Self-organized hetero-nanodomains actuating super Li+ conduction in glass ceramics

Easy-to-manufacture Li2S-P2S5 glass ceramics are the key to large-scale all-solid-state lithium batteries from an industrial point of view, while their commercialization is greatly hampered by the low room temperature Li+ conductivity, especially due to the lack of solutions. Herein, we propose a nanocrystallization strategy to fabricate super Li+-conductive glass ceramics. Through regulating the nucleation energy, the crystallites within glass ceramics can self-organize into hetero-nanodomains during the solid-state reaction. Cryogenic transmission electron microscope and electron holography directly demonstrate the numerous closely spaced grain boundaries with enriched charge carriers, which actuate superior Li+-conduction as confirmed by variable-temperature solid-state nuclear magnetic resonance. Glass ceramics with a record Li+ conductivity of 13.2 mS cm−1 are prepared. The high Li+ conductivity ensures stable operation of a 220 μm thick LiNi0.6Mn0.2Co0.2O2 composite cathode (8 mAh cm−2), with which the all-solid-state lithium battery reaches a high energy density of 420 Wh kg−1 by cell mass and 834 Wh L−1 by cell volume at room temperature. These findings bring about powerful new degrees of freedom for engineering super ionic conductors.

refinements given in the Supplementary info -no structural models, no refinement algorithm (sequence for refining different parameters), etc.
-The authors say that "a 2D 31P refocused INADEQUATE experiment was performed for LPS7228". I can not understand this sentence. Please rephrase.
-When the authors describe the compositions of the glass-ceramics, the sum of mass concentrations is higher that 100%.While reading it becomes more clear what do the authors mean. However, I suggest more straightforward description of the compositions through all the manuscript and supplementary info.  Chem. Mater. 2016, 28, 23, 8764-8773) and even non-Li + conductor aluminum thiophosphates, blocking the Li + migration and thus leading to a large Ea.
For DFT results, Ea is 0.21 eV when Li + migrating along grain boundary and 0.37 eV when Li + migrating across bulk (grain). Given that the experimental Ea is derived from both the Li + migration along grain boundary and across bulk, the value 0.291 eV (LPS7228) agrees well with the DFT results. Response: Thank you for the suggestion. To evaluate the moisture stability of Al-GCs, total generation amount of H2S is characterized by on-line gas analysis mass spectrometry (MS) system (HPR-20, Hiden Analytical Ltd.). A saturated KI solution (25 °C) is used to simulate a room environment with a 68 % RH (relative humidity). As shown in Figure R3a, 150 mg Al-GCs powders are placed in a glass 6 bottle with continuous flow of humid argon gas. The generated H2S gas is instantly blown out by the flowing argon gas and then detected by the MS system ( Figure R3b). In addition, dry argon gas is used to clean the entire pipeline before the experiment. Figure R3c shows the curves of the ionic current of H2S as a function of exposure duration, while the integral area of the curves is proportional to the total generation amount of H2S. As shown in Figure R3c, H2S is generated once the Al-GCs are exposed to moist argon gas, and the total generation amount of H2S continuously increased with decreasing the Li2S content. Response: Experimentally, the electrochemical window (EW) is usually probed via the cyclic voltammetry (CV) method. According to the approach reported by Han et al (Adv. Energy Mater., 2016, 6, 1501590), an asymmetric cell setup ((Al-GCs)-VGCF|Li6PS5Cl|Li/In) with lithium-indium alloy (acting as both counter and reference electrode) and a composite of Al-GCs with VGCF (weight ratio of Al-GC to VGCF is 70 : 30) is used as working electrode to obtain the intrinsic EW of Al-GCs.
Specifically, to probe the upper potential of the EW, the cell is cycled at a scan rate of 0.1 mV/s, starting at the open-circuit potential, up to 3.5 V and back to 1 V vs. Li + /Li and the cycle ends at the open-circuit potential. To probe the lower potential of the EW, the cell is cycled at a scan rate of −0.1 mV/s, starting at the open-circuit potential, down to 0.5 V and back to 2.0 V vs. Li + /Li. The EWs of Al-GCs extracted from Figure R4a and R4b are shown in Figure R4c. electrolyte and Li/In anode were fabricated and cycled at 0.1 C rate between 3.0 and 4.5 V vs. Li + /Li. Figure R4d and R4e, the bare-NCM811|Li6PS5Cl|Li/In battery exhibits fast capacity fade with 52.5 % capacity retention over only 70 cycles, similar to the results reported by Koerver et al (Chem. Mater., 2017, 29, 5574−5582). This common phenomenon is cuased by the narrow thermodynamic intrinisic EW of sulfide electrolytes ranging from 1.7 to 2.9 V vs. Li + /Li (Chem. Mater., 2019, 31, 8328−8337;Nat. Mater., 2020, 19, 428−435), and the tendency for the high-voltage cathode oxidizing the sulfide electrolytes in physical contact, in particular at high charging potential (Nano Lett., 2020, 20, 1483−1490Nat. Energy, 2022, 7, 83−93).

As shown in
Coating cathode active material particles with an electronically insulating/ionically conductive, chemically compatible material has been verified to address this problem effectively (ACS appl. Mater. Interfaces, 2021, 13, 41669−41679;Adv. Energy Mater., 2021, 2100126). Accordingly, the ASSB with LiNbO3 coated NCM811 cathode composite ( Figure R4f) demonstrates much better electrochemical performance, maintaining 80 % capacity retention over 70 cycles. Consequently, it can be concluded that although LPS7228 has a narrow EW, it can still match high-voltage cathode materials with stable coating.
8 Figure R4. CV of (Al-GCs)-VGCF|Li6PS5Cl|Li/In cells at a scan rate of 0.1 mV/s in the voltage range of (a) 1.0 − Al-GCs extracted from Figure S20a and S20b are shown in Figure S20c. Response: Here, to test whether the conductivity of LPS7228 film decreases after bending, we fix one side of an LPS7228 film on a glass bottle with a diameter of 2.76 cm, and hold the other side with tweezers for bending test in a glovebox filled with argon ( Figure R5b). Then, the conductivity of the LPS7228 film after the 10 th and 20 th bending are evaluated. As shown in Figure R5e, the LPS7228 film shows an ionic conductivity of 8.69 and 8.40 mS cm −1 after the 10 th and 20 th bending, respectively, which are slightly lower than that of the pristine LPS7228 film (10.1 mS cm −1 ). The slight decrease in conductivity may be due to the formation of micro-cracks within the LPS7228 film caused by bending.  Response: That is a good question. As for the electrolytes, the of LPS7228 powders and LPS7228 film reaches 35.5 and 25.8 mS cm −1 at 60 °C, respectively ( Figure R6d).
D.C. galvanostatic test is conducted to further evaluate the compatibility of LPS7228 with lithium metal anode at 60 °C. Figure R6a displays an overpotential gradually increases from 4.8 to 5.6 mV during the first 20 hrs at 0.1 mA cm −2 /0.1 mA h cm −2 , which corresponds to the formation of a stabilized Li/LPS7228 interface. Subsequently, this stability interface can support a smooth Li plating/stripping process for over 300 hrs at 0.2 mA cm −2 / 0.2 mA h cm −2 (this battery is still cycling steadily). Compared with the symmetric cells at room temperature ( Figure 4a in the manuscript), high temperature significantly improves the kinetics of ion transports and Li plating/stripping, leading to a lower overpotential.
Then, the LiNbO3-coated NCM622|LPS7228 film|μSi battery with an areal loading of 44.6 mg cm −2 operates in a voltage range of 2.6 − 4.4 V at 60 °C. As shown in Figure R6b, an initial discharge capacity of 175.8 mAh g −1 is obtained at 0.1 C, corresponding to an areal capacity of 7.84 mAh cm −2 .
Due to the improved kinetic at 60 °C, a high value of 6.97 mAh cm −2 is still obtained when increasing the C-rate to 0.2 C. Remarkably, a high reversible areal capacity of 6.04 mAh cm −2 is maintained after 100 cycles, showing a good retention of 86.6 % with an average Coulombic efficiency (CE) of 99.5 %.
Compared with the performance at room temperature (Figure 4j), higher discharge capacity and capacity retention can be achieved at 60 °C.
Also, the high-temperature performance of LiNbO3-coated LiCoO2|LPS7228|Li is evaluated at 60 °C, as shown in Figure R6c. Due to the high and anodic stability of LPS7228, a reversible specific discharge capacity of 143.1 mAh g −1 is achieved for the first cycle with high initial of CE 96.8 %, and maintains 132.9 mAh g −1 after 100 cycles at 0.2 C, showing a good retention of 92.8 %, which is better than that of cycling at room temperature.
To verify whether the long-time heating is harmful to the hetero-nanodomains, the LPS7228 electrolytes are extracted from the LiNbO3-coated LiCoO2|LPS7228|Li ASSB after 100 cycles at 60 °C (heat treated at 60 °C for 1061 hours). As for the , the heated-treated LPS7228 still remains a high value of 34.8 mS cm −1 at 60 °C, which is almost identical to that of the pristine LPS7228, 35.5 mS cm −1 (Figure R6d). To check whether long-time heating is harmful to the hetero-nanodomains (e.g., making them homogeneous), the microstructure of heated treated LPS7228 is observed by cryo-TEM.
As shown in Figure R6e and R6f, a large amount of hetero-nanodomains still exist. Consequently, it can be concluded that the long-time heating can not damage the hetero-nanodomains, which are selforganized when annealing at 300 °C as indicated by in situ temperature-dependent XRD ( Figure S5).   , 1986, 90, 26−33), the transportation of ions or carriers in grain boundary is anisotropic, that is, ions or carriers move more rapidly along grain boundaries (parallel) than across them (perpendicular). It is also well acknowledged that because grain boundaries possess the two key characteristics necessary for enhanced ionic diffusion, i.e., high defect densities (displaced atoms) and high mobilities (interconnected excess free volume), ions or carriers move more rapidly along grain boundaries than across grains or bulk (Ref 2, Phys, Chem., 1984, 88,1057-1062. For example, Zhu et al. employed the conductive atomic force microscopy technique to study the local Li-ion diffusion induced conductance change in LixCoO2 grains, discovering that the grain boundaries have a much lower diffusion energy barrier for Li + migrating along (Sci Rep., 2013, 3, 1084. In microcrystalline materials, the size of grain boundaries is only ~ 1/1000 of the grains (given the size of grain is several microns and the width of grain boundary is several nanometers, which are typical values in microcrystalline materials). In this case, ions or carriers mainly transport among grains.
When ions migrate from one grain to another, they must move across (perpendicular to) the grain boundaries, which act as barriers due to the anisotropy of ions migration in grain boundaries. Therefore, in microcrystalline materials, grain boundaries increase the overall resistance.
In nanocrystalline materials, the situation is quite different since the width of grain boundaries becomes comparable to the grain size. In such case, carriers mainly transport along (parallel) grain boundaries with enhanced ionic diffusion than the grains (bulk), leading to the significant conductivity enhancement. For example, fast transportation along grain boundaries in nanosized polycrystal silver halides has been reported by Maier (Ref 4, Phys. Chem., 1986, 90, 26). Bellino has also shown that the total ionic conductivity of nanostructured, heavily doped ceria solid electrolytes increases by about one order of magnitude compared with the conductivity of conventional microcrystalline materials (Adv. Funct. Mater., 2006, 16, 107-113).
Consequently, in our work, the self-organized hetero-nanodomains, which compose a considerable number of grain boundaries, can actuate super Li + -conduction, leading to a high room temperature of 13.2 mS cm −1 .
Response: We measured the size of self-organized hetero-nanodomains using Gatan Digital Micrograph software and gave a description in the original manuscript. Please find in Page 8, line 195, "As shown in Figure 2d, a large amount of nanodomains (the dark area) with a size of 20 -40 nm are observed, indicating the burst nucleation." Figure S1a for SXRD patterns. With increasing the nucleation-accelerant "Al2S3" phases, e.g., P2S74-and P2S64-are produced in GCs. 1. "PDF of standard Li3PS4 needs to be provided in Figure S1a for SXRD patterns."

Comment 3 PDF of standard Li3PS4 needs to be provided in
It should be noted that the patterns demonstrated in Figure  2. "With increasing the nucleation-accelerant "Al2S3" phases, e.g., P2S74-and P2S64-are produced in GCs. Why?" We have explained the reasons in the original manuscript, please see Page 7, line 161-179 for details, or as follows: The solid-state chemistry of the Li2S-P2S5 binary system indicates that the crystalline substances depend on the fraction of Li2S or P2S5, i.e., the P/S ratio (J. Mater. Chem. A, 2017, 5, 18111−18119). For example, when the molar ratio of Li2S to P2S5 is 75 : 25 (P/S = 0.25), pure Li3PS4 crystalline can be obtained, within which all sulfur atoms are terminals to build PS4 3− polyhedra.
In this work, the P/S ratio of all the samples' reagents is fixed to 0.25, where P source is from P2S5 According to the phase composition analysis, a Li3PS4 analogy phase is identified in LPS7426. Hence, to figure out the local environment of Al in Al-GCs, XPS of the as-prepared LPS7426 is conducted with Al2S3 reagent for a fair comparison. As shown in Figure R7, for the S 2p spectrum of Al2S3, the peak located at 161.86 eV represents the Al−S species, which is also verified in Al 2p spectrum (74.39 eV) (Angew. Chem. Int. Ed., 2016, 55, 9898−9901). For LPS7426, in addition to the peaks located at 161.78, 162.08 and 163.28 eV representing P−S−Li, P=S, P−S−P, respectively, the peak at 161.92 eV is assigned to Al−S species, which corresponds to the peak located at 77.24 eV in Al 2p spectrum.
Consequently, it can be concluded that "Al" has a chemical bonding in the crystal structure of Li3PS4 rather than a physical mixture.      Interfaces, 2019, 11, 42280−42287). Though P2S6 4− possesses very low conductivity of 10 −7 to 10 −6 S cm −1 at room temperature (J. Power Sources, 2006, 159, 193−199;Solid State Ionics, 2016, 284, 61−70), its impact on the conductivity of LPS7228 is negligible due to the low content of ~ 3 wt%  Chem. Mater. 2016, 28, 23, 8764-8773) and even non-Li + conductor aluminum thiophosphates, blocking the Li + migration and thus leading to a large Ea.  Figure R8, the results indicate that the reduction of LPS7525 starts at 2.02 V while the oxidation starts at 2.70 V. For LPS7228, the electrochemical voltage window is 1.53 to 2.94 V, which is wider than that of LPS7525, 2.02 − 2.7 V.
Although LPS7228 has a narrow EW, it can still match high-voltage cathode materials with stable coating. More detailed discussion has been added in the Supporting Information related to Figure S20 in the revised version. Pan's work (Sci. Adv., 2022, 8, eabn4372), a Li-Al alloy anode shows excellent compatibility toward the Li10GeP2S12 electrolyte, which is widely accepted to be easily reduced when contacts directly with Li metal. In addition, compared with Li metal, the Li-containing alloys such as Li-In, Li-Al, Li-Zn and Li-Sn have higher Li + diffusion coefficients, and better wetting on sulfide electrolyte, which can suppress the growth of Li dendrites (Chem. Mater., 2017, 29, 10, 4181-4189;Joule, 2019Joule, , 3, 2165Adv. Energy Mater., 2020, 10, 2000945;J. Mater. Chem. A, 2020, 8, 1247−1253.

Comment 7
The value of critical current density (CCD) has been widely adopted to quantify the dendrite suppression capability of the SSEs. Thus, the CCD of all the composition need to be reported as given in the recently published reference; (Adv. Funct. Mater. 2022, 32, 2201528). Also, provide the galvanostatic cycling performance of Li//Li symmetric cells with counterparts.
Response: Thanks for your good suggestion. Symmetric Li|Al-GCs|Li cells are assembled to test the critical current density (CCD) at room temperature, and the results are shown in Figure R9. Then, the galvanostatic cycling performance of Li|Al-GCs|Li cells at 0.2 mA cm −2 /0.2 mAh cm −2 and room temperature is depicted in Figure R10. A sudden voltage drop appears in the Li|LPS7525|Li symmetric cell after 123 h, illustrating that a short circuit occurs in the cell, while the Li|LPS7426|Li symmetric cell can only deliver stable cycling for 66 h. The Li|LPS7327|Li symmetric cell exhibits a stable cycle for over 160 h at 0.2 mA cm −2 , and it is still cycling steadily. The overpotential of Li|LPS7129|Li symmetric cell shows obvious fluctuations, which could be caused by the interfacial reactions between LPS7129 and Li metal anode (Adv. Energy Mater., 2020, 10, 1903422  Response: Thanks for your good question. To explain the resistance evolution with cycling, we should first figure out the correspondence between the EIS profile and the resistance of the Li|LPS7228|Li symmetric cell, for example, the bulk resistance, the interfacial resistance, etc. As shown in Figure   R11 ( Figure S16 in the revised version), the EIS plots of the Li|LPS7228|Li symmetrical cell after various cycles show the similar appearance, which composes a suppressed semicircle with a tail connected. According to the previous reports (Chem. Mater., 2016, 28, 2400−2407Sci. Adv., 2022, 8, eabn4372), the interfacial resistance of Li-Li symmetric cell is determined by the span of the semicircle and the ohmic (bulk) resistance is determined by the high frequency intercept with the real axis.

Mater., 2017, 29, 1605531).
According to the above analysis, the bulk resistances (R1) of Li|LPS7228|Li after the 1 st , 200 th , 500 th , 1000 th cycles are 57.5, 58.7, 61.5, 58.5 Ω, respectively. Hence, the bulk resistance remains nearly unchanged during cycling. It should be noted that after 200 cycles, the interfacial resistance is almost two times larger than that of the 1 st cycle, and then remains stable, which agrees well with the overpotential evolution with cycling (Figure 4a in the manuscript).   (Figure R12b), respectively, which are comparable to that of previously reported GCs (Adv. Mater., 2021, 2006577). According to your suggestion, the cycling performance of LiCoO2|LPS7228|Li ASSB at 0.5 C, room temperature is evaluated. As shown in Figure R13b   Changes in the Supporting Information related to Figure S22 in the revised version: " Figure S22a shows the cyclic voltammetry (CV) of the LiNbO3 coated LiCoO2|LPS7228|Li at a scan rate of 0.1 mV s −1 , which exhibits well-defined redox peaks, corresponding to the main lithiation/delithiation process.
The symmetry of oxidative/reductive peaks indicates a good reversibility of the intercalation/deintercalation process. As shown in Figure S22b and S22c, the cycling performance of LiCoO2|LPS7228|Li ASSB at 0.5 C is evaluated. A reversible specific discharge capacity of 120.5 mAh g −1 is achieved at 0.

Response:
We thank the reviewer for the strong endorsement of our work. interpretation of the spectra can be strongly affected by charging and differential charring effects.

Comment 1 The authors claim that the interface between lithium and glass-ceramics with
Response:

"First, it is not clear what happens with the interface when other sulfide modifiers are used."
To figure out what happens with the interface when other sulfide modifiers are used, symmetric Li-Li cells using Si-LPS7327 and Ga-LPS7228 GC electrolytes, due to the highest ionic conductivity among their respective components, are assembled and the Li stripping/plating behaviour is tested at a current density of 0.2 mA cm −2 (0.2 mAh cm −2 ). As shown in Figure R14a, Li|Si-LPS7327|Li exhibits a continuously increasing overpotential from 32 to 521 mV within 135 h, corresponding to a significant increase of the interfacial resistance from 118 Ω to 1925.8 Ω (Figure R14b and R14c).
Another impurity at 103.07 eV is caused by SiO2, which could be due to the side reactions with oxygen during the sample transfer to the chamber. Consequently, at the Li/Si-LPS7327 interface, formation of large amount of poorly conductive Li2S and Si 0 , which also undergoes severe volume change during lithiation/delithiation to result in contact loss (ACS Nano, 2014, 8, 8591-8599;J. Power Sources, 2007, 163, 1003-1039, accounts for the continuous overpotential increase with Li plating/striping. The X-ray microtomography is also conducted to observe the Li/Si-LPS7327 interface evolution with cycling. Figure R14g shows the sliced images of the reconstructed tomographic volumes of pristine and as-cycled Li|Si-LPS7327|Li. A spiky interface and cracks (marked as yellow arrows) are clearly observed. What's worse, the Si-LPS7327 particles at the interface have been seriously pulverized (marked as yellow dash line). All the experimental analysis demonstrates the unstable interface between SiS2 tuned GCs and lithium metal.
In contrast, the Li|Ga-LPS7228|Li symmetric cell shows an initial overpotential of 15.8 mV, which remains stable for 135 h, in line with the nearly unchanged interfacial resistance ( Figure   R14d). Figure R14f shows the Ga 2p3/2, S 2p, and P 2p XPS spectra for pristine Ga-LPS7228 and the as-cycled Ga-LPS7228/Li interface. The emerging signals at 1119.6 and 1117.8 eV of Ga 2p can be attributed to the formation of Li-Ga alloy and Gallium (J. Electrochem. Soc., 2016, 163, A2488−A2493), indicating the reduction of Ga 3+ to lower-valency Ga species, while no other 31 obvious reductive species, e.g., Li2S, are distinguished from P 2p or S 2p spectra. It should be noted that trace amounts of residual liquid metal Ga can relieve the stress generated from Li plating, while the solidified Li-Ga alloy with a low surface ion diffusion barrier can guarantee a rapid and homogeneous Li deposition (J. Mater. Chem. A, 2020, 8, 17415-17419;Nat. Energy, 2018, 3, 227-235), which is verified by the smooth and integrated interface between Ga-LPS7228 and Li metal anode after cycling as observed in Figure R14h. All the experimental analysis demonstrates the stable interface between Ga2S3 tuned GCs and lithium metal. cells are assembled and galvanostatic Li stripping/plating tests are conducted to assess the stability of Li-Al alloy/Li6PS5Cl interface. As shown in Figure R15f, the Li-Al alloy 1#|Li6PS5Cl|Li-Al alloy 1# symmetric cell exhibits an overpotential of ~ 38 mV, which maintains stable for over 200 h. On the contrary, the Li-Al alloy 2#|Li6PS5Cl|Li-Al alloy 2# symmetric cell survives only 24.8 h with a short circuit appearing afterwards. Then, the cells are imaged using X-ray microtomography and the corresponding slices through typical tomograms are shown in Figure R15g and R15h, respectively. The Li-Al alloy 1#/Li6PS5Cl interfaces are devoid of any noticeable features, and no dendritic structures are observed ( Figure R15g). In comparison, the interface of Li-Al alloy 2#/Li6PS5Cl becomes spiny. What's worse, dendritic structures are clearly seen in Figure R15h marked by yellow arrows.
According to the above results and previous reports (Sci. Adv., 2022, 8, eabn4372), whether the Li-Al/argyrodite interface is stable depends on the molar ratio of Li : Al (or the Li proportion) in Li-Al alloy. In terms of the physical/chemical properties, the Li-rich Li-Al alloy 2# is closer to the pure Li metal and hence the Li-rich Li-Al alloy 2# has a lower electrode potential than that of Li-poor Li-Al alloy 1# (i.e., 0.23 vs. 0.38 V vs. Li + /Li, respectively) (Energy Storage Materials, 2020, 25, 93−99). Thus, Li6PS5Cl electrolytes are prone to be chemical reduced by Li-Al alloy 2#, leading to uneven Li deposition thus further resulting in the formation of Li dendrites to cause short-circuits eventually. Consequently, Li6PS5Cl is not compatibility with the Li-Al alloy with 33 lower Al proportion, and the Li-Al alloy with lower Al proportion still undergoes severe dendrite formation during cycling, while the Li-Al alloy with higher Al proportion does not.

As for the question "Is there any difference between in situ formed Li-Al layer and Li-Al
alloy used as anode?". Yes, we think the in-situ formed Li-Al alloy layer functions quite differently than the Li-Al alloy used as anode.
For Li-Al alloy anode, volume swells/contracts concomitantly with lithiation/delithiation, which can lead to serious contact loss during cycling (Nat. Energy, 2017, 2, 17119). In comparison, because the in-situ formed Li-Al alloy is line phase, i.e., not solid solution, it is compositionally dynamic balance on cycling by virtue of contact with the lithium foil. Hence, it does not undergo severe volume changes. In addition, as indicated by the atomic force microscopy (AFM) tests, Li metal is softer than Li-Al alloy (Li : Al, 1 : 1). Consequently, it can be speculated that Li metal, and hence the in-situ formed Li-Al alloy layer, contacts more closely with the electrolyte than Li-Al alloy anode does. The close contact will uniformize the Li + flux during the Li plating/stripping processes, which can suppress the formation of Li dendrites. Consequently, the in-situ formed Li-Al alloy layer functions quite differently compared to the Li-Al alloy used as anode.
The detailed illustration of the AFM measurement in Figure R15i-k: Firstly, the AFM tip is brought to approach to the metal surface. An attraction force is received for the tip as it approaches to the surface infinitely, resulting in the cantilever getting bent. Further approach of the AFM tip towards the surface of metal surface causes loading of the tip onto the metal surface. The cantilever is slowly retracted from the surface while the tip is still in contact with the metal surface until it completely separates from the metal surface. Figure R15j and R15k shows the force-distance curves of pure Li film and Li-Al alloy 1#, respectively. According to previous reports (Membranes, 2012, 2, 783−803;J. Colloid Interface Sci., 1975, 53, 314-326), the slope of the forcedistance curve measured during the loading step that can be used to qualitatively describe the hardness of the measured sample. Clearly, Li-Al alloy 1# has a higher slop's value than that of pure Li metal, indicating that lithium metal is softer that Li-Al alloy. Thanks for your comment. We are sorry that we omitted the experimental details of XPS in the original manuscript. We have added this in our revised manuscript.
Changes in the manuscript, Page 25, line 597-606: "The samples are placed onto double-sided tape fixed to clean glass slides and placed in a vacuum transfer holder inside an Ar-filled glovebox.

35
An X-ray photoelectron spectroscopy spectrometer (Thermo Scientific Model K-Alpha XPS) with a monochromatized Al Kα source (1486.6 eV) is used to obtain the surface chemistry of the samples.
A 400 μm X-ray spot size is used to maximize the signal intensity and to obtain an average surface composition over a large area. The sample surface is cleaned via Ar + sputtering with an acceleration voltage of 0.5 kV for 200 s. A dual beam charge neutralization is applied for charge compensation.
The base pressure in the analysis chamber is 3 × 10 −10 mbar. The pass energy is 23.5 eV. Spectra are charge corrected using the C 1s core level peak set to 284.8 eV. Data evaluation is performed with the software Thermo Avantage XPS software package (version 5.976)."

Comment 2
The understanding of the phase composition and evolution in the system is of a high importance. The authors put a lot of efforts in the XRD analysis. However, there are no details on Reitveld refinements given in the Supplementary info -no structural models, no refinement algorithm (sequence for refining different parameters), etc.
Response: Thanks for your suggestions. Rietveld refinements were performed using GSAS II program.
Initially, the structural information obtained from the Materials Project database (Sci Data, 2015, 2, 150009). These steps were used during the refinements: (i) scale factor, (ii) 12 coefficients for a Chebyshev function background, (iii) peak shape parameters, (iv) lattice parameters, (v) zero error, (vi) fractional atomic coordinates, (vii) atomic occupancies. Then, multiple correlated parameters were refined simultaneously, and the related constraints Tables can be found separately in Supporting Information.
Changes in the manuscript, Page 24, line 575-580: "Initially, the structural information obtained from the Materials Project database. These steps are used during the refinements: (i) scale factor, (ii) 12 coefficients for a Chebyshev function background, (iii) peak shape parameters, (iv) lattice parameters, (v) zero error, (vi) fractional atomic coordinates, (vii) atomic occupancies. Then, multiple correlated parameters are refined simultaneously, and the related constraints Tables can be found separately in Supporting Information." Changes in the Supporting Information:        Response: To make this sentence more explicit, we have rephrased the wording of lines 140-142 to: "a 2D 31 P− 31 P INADEQUATE NMR experiment was performed for LPS7228. The 2D experiment is based on J-coupling which allows us to probe P−P and/or P−S−P bond connectivity." Changes in the manuscript: "a 2D 31 P− 31 P INADEQUATE NMR experiment was performed for LPS7228. The 2D experiment is based on J-coupling which allows us to probe P−P and/or P−S−P bond connectivity". All my questions have been clearly answered, I recommend publication of this manuscript in Nature Communications.
Reviewer #2 (Remarks to the Author): The resulting self-organized hetero-nanodomains, composed of a considerable number of grain boundaries, can actuate super Li+ conduction, leading to a high room temperature of 13.2 mS cm−1 in the cold-pressed state. Also, high energy density all-solid stated sulfide-based batteries had been achieved.
Yes. This work is of great significance to the field and the related fields and presents a new approach