cells at 60°C. The yellow and dark gray lines correspond to the operation at 1.0 C (~0.5 mA cm⁻²). The green and light gray lines correspond to the operation at 2.0 C (~1.0 mA cm⁻²). (F) Cycle performance of Li|Ag/LiF-coated|NCM111 cells at 0.5 C (~0.25 mA cm⁻²) at room temperature. (G) Cycle performance of our LFP|Li hybrid full cells and its comparisons with recent literature reports regarding various interfacial modifications for similar hybrid full cells as a function of the current density. The references for the reports are shown in table S1.

In Fig. 5E, we further investigated the extended cycle performance and the corresponding coulombic efficiency of the hybrid solid-state full cell using the LFP cathode/lithium anode and the Ag/LiF-coated LLZTO electrolyte. It shows that it can retain a discharge capacity of 141.4 mAh g⁻¹ and a coulombic efficiency of 99.7% over 2600 cycles under 1C_cathode (0.5 mA cm⁻² for lithium anode) at 60°C (green line). More remarkable is that the superior cycle stability could be achieved even at a higher rate of 2C_cathode (1.0 mA cm⁻² for lithium anode) over 3000 cycles, which, to the best of our knowledge, records one of the highest cycle stabilities reported to date for cells using lithium anode and LLZTO solid electrolyte. To further verify the general applicability of our lithium metal/LLZTO couple, we fabricated a hybrid full cell with a commercial LiNi_1/3Co_1/3Mn_1/3O₂ cathode. Figure 5F shows that respectable cycling stability could still be achieved with the representative layered cathode at the voltage range of 2.8 to 4.2 V, which successfully retained a discharge capacity of 135.2 mAh g⁻¹ at 0.5C_cathode (0.25 mA cm⁻² for lithium anode) at room temperature after 1700 cycles. For reference, we plotted the relative performance standing of our hybrid full cell in comparison with those of recently reported relevant solid-state cells that have gone through various interfacial modifications (e.g., the introduction of interlayers or removal of interfacial impurities as denoted in the figure) with respect to two key demonstrated indicators of current density and long-term cycling in Fig. 5G (17, 18, 20–25, 48, 50–55) It clearly manifests that the modification with the Ag/LiF multilayer in the current work excels the previous reports, not only in the high-rate capability (2C_cathode) but also in the long-cycle durability (over 3000 cycles). It is our belief that it is attributed to the orthogonal multifunctions of the Ag/LiF layer, which has enabled the homogeneous interfacial property during charge/discharge and allowed the operation of the lithium cell even at high current density by preventing the electronic leakage through the cell.

DISCUSSION

We proposed a layer-by-layer strategy that adopts a lithiophilic and electron-blocking multilayer to simultaneously achieve homogeneous lithium-ion conduction at the interface of lithium/solid electrolyte and block the undesirable electron leakage across the solid electrolyte. It was demonstrated that the lithiophilic Ag layer offers a remarkably low interfacial resistance of 13.4 ohm cm² by enhancing the wettability of lithium metal on the LLZTO solid electrolyte, whereas the electron-blocking LiF layer of the unique composite structure could mitigate the electronic conduction through the solid electrolyte even at high temperature/high voltage while maintaining the sufficiently high lithium-ion conductivity. Our orthogonal approach was verified to be successful; the lithium symmetric cell using the multi-interlayer exhibited a record-high CCD of 3.1 mA cm⁻² at 60°C, and the hybrid full cell made with a commercial LFP cathode and LiNi_1/3Co_1/3Mn_1/3O₂ cathode could show the outstanding long-term durability (>3000 cycles) at a high current density (2 C; ~1.0 mA cm⁻²), which is the highest performance reported to date. Our discovery indicates that the combined strategy to regulate both lithium and electron transport is indispensable to address the multi-origin lithium dendrite issues, opening an avenue toward the future solid-state batteries by the layer-by-layer approaches through the exploration of various combinations of lithiophilic and electron-blocking layers.

MATERIALS AND METHODS

Fabrication of LLZTO pellets

The Ta-doped LLZTO powders were synthesized using a solid-state synthesis method with LiOH·H₂O (99.99%; Sigma-Aldrich), La₂O₃ (99.99%; Sigma-Aldrich), Ta₂O₅ (99.99%; Sigma-Aldrich), and ZrO₂ (99%; Sigma-Aldrich). Note that 20 weight (wt) % excess LiOH·H₂O was added to compensate for the volatile Li components during the sintering process. We combined the precursors at stoichiometry and calcined them at 900°C for 12 hours in an alumina crucible to form a cubic LLZTO phase. The obtained LLZTO powders were ball-milled at 200 rpm for 10 hours to attain a fine powder with a particle size of 5 to 10 μm. We pressed them into 10-mm-diameter pellets, which were sintered at 1100°C for 10 hours in air with 0.2 wt % γ-Al₂O₃ (99.9%; Sigma-Aldrich) as the sintering agent. The obtained pellets were polished with 600, 1500, and 2000 grit sandpaper to clean the surface and remove possible surface impurities. The final thickness of the LLZTO pellets was approximately 700 μm, and their relative density was approximately 93%.

Deposition of Ag/LiF multi-layer

Both Ag and LiF layers were deposited on the LLZTO pellets using a vacuum thermal evaporator integrated in a glove box (Korea Vacuum Tech) with a base pressure of <1 × 10⁻⁶ torr. A LiF thin film with a thickness of 15 or 150 nm was evaporated/deposited on the surface of LLZTO pellets at a deposition rate of 0.5 Å s⁻¹ and was further annealed at 600°C for 2 hours in an Ar atmosphere to improve the contact between the LiF film and LLZTO pellet. The Ag thin film with a thickness of 100 nm was evaporated/deposited on the LiF-coated surface of the LLZTO pellet at the same deposition rate.

Material characterization

Structural characterization of the LLZTO samples was conducted using XRD (D8 Advance, Bruker) with Cu Kα radiation in the 2θ range of 10° to 60° with a step size of 0.02°. The phase of the cubic LLZTO phase was examined using the FullProf program in fig. S1. The surface and cross-sectional morphologies of the pellets were investigated using SEM (SUPRA 55VP, Carl Zeiss) with secondary electron imaging and backscattered electron detection (BSD) imaging. To investigate the cross-sectional microstructure of the multilayers, a TEM sample was prepared using focused ion beam SEM (Helios G4 UC, Thermo Fisher Scientific) with carbon and Pt protective layers. HRTEM (JEM-2100F, JEOL Ltd.) and FFT analysis were performed to identify the crystallinity of the coating layers and LLZTO. To characterize the chemical composition of the multilayer, XPS (Nexsa, Thermo Fisher Scientific) analysis was carried out with monochromatic Al Kα radiation (1486.6 eV) without any exposure to moisture and air. The XPS spectra were calibrated based on the C 1s peak at 284.5 eV, which corresponds to C─C bonding.

Electrochemical measurements

To evaluate the ionic and electronic conductivities of the LLZTO pellets, Ag and In electrodes were used as a lithium-ion blocking electrode, and the as-assembled In/Ag/LLZTO/Ag/In cells were heated at 120°C to enhance the contact between the electrode and LLZTO. The EIS measurements to determine the ionic conductivity were performed using a potentiostat (VSP-300, Bio-Logic Science Instruments) with an AC amplitude of 10 mV in the frequency range from 3 MHz to 100 mHz at room temperature. The DC polarization for the electronic conductivity measurement of the LLZTO was conducted at various applied voltages from 0.1 to 2.5 V for 1 hour and with different temperatures.

The lithium symmetric cells were fabricated by attaching 300-μm-thick lithium foil with a diameter of 5 or 9 mm on both sides of the LLZTO pellets and assembling 2032-type coin cells. To reinforce the good contact with Li metal, the as-assembled coin cells were placed in an oven at 120°C overnight. The galvanostatic tests were performed using a potentio-galvanostat (WBCS 3000, WonATech). The interfacial ASR was fitted with an equivalent circuit in fig. S4 and calculated for the fitted resistance by dividing two of the symmetric lithium cells and normalizing with the electrode area (0.15 cm²).

To fabricate the hybrid full cells, the LFP composite cathode was prepared with a mixture of LFP powder, Super P carbon, and polyvinylidene fluoride binder in a weight ratio of 80:10:10 with N-methyl-2-pyrrolidone solvent. The obtained slurry was coated on Al foil and dried at 70°C in a vacuum oven. The areal loading mass of active materials was 3 to 3.5 mg cm⁻². A small amount of ionic liquid (2 M LiFSI in Pyr13FSI) was applied to improve the interfacial contact between the cathode particles and LLZTO.

Supplementary Materials

This PDF file includes:

Figs. S1 to S8

Table S1