NASICON-type air-stable and all-climate cathode for sodium-ion batteries with low cost and high-power density

The development of low-cost and long-lasting all-climate cathode materials for the sodium ion battery has been one of the key issues for the success of large-scale energy storage. One option is the utilization of earth-abundant elements such as iron. Here, we synthesize a NASICON-type tuneable Na4Fe3(PO4)2(P2O7)/C nanocomposite which shows both excellent rate performance and outstanding cycling stability over more than 4400 cycles. Its air stability and all-climate properties are investigated, and its potential as the sodium host in full cells has been studied. A remarkably low volume change of 4.0% is observed. Its high sodium diffusion coefficient has been measured and analysed via first-principles calculations, and its three-dimensional sodium ion diffusion pathways are identified. Our results indicate that this low-cost and environmentally friendly Na4Fe3(PO4)2(P2O7)/C nanocomposite could be a competitive candidate material for sodium ion batteries.

The manuscript from Chen et al. reported the development of a low cost scalable polyanionic cathode material for sodium-ion batteries. Different synthetic strategies were adapted to fabricate the cathode materials of tuneable morphology and particle size. Good rate capability and long-term cycling stability were demonstrated in both half cell and full cell designs. The air stability as well as the high/low temperature properties also were investigated. It is a big step for Na-ion battery development towards practical application. Furthermore, the authors conducted in-situ synchrotron XRD/XAS characterization and theoretical simulation to reveal the intrinsic nature of this material, which brings further in-depth understanding of the material. I believe the work is significant in the field and has potential broad impact. Hence, I recommend the paper to be published in Nature Communications. Below are some suggestions and minor questions for the authors to consider. Hopefully they will find them useful in further improving the paper quality. 1. Maybe I missed it but I did not find the rational of using Fe3O4 as the anode for the full cell demonstration. Hard carbon is usually the anode material for Na-ion battery full cells. I am curious why the authors did not choose that and what the performance will look like if hard carbon was used as anodes.
2. The full cell performance in figure 3 is really impressive. However, I just would like to remind the author that it was cycled between 0 and 4V, which will make it not very practical because of the very low discharge cut off voltage. 3. The performance at low and high temperatures is impressive. For the test at -20C, I am wondering whether the cells were charged and discharge at the temperature. I would like to suggest the authors to add some detail information in the Method part. 4. With the excellent electrochemical performance, I am wondering whether the authors have tried high loading electrodes. 5. In Figure 2b, the CV curves are slightly different with 1st cycle and following cycles. Can the authors give more explanations about this phenomenon?

Responses to Reviewers:
We have carefully considered all the comments and questions raised by the reviewers. We took time to plan and carry out additional experiments, which helped to address the reviewers' comments and questions. Newly obtained data are included in the revised manuscript or supplementary information and the relevant discussions have been amended in the manuscript.
We sincerely thank the reviewers for raising relevant questions and constructive comments which have, in our opinion, greatly helped us to improve the quality of the present work. The specific point to point responses are displayed below. Corresponding changes are marked in turquoise in the copy of the revised manuscript.

Reviewer #1 (Remarks to the Author):
I must admit the work is well carried out combining many sophisticated techniques. However, the manuscript lacks sufficient novelty for publication in Nature Comm. 1. In abstract and Introduction, you mention using earth-abundant element like Fe will reduce the cost. Think carefully....it is NOT true. There are two things: materials/precursor cost and processing cost. Fe is prone to oxidation. So we need to use Fe(II) based precursors which is not so cheap. More importantly, you need to anneal the Fe-based compounds in Ar-atmosphere to avoid oxidation. Making large-scale production using Ar-ambience is cumbersome and processing cost will be high (due to Ar). So, the net cost is not cheap. Fe-based compounds can be low-cost only if it is Fe(III) based composition (can be annealed in air) e.g. NaFeO 2 , Na x Fe 1/2 Mn 1/2 O 2 (Komaba et al, Nature Mater). You can get economic system with Mn-analogue e.g. Na 4 Mn 3 (PO 4 ) 2 P 2 O 7 . So, your claim of low-cost cathode (with Fe orthopyrophosphate) is

WRONG.
Answer：Thanks for your profound comment. Indeed, iron is the most earth-abundant of the 3d elements, and it has been widely used since the beginning of human civilization. Since the sodium sources are almost unlimited, the utilization of Fe can further reduce the overall cost of electrode materials. [1][2][3] For large-scale production, the costs of the final cathode products comprise several part, such as the price of the raw material, energy consumption, etc. In the case of Febased layered oxide materials, almost all of them have to be sintered at high temperature (commonly above 850 °C). 4,5 Thus the heat consumption at high temperature (commonly generated by electricity) cannot be neglected. Meanwhile, in our manuscript, the Na 4 Fe 3 (PO 4 ) 2 (P 2 O 7 ) material only underwent a 500 °C calcination process, which is considerably lower than that needed for the Fe-based layered oxides. On the other hand, those Fe-based layered oxide materials also face critical issues, such as their low capacity retention, continuous working voltage drop and capacity drop, air stability, initial cycle Coulombic efficiency, etc. [6][7][8][9] Polyanionic-based materials possess a robust crystallised framework, very stable voltage-capacity performance, and high initial cycle Coulombic efficiency, although their theoretical capacities are generally limited because of their large molecular weights. 3,10 Therefore, the polyanionic-based materials, especially Fe-based polyanionic materials, are excellent supplements to the various kinds of cathode materials in progress towards real applications of SIBs.
Another important issue that concerns the reviewer is the price of Fe(II) raw materials. Indeed, Fe is prone to oxidation, but we checked the prices on websites and found that the price of raw material containing only Fe(II) is just slightly higher than for those containing Fe(III). For instance, Fe (II) oxalate is common iron source for LiFePO 4 , and we checked the price -USD $1760 per metric ton for Fe(II) oxalate and USD $1500 per metric ton for Fe (III) oxalate (please refer to www.alibaba.com). Moreover, there are three aspects that the reviewer might have neglected.
(a) It is necessary to analyses the oxidizability under specific conditions. Fe 2+ is easily oxidised to Fe 3+ under aqueous conditions, but solid state Fe(II) raw materials are relatively stable and not easily oxidised when the Fe(II) raw materials are kept in a dry and inert atmosphere (such as N 2 ). N 2 is much cheaper compared to Ar (Please refer to www.boc.com.au). So, it will not cost much to carefully store these solid state Fe(II) raw material sources.
(b) In our manuscript, we used argon gas as the protective gas, since we consider that the high purity Ar gas can offer a completely inert protective atmosphere during sintering. In the real situation of industrial manufacturing, however, N 2 is normally adopted, it is generated from liquid nitrogen, and the purity of N 2 can reach 99.999 % or even higher. Also, since this Na 4 Fe 3 (PO 4 ) 2 (P 2 O 7 ) material only needs a 500 °C sintering, there is no possibility for any reaction between the raw Fe sources and N 2 . In fact, the commercial LiFePO 4 material needs 600 °C or more to form the final phase, and N 2 is widely used in many world famous LIB factories. Nevertheless, we have redone the identical calcination procedure with the same precursor using high purity N 2 as the protective gas. The charge-discharge curve shows no difference from the one where Ar was used as the protective gas ( Figure Ia and b). So, there is no need for reviewer to be concerned about the problems that Ar may bring.
(c) In iron-based polyanionic materials for both LIBs and SIBs, both iron (II) sources and iron (III) sources are applicable for synthesizing the final products, such as LiFePO 4 and Na 3.12 Fe 2.44 (P 2 O 7 ) 2 .
The Fe (III) can be reduced to Fe (II) in the presence of carbon within an inert sintering atmosphere. So, we think that our Na 4 Fe 3 (PO 4 ) 2 (P 2 O 7 ) material is also suitable for raw Fe (III) sources. We have redone the identical calcination procedure with the precursor using Iron (III) acetate as iron source (with the others identical as described in the Method part) and using high purity N 2 as the protective gas. Surprisingly, the charge-discharge curves showed no obvious difference from the ones using Fe (II) acetate ( Figure Ia and c). Moreover, we assume that the other Fe-based polyanionic materials are amenable to sintering using raw Fe (III) sources, as long as the presence of carbon sources can be guaranteed during sintering (maybe at or above 500 °C). Therefore, this Na 4 Fe 3 (PO 4 ) 2 (P 2 O 7 ) material, or other Fe-based orthopyrophosphate cathode, is really cheap and can be manufactured on a large scale at rather low-cost for the commercial SIB market. 2. This material is well known and well-studied. So, materials point-of-view there is no novelty.
The authors also agree with this fact.
Answer: Thank you for your valuable comment. Kang's group firstly studied the Na 4 Fe 3 (PO 4 ) 2 P 2 O 7 material and obtained its structural details in 2012. 11 Then, they characterized its physical and chemical properties with more techniques. 12 Since then, there have been a few reports on this material from different aspects or perspectives, such as sodium ion diffusion and voltage trends, 13 Mn-doping, 14 potential application in aqueous systems, 15 surface modifications, 16,17 etc. Nevertheless, this material possesses a very stable charge/discharge curve, small volume shrinkage, and outstanding 3D sodium diffusion pathways, so it should receive more attention and more in-depth or comprehensive investigations apart from the reports mentioned above.
In our manuscript, our aim was to discover a polyanionic material that can be generally utilized over a wide range of temperatures. So, we initially focused on the electrodes that possess 3D sodium diffusion pathways. After screening the existing polyanionic material compounds, we found that only two or three materials have 3D diffusion pathways. Therefore, it is meaningful to highlight their unique properties and provide a more complete investigation. Also we nanosized the material and gave the particles a uniform carbon coating, which is important for helping cathodes to demonstrate their best performance. More importantly, we tested its ionic conductivity, and we found that this material has a very similar value to that of the well-known sodium superionic conductor (NASICON)-type Na 3 V 2 (PO 4 ) 3 material. Considering its highly competitive ionic conductivity and 3D diffusion pathways, we think that it is appropriate to assign this material to the NASICON family, which will surely arouse extensive interest from researchers for more integrated investigations. Therefore, although this material has been previously reported, there is still considerable room for further explorations with wider applications, morphology control, and novel concepts.   Fig. 1) are very routine work. It is well performed but does not give any novel finding.

All characterisations (in
Answer: Thank you for your valuable comment. The physical properties and nanoscale characterizations were carried out as shown in Figure 1. Synchrotron XRD is a powerful tool to perform Rietveld refinements with detailed atomic fractions, site positions, etc. Also, FT-IR, XPS, and Raman spectroscopy are the commonly used tools to characterize basic physical and chemical properties. These techniques are important to provide a whole picture of this material from different aspects. We found that the intensity ratio of the D band to the G band (I D /I G ) of the NFPP-E sample (1.04) is larger than that of the NFPP-C sample (0.94), which indicates that the carbon conductivity of the NFPP-E particle surface would be higher than that for NFPP-C.
Besides, by employing the sol-gel method with an appropriate complex agent, nanosized Na 4 Fe 3 (PO 4 ) 2 (P 2 O 7 ) plates can be obtained, which can be verified by SEM and TEM tools. The AFM technique supports our findings in another dimension. The HAADF images firstly present the clear arrangements of atoms at the atomic level, which can help the readers to better understand this material via an authentic and visual approach. Therefore, although the characterisations in Figure 1 are commonly seen and routine work, they are providing indispensable information on the as-obtained nanosized carbon coated Na 4 Fe 3 (PO 4 ) 2 (P 2 O 7 ) material. 5. The structure refinement, magnetic data, XPS, Raman spectroscopy, BET are all very routine work with expected results. Nothing novel.
Answer: Thank you for your valuable concerns. Structure refinement is a basic and preliminary approach to provide detailed analysis of crystal structures. The magnetic data provides supplementary proof that can determine that whether there is any ferromagnetic or antiferromagnetic impurity in the main phase. Also, the effective magnetic moments can be calculated to detect the high/low spin state of iron in this material, which is important to further improve the voltage platform with appropriate element doping. The XPS, Raman spectroscopy, and BET techniques are offering a complete picture of the different properties of this as-prepared material. We think that these tests and data are basic and necessary. We have also refitted the XPS data with more proper Fe deconvolution curves of both samples, as shown below: Figure S4. X-ray photoelectron spectroscopy (XPS) results on Fe for the (a) NFPP-E and (b) NFPP-C samples.
We cannot anticipate what we will get, however, until we run the tests. We need to use the routine work and techniques to obtain a deeper understanding step by step until we find something novel and interesting. These thorough characterizations are required to obtain a deeper understanding of our as-prepared material. 6. The TEM work is good, but again nothing never. All results are expected.
Answer: Thank you for your valuable concerns. The electron transmission tests are based on scanning transmission electron microscopy, which is one of the most advanced techniques and involves state-of-the-art advances. The TEM work in Figure 1 and Figure S6 clearly show the lattice fringes and the atom array in the HAADF mode. It is very helpful for researchers who are not very familiar with this orthorhombic symmetry and also offers a visual method to more conveniently gain an understanding of the atomic arrangement. With the help of TEM, we can also better understand how the synthesis conditions influence the final morphology and the carbon coating situation. Also in Figure 4c and d, we look at the charged and discharged electrodes, and the lattice fringes can be clearly observed, which represent complete proof of the single phase transition during cycling. In addition, we employed EDS mapping to see whether there are any particular element-rich aresa, and we found that both before and after cycling, charged and discharged samples all have a homogeneous element distribution within the nanosized particles. Thus, the obtained TEM related work is in a close accordance with the revealed electrochemical properties. It is important to employ TEM to characterise atomic level properties, and it is also a meaningful and innovative strategy to explore what we cannot see directly in the nano world. You have constructed full cell, but the choice of Fe 2 O 3 is not good (Fig 3). So, the full cell profile is very sloping and have low energy density.
Answer： Thanks for your insightful comment. Since this material possesses the 3D sodium diffusion pathways, outstanding rate performance can be achieved under various synthesizing conditions. Kang's group reported its preliminary electrochemical performance without further particle or morphology optimization. 11,12 Since then, Islam's group explored the potential of other redox centres such as Ni and Co with calculated high voltage trends, 13 and Rojo's group investigated its performance in an aqueous system. 15 Other researchers also did some relevant work on this material. 16,17 Very recently, Barpanda's group explored its potential for thin-film SIBs and potassium intercalation possibilities. 18,19 All the pioneering work is encouraging and shows the outstanding promise of this Na 4 Fe 3 (PO 4 ) 2 P 2 O 7 material for SIBs. We have cited their work in the revised manuscript. In our manuscript, we present the long-term cycling stability up to 4400 cycles, and we have compared all the Fe-based polyanionic cathode materials in Figure   S8. We also emphasised the mid-working voltage retention, which is a vital parameter for energy retention that is often neglected by other researchers. We examined long-cycled electrodes, and good crystallinity was still well maintained. Based on these pioneering results, we developed a wider picture of its outstanding performance, as well as incorporating insigts from GITT study and pseudocapacitance study. For cathode materials, these electrochemical results are meaningful and competitive with other researchers' results.
The reason for using Fe 3 O 4 is that we initially were inspired by Prof. Khalil Amine's work published in Nano Lett. 20 They used Fe 3 O 4 as the anode since hard carbon is severely limited by the applied current density. A high capacity over 250 mAh g -1 can be achieved at 0.05 C or 0.1 C, although, if the current density increases to 1 C or above, only less than 50 % capacity can be obtained. The PPy-coated Fe 3 O 4 nanospheres in our manuscript have shown relative good rate capability. Also the density of Fe 3 O 4 is more than 2 times higher than for those hard carbons and amorphous carbons. As a result, the loading of Fe 3 O 4 in the electrode is far higher than for hard carbon. This translates to a significant increase in the energy density of Fe 3 O 4 on the cell level compared to hard carbon. So, we adapted the PPy-coated Fe 3 O 4 is our manuscript even though its initial cycle Coulombic efficiency is relatively low. Nevertheless, we admit that hard carbon is more widely used to construct SIB full cells. So, we also fabricated full cells using purchased hard carbon (KURARAY Co., Ltd., Japan (Type 2)). SEM images and electrochemical properties of this hard carbon are shown below: It can be seen that the charge/discharge curves of the Na 4 Fe 3 (PO 4 ) 2 (P 2 O 7 )//Hard carbon full cell are not as sloping as for those using PPy-coated Fe 3 O 4 , so the energy density can be improved with an elevated mid-working voltage platform. Nevertheless, the capacity is continuously dropping within the initial 10 cycles. We cycled the fabricated full cell up to 120 cycles at 100 mA g -1 and found that the capacity retention is less than 50 %, and the charge capacity is always higher than discharge capacity for every cycle, which might be the main reason for the continuous capacity drop. The Coulombic efficiency of commercial hard carbon (97 % each cycle) is likely to be another reason for this. Nevertheless, we are still putting much effort into working towards better full cell performance using commercial hard carbon as anode with an optimised electrolyte system, more suitable loading ratio, etc. to be used in our future work.
We have added the corresponding figures and expressions to the revised manuscript. We hope that our reply can answer your questions and concerns.
8. The battery study at high and low temp is new. Good data presented.
Answer： Thank you for your positive encouragement. We focus on the real applications of SIBs, so we need to explore the all-climate performance of the cathode material. We will continuously test this sort of performance under various conditions. 9. Fig 4: Sodium storage mechanism is not novel. Its already published by Kang group (Chem Mater) and it is well known this system undergoes solid-solution (single phase) redox reaction. Answer: Thank you for your critical comment. Indeed Kang' s group has previously conducted ex-situ XRD testing during cycling, and they found that this system underwent a solid-solution redox reaction. It would be more appropriate and direct, however, to exhibit an in-situ test with consecutive variations. We examined the electrodes in both fully charged and fully discharged states with HRTEM, which shows detailed information on the crystal structure, as displayed in Figure 4c and d. We further confirmed the topotactic single phase transition process during the highly reversible cycling. In addition, we found that when the electrode is below the main discharge/charge platform, almost no two theta shifts can be observed, although after the main platform finishes, a relatively dramatic variation appears. The calculated lattice parameter shifts further reflect this phenomenon. Therefore, we think that this in-situ synchrotron-based XRD is of value to provide readers with more subtle relevant information. As for XAS, we also employed in-situ XAS, which provided the first reported such results, to the best of our knowledge. The remaining Na + ions in the crystal structure (about 25 %) can be regarded as belonging to the binding pillars. We noticed that the spectra of both the initial state and discharge state were almost the same, but with an obvious discrepancy compared to the LiFePO 4 reference sample.
This can be ascribed to the individual finger print information on P 2 O 7 and PO 4 groups. Some of the oxidized Fe 3+ remains at +3 and cannot be reversed back to Fe 2+ upon discharging. Their achievable reversibility is excellent, however. These phenomena could not be expected before these XAS experiments, so it was important and meaningful to conduct these experiments with the synchrotron radiation sources. So, this section is also Not novel.
Answer: Thank you for your valuable comments. We have carefully looked at the published results from both Kang's group and Saiful Islam's group, and compared our findings to their results. We found that there are some novel discoveries and methods, as well as new concepts that will be interesting to the readers. We have cited their work in the revised manuscript. Firstly, we employed bond valence method (BVS) calculations for a preliminary examination of the ionic states and diffusion pathways, since it is a well-established tool against experiment. It was found that the isosurface near -3.5 eV (lowest energy regions) is all connected, and it is a very important representation for the 2D or 3D diffusion pathways ( Figures S19-S20). These results provide strong support for conclusions of the following DFT study. Secondly, we found that all 16 sodium sites can be divided into three different types of sodium positions. Their specific E f (binding energy) values have been summarised in Table S3. Unlike the report of Kang's group, we found that the lowest energy barrier is located on the a-axis. Besides, we provided detailed information on the energy barriers with the same and different sodium types, and the 3D diffusion pathways are clearly observed in Figure 5c and d. Thirdly, based on the 3D diffusion pathways and ionic conductivity test results, we consider that this type of material can be regarded as a new member of the existing family of NASICON-type materials. We believe that this new concept will throw light on the researchers' motivations to discover more 3D diffusion-enabled materials in the near future.
Overall, the work is well carried out but lacks sufficient novelty. It can be submitted to some field journal like J Electrochemical Soc/ J Power Sources etc.
Generally speaking, our manuscript reports tuneable nanosized Na 4 Fe 3 (PO 4 ) 2 P 2 O 7 plates via a facile one-step sol-gel method with high phase purity and uniform carbon coating. We also have conducted multiple physical and electrochemical tests, which has provided a comprehensive picture of this material with much potential to be manufactured on a large scale. In addition, the low/high temperature testing and air stability investigation have brought more exciting properties to light, and both the BVS calculations and the DFT study clearly show the 3D sodium diffusion pathways, indicating its high rate capability resulting from various synthetic conditions. It is important to expand the family of the well-known NASICON-type materials because of its competitive ionic conductivity and 3D low energy barriers. We believe that our research will throw light upon the polyanionic-based materials and attract wide attention from more researchers. All the new aspects and characterizations mentioned above point to the sufficient novelty of our study. Hence, the authors think that this study should be published in Nature Communications. We really appreciate the overall reviewing work that has taken up your precious time, and we hope that our revised manuscript will answer your questions and meet your requirements. Thank you again for your reviewing work and patience.   with bipolar MO 6 F 6-x octahedra, while the MO 6 F 6-x octahedra can offer the main electrostatic repulsion that accounts for the various Na + de-/intercalation voltage platforms. The critical factor for a NASICON-type structure is the arrangement of sodium sites and whether the 3D diffusion pathways of Na + can be achieved with relatively low energy barriers. As for Na 4 Fe 3 (PO 4 ) 2 (P 2 O 7 ) material, the P 2 O 7 ditetrahedra are constituted of two corner-linked PO 4 tetrahedra, and the PO 4 tetrahedra share edges with the FeO 6 octahedra. This structure is very similar to those pure NASICON-type structures, and all the sodium ions are located in all the connected channels that enable 3D diffusion, as illustrated in our manuscript. Therefore, we consider that it is appropriate and necessary to expand the family of the NASICON materials. Moreover, we are aiming to add more families of the polyanionic-based materials to the expanded family of existing NASICON structures, so long as they can meet the requirements mentioned above. We also added the corresponding discussion to the revised manuscript.
2. There are several corrections required. (i) The XRD pattern in Figure 3a should be explained what wavelength was used and the XPS data in Figure 3b and c should be fitted to give more detailed information. (ii) The EIS data in Figure S9b and c are not fitted well and correctly.
Please modify them.
Answer：Thank you for your valuable suggestions. We are sorry for the inaccuracies in the manuscripts. We have revised the manuscript according to your comments.  3. Recent publications in Fe-based/polyanionic materials should be cited, compared and discussed appropriately as well.
Answer: Thank you for this valuable comment. We have added the recently published papers relevant to the Na 4 Fe 3 (PO 4 ) 2 (P 2 O 7 ) material (Refw. 50-54). We are glad to see that this material is being paid more and more attention by researchers worldwide. Figure S4    Below are some suggestions and minor questions for the authors to consider. Hopefully they will find them useful in further improving the paper quality. Coulombic efficiency is relatively low. Nevertheless, we agree that hard carbon is the more common choice for SIBs. We fabricated the cells using hard carbon as anode. The anodes were purchased from KURARAY Co., Ltd., Japan (Type 2). The SEM images and electrochemical

Further explanation of
properties of this hard carbon are shown below: We then fabricated the Na 4 Fe 3 (PO 4 ) 2 (P 2 O 7 )//Hard carbon full cell with the loading mass ratio of 1.8 :1 to make the capacity balance according to the individual specific capacities of the electrodes. The anode electrodes were presodiated to reduce the dramatic initial irreversible capacity loss.
The electrochemical performance is displayed below: It can be seen that the charge/discharge curves of the Na 4 Fe 3 (PO 4 ) 2 (P 2 O 7 )//Hard carbon full cell are not as sloping as those using PPy-coated Fe 3 O 4 , so the energy density can be improved with an elevated mid-working voltage platform, although the capacity is continuously dropping within the initial 10 cycles. We also cycled the fabricated full cell for 120 cycles at 100 mA g -1 , finding that the capacity retention is less than 50 %, and the charge capacity is always higher than discharge capacity for every cycle, which might be the main reason for the continuous capacity drop. The Coulombic efficiency of commercial hard carbon (97 % each cycle) is likely to be another reason for this. Nevertheless, we are still putting much effort into achieving better full cell performance using commercial hard carbon as anode with an optimised electrolyte system, more suitable loading ratio, etc. in our future work.
We have added the corresponding figures and expressions in the revised manuscript. We hope that our reply can answer your questions and concerns.
2. The full cell performance in figure 3 is really impressive. However, I just would like to remind the author that it was cycled between 0 and 4V, which will make it not very practical because of the very low discharge cut off voltage.
Answer: Thanks for your professional comment. We are sorry that we did not notice that if we selected the very low discharge cut off voltage, the practical value would be reduced due to the required voltage range. We fabricated the full cells using hard carbon as anode with an appropriate voltage window (4.0 V-0.5 V). We will keep this issue in mind in our further work, including on other types of electrodes when making full cells for SIBs.
3. The performance at low and high temperatures is impressive. For the test at -20 , I am wondering whether the cells were charged and discharge at the temperature. I would like to suggest the authors to add some detail information in the Method part.
Answer: Thank you for your valuable comments. We apologize for the missing explanations of high/low temperature tests in the Methods part. We have modified the manuscript according to your comments. In order to remove your doubt, we have retested the cells at -20 °C using the high/low temperature test box. We have retested four cells, and the data are summarized below: Figure  It can be seen that there are some deviations for the four individual cells, although similar capacity values to the data presented at Figure 3e can be achieved at various current densities ( Figure IIa). Also we have plotted the corresponding charge-discharge curves for the NFPP-E electrode at 1 C, and around 60 mAh g -1 can be obtained at -20 °C ( Figure IIb). We have revised Figure S16 with the newly measured average curves: Figure S16 Comparison of the charge/discharge curves of NFPP-E electrodes at room temperature, 50℃, and -20 ℃, respectively.
We also added the recently published papers to the revised manuscript, although they have obtained similar results. 2 We hope that our added work and explanations can remove your doubts.
4. With the excellent electrochemical performance, I am wondering whether the authors have tried high loading electrodes.
Answer: Thank you for this practical concern. The loading density in the manuscript is about 2.0 mg cm -2 , which is in the middle among the reported electrochemical results for various positive electrodes. To address more practical concerns, we increased the loading mass to ~3.5 mg cm -2 , and the corresponding electrochemical performance is shown below: It can be seen that with the increased loading mass, the NFPP-E electrodes (~3.5 mg cm -2 ) showed slightly decreased C-rate capacities compared to the lower one (2.0 mg cm -2 ) (Fig. IIIa).
Almost no obvious discrepancy can be seen at small current densities (0.1 C and 0.2 C), meaning that the full capacity can be achieved at tiny (negligible) electrochemical polarizations. With high current densities, however, large polarizations can be seen in Fig. IIIb, since the thicker electrode will result in longer distances for sodium ion transport. The noticeable capacity drop is reasonable and predictable, and around 77 mAh g -1 still can be obtained with this high loading mass. We will remember to increase the loading mass on prepared electrodes in our further or other works. Figure 2b, the CV curves are slightly different with 1st cycle and following cycles. Can the authors give more explanations about this phenomenon?

In
Answer: Thank you for your professional question. In Figure 2b, we also noticed that there is a small difference between the first cycle and the following cycles. Generally speaking, this phenomenon can be ascribed to several main reasons: (i). Phase transition during the first charge/discharge process. Some of other polyanionic-based material will undergo this phase transition during the first charge process, such as alluaudite-type Na 2 Fe 2 (SO 4 ) 3 material 3,4 and Na 2 FeP 2 O 7 material. 5 Obvious peak shape changes can be seen directly. These phase transitions are normally driven by the strong electrostatic repulsion created via edge-sharing Fe 2 O 10 dimers.
(ii).The formation of the solid-electrolyte interphase (SEI) layer during the first cycle. The decomposition of electrolyte and side reactions take place in the first cycle, forming a consolidated SEI layer to buffer the subsequent cycles. For this reason, normally there is no peak shape change, while just detectable peak position shifts appear. (iii). Different scan rates. A larger scan rate will result in a stronger current response (more non-faradaic contribution to the total current), whereas some tiny peaks will not be seen due to the generation of larger polarization or stronger double-layer capacitance. In our material in the manuscript, it can be seen that there are no obvious peak shape changes (both in Figure 2b and Figure S9a), and only small peak position shifts can be detected. We also used the same scan rate (0.05 mV s -1 ), so the main reason for the peak discrepancy is the formation of the SEI layer during first charge process. In order to further confirm our proposed reasons, we present the charge-discharge curves of the first two cycles and the corresponding dQ/dV plots ( Figure IVa and b). Clearly the peaks have the same shape with small position shifts. So, we believe that there is no detectable phase transition during the first/second cycle apart from the SEI layer formation, which is the dominant reason for the slight difference between the 1 st cycle and following cycles.