Ultra-Thin Lithium Silicide Interlayer for Solid-State Lithium-Metal Batteries

charged at a current of 0.92 mA cm


Introduction
The demand for safe rechargeable batteries with high energy density is evergrowing. [1] To overcome the low theoretical capacity of graphite (372 mAh g −1 ) in lithium-ion batteries (LIBs), attempts have been made to use metallic lithium in the body-centered cubic crystal (Li BCC ) phase as an anode material due to its high theoretical specific capacity (3860 mAh g −1 ) and low reduction potential (−3.04 V vs standard hydrogen electrode). [2] However, irregular growth of Li BCC in the plating process can penetrate through the separator, bringing about short-circuiting and eventually fire and explosion. [3] The use of liquid electrolytes with volatility and flammability aggravates safety concerns. [4] Even if short-circuiting was no concern, low Coulombic efficiency limits the cyclability of the full-cell. This is probably related to not only side reactions at the Li BCC /electrolyte contact during Li BCC deposition but also Li BCC stripping-induced true-contact-area reduction that often becomes limiting kinetically. Since almost all Li + -conducting electrolytes are thermodynamically unstable when in contact with Li BCC , the question is how to prevent the Li BCC /electrolyte true contact area (and thus side reactions) from growing limitlessly (as in liquid electrolytes) while maintaining sufficient true contact area as well as connectivities for long-range ionic and electronic percolation. This is a key design problem that has to do with both stress and corrosion. Accordingly, all-solid-state battery (ASSB) configurations adopting solid electrolytes have been anticipated to be fundamentally different from liquid-electrolyte-based batteries in both the stress and corrosion aspects by virtue of their load-bearing capability and lack of fluidity. As such, the architectural design of ASSB is highly consequential.
Among a wide range of solid electrolytes (SEs), sulfide-based SEs have captivated researchers because it features high ionic conductivity of 10 −2 to 10 −4 S cm −1 and mechanical ductility that enables the fabrication of densified electrodes on a large scale via a simple roll-to-roll slurry coating and calendering process. [5] That being said, argyrodite SE has severe intrinsic limitations from both thermodynamic and mechanical considerations. Thermodynamically, the reductive decomposition (side reaction) at the interface with Li BCC (originating from its narrow electrochemical stability window [6] ) reduces the meter) can just displace and penetrate the soft argyrodite, like a hypodermic needle can penetrate the soft skin since, in deposition, a large mechanical pressure can be generated. [9] Such high Li Metal hardness-induced mechanical penetrations can increase the side-reaction area, thereby further reducing Coulombic efficiency and eventually resulting in internal short-circuiting. To address these concerns, various strategies, such as hybrid electrolytes [10] (Li 6 PS 5 Cl/poly(ethylene oxide) composite), alloys [11] (In-Li and Ag-Li), and interlayers [12] (e.g., polymer with Li salt, graphite, and Ag-C layer), have been proposed to "buffer" the direct contact of Li BCC and SE. In particular, the Ag-C interlayer design developed by Lee et al. [12e] manifested outstanding cycling stability with the aid of Ag, enabling the formation of dense Li BCC at the interlayer's interface with the current collector (CC) rather than the SE. Nevertheless, the use of costly noble metals is not desirable; thus, exploring the utilization of earth-abundant elements is necessary.
This exploration then calls for careful and fundamental considerations of the requirements that an interlayer for ASSB needs to meet. First, a material for the interlayer that is both thermodynamically and mechanically stable in contact with both Li BCC and SE should be identified. An appropriate electrochemical stability window for the designed charging/ discharging regiment is crucial; for lithium metal batteries, we are mainly interested in thermodynamic stability around ≈0 V versus Li + /Li. In other words, the MIEC phase must be absolutely stable in naked contact with Li BCC . On the phase diagram (could be binary, ternary, quaternary, etc.) containing Li element, material selection should be determined by the presence of a direct tie-line with the Li BCC phase. The terminal lithiated phases of carbon, aluminum, silicon, etc., fit this requirement. Second, since the deposition/stripping generated cyclic stresses inside Li BCC could fracture the interlayer or penetrate the SE, such stresses should be mitigated by the creeping of Li BCC . For these reasons, we believe that an interlayer material should have properties of a mixed ionic and electronic conductor (MIEC) with a high surface area of 3D open porous structure and high lithiophilicity, thereby being capable of 1) nucleating Li BCC homogenously upon the initial lithiation and for later stripping/deposition; 2) offering high true contact area (i.e., numerous active sites) for chargetransfer reaction-faradaic reaction Li* ↔ Li + + e − , where Li* is a surface ad-atom; [13] and 3) maintaining long-range electronic/ionic percolations for high Li inventory reversibility during (de)lithiation [9,14] (and significant Li ad-atom diffusivity, since Li ad-atom can be considered a composite of e − and Li + ).
In particular, a length scale of <≈100 nm is desired for the pores of the MIEC interlayer. When a metal's characteristic dimension is sufficiently confined to the nanoscale, the "smaller is stronger" trend of dislocation plasticity [7] breaks down, and the metal falls into a "smaller is much weaker" regime governed by diffusion-enabled creep, as shown in Figure 1a. This suggests that the Li BCC may behave like an "incompressible work fluid" during battery cycling. [15] That is to say, nanosized pore confinement inside the 3D openporous interlayer can drive the Li BCC to "fluidize" sufficiently and reduce its plastic hardness <<10 MPa. This softening of the Li Metal "hypodermic needle tip" would mitigate its penetration into the soft SE and also enable Li Metal flow toward the reserved porosity. As such, both the SE and the interlayer can sustain less cyclic stress and retain their morphological integrity ( Figure 1b).
Ideally, the working mechanism of the designed 3D openporous MIEC interlayer will be as follows. Upon being intermeshed with the fluidized, tidal-wave-like Li BCC on one end, and with the SE on another end, the interlayer would maintain long-range electronic and ionic contact. Specifically, the large internal surface areas of the open porous MIEC offer plenty of reaction area for the charge-transfer reaction Li*(MIEC) ↔ Li + + e − , where Li*(MIEC) is an MIEC-surfaceadsorbed Li ad-atom ( Figure 1c) that is charge-neutral. Once the charge-transfer reaction is complete, the Li*(MIEC) can diffuse on the inner surface of the open-porous MIEC and join the Li BCC phases quite a long distance away, i.e., Li BCC on the CC side as well as Li BCC intermeshed inside the MIEC. This is similar to the daily routines of millions of people commuting on highways between the city center (Li*(MIEC@SE)) and sleepercommunity suburbs (Li BCC (MIEC@CC)). The charge-transfer reaction Li*(MIEC@SE) ↔ Li + + e − is analogous to a salaryman (Li + ) finding his car (e − ) near the city center, which can be a real headache but the open-porous MIEC provides plenty of parking spaces to accomplish this, as well a highway to the suburbs ( Figure 1c).
Herein, we propose an open 3D nanoporous MIEC design of ultra-thin hybrid lithium silicide (Li 3.75 Si)-carbon nanotubes (CNTs) interlayer (denoted as Li 3.75 Si-CNT) through the prelithiation of cheap silicon (Si). Figure 1d schematically illustrates the 3D MIEC interlayer design in this work, where the speculated pathways of Li* formation and transport and Li BCC phase nucleation and deformation/coarsening are identified in the designed configuration. Interestingly, the massive volume expansion of the Si host during the prelithiation, which was a well-known cause of capacity degradation in liquid-electrolyte-based LIBs, [16] turned out to be beneficial in this work. This is because the expansion strain allows the formation of small-sized pores (<100 nm) in between Li 3.75 Si, and the soft metallic-alloy nature of Li 3.75 Si helps to consolidate the Li 3.75 Si particles (cold sintering) and improve the adhesion with SE. With this in situ consolidated and SE-adherent ultra-thin MIEC interlayer, we obtained stable reversible plating/stripping behavior on the stainless steel (SUS) CC in asymmetric solid-state cell configuration using sulfide-based argyrodite (Li 6 PS 5 Cl). A full-cell with NCM811 cathode achieved excellent cycling stability (97.65% after 49 cycles) at room temperature.

Fabrication of Li 3.75 Si-CNT Interlayer
We first prepared the commercial Si nanoparticles (SiNPs) with a particle size of ≈50 nm, as shown in a scanning electron microscope (SEM) in Figure S1a in the Supporting Information. Transmission electron microscopy (TEM) image and its fast Fourier transform show a highly crystalline structure of Si nanoparticles (with 0.31 nm d-spacing for (111) plane) which coincides with X-ray diffraction (XRD) patterns (Figure S1b,c, Supporting Information). Then, a half-cell containing SiNP-CNT, Li 6 PS 5 Cl as an SE separator, and Li BCC foil as a counterelectrode was prepared for fabrication of the Li 3.75 Si-CNT interlayer (see details in the Experimental Section). Under a controlled external stack pressure, the prepared cell of SUS/SiNP-CNT/SE/Li BCC was discharged in constant current mode until the cut-off voltage of 5 mV was reached; the voltage profile is shown in Figure 2a. The cross-sectional SEM images of this cell displayed in Figure 2b-d and Figure S2 in the Supporting Information show that the SiNP-CNT layer with a thickness of 2 µm was transformed to the Li 3.75 Si-CNT layer with a thickness of 3.25 µm after the first discharging. Such thin interlayer design is effective on increasing energy density of ASSB. Table S1 in the Supporting Information shows the comparison of Li 3.75 Si-CNT interlayer with previously reported interlayer designs. Figure 2e shows that the large-sized pores (>1 µm) in the original SiNP-CNT layer shrank to nanopores after the first electrochemical lithiation. The nanopores with reduced diameter after lithiation (<100 nm pores) can constrain Li BCC to be deposited in small characteristic sizes inside the Li 3.75 Si-CNT layer, thereby making diffusion-enabled creep a dominant flow mechanism. In addition, this nanoporous Si can provide uniform lithiophilic sites to depress nucleation barrier and suppress dendritic lithium formation. [17] Two levels of stack pressure of 5 and 65 MPa are applied to cells to evaluate the effect of stack pressure on cell performance. Figure 2g,e shows the size of the resultant pores (the interparticle gap between Li 3.75 Si) under stack pressures of 5 and 65 MPa, respectively. The pore sizes were smaller when the applied external stack pressure was higher. For the observation of pores more clearly for the sample with stack pressure of 65 MPa, additional SEM images are shown in Figure S3 in the Supporting Information and a cross-sectional SEM image of Li 3.75 Si-CNT layer after ion beam milling process is shown in Figure S4 Li 3.75 Si-CNT interlayer in ASSB. a) Semiquantitative illustration of the deformation mechanism map for metal (i.e., Sn). [7] The gray curve denotes displacive deformation (Hall-Petch strengthening predicts the ''smaller is stronger'' trend). Surface diffusional deformation is governed by Coble creep. At a fixed strain rate, metal shows ''smaller is weaker'' (red curve). When reducing sample size, the competition between displacive deformation and diffusional deformation brings about a crossover which represents the transition between the two deformation mechanisms. b) Schematic showing porous MIEC structure having 100 nm Li BCC within the pores, which indicates that softening of the Li Metal prevents displacement and penetration of the soft SE and MIEC. c) Li BCC deposition mechanism at interfaces (MIEC@SE and MIEC@CC) in porous MIEC interlayer. d) Schematic illustration (cross-section) of Li 3.75 Si-CNT interlayer in terms of Li BCC deposition mechanism. Increase of state of charge (alloying reaction of Si, xLi + + xe − + Si → Li 3.75 Si and Li metal deposition, Li + + e − → Li BCC ) is shown to the right direction. Li 3.75 Si-CNT layer and the SE remained fully adherent after the disassembly of the cell for both stack pressures, while the interface between the Li 3.75 Si-CNT and the SUS CC showed a small gap (Figure 2f). Strong and spatially uniform adhesion of MIEC interlayer to SE is required to prevent localization of current densities. The formation of such a seamless interface has been assisted by the volumetric expansion caused by the lithiation of Si to Li 3.75 Si during charging, which introduces compressive stress to the SE, combined with the ductile (soft) properties of Li 3.75 Si and Li 6 PS 5 Cl.
To check compatibility between SiNP-CNT (or Li 3.75 Si-CNT) with SE (Li 6 PS 5 Cl), XRD analyses of SUS/SE/Li 3.75 Si-CNT, SUS/SE/SiNP-CNT stacks, and SE were conducted as shown in Figure S5 in the Supporting Information. For all three XRD results, we observed no changes in Li 6 PS 5 Cl crystalline structures regardless of the lithiation status of the SiNP-CNT layer. The stack with the SiNP-CNT layer showed XRD peaks indicating crystalline-Si. For the stack with the Li 3.75 Si-CNT, those peaks no longer exist; instead, the peaks indicating the crystalline-Li 3.75 Si appeared in XRD analysis as shown in Figure S6 in the Supporting Information-since the formation of Li 22 Si 5 requires a much higher temperature, Li 3.75 Si is the fully lithiated state at the testing temperature. [16,18] Despite no electrolytes in the SiNP-CNT layer, SiNPs can be fully lithiated because 1) volume expansion accompanied by Li-Si alloying reaction increases the true contact area between SiNPs, thus securing percolating pathways for charge carriers, 2) Li 3.75 Si itself has a higher ionic and electronic conductivities than pure Si, which makes it a "good" MIEC. During discharge of the SUS/SE/Li 3.75 Si-CNT, we also can see the voltage plateau indicating delithiation of crystalline-Li 3.75 Si ( Figure S7, Supporting Information) which is commonly seen in LIBs. [19] Based on these results, the interlayer thickness and material density of the Li 3.75 Si phase were obtained, and we calculated the porosity of the SiNP-CNT to be 58.5 vol% and the Li 3.75 Si-CNT layer to be 11.0 vol%, as shown in Figure 2h and Table S2 in the Supporting Information.  h) The porosity of SiNP-CNT (58.5%) and Li 3.75 Si-CNT layer (11.0%) is calculated based on the thickness and density of each layer.

Behavior of Li Metal Deposition/Stripping with Li 3.75 Si-CNT MIEC Interlayer
To check the impact of the Li 3.75 Si-CNT interlayer on the ASSB, electrochemical characterization of half-cells with and without the Li 3.75 Si-CNT interlayer was conducted. The configuration of the cells compared was as follows: SUS/Li 3.75 Si-CNT/SE/Li BCC (same as the cell with SUS/SiNP-CNT/SE/Li BCC after lithiation of Si) and SUS/SE/Li BCC , respectively, where SE is Li 6 PS 5 Cl. After electrochemical lithiation of SiNP-CNT to form Li 3.75 Si-CNT (cut-off discharge voltage: 5 mV), it was further discharged at a current density of −0.13 mA cm −2 for Li plating for 20 h (areal capacity: 2.55 mAh cm −2 ) and charged at a current density of 0.25 mA cm −2 for Li stripping. The cut-off voltage of 200 mV was applied in Li stripping for Li 3.75 Si-CNT to avoid Li x Si delithiation that potentially can cause severe morphological change due to the volume shrinkage.
At the beginning of Li deposition in the first cycle, there were significant voltage dip and flat voltage plateau (Figure 3a). The difference between the bottom of the voltage dip and the voltage plateau corresponds to Li BCC nucleation overpotential. The nucleation overpotential of SUS/Li 3.75 Si-CNT/SE/Li BCC and SUS/SE/Li BCC was 1.5 and 14.2 mV, respectively. Almost ten times lower nucleation overpotential in SUS/Li 3.75 Si-CNT/SE/Li BCC is probably attributed to the high surface area of porous lithiophilic Li 3.75 Si-CNT interlayer, which offers numerous sites for accomplishing the charge-transfer reaction Li*(MIEC) ↔ Li + + e − , as well as Li BCC nucleation sites, effectively increasing the "true contact area" between SE and CC (see Figure 3a insets). On the other hand, in SUS/SE/Li BCC cell configuration, the chargetransfer reaction and nucleation sites are limited to physical contact points (area) at the interface between SUS and SE, and microscopically rough surfaces make the true contact area potentially much smaller than nominal contact area. [20] Notably, there were two voltage dips in SUS/Li 3.75 Si-CNT/ SE/Li BCC as shown in the inset of Figure 3a. Since there exist pores inside the Li 3.75 Si-CNT interlayer and Li BCC gets deposited at the MIEC/CC interface, these could correspond to different physical processes, such as Li* clustering on MIEC inner surface, Li BCC nucleation on MIEC inner surface (Li BCC (MIEC)), Li BCC nucleation closer to the CC (Li BCC (MIEC@CC)), etc. If we hypothesize that the second voltage dip arises upon Li BCC nucleation at the SUS/Li 3.75 Si-CNT interface, the Li BCC (MIEC@CC) is expected to be observed only after the later voltage dip. To explore the Li BCC nucleation characteristics, we examined two sets of Li 3.75 Si-CNT interlayer, each of which was detached from SUS after Li plating up to 0.05 and 0.1 mAh cm −2 , respectively. Only the latter cell manifested the second voltage dip. SEM images in Figure S8 in the Supporting Information display the SUS-side surface of the former interlayer and exhibit a relatively smooth surface with a limited number of apparent pores, which implies that Li BCC growth inside the Li 3.75 Si-CNT layer fills out the pores. In contrast, SEM images of the latter interlayer shown in Figure S9 in the Supporting Information exhibit some black plates covering Li 3.75 Si-CNT, indicating that the Li BCC deposition occurred between the Li 3.75 Si-CNT layer and SUS, while some pores of Li 3.75 Si-CNT layer may remain not completely filled. In line with this finding, XRD analysis (Figure 3b) shows that Li 2 S derived from the decomposition of Li 6 PS 5 Cl is clearly present in the SUS/SE/Li BCC configuration after Li deposition (−2.55 mAh cm −2 ), while there are less Li 2 S in SUS/Li 3.75 Si-CNT/SE/Li BCC . Lessdecomposition of SE in the interlayer-containing half-cell indicates that the supply of fresh neutral Li atoms to the Li 3.75 Si-CNT/SE interface is kinetically limited; an alternative explaination is that the porous Li 3.75 Si-CNT interlayer acts as a filter that forbids Li 2 S or broken-off argyrodite particles to be transported into the suburbs, while still allowing Li + /Li* to pass in and out.
Given the post-mortem SEM images and XRD results, we believe that the newly formed Li* accumulates mainly at the SUS/Li 3.75 Si-CNT interface rather than the Li 3.75 Si-CNT/SE interface. Preferential deposition at the SUS/Li 3.75 Si-CNT interface may have become possible by not only the enhanced adhesion of the interface between Li 3.75 Si-CNT and SE but also sufficient Li* ad-atom diffusion rate through the Li 3.75 Si-CNT interlayer, like millions of commuters choosing to sleep every night in suburbs tens of miles away from their workplaces. Such stable and "off-city-center" Li BCC deposition behavior in SUS/Li 3.75 Si-CNT/SE/Li BCC brings about high reversibility of Li inventory "tidal waves" during the charging and discharging process. Accordingly, when the charge capacity portion coming from the lithiation of Si (0.68 mAh cm −2 ) is subtracted, the SUS/Li 3.75 Si-CNT/SE/Li BCC cell has an initial Coulombic efficiency of 96.4%, which is much higher than the 89.4% in SUS/SE/Li BCC configuration (Figure 3c).
To further prove the interpretation above, we investigated the Li BCC morphology and location in each cell during the Li BCC plating and stripping process. The cross-sectional SEM images and its corresponding energy-dispersive X-ray spectroscopy (EDS) for SUS/Li 3 Figure 3l,m. Such random deposition can increase localized current density and consequently vary the local pressure, which encourages Li Metal dendrite growth at the interface with SE in ASSB during further cycling of the cell (Figure 3j).

Identifying Functions of Each Component in Li 3.75 Si-CNT MIEC Interlayer
To further understand the role of the silicide and CNTs in the Li 3.75 Si-CNT layer, we compared the SUS/Li 3.75 Si-CNT/SE/Li BCC and other asymmetric cell configurations, including only the CNTs (denoted as SUS/CNT/SE/Li BCC ) and Li 3.75 Si layer without CNTs (denoted as SUS/Li 3.75 Si/SE/Li BCC ). The SUS/CNT/SE/Li BCC cell showed a lower Li BCC nucleation overpotential than that of SUS/SE/Li BCC , likely due to the high surface area of CNTs with many contact points with SE which offers numerous nucleation sites in the first cycle (Figure 4a). However, CNTs with relatively low ionic conductivity (compared to SE) likely cause excessive Li BCC plating at the CNT/SE interface, where Li BCC can directly contact and react with SE, causing severe decomposition similar to the SUS/SE/Li BCC cell. This is similar to the salaryman (Li + ) easily finding his car (e − ) near the city center, but there is no good highway to the suburbs, causing traffic jams near the city center. The decomposition reactions concentrated at the interface can induce continued stress concentration and the formation of large cracks in SE, eventually leading to cell short-circuiting. [6] On the other hand, we observed that SUS/Li 3.75 Si/SE/Li BCC exhibits outstanding stability and cyclability without any short-circuiting for    Figure S10, Supporting Information). Both Si and carbon are earth-abundant elements that are commercially available at low cost, and the aforementioned results highlight the synergistic roles of CNT and silicides in improving the ASSB performance.
Generally, interparticle contact between SiNPs as well as their adhesion to the SUS CC is poor without a binder; thus, SiNP compact often experiences particle contact loss and SiNP/SUS interface decohesion. Such phenomena could hinder the interlayer from functioning even after the lithiation of the SiNPs. For ASSBs, even though polymer-based binders can be used to glue the Li 3.75 Si particles together to form a tougher compact, they could increase the interfacial resistance owing to their low electrical and ionic conductivities. Herein the wrapping-around CNTs act as a binder during the electrode fabrication process of SiNP-CNT because their outstanding mechanical characteristics (high modulus of about 1 TPa and high tensile strength of more than 10 GPa) improve the integrity of the networking structure of the interlayer during fabrication as well as electrochemical cycling process. [21] In other words, CNT-added SiNPs work like reinforced concrete, in which the relatively low strength and ductility of pure concrete, especially in tension, are enhanced by steel bars having higher tensile strength and ductility. Such mechanically reinforced Li 3.75 Si-CNT interlayer preserving morphological integrity further strengthens its function as a strain buffer. As proof of this assertion, under the harsh mechanical environment of bent Li 3.75 Si-CNT interlayers (see Figure S11 in the Supporting Information. The interlayer was repeatedly bent by about 40°-50° ten times), the fact that CNTs seem to be pulled taut to prevent the fracture of Li 3.75 Si was discovered in the SEM top-viewed images. Moreover, the homogenous distribution of 1D CNTs with high electronic conductivity not only helped the formation of uniform Li 3  lithiation but also induced better plating and stripping of Li BCC phases inside and in contact with Li 3.75 Si-CNT compared to the cell without CNTs (SUS/Li 3.75 Si/SE/Li BCC ). For these reasons, we believe that SUS/Li 3.75 Si-CNT/SE/Li BCC cell showed the longest cycle life without short-circuiting.

Full-Cell Evaluations
To check the practical feasibility of the Li 3.75 Si-CNT interlayer, a full ASSB cell containing a cathode needs to be evaluated. Moreover, electrochemical lithiation of SiNPs before operating a cell is required for the formation of our Li 3.75 Si-CNT interlayer, which means that an additional Li source is required. Therefore, it is necessary for the full-cell to include an additional amount of cyclable Li; such an additional Li inventory can come from the cathode. However, an extra amount of cathode material that is used only for the formation of Li 3.75 Si-CNT will decrease the full-cell energy density and increase resistance in the aspect of cell design (e.g., the high areal capacity of the electrode could result in high cell resistance) due to thicker cathode electrode. Since we have noticed that the Si functions as a lithiophilic material which promotes good wetting with molten Li, [22] nonelectrochemical lithiation or mechanical lithiation was considered. Briefly, Si physically contacting Li BCC could be lithiated and transformed to lithium silicide in the absence of electrochemical current to avoid the extra cathode loading in a full-cell. To confirm that solid-state Li Metal also can give rise to smooth lithiation of the microns-thick SiNP-CNT layer, in situ SEM analysis was conducted using a single micron-sized Si particle (Figure 4d). We monitored that the Si particle was lithiated for a few seconds and expanded, which is commonly observed in the LIB Si anode material. [16] Based on this experimental observation, we adopted an approach to fabricate a full-cell by inserting thin Li BCC foil (≈30 µm thick as shown in Figure S12, Supporting Information) between the CC and SiNP-CNT layer. The full-cell with NCM811 cathode (denoted as Li BCC /SiNP-CNT/SE/NCM811) was investigated and the cross-sectional SEM images and its corresponding EDS (Figure 4e) of the fullcell before cycling (i.e., without any electrochemical current) confirmed that SiNP-CNT layer was transformed to lithium silicide interlayer upon the cell assembly where the uniaxial pressure was applied. The magnified image in Figure 4f shows a highly densified lithium silicide-CNT layer similar to the one obtained by electrochemical lithiation of the half-cell of SUS/Li 3.75 Si/SE/Li BCC (Figure 2d-g). In addition, the voltage plateau assigned to the alloying reaction of Si with Li is not seen in the voltage profile of a symmetric cell (Li BCC /SiNP-CNT/SE/Li BCC ), as shown in Figure 4g, which also indicates that SiNP-CNT can be lithiated without external current by contacting thin Li BCC foil.
Electrochemical characterization of the prepared full-cell (Li BCC /SiNP-CNT/SE/NCM811) was conducted, and a fullcell of Li BCC /SE/NCM811 was also fabricated for comparison. The cells were cycled between 2.8 and 4.3 V at 0.1 C for the first cycle and 0.35 C for the rest of the cycles (see the Experimental Section for the details of cell preparation and electrochemical measurements). The voltage profiles show that the initial specific capacity of cathode materials was similar in each cell (207.8 mAh g −1 for Li BCC /SiNP-CNT/SE/NCM811 and 205.1 mAh g −1 for Li BCC /SE/NCM811) (Figure 5a). The initial Coulombic efficiency of Li BCC /SiNP-CNT/SE/NCM811 was 92.0%, being higher than that of Li BCC /SE/NCM811 (86.5%) (Figure 5b). This result indicates that Li 3.75 Si-CNT pre-formed by reacting SiNP-CNT with the Li BCC foil prevented direct contact between Li BCC and SE and chemical reduction of Li 6 PS 5 Cl SE, which matches with the results of the half-cell. More interestingly, the de-alloying reaction of Li 3.75 Si (assigned to a range of 3.2-3.4 V in the full-cell) does not show up in the voltage profile during discharging to a cut-off voltage of 2.8 V. This is because Li BCC foil used for lithiation of SiNP-CNT supplies enough source of Li to the cathode, and the delithiation potential of Li 3.75 Si is higher than the stripping potential of Li Metal (i.e., the electrochemical de-alloying reaction of Li 3.75 Si would start if there is no other lithium source in anode). Therefore, the Li 3.75 Si-CNT MIEC layer here functions not as an anodeactive material, but as a ground for charge-transfer reaction Li* ↔ Li + + e − ("parking lot"), a highway for transportation of Li* ad-atoms and e − , and nucleation site and creep channel of Li BCC phases. In contrast, delithiation of Li 3.75 Si occurred during discharging in SiNP-CNT/SE/NCM811 without excess Li inventory, as can be seen from the voltage profile displayed in Figure S13 in the Supporting Information.
Full-cell cycling performance of 30 µm-Li BCC /SiNP-CNT/SE/ NCM811 shows capacity retention of 94.6% after 100 cycles and 88.9% after 200 cycles, while the Coulombic efficiency reaches 99.9% after tens of cycles. On the other hand, drastic capacity fading was observed for the full-cell without the Li 3.75 Si-CNT interlayer (30 µm-Li BCC /SE/NCM811) just after 30 cycles. The stack pressure was found to highly affect cell degradation for the 30 µm-Li BCC /SE/NCM811. The cell under a high stack pressure of 65 MPa showed a sudden capacity drop due to short-circuiting (see Figure S14a, Supporting Information) as compared to low stack pressure of 34 MPa. However, the cell containing the Li 3.75 Si-CNT interlayer shows outstanding electrochemical performance regardless of stack pressure (voltage profiles of 30 µm-Li BCC /SiNP-CNT/SE/NCM811 at stack pressure of 5 MPa are shown in Figures S14b,c, Supporting Information). These results indicate that the Li 3.75 Si-CNT interlayer enables the ASSB systems to operate at lower stack pressures. To investigate the effect of the pore size inside the Li 3.75 Si-CNT interlayer on the cycling performance of the full-cells, we applied different stack pressures of 5 MPa as from the fabrication process of ASSB. Then the stack pressure of 65 MPa was applied after first cycle. This is because 5 MPa stack pressure during the formation of Li 3.75 Si-CNT interlayer (lithiation of SiNPs) leads to relatively large-sized pores than those formed under 65 MPa as discussed before (see Figure 2j). The cell (Li BCC /SiNP-CNT/SE/NCM811) under such evaluation condition shows short-circuiting after 21 cycles, as shown in Figure S15 in the Supporting Information. These results indicate that relatively large-sized Li BCC formed inside large pores has higher hardness than nanosized Li BCC (according to the "larger is much harder" trend as illustrated in Figure 1a), and therefore such Li BCC @MIEC can induce mechanical displacement and penetration of SE during lithium deposition.
The rate capability of the full-cell with Li 3.75 Si-CNT interlayer (30 µm-Li BCC /SiNP-CNT/SE/NCM811) was studied by varying the discharge rate from 0.05 C to 5 C at a fixed charge rate of 0.05 C in a constant current mode, and we observed that the discharge capacity was about 82.8% at 2 C and 76% at 5 C as compared to the original capacity at 0.05 C (Figure 5c,d). Furthermore, the discharge capacity and cyclability at low temperatures were investigated, which is a critical decision point for battery application in the electric vehicles industry. As the temperature decreased to 20 °C, the areal capacity of the 30 µm-Li BCC /SiNP-CNT/SE/NCM811 was 2.4 mAh cm −2 , which is 85.8% of the areal capacity obtained at 60 °C (Figure 5e). The capacity was sustained without a significant decrease and remained at 97.65% after 49 cycles (Figure 5f).
Post-mortem analysis using each full-cell (after 200 cycles for 30 µm-Li BCC /SiNP-CNT/SE/NCM811 and 73 cycles for 30 µm Li BCC /SE/NCM811, respectively) in the discharged state was conducted to examine the morphological change of Li BCC , Li 3.75 Si-CNT layer, and their interfaces. As clearly shown in Figure 6a,b, the Li 3.75 Si-CNT layer not only separates the SE and Li BCC but also preserves its morphological integrity without any spallation or deformation for the Li BCC /SiNP-CNT/SE/NCM811 cell.
According to previous works on the inelastic deformation behavior of metals, interfacial diffusion-mediated Coble creep, which is one of the deformation mechanisms dominant at very small sample sizes, [15c] can cause drastic softening of Li BCC . The fracture of Li 3.75 Si-CNT can be avoided by such small-sized Li BCC formation within tiny pores during the Li plating process, which ensures the mechanical stability of both Li 3.75 Si-CNT and the soft SE. Furthermore, in order to cause a morphological change of Li 3.75 Si (Young's modulus: 41 GPa and hardness: 1.9 GPa, absolute value), Li BCC nucleated in the Li 3.75 Si layer needs to afford the stress. [23] However, the hardness and strength of the Li 3.75 Si layer are much higher than the hardness (≈7-43 MPa) and strength of Li BCC . [8] Therefore, the creep of Li BCC located inside the Li 3.75 Si porous layer is inevitable. In addition, the tensile strength of Li 3.75 Si-CNT enhanced by adding CNTs as well as the significantly reduced hardness of Li BCC derived from its size effect (Figure 1a) further help to avoid the deformation and damage of the Li 3.75 Si-CNT interlayer.
Densified Li BCC deposit with a smooth surface exists even after 200 cycles because of the outstanding mechanical and transport properties of the Li 3.75 Si-CNT interlayer ( Figure S16a, Supporting Information), which coincides with the reversible Li plating/stripping phenomena described in Figure 3. In the case of the Li BCC /SE/NCM81, there is a porous structure at the boundary between the SE and Li BCC that likely is a side reaction product, and Li BCC also has the feature of a porous and moss-like structure that might come from the inhomogenous Li BCC extrusion into SE and also inhomogenous stripping upon repeated cycling (Figure 6c,d; the magnified cross-sectional SEM images of the Li BCC side are shown in Figure S16b, Supporting Information).

Conclusion
In summary, we showed the significant role of an ultra-thin open 3D nanoporous Li 3.75 Si-CNT MIEC interlayer (3.25 µm) in the electrochemical performance of an ASSB with argyro-dite (Li 6 PS 5 Cl) SE. In the asymmetric cell of SUS/Li 3.75 Si-CNT/ SE/Li BCC (achieved by electrochemical lithiation of SUS/SiNP-CNT/SE/Li BCC ), we observed that dense metallic Li forms and grows out between the CC and the Li 3.75 Si-CNT interlayer, thereby the direct contact of Li BCC to SE and decomposition of SE being limited. Such deposition behavior is attributed to the interlayer's 1) mixed conducting properties, maintaining long-range ionic/electronic percolations, 2) 3D nanoporosity, which offers numerous active sites for charge-transfer reaction (Li*(MIEC) ↔ Li + + e − ), and 3) high lithiophilicity, which reduces nucleation barrier and enables facile surface diffusion through the porous MIEC. In particular, small-sized pores constraining the deposited Li BCC to be <≈100 nm (smaller is much softer) mitigate the formation of Li Metal -induced mechanical instabilities in the soft SE and the MIEC interlayer, while facilitating diffusional Li Metal flow toward the reserved porosity. On the contrary, the cell without our MIEC interlayer experiences inhomogenous Li Metal deposition and massive formation of unwanted residues such as Li 2 S during plating/ stripping, which leads to poor Coulombic efficiency. The Li 3.75 Si-CNT interlayer may also act as a filter of such brokenoff argyrodite or Li 2 S residues, while still allowing Li + /Li* to pass in and out. A full-cell made of 30 µm thick Li BCC foil as Li inventory in contact with the MIEC interlayer (30 µm-Li BCC / SiNP-CNT/SE/NCM811) demonstrated stable cycling stability of 88.9% after 200 cycles and 97.65% after 49 cycles at 60 and 20 °C, respectively. On the other hand, drastic capacity fading was seen in the cell with the same configuration but without the MIEC interlayer (30 µm-Li BCC /SE/NCM811). The nanoarchitectured ultra-thin Li 3.75 Si-CNT MIEC interlayer designed and tested in this work showed significant improvement in our ASSB cell. We believe that the results of this work may pave a path for achieving commercially feasible ASSB with high-energy density and improved safety using earth-abundant elements.

Experimental Section
Electrode Preparation: The SiNP-CNT electrode was fabricated using commercial SiNPs (Alfa Aesar), CNTs, and deionized water solvent (the mass ratio of SiNPs:CNTs was 90:10). The dispersion solution was mixed and stirred vigorously in a planetary centrifugal mixer (Thinky Corporation, AR-100), and it was cast on the stainless steel foil (thickness was ≈0.01 mm) up to ≈0.21 mg cm −2 . The electrode was then dried at 110 °C for 6 h in the vacuum oven. The other electrodes (SiNP and CNT) were fabricated under the same conditions described above.
Cell Preparation: All ASSBs in this study were fabricated using pellet cells (Canrd, China). For the half-cell, 90 mg powder-typed Li 6  Material Characterization: Morphological investigation for the materials, electrode, and cells was conducted using SEM and EDS (Zeiss Merlin) and high-resolution TEM (JEM-2100F, JEOL). Sample preparation for cross-sectional view ( Figure S4, Supporting Information) was carried out using ion milling system (IB-09020CP, JEOL). The two X-ray diffractometers (AERIS, Malvern PANalytical with Cu Kα of wavelength 1.542 Å and Empyrean, Malvern PANalytical Mo Kα of wavelength 0.7107 Å) were used for the physicochemical characterizations. The cells (SUS/SE/Li 3.75 Si-CNT and SUS/SE/SiNP-CNT) were disassembled in an argon-filled glove box (H 2 O < 1 ppm, O 2 < 1 ppm) and sealed by polyimide film to prevent its exposure to air, and the top side of the Li 3.75 Si-CNT and SiNP-CNT was detected. To examine the morphological changes of the samples after cycling, the cycled cells were disassembled in an Ar-filled glove box (H 2 O < 1 ppm, O 2 < 1 ppm) and then the electrode was cut by scissors for cross-sectional SEM. For observation of micron-sized silicon lithiation (i.e., nonelectrochemical lithiation as mentioned in the Results and Discussion), in situ technique was conducted using SEM (Verios 460, FEI).

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.             Table S1. Comparison of the previously proposed interlayers and LiSi-CNT interlayer for all-solid-statebattery using lithium anode.