Rational design of spontaneous reactions for protecting porous lithium electrodes in lithium–sulfur batteries

A rechargeable lithium anode requires a porous structure for a high capacity, and a stable electrode/electrolyte interface against dendrite formation and polysulfide crossover when used in a lithium-sulfur battery. Here, we design two simple steps of spontaneous reactions for protecting porous lithium electrodes. First, a reaction between molten lithium and sulfur-impregnated carbon nanofiber forms a fibrous network with a lithium shell and a carbon core. Second, we coat the surface of this porous lithium electrode with a composite of lithium bismuth alloys and lithium fluoride through another spontaneous reaction between lithium and bismuth trifluoride, solvated with phosphorous pentasulfide, which also polymerizes with lithium sulfide residual in the electrode to form a solid electrolyte layer. This protected porous lithium electrode enables stable operation of a lithium-sulfur battery with a sulfur loading of 10.2 mg cm−2 at 6.0 mA cm−2 for 200 cycles.

which we wish could show our work's novelty. First, the two papers mentioned by the reviewers were indeed fine attempts of transformative impacts on the field of Li protection. However, we would argue that they would both fell short in providing an adequate protection for Li to survive hundreds of cycles in a Li-S battery. Specifically, 10.1021/acsami.8b07248) employed the polymerized 1,3-dioxolane to shield the planar Li 2 . With this protection, however, the battery still lost 37.5% of its capacity after 100 cycles at 0.1 C, as polymeric protective layers often suffer from polysulfide permeation and dendrite penetration.
As suggested by the reviewer, we compare our work to representative papers in the literature on the protection of porous Li electrodes (Table R1), with an emphasis on their effectiveness in achieving stable cycling in a Li-S battery, which is the ultimate goal of our work. Among these efforts, Liu et al. further employed inorganic solid electrolyte pellet based on Garnet-type Li 6.5 La 3 Zr 0.5 Ta 1.5 O 12 (LLZTO) for protecting the Li/rGO electrode 3 . Xu et al. attempted to thinner the surface solid electrolyte layer by building a bilayer dense-porous structure 4 . Though effective for the suppression of side reactions and dendrite, the use of inorganic solid electrolyte either as the surface protective layer or the scaffold persistently suffers from the fragility, the high Li + transport resistance, and the relatively low volumetric energy density.
In this work, instead of using the Garnet-type solid electrolytes, we turned to use the amorphous Li 2 S-P 2 S 5 solid electrolyte for surface protection. Unlike the high sintering temperature required for Garnets (>1100 • C), the sulfide-based electrolyte can be sintered at a low temperature (160 • C), without side reaction (with carbon) or Li melting. Thus the solid electrolyte can be in-situ formed onto the Li-coated CNF matrix as a surface layer. Supported by the compressible and flexible interwoven CNF matrix, the fragility issue of the solid electrolyte can be much addressed, leading to a thin/dense protective layer with a low Li + transport resistance. Lin et al. 1 Unknown for Li-S excessive addition of liquid electrolyte Liu et al. 3 Unknown for Li-S high resistance; fragility of the solid electrolyte Xu et al. 4 4.3 mg cm -2 , 36.6% (39 mA g -1 ) high resistance; high scaffold weight Lin et al. 5 4.14 mg cm -2 , 12.7%(161 mA g -1 ) non-continuous surface protective layer This work 3.42 mg cm -2 , 41.2%(248.5 mA g -1 ) Moreover, inside the porous Li scaffold, liquid electrolyte is usually necessary for wetting the Li surface to address the interphase problem. A robust artificial SEI for smoothing the Li deposits and suppressing the consumption of liquid electrolyte thus would be necessary. Lin et al. treated the Li/rGO electrode in the Freon gas atmosphere to form the LiF-based artificial solid electrolyte interphase (SEI) across the electrode for protection 5 . There are also some other reported methods to derive LiF-rich artificial SEI as shown in Table R2. However main concerns are raised for the coating condition (vapor deposition and atomic layer deposition), the quality of the coating layer (solution deposition, electrolyte additive, and blade casting), and the ionic conductivity of LiF (defect-free LiF is highly insulating) 6 . As a result, few of them can be readily applicable for coating the artificial SEI inside the porous Li scaffold at this stage.   Table R1). The open symbols represent the specific capacities at initial cycles, while the filled symbols represent the specific capacities after cycling. Some of the reported planar anodes such as the bare Cu foil, and the Cu foil protected by double-layered nanodiamond are also included for comparison 13,14 . The porosity of the electrode can be tuned by adjusting the density of carbon nanofibers as well as the initial sulfur loading.
The Li-S/CNF electrode was compressed before characterization and testing. Before compression, the electrode is more porous and allows for the solution deposition to occur uniformly through it.  from either the porous S/C structure or from the protective coatings of LiF, and Li 3 Bi.
Response：We agree with the reviewer that it is important to compare MFC-Li/CNF with MFC-Li. The symmetric cell with the MFC-Li electrode was charged and discharged with an areal capacity of 1.0 mAh cm -2 at a current density of 1.0 mA cm -2 or 2.5 mA cm -2 , respectively. Stable operation without voltage hysteresis lasted for 300 cycles (600 hours) during 1.0 mA cm -2 cycling, considerably better than the bare Li metal as shown in Fig. R4. However, for 2.5 mA cm -2 , within 300 cycles, the overpotential of the symmetric cell with the MFC-Li electrode promoted to over twice of its initial value as shown in Fig. R5. This can be attributed to the fact that the SEI layer thickens during cycling, which would block the ion transport across the interphase.  Further, for Li-S battery testing, we have operated the MFC-Li anode under a low sulfur loadings (2.8 mg cm -2 ). Its performance at a 0.5 mA cm -2 was relatively stable (0.134% decay per cycle), but it suffered from a relatively fast decay at 2.0 mA cm -2 (0.355% decay per cycle), due to the significant deformation of the Li/electrolyte interphase as shown in Fig. R6a. Under the same sulfur loading, the specific capacity achieved by MFC-Li/CNF was remarkably higher than the MFC-Li, due to the decrease of overpotential in the anode side, as shown in Fig. R6b. Through the comparison, we can confirm that combining the protective coating and porous Li scaffold is essential for stabilizing the Li/electrolyte interphase.

Comment 5:
What is the C.E. for Li-S full cell? In figure 5a-b, at 1mA/cm2, the C.E. seems to be larger than 100%, please confirm.
Response：Thanks for this constructive comment. In Fig. 6c, we showed both charge and discharge capacities. In the revised manuscript, we further included the coulombic efficiency in Fig. 6c (as can be also seen in Fig Response：Thanks for this very helpful comment. We have operated the batteries at room temperature for comparison before. For clear viewing, we didn't include them in the manuscript. The average capacity fading rates were 0.033% and 0.092% for 23 • C cycling at 1.0 mA cm -2 and 4.0 mA cm -2 , respectively. While the average capacity fading rates were 0.035% and 0.123% for 45 °C cycling at 1.0 mA cm -2 and 4.0 mA cm -2 , respectively ( Fig. R7). At a higher temperature, the thickening of native SEI will be accelerated, leading to the slight increase of capacity fading, as shown in our test results. Response：We appreciate this very constructive suggestion from the reviewer.
Specifically for the purpose of performance evaluation, we screened out the pristine carbon cloth as the cathode current collector, which is of a low tortuosity, and a robust interwoven structure. The untreated carbon fiber in the carbon cloth is highly graphitic. As a result, there would be minor structural or compositional change for the carbon cloth over cycling. Surface engineering (e.g. decoration of nanostructured polysulfide binding materials) of the electrode can enhance the sulfur utilization, however, there presumably would be the clogging of micropores or passivation of the active sites during cycling. So herein carbon cloth electrode was not subjected to any pretreatment. The sulfur/carbon cloth electrode with a high loading was reported before by our group and other researchers. For example, with an activated carbon cloth, Elazari et al. attained an areal loading of sulfur around 6.5 mg cm -2 at 0.98 mA cm -2 , reaching a maximum specific capacity of 1050 mAh g -1 . Though their carbon cloth possessed abundant micropores and there was intensive addition of LiNO 3 (2.0 wt%) as the side reaction inhibitor, their battery's fading rate was still relatively fast (0.25% decay per cycle for 80 cycles), indicating the necessity of rationally modifying the Li anode. We have included the corresponding discussion in the main context.

Response to the second reviewer
General comment: This work claims to fabricate a porous lithium electrode which is protected with LiF and Li 3 Bi Layer and is in direct contact with the solid electrolyte layer. The authors showed some nice results and interesting cycling data. However, after carefully going through the manuscript, I regret to say that I won't recommend this work for publishing in Nature Communication keeping the lack of novelty and experimental support provided in this work.
Response: Firstly, we would like to thank the reviewer for carefully reviewing our manuscript and acknowledging the performance achieved in this study. We also understand the reviewer's concern regarding to the novelty and experiment support.
We sincerely hope that we could address the reviewers' concerns from the following perspectives.

Response:
We thank the reviewer for raising this concern. The previous work (New nanostructured Li 2 S/silicon rechargeable battery with high specific energy, DOI: 10.1021/nl100504q | Nano Lett. 2010, 10, 1486-1491) proposed a silicon/Li 2 S full battery configuration, in which an organic lithium superbase (n-butyllithium) was employed for the prelithaition of sulfur/carbon composite. The previous work was impressive, however when preparing the manuscript, we didn't realize it was strongly correlated with our work. Much different from the previous work, the motivation of our work is to create a protected porous Li anode, which can be deeply cyclable in the Li-S full battery. We further try to illustrate the strength of this work, by systematically comparing our work with the literature, as could be found in our response to the first question of the first reviewer. showing that Li-rich surface doesn't contain any residue of sulfur.

Response:
We appreciate the valuable suggestion from the reviewer. The Li 2 S-rich surface was formed from the gravitational setting of Li 2 S during the formation process of Li-S/CNF composite. After the reaction and Li infusion, the CNF can be largely lithiated, which possesses a much improved lithiophilicity 15,16 . The lithiated CNF thus can trap the molten Li to form a fibrous network. In contrast, Li 2 S has poor affinity with both carbon and Li, so it gradually condenses on the electrode's bottom to form a Li 2 S-rich surface.
The formation of Li 2 S-rich surface allows us to create a conformal solid electrolyte layer to shut down the polysulfide crossover. We support this claim by providing the elemental mapping of the electrode's cross section, and a gradient distribution of sulfur can be visualized as shown in Fig. R8(a, b). Despite the cross section, the surface SEM also showed the designed Li 2 S-rich surface exhibited a dense morphology due to the condensation of Li 2 S and the surface sulfur fraction (atomic) reached 79% based on the EDX mapping result (Fig. R8(c, d)). On the designed Li-rich surface, we further observed the Li-coated fibers with micrometer diameter, with an atomic fraction of sulfur as low as 6% (Fig. R8(e, f)).  Fig. R9(c, f, i, l).
Assuming complete Li stripping, the volume fraction taken up by Li can be derived.
Herein, we specifically chose an initial sulfur loading of 2.5 mg cm -2 , which was almost the minimum loading value to trigger the uniform Li filling (Fig. R10).
Besides, under this condition, the volume expansion ratio of the anode can be also maintained at a reasonable value (~70%).

Response:
We sincerely thank the reviewer for this very constructive comment. Both structural and compositional analysis were conducted. There was negligible electrode level deformation after high rate cycling (4.0 mA cm -2 ) as shown in Fig. R11(a, b).
From XPS depth profiling (Fig. R11c), we observed the coverage of a native passivation layer (SEI) on the solid electrolyte surface, likely formed from the decomposition of LiTFSI. If we further probed inside the solid electrolyte, we observed that the peaks of native passivation layer diminished, indicating that the surface solid electrolyte layer prevented the permeation of liquid electrolyte. On the Li-rich surface, we could observe the Li-coated fiber maintained its morphology over cycling ( Fig. R11(d, e)). XPS depth profiling further indicated the Li-coated fibers were passivated by both the Li 3 Bi/LiF-based protective layer and the native SEI (Fig.   R11f). These results testify that the structure of the porous electrodes and the Ar sputter speed was 13 nm min -1 .

We sincerely thank the reviewer for the constructive comments, which are really helpful for improving the quality of this paper!
This manuscript is much improved by replying reviewers' comments adequately. The major issue for this work is the lack of novelty. The protection layer forms by a spontaneous reaction is not a new idea, and the metal fluoride protection idea has been published several times in the literature.
The work provides good insights on exploring a specific reaction that serves the purpose, which should be published elsewhere. But I won't recommend its publication in Nature Communication.
Reviewer #2 (Remarks to the Author): After going through the significant efforts by the authors, I am satisfied with the revisions and recommend the article for publishing.
Reviewer #3 (Remarks to the Author): The authors present a multistep procedure for the protection of Li anode in Li-S battery, indeed a key for enabling high Coulombic efficiency and long-term cyclability in practical Li-S batteries. The approach is rather novel resulting in a pretty reasonable performance in the absence of LiNO3 in the electrolyte. I can recommend it for publication in Nat Comm if the authors demonstrate that amongst those approaches which deal with the ex-situ formation of protective layers on Li metal their approach is the best. They should be able to that by comparing the 1) Areal capacity of different test cells and 2) the presence/absence of LiNO3. Reason being that the suitability of any sort of in-situ/ex-situ protection layer can be best examined in the absence of LiNO3 and at practically high gravimetric/areal capacities.
1-If I got it right (please confirm), the presented cycling data in this work is in the absence of LiNO3 which is impressive. In that case the authors should highlight that and compare their work with recent literature such as NATURE NANOTECHNOLOGY | VOL 13 | APRIL 2018 | 337-344 "2D MoS2 as an efficient protective layer for lithium metal anodes in high-performance Li-S batteries" where LiNO3 is used in the electrolyte and elucidating the effect of the protection layer is rather difficult. 2-The authors mention that their "protected porous Li electrode enables stable operation of a Li-S battery of a high sulfur loading (6.8 mg cm-2) at a high current density (4.0 mA cm-2) for more than 200 cycles" and "At 23 °C, the discharge capacity achieved at a cathode current density of 4.0 mA cm-2 (anode current density 3.4 mA cm-2) was 4.14 mAh cm-2". Again, while this result is impressive in the absent of LiNO3, 4.14 mAh cm-2, is not going to be a practical areal capacity for the future Li-S battery. The math is simple-the areal capacity of the graphite anode in commercial LIB is 2.5-3.5 mAh cm-2 and for a Li-S battery with a lower operating voltage (≈2.1 V compared to ≈3.7 of that of LIB) to rival LIB, the areal capacity should be at least 6 mAh cm-2. Given the excess of pretty much every single component in the Li-S battery (for example you are using 100 μm liquid electrolyte, additional protection layers, a carbon cloth current collector), even at 6 mAh cm-2 and gravimetric capacities above 1000 mAh g-1s, translation to LIB is questionable. I suggest avoid using terms such as high current density when both the areal and gravimetric capacity are not satisfactory compared to the practical levels.
3-Even though 6.8 mg cm-2 is certainly a high sulfur loading, a gravimetric capacity of 800 mAh g-1s at such a low current density (1 mAh cm-2) means that two critical issues of polysulfides shuttle and Li anode stress are not truly examined in these batteries, the latter in particular. I suggest assembling test cells with high areal capacity cathodes (> 6 mAh cm-2) over a few tens of cycles to ensure true examination of the full range of technical challenges that are present in a practical Li-S battery. Perhaps use a higher sulfur-loading cathode and monitor the performance at room temperature while areal capacity is higher than 6 mAh cm-2. And please plot the Coulombic efficiency data in a more visible fashion, perhaps magnify it in the range of 90 to 100 %. 4-Given that your approach is a two-step process, have you conducted cycling tests with 1) just porous Li electrode, comprising fibrous networks of a Li shell and a carbon core with no coating on the surface, and 2) a lithium foil coated with your proposed composite layer of Li3Bi and LiF (if its formation is at all possible on a bare lithium foil). Also, it would be interesting to examine the performance in the presence of LiNO3 additive in the electrolyte. 5-The authors should present discussions and argument on why they think their approach is a better alterative for the simple and commonly used approach of using LiNO3 as the electrolyte additive. As far as I'm concerned, the literature of Li-S is overwhelmed with similar performance metrics as the authors present in this work where an in-situ SEI layer is formed on the Li metal by using LiNO3 or other additives in the electrolyte composition. In summary, I find the approach interesting and novel. Battery manufacturers are not going to be fond of formation of ex-situ protection layers or composite Li electrode given the complexity of the procedures and the demand for the whole process to be conducted under Ar. Nonetheless, I believe this work would be of the interest to the scientific community. The results even though not impressive in the general literature of Li-S can be considered as very good given the absence of LiNO3. To ensure the true suitability of this Li coating, test cells must be configured with high areal capacity cathodes to allow for passage of realistic current densities.

Reviewer #1 (Remarks to the Author):
This manuscript is much improved by replying reviewers' comments adequately. The major issue for this work is the lack of novelty. The protection layer forms by a spontaneous reaction is not a new idea, and the metal fluoride protection idea has been published several times in the literature. The work provides good insights on exploring a specific reaction that serves the purpose, which should be published elsewhere. But I won't recommend its publication in Nature Communication.
Response：We appreciate the comments by the reviewer, which has helped improve this manuscript tremendously. With the points elaborated below, we hope to clear the doubt over the novelty of the work. Li metal 6 . However, none of these reactions could generate a sufficiently robust and conductive layer that is also immune to the attack by polysulfides; InF3 has a rather low solubility; CuF2 produces Cu which is unstable towards polysulfides, and DMF, though initiating the LiF formation, is known to attack Li [7][8][9] .
In this work, we present a new design strategy that builds upon the need for all the ingredients to work together before and after the spontaneous reaction. After the reaction, we need LiF, the most effective component in SEI for Li protection, but given the brittleness of LiF, we also need a stronger component to form a composite, likely a Li alloy. We need another layer on top to prevent the attack of polysulfide, likely a solid electrolyte, and this protection has to go onto a porous Li electrode for a practical areal loading. Meeting all these requirements are challenging, but intriguingly, they also bring us more ingredients, which opened up the space of the reaction design. Instead of using LiF directly, which is notoriously insoluble for a solution-phase reaction, we exploited Bi as needed in the alloy component, in the form of BiF3 as the fluoride source.
The solvation of BiF3 required P2S5, a complexant known for forming solid electrolytes when reacting with Li2S 10 . Sulfur for the Li2S then became the perfect 'lubricant', whose affinity to carbon matrix and exothermic reaction with Li marry the two. Such a design rationale eventually led to the good battery performance, the strongest evidence is that the work is a viable solution to the Li protection in Li-S batteries.
Second, we would like to argue that the present work provides more than a specific reaction, but a general approach towards an effective metal protection in alkaline metal batteries by offering the following mechanistic insights: range of anodes, including metallic Na, K, and alloys. We have now included the following figure into the Supporting Information to prove the generality of the solvation mechanism for two more metal fluoride (Fig. R1), and to show the general feasibility of the corresponding coating strategy (Fig. R2). LiF and Li3Bi at a nanoscale along the in-plane direction (Fig. 3(d-f)). We also used density functional theory to reveal that the low energy barrier of surface diffusion would promote uniform Li deposition through the artificial SEI. To reveal more structural details, we have now characterized the SEI layer with HR-TEM, confirming nanometersized crystallites of LiF and Li3Bi (Fig. R3). We have also quantified the value of the ionic conductivity of the surface layer with a careful set of experiments given its critical role in stabilizing the anode (Fig. R4). We first combined cross-sectional SEM and TOF-SIMS to measure the thickness of a Li3Bi/LiF layer on the surface of Li metal ( Fig.   R4(a, b)), and then measured its ionic conductivity in a dry, symmetric cell to be 6.9×10 -corroborates the suggested benefits of our design strategy. At last, we would like to thank the reviewer again for the reminder of previous work on spontaneous reactions and metal fluoride coating. Although we had a summary in Introduction on metal fluoride coating, we have now revised Introduction to give a review on liquid-phase spontaneous reactions, and discuss how our approach differ from the preceding work.
Reviewer #2 (Remarks to the Author): After going through the significant efforts by the authors, I am satisfied with the revisions and recommend the article for publishing.
Response：We sincerely thank the reviewer for the kind acknowledgement and we will move on to carefully revise the manuscript.
Reviewer #3 (Remarks to the Author): The authors present a multistep procedure for the protection of Li anode in Li-S battery, indeed a key for enabling high Coulombic efficiency and long-term cyclability in practical Li-S batteries. The approach is rather novel resulting in a pretty reasonable performance in the absence of LiNO3 in the electrolyte. I can recommend it for publication in Nat Comm if the authors demonstrate that amongst those approaches which deal with the ex-situ formation of protective layers on Li metal their approach is the best. They should be able to that by comparing the 1) Areal capacity of different test cells and 2) the presence/absence of LiNO3. Reason being that the suitability of any sort of in-situ/ex-situ protection layer can be best examined in the absence of LiNO3 and at practically high gravimetric/areal capacities.

Response:
We are grateful to the reviewer for the constructive suggestions. In the response below, we follow the reviewer's suggestions and compare our work systematically to others to show that we have achieved a deeply cyclable Li anode with a high areal capacity in the Li-S full battery, in the absence of LiNO3 electrolyte additive. However, the areal capacity, current density and the Li utilization ratio of their work were lower. At a higher current density, for example, the use of LiNO3 may not bring the same benefit as we will show in the response to the 4th comment of the reviewer's.

Table R1
A performance comparison between our Li-S battery and those using LiNO3 as electrolyte additives.  14 Liu et al. 12 Chen et al. 15 Cha et al. 13 Tang et al. 16 Liu et al. 17 Cai et al. 18 Chang et al. 19 This work * The capacity exhibited slight increase in the initial tens of cycles (Fig. R5). Therefore, the capacity retention is calculated based on the ratio of capacity at the last cycle and the maximum capacity achieved over cycling.
This table could be found in Table S1.
2. The authors mention that their "protected porous Li electrode enables stable operation of a Li-S battery of a high sulfur loading (6.8 mg cm -2 ) at a high current density (4.0 mA cm -2 ) for more than 200 cycles" and "At 23 °C, the discharge capacity achieved at a cathode current density of 4.0 mA cm -2 (anode current density 3.4 mA cm -2 ) was 4.14 mAh cm -2 ". Again, while this result is impressive in the absent of LiNO3, 4.14 mAh cm -2 , is not going to be a practical areal capacity for the future Li-S battery.
The math is simple-the areal capacity of the graphite anode in commercial LIB is 2.5-3.5 mAh cm -2 and for a Li-S battery with a lower operating voltage (≈2.1 V compared to ≈3.7 of that of LIB) to rival LIB, the areal capacity should be at least 6 mAh cm -2 .
Given the excess of pretty much every single component in the Li-S battery (for example you are using 100 μL liquid electrolyte, additional protection layers, a carbon cloth current collector), even at 6 mAh cm -2 and gravimetric capacities above 1000 mAh g S -1 , translation to LIB is questionable. I suggest avoid using terms such as high current density when both the areal and gravimetric capacity are not satisfactory compared to the practical levels.
Response：We appreciate the suggestion, and we have revised the discussion using more appropriate terms. We have also assembled full batteries with an even higher sulfur loading (10.2 mg cm -2 ) and tested the battery at practically high current densities of 4.0, 6.0 and 8.0 mA cm -2 , with details in the following response.
3. Even though 6.8 mg cm -2 is certainly a high sulfur loading, a gravimetric capacity of 800 mAh g-1s at such a low current density (1 mAh cm -2 ) means that two critical issues of polysulfides shuttle and Li anode stress are not truly examined in these batteries, the latter in particular. I suggest assembling test cells with high areal capacity cathodes (> 6 mAh cm -2 ) over a few tens of cycles to ensure true examination of the full range of technical challenges that are present in a practical Li-S battery. Perhaps use a higher sulfur-loading cathode and monitor the performance at room temperature while areal capacity is higher than 6 mAh cm -2 . And please plot the Coulombic efficiency data in a more visible fashion, perhaps magnify it in the range of 90 to 100 %.
Response：We agree that a high areal capacity and a high current density would be required to examine all the challenges in practical Li-S batteries. We have now tested an additional battery with a 50% higher sulfur loading (10.2 mg cm -2 ) at even higher current densities, as shown in Fig. R5 20 .
At 45 °C , this battery delivered discharge capacities of 7.53 mAh cm -2 and 7.05 mAh cm -2 at 4.0 mA cm -2 and 6.0 mA cm -2 respectively. Even at 8.0 mA cm -2 , it could achieve 6.6 mAh cm -2 (Fig. R5b). When we lowered the temperature to 23 °C as suggested, the discharge capacity decreased likely due to the slower reaction kinetics in the high-loading electrode (Fig. R5a), as previous shown by Xiao et al. 21 . As shown in Fig. R5(c, d), the high-loading battery (10.2 mg cm -2 ) could be cycled at 4.0 and 6.0 mA cm -2 at 45 °C for more than 200 cycles stably, with an areal capacity constantly above 7.5 and 6.8 mAh cm -2 , respectively. If cycled at 8.0 mA cm -2 , the battery could last for about 180 cycles. We hope that with this additional set of data, we have shown the feasibility of our design to address the issues in a practical Li-S battery.
As also suggested, we have plotted all the C.E. values within a smaller range to make them more visible as shown in Fig. R5d and Fig. 6(d, e).  Fig. 6 and Fig. S26. 4. Given that your approach is a two-step process, have you conducted cycling tests with 1) just porous Li electrode, comprising fibrous networks of a Li shell and a carbon core with no coating on the surface, and 2) a lithium foil coated with your proposed composite layer of Li3Bi and LiF (if its formation is at all possible on a bare lithium foil). Also, it would be interesting to examine the performance in the presence of LiNO3 additive in the electrolyte.
Response：We have indeed tested the first type of electrode before as the reviewer suggested, using sulfur cathode with a moderate loading (2.8 mg cm -2 ) and current (2.0 mA cm -2 ). As shown in Fig. R6a, a Li-S/CNF electrode, which comprised only the network of Li shells and carbon cores, did not perform well, as lithium polysulfides could corrode the electrode quickly. We have now also tested the second type, a planar Li electrode coated with Li3Bi (MFC-Li). Although this electrode was much more stable than bare Li, it suffered from a severe capacity decay in the first 100 cycles (Fig.   R6a). The results are consistent with our expectation; neither the porous structure nor the coating alone would stabilize the anode in a practical Li-S battery.
Adding LiNO3 did improve the cycling stability (Fig. R6b), likely because of the healing of SEI by LiNO3. However, the stabilizing effect is limited and short-living for both bare Li and Li-S/CNF, likely due to the continuous consumption of LiNO3. MFC-Li showed an improved stability with LiNO3 at 2.0 mA cm -2 , but the combination would be less effective when the current density rose to 4.0 mA cm -2 . Nonetheless, the performance is all inferior to that of MFC-Li/CNF, confirming the effectiveness of our design strategy. leading to the battery package swelling 23 . Also, the combination of LiNO3, sulfur and carbon forms a hazardous composition similar to the black gun powder 24 . Considering these potential drawbacks, the LiNO3 additives might not be relied heavily upon in the future.
We have now included this brief discussion on LiNO3 at the end of the result section.
In summary, I find the approach interesting and novel. Battery manufacturers are not going to be fond of formation of ex-situ protection layers or composite Li electrode given the complexity of the procedures and the demand for the whole process to be conducted under Ar. Nonetheless, I believe this work would be of the interest to the scientific community. The results even though not impressive in the general literature of Li-S can be considered as very good given the absence of LiNO3. To ensure the true suitability of this Li coating, test cells must be configured with high areal capacity cathodes to allow for passage of realistic current densities.
Response：We do agree with the reviewer that it would be important to look at the scalability of the method. As indicated, the porous Li electrode (Li/S-CNF) needs to be fabricated in the inertial Ar gas atmosphere to prevent side reactions between Li and N2/O2/CO2. Inspired by the reviewer, we further studied the dry air stability of the precursor and the as-fabricated Li electrode to see whether they are stable during the protective coating and the battery assembly process. We show that BiF3-P2S5 in DME precursor remained stable after the dry air exposure as shown in the UV-Vis spectra ( Fig.R7 (a, b)). Meanwhile, for the MFC-Li/CNF, the electrode components (LiF, Li3Bi and Li2S-P2S5 solid electrolyte) are stable compounds, and the Li3Bi/LiF coating could to some extent prevent the side reactions between Li and N2/O2/CO2, which lowers the Li loss from 46% in the case of air-exposed Li-S/CNF to 7.4% (Fig. R7(c, d)). Anhydrous CaCl2 (200 mg) was put in the vessel (20 mL) to create dry air environment.
The electrodes (8 mm in diameter) for testing were punched from the same Li/S-CNF or MFC-Li/CNF pellet (18 mm), to guarantee that their properties were almost identical.
The nitrogen and oxygen contained in the vessel are excessive, which can fully react ~34 mg Li to form Li3N and Li2O.
We are beyond grateful for the reviewers' comments, which are really helpful for improving the quality of this paper.