Development of efficient aqueous organic redox flow batteries using ion-sieving sulfonated polymer membranes

Redox flow batteries using aqueous organic-based electrolytes are promising candidates for developing cost-effective grid-scale energy storage devices. However, a significant drawback of these batteries is the cross-mixing of active species through the membrane, which causes battery performance degradation. To overcome this issue, here we report size-selective ion-exchange membranes prepared by sulfonation of a spirobifluorene-based microporous polymer and demonstrate their efficient ion sieving functions in flow batteries. The spirobifluorene unit allows control over the degree of sulfonation to optimize the transport of cations, whilst the microporous structure inhibits the crossover of organic molecules via molecular sieving. Furthermore, the enhanced membrane selectivity mitigates the crossover-induced capacity decay whilst maintaining good ionic conductivity for aqueous electrolyte solution at pH 9, where the redox-active organic molecules show long-term stability. We also prove the boosting effect of the membranes on the energy efficiency and peak power density of the aqueous redox flow battery, which shows stable operation for about 120 h (i.e., 2100 charge-discharge cycles at 100 mA cm−2) in a laboratory-scale cell.


Response to reviewers' comments
We are grateful to all reviewers for their time and constructive comments. Below are the point-topoint response. Reviewers' comments (Blue); Authors' response: black; Revised text in manuscript and ESI (Red).

Reviewer #1 (Remarks to the Author):
The work by Neil B. McKeown, Qilei Song et al constitutes an interesting study addressing a very important topic of our days: storage of energy. The work is basically totally devoted to the use of a spirobifluorene based polymeric membrane as an efficient substituent for the conventional Nafion membrane. Whereas the topic is of relevance since the storage of energy, in particular with RFBs, is nowadays a very important topic driven by the Climate Change Agenda in the search for sources of clean energy. It would be worth to complement with some additional work. Otherwise it would be more likely to be find in a more specific journal as J. Membrane Science. The work is almost totally devoted to the study of a spirobifluorene based polymer of intrinsic microporosity (PIM-SBF) membrane to replace the conventional Nafion. Response: We thank the reviewer for offering valuable comments and suggestions. As the reviewer pointed out, we believe our work will contribute substantially to the development of new ionconductive membranes and advance their applications in a range of energy and environmental processes, particularly those in the renewable energy conversion and storage including redox flow batteries (RFBs). The reviewer had concerns about the cost of production, recovery and reuse and the universality of our membranes in different RFB chemistries. However, we need to explain that the novelty of this work does not lie in the synthetic chemistry itself or membrane manufacturing. The design strategy of making ion-conductive membranes that combine fast ion transport with precise molecular sieving functions enable long-life organic RFBs under mild and environmentalfriendly operation conditions, based on the sulfonation of a spirobifluorene based polymer of intrinsic microporosity (sPIM-SBF). sPIM-SBF is the prototypical PIM-based polymer demonstrating its exceptional performance as an RFB membrane separator. Furthermore, we and other research groups inspired by this work will make cheaper and simpler sulfonated PIM-based polymers for membrane process to match the scales required by the commercial and residential deployment of RFBs.
Detailed explanations and additional experiments have been given to address all the major and minor issues raised by the reviewer. We hope the reviewer will value our efforts and recognize the significance of our work in a broad context of a new generation of ion-conductive PIM membranes and the broad impact on energy conversion and storage.
1) It would be important to estimate the cost of production (with synthesis included) of this polymer so that it can be compared with Nafion. This can be done and added to SI. Response: The cost of polymer synthesis and membrane manufacturing will become a critical issue when we scale up the membranes for large scale applications. The primary goal of our work is to demonstrate the concept of sulfonated PIM membranes to achieve fast and selective ion transport in redox flow batteries. At this early stage, we demonstrated the concept in laboratory scale and our polymer design is still at low TRL level. Further work will be required to scale up the synthesis or developing alternative synthetic chemistry using low-cost monomers. Nevertheless, the production cost of sPIM-SBF-1.40 was estimated as an example based on the cost of raw materials, reagents and catalysts used in the synthesis (cost data were accessed on websites of common chemical vendors on 01 Dec. 2021) as shown in Supplementary Table 13. The final cost would be £ 2.4 per gram for sPIM-SBF polymers and £ 120 per m 2 for sPIM-SBF membranes (50 µm thick). It should be noted that this simple estimation does not include the costs associated with manufacturing and operational costs. We envision that large scale synthesis and manufacturing will further reduce the cost of these polymer membranes.
Furthermore, thanks to the solution processability, PIM polymers can be processed into thin-film composite membranes with a sub-micron layer of sPIM-SBF supported on a low-cost substrate, through well-established manufacturing techniques such as roll-to-roll casting (Jue, M. L. & Lively, R. P. Curr. Opin. Chem. Eng. 2022, 35, 100750). Several m 2 of thin film composite membranes could be prepared from 1 g of polymer, hence reducing the total cost of the composite membranes. Our group has demonstrated the feasibility of making PIM-based thin-film composite membranes for redox flow batteries (Tan, R. et al. Nat. Mater. 2020, 19, 195-202).
Recently, Lively and Finn groups reported SBF-based polymers through the same synthetic procedures for making SBF monomers, and also highlighted the scalability and translatability of these polymers for membrane separation (Thompson, K. A. et al. Science 2020, 369, 310-315;). In addition, sPIM-SBF is the prototypical PIM polymer with exceptional performance as RFB membranes. The structure-property-performance correlations established in the work will provide design rules for further development of sulfonated PIMs, for example, through alternative synthetic chemistries towards more cost-effective PIMs. We have added the following explanation in the revised manuscript: Revised text "(Manuscript, page 15-16) Low-cost membrane separators would have significant practical benefits for the widespread deployment of large-scale energy storage technologies.
Although, our small-scale, multi-step synthesis of sPIM-SBF would not be cost-effective as it stands, the synthesis is scalable (e.g., there is no requirement for chromatographic separation of intermediates) and the cost per unit mass of the polymer would benefit greatly from large-scale production. Based on a simplified cost analysis (Supplementary Table 13), sPIM-SBF membranes are promising as a more economically competitive alternative to Nafion membranes. In addition, we anticipate that the sPIM-SBF would be used as the thin selective layer (< 1µm) on a cheap microporous support, therefore, the important cost per unit area of membrane will be reasonable. Our sulfonation method will also be applicable to other PIMs which will allow for optimization of materials for more economically competitive membranes. …"

Revised note in supplementary information:
The production cost of sPIM-SBF-1.40 was estimated as an example based on the cost of raw materials, reagents and catalysts used in the synthesis (cost data were accessed on websites of common chemical vendors on 01 Dec. 2021). The final materials cost would be £ 2.4 per gram for sPIM-SBF polymers and £ 120 per m 2 for sPIM-SBF membranes (50 µm thick). It should be noted that this simple estimation does not include the costs associated with manufacturing and operational costs. We envision that large-scale synthesis and manufacturing will further reduce the cost of these polymer membranes. 2) One of the advantages of the Nafion membrane is the easiness of its full recover for subsequent uses. The indication that the PIM membranes afford stable battery operations over 2000 charge and discharge cycles does not guarantee that the membranes are recoverable. This should be discussed. Response: We thank the reviewer for the comment. We performed additional experiments to confirm the recyclability of used sPIM-SBF membranes. The examined samples are post-cycling membranes that were then stored in supporting electrolyte solutions (pH = 9) after being disassembled from batteries before lockdown due to COVID-19 pandemic in March 2020 (cycling performance shown in Fig. 4). No sign of degradation or structural change can be observed from NMR, FT-IR spectra and GPC traces (supplementary Fig. 38). These membranes remain mechanically robust after being bent or folded repeatedly. Moreover, they are readily soluble in DMSO and can be re-processed into robust free-standing membranes (supplementary Fig. 39). Further, we have confirmed the stable ionic conductivity of sPIM-SBF membranes through ageing tests of ~400 days (supplementary Fig.  22). These indicate that used PIM membranes can either be directly re-used in a new battery cell or be re-dissolved to cast a fresh film. We have added the following explanation in the revised manuscript:

Supplementary
Revised text "(Manuscript, page 16)…Furthermore, the ease of recovery of membranes from endof-life batteries for subsequent uses constitutes a critical part in the life cycle assessment of a new membrane product. We characterized, by NMR, FT-IR and GPC, sPIM-SBF membranes that were initially used in low-concentration RFBs for more than 2000 cycles (performance shown in Fig. 4) and then stored in supporting electrolyte solutions for around 20 months. No sign of degradation or structural change can be observed from these measurements ( Supplementary Fig. 38). In addition, these membranes remain mechanically robust after being bent or folded repeatedly. Moreover, used sPIM-SBF membranes remain readily soluble in DMSO and can then be re-processed into robust free-standing membranes ( Supplementary Fig. 39). These results indicate that PIM membranes from spent RFB cells could either be directly re-used in a new battery cell or be re-dissolved to cast a fresh film. 3) The sulfonated polymers are well characterized, no doubt. However, characterization of the 2,6-DPPAQ along the cycles (beginning, middle and final) would be helpful to the interpretation of the viability of the RFB working with the PIM-SBF membrane. This is partially done with data in Supplementary Fig. 26, but not under working conditions. It is also specifically mentioned on page 12: "The analysis of the positive and negative electrolyte solutions after battery cycling, using NMR spectroscopy, cyclic voltammetry, and ICP showed that the decomposition of these molecules in near neutral pH conditions is relatively slow.", but then it reports to fig 4 e and f which are schematically and to Supplementary Fig. 36 that shows the CV profiles of redox active molecules after RFB full cell cycling tests. These seem to show some level of changes but are not directly compared to the initial CVs.

Response:
We agree with the reviewer that the stability of redox-active species is of crucial importance for ensuring long lifetime of flow battery systems, and that characterization of redoxactive species along the cycles would provide useful information. 2,6-DPPAQ was initially reported by Aziz and Gordon groups (Ji, Y. et al. Advanced Energy Materials 2019, 9,1900039). Critical properties of 2,6-DPPAQ had been thoroughly studied in their work, particularly the chemical and electrochemical stability. Based on a combination of characterization techniques (e.g., NMR, CV, UV-Vis, symmetric and full cell testing), 2,6-DPPAQ was found to be the most stable redox-active organic molecules for aqueous flow battery in their work. We employed the same testing conditions as the Aziz and Gordon work, including electrolyte pH (i.e., 9) and cut-off voltages (0.5 -1.5 V). Hence, the stability of 2,6-DPPAQ under working conditions would be comparable to that reported previously. Nevertheless, we performed CV measurement for the fresh positive and negative electrolyte solutions (i.e., initial CV) and made comparisons with post-cycling electrolyte solutions. Despite vanadium flow batteries being the most mature flow battery technology, their market penetration is limited by the high cost of Nafion membranes and vanadium. Very few hydrocarbonbased membranes remain stable in vanadium flow batteries due to the presence of highly oxidative VO2 + . In contrast, these low-cost membranes can maintain their long-term chemical stability and low ionic resistance in aqueous organic flow batteries, where Nafion membranes are not selective enough to avoid crossover-induced capacity fade. We envision that sPIM-SBF membranes are likely not sufficiently stable for long-term operation of vanadium flow batteries (e.g., cleavage of the dioxinlinkages and hydrolysis of nitrile groups), but we have demonstrated their superior ionic conductivity, unprecedented selectivity towards organics, as well as stability and recyclability in near-neutral pH flow battery, which would advance the development and commercialization of aqueous organic flow batteries. We have added the following discussion to the main text.
Revised text "(Manuscript, page 16)... In recent years, a wide range of flow battery chemistries have been developed beyond conventional vanadium flow battery, such as neutral or alkaline aqueous organic RFBs, semi-solid RFBs, metal-air RFBs, and non-aqueous RFBs 2-7 . It is challenging to find a one-size-fits-all solution and develop one specific type of membrane separators that finds universal Nafion 115 Nafion 115 applications in all types of flow battery systems. Membrane separators should be designed and tailored to meet specific requirements of each individual RFB chemistry. We envision that sPIM-SBF membranes are likely to be not sufficiently stable for long-term operation of vanadium flow batteries (e.g., cleavage of the dioxin-linkages and hydrolysis of nitrile groups), but we have demonstrated their superior ionic conductivity, unprecedented selectivity towards organics, as well as stability and recyclability in near-neutral pH flow battery, which would advance the development and commercialization of aqueous organic flow batteries. 5) Indication that pH 9 is near neutral pH is erroneous and should be avoided. Neutral pH is 7 -7.5; 9 is an alkaline pH value. Response: We understand the reviewer's comment that the pH value of 9 is generally defined as an alkaline value based on Brønsted-Lowry acid-base theory (Petrucci, R.H., Harwood, W.S. & Herring, F.G. General Chemistry Prentice-Hall 2002, 8th edition, 666). In aqueous RFB settings, however, the pH value of 9 can be regarded as a near-neutral pH value compared with strongly alkaline solutions with pH over 14. Actually, it is commonly found in RFB publications that pH 9 or even pH 12 is classified as the near-neutral region (Ji, Y. Classic aqueous organic flow batteries are operated at a pH=14 (e.g., 2,6-dihydroanthraquinones), while most organic species undergo a number of side reactions that lead to irreversible loss of battery capacity at such a high pH condition. Hence, as an important research direction, efforts have been made to reduce the pH of the electrolyte solution towards neutral or near-neutral pH conditions. As reported by Aziz and Gordon groups, 2,6-DPPAQ suffers from severe alkyl chain cleavage side reactions at pH 14 while becoming more stable at pH 12, and exhibits adequate chemical stability at near neutral pH of 9 (Ji, Y. et al, Adv. Energy Mater. 2019, 9,1900039). Therefore, the operating pH of 9 is classified as near-neutral pH, where redox-active molecules show greatly improved stability as compared with that in strong alkaline condition (e.g., pH 14).

6) BET (Brunauer-Emmett-Teller) surface should not always be abbreviated
Response: Full words of BET has been added when it was first appeared in the main text.
Revised text "(Manuscript, page 6)…although the apparent Brunauer-Emmett-Teller (BET) surface area decreases from 692 m 2 g -1 to 482 m 2 g -1 on increasing the degree of sulfonation (Fig. 2d)…" 7) In the Supplementary information section when the battery lifetime is predicted an energy storage capacity value is given of 2kWh for an active surface area of 1m^2. How the value of energy storage capacity is obtained should be better elucidated.

Response:
The energy storage capacity of 2 kWh is one of our assumptions/targets in the lifetime estimation calculations, instead of any experimental value. Based on the materials utilization ratio and capacity decay rate as measured in our laboratory-scale cells, we calculated how much electrolyte would be needed to achieve an energy storage capacity of 2 kWh as well as how long the battery stack would remain in service before reaching the loss of 20% of the total capacity.

Reviewer #2 (Remarks to the Author):
The authors report on a nano-porous ion-selective membrane for redox flow batteries using large organic molecules as the active species, operating with different electrolytes on either side of the membrane. They claim low crossover of active species through size exclusion as well as Donnan exclusion. They report characterization data of the membrane as well as data obtained by cycling a flow battery. The polymer chemistry reported as well as the polymer characterization is sound. However, there are several aspects of this paper that detract from its publication in Nature Communications: 1. The reported conductivities are quite low in spite of measurement in KCl solution. Response: We respectfully disagree with the reviewer's comment. The values reported in this work correspond to potassium ionic conductivity not proton conductivity, which is normally much higher, due to the higher ionic mobility of proton (36.23 × 10 -8 m 2 s -1 V -1 ) relative to potassium (7.62 × 10 -8 m 2 s -1 V -1 ). In fact, conductivity with values in the range of 0.02-0.04 S cm -1 for sPIM-SBF membranes represents the highest level for potassium conduction as compared to the state-of-the-art ion exchange membranes and ion sieving membranes ( Table 1, Supplementary Table 8 and Table R1).  2. Most data is presented without any error bars (something which absolutely needs to be remedied) Response: We appreciate the reviewer for the suggestion to include error bars in figures. We have updated relevant figures and provided/highlighted the repeated details of corresponding measurements in method section.    3. The conclusion of Donnan exclusion coupled with size-exclusion does not appear to be supported by evidence. The effect appears to be purely that of size exclusion, which is well known and has been previously reported. Response: We agree with the reviewer that size-sieving, instead of Donnan exclusion, is the main feature of sPIM-SBF membranes that gives high performance and distinguishes them from previously reported polymers.

Revised figures in main text:
The size exclusion mechanism dominates the selective transport of anions for sPIM-SBF membranes. A size-exclusion cut-off of ~7 Å was found for sPIM-SBF membranes among a range of anions (i.e., Cl -, NO3 -, SO4 2-, CO3 2-, and Fe(CN)6 4 ), which agrees well with the pore size distribution of sPIM-SBF as measured by gas sorption and molecular simulation. In contrast, the size sieving phenomenon was not obvious for commercial benchmark Nafion® membranes that possess similar amount of sulfonate groups ( Supplementary Fig. 24, Fig. 25, Table 9 and Table 10).
We attribute the superior ion sieving properties of sPIM-SBF membranes to the well-defined ion transport pathways as well as their polymer chains of extreme rigidity. It has been previously reported that semi-rigid glassy polymers show improved selectivity (e.g., sPSF, sPEAK, sDAPP, s-tripPEEK) as compared to Nafion. However, these polymers have negligible microporosity and the formation of ion transport channels still rely on the uncontrolled microphase separation between different polymer chain segments. Importantly, one of our design strategies is the use of PIM polymers to provide rigid backbones and well-defined micropores with fixed chain conformation so as to arise from microporosity with narrow size distribution. We have revised text in the revised manuscript (shown below).
Revised text "(Manuscript, page 11)…Indeed, such low permeability values of ferrocyanide and 2,6-DPPAQ can be considered as crossover-free are mainly attributable to efficient size sieving through the rigid PIM backbones that restrict the thermal motions that result in the opening of a void with sufficient size for these large redox active species to move between free volume elements. Whilst the size-sieving is the primary reason for the highly selective ion transport, Donnan exclusion from the negatively charged sulfonate groups may further reduce the permeability of ferrocyanide and 2,6-DPPAQ anions and enhance selectivity (Supplementary Fig. 18). 4. While the cycling numbers seem to be impressive, they also are more a bit disingenuous -the cycle duration seems to be extremely low -perhaps only minutes -while multiple hours / cycle would be the norm. Unless data can be presented at this scale, not many conclusions about suitability can be drawn. Response: Thank the reviewer for raising this concern. We agree cycle number itself is indeed not informative enough. Hence, following the reviewer' suggestion, we have added cycle time into the figures.
For certain types of RFB systems, such as non-aqueous RFB, hybrid RFB and metal-ion RFB, multiple hours/cycle is the norm due to the high concentration of redox couples (e.g., multiple mole per litre) and/or very low current density that is limited by the sluggish reaction kinetics and slow ion transport. In contrast, for aqueous organic RFBs, lab-scale demonstration of several thousands of cycles with short cycle duration (< 0.5 hour/cycle) is common practice for evaluating new membranes and screening redox-active species (Lin, K.  19,[195][196][197][198][199][200][201][202]. In our work, laboratory scale RFB cell was constructed to evaluate different membranes within a relatively short duration period at a high current (700 mA for 7 cm 2 of membrane). As shown in the main figures, a clear cycling-performance difference can be observed within a relatively short duration period for both lowconcentration and high-concentration RFBs, where sPIM-SBF-1.40 membrane greatly outperformed benchmark Nafion® membranes. We believe the results have proven the key hypothesis and conclusions of this work.

Reviewer #3 (Remarks to the Author):
This manuscript reports a new membrane consisting of spirobifluorene-based polymers with intrinsic microporosity (PIM-SBF) for aqueous organic redox flow batteries (AORFBs). By modulating the sulfonation degree of the polymers, the authors attempted to control their porosity and ion conductivity as well as ion selectivity in the crossover. The new membrane fabricated by one polymer (sPIB-SBF-1.40) showed higher ionic conductivity, KCl permeation rate, and anion selectivity than commercial Nafion 115 membrane, which finally led to higher cycle stability in AORFB using 2,6-DPPAQ and K4Fe(CN)6. This work is well written, and the authors systematically investigated the structure-property relationship between sulfonation degree of the polymers, their physicochemical properties, and membrane characteristics. However, a few issues should be addressed before acceptance. Response: We thank the reviewer for positive comments and clearly pointing out the design strategy of our work.
1. The molecular weights and the PDI values of the polymers used in this study should be reported.  Fig. 2h. The stress-strain curves show adequate mechanical robustness of these membranes. We added GPC characterization in Supplementary Fig.5 and Table  1, and measurement detail in method section (shown below).
Revised text "(Manuscript, page 23)…Gel permeation chromatography (GPC). GPC measurement of original PIM-SBF was carried out by a GPC MAX 1000 system equipped with two Viscotek columns (CLM3012 LT 5000L) and a RI detector (VE3580) using CHCl3 (~ 1 mg mL -1 ) at a flow rate of 1 mL min -1 as an eluent. GPC traces of sPIM-SBF polymers were obtained by a GPC Agilent 1260 system equipped with two PLgel MIXED-C columns (200-2,000,000 g mol -1 , 5 µm) and a RI detector using DMF containing 0.1% w/v lithium bromide (~ 5 mg mL -1 ) at 60 °C at 1 mL min -1 as an eluent. Calibration was performed using narrow dispersity poly(methyl methacrylate) standards." sPIM-SBF polymeric backbones by using TMSCS as the sulfonating agent and performing sulfonation functionalization under mild conditions. Response: Initially, we did attempt liquid state NMR but only got weak signals presumably due to their high viscosity, which is a common observation for PIMs with rigid chains. Hence, we performed solid state NMR instead to characterize the structure of sPIM-SBF. We have now tried liquid state NMR again but adjusted some key parameters to enhance the signal, i.e., increasing the sample concentration (50 mg of sPIM polymers is dissolved in 0.75 mL DMSO-d6), increasing the temperature up to 333.0 K and increasing the number of scans up to 4096.

Supplementary
The obtained liquid state 13 C NMR spectra show similar conclusion as that from solid-state results in terms of sulfonation site, but are indeed more accurate and quantitative. We have replaced the solid state 13 C NMR spectra with the liquid state 13 C NMR spectra in Supplementary Fig. 3 and provide the detailed characterization information in method section (shown below). We observed significantly broadened peaks in 1 H NMR spectra for sPIM-SBF, which is a distinct characteristic for PIM polymers. The broadened 1 H NMR peaks is related to stronger spin-spin interactions between 1 H nuclei in PIMs with fixed polymeric chain conformation leading to shorter spin-spin relaxation times (T2) (McKeown, N. B. Polymer 2020, 202, 122736).
Revised text "(Manuscript, page 22)… 1 H and 13 C liquid state NMR spectra of original PIM-SBF, sPIM-SBF polymers and 2,6-DPPAQ redox molecules were recorded in deuterated solvents using a Bruker AVA 500 spectrometer (500 and 125 MHz, respectively). PIM-SBF and 2,6-DPPAQ redox molecules were tested at 298 K with number of scans of 8 and 128 for 1 H and 13 C NMR. Specifically, high-concentration solution samples of sPIM-SBF polymers (50 mg) in DMSO-d 6 (0.75 mL) were prepared and tested at 333 K. 1 H and 13 C spectra were typically compiled from 32 and 4096 scans, respectively." Supplementary Fig. 3 | Liquid state 13 C NMR spectra and peak assignment of sPIM-SBF polymers. a, Chemical structure of sPIM-SBF polymers. b, Liquid state 13 C NMR spectra of PIM-SBF and sPIM-SBF polymers with varied degree of sulfonation. The resonance at 144.8 ppm (4') was attributed to carbon environment directly bound to sulfonate groups. Gradually increased resonance at 148.2 ppm (4') and decreased resonance at 128.0 ppm (4) indicated that the degree of sulfonation could be controlled by regulating the molar ratio between TMSCS and PIM-SBF in the reaction 4 . 3. The raw data for all NMR measurements should be provided in the supplementary information. Response: We thank the reviewer for pointing out the raw data omission of NMR measurements. The 1 H NMR chemical shifts of monomer 2,2',3,3'-tetramethoxy-9,9'-spirobifluorene (SBF-OMe) and the corresponding sulfonated model compound (sSBF-OMe), and the 13 C NMR chemical shifts of sPIM-SBF polymers were added in the method section (shown below). To make the display of NMR raw data clearer, all the 1 H and/or 13 C chemical shifts of materials were positioned after synthesis procedure in the method section of main text, and their corresponding 1 H and/or 13 C spectra were shown in the supplementary information ( Supplementary Fig. 2, Fig. 3, and Fig. 28). The NMR characterization of the synthesis from 2,2',3,3'-tetrahydroxy-9,9'-spirobifluorene (SBF) monomer precursors to PIM-SBF polymer has been reported on our published papers (Bezzu, C. G. et al. Adv. Mater. 2012, 24, 5930-5933; Bezzu, C. G. et al. J. Mater. Chem. A. 2018, 6, 10507-10514). 4. In this study, it was found that the sulfonation clearly decreased the BET surface area and pore volume of the polymers. However, as the sulfonation degree increased, the ion permeation rate also increased (Fig 3). It is surprising that the increasing rate is steeper in the case of larger anions (Table  S8). Furthermore, the ion selectivity was lowered in the more sulfonated polymers (Fig 3 and Table  S9), despite lower BET surface, pore-volume, and highly negatively charged backbone structures. The authors should provide a more detailed and reasonable explanation for these behaviours. Response: We must emphasise that the BET surface area measurements were obtained with dried membranes, while the ion permeation rates were measured with water-swollen membranes. With the degree of sulfonation increases from sPIM-SBF-0.53 to sPIM-SBF-1.86, the BET surface area decreases from 692 to 482 m 2 g -1 , and micropore volume from 0.217 to 0.167 cm 3 g -1 . The decrease of BET surface area and narrower pore size distribution is likely due to the bulkiness of sulfonate and stronger inter-/intra-chain interactions. However, presence of hydrophilic micropores in aqueous solution leads to swelling of membranes and the expansion of pore dimension. Despite lower BET surface area and pore volume are found in heavily sulfonated PIMs, these PIMs tend to swell more significantly, as supported by the electrolyte uptake experiments. Hence, the size of percolated micropores in sPIM-SBF of high IEC becomes larger and less selective towards different ions of varied size. Therefore, we have added the following explanation in the revised manuscript: Revised text: "(Manuscript, page 7)…Combining the results from the experimental measurements and molecular simulation, we attribute the decrease of BET surface area and narrower pore size distribution to the bulkiness of the sulfonate groups and stronger inter-/intra-chain interactions. Despite their lower BET surface areas and pore volumes, the sPIM-SBF polymers tend to swell following absorption of electrolyte solution. A greater degree of sulfonation enhances the degree of swelling of membranes and the expansion of the pore dimensions. Hence, moderately sulfonated sPIM-SBF is optimal to maintain the sub-nanometer size of electrolyte-filled micropores to provide the exquisite size-selectivity that favours charge carrier ions over larger redox-active species.

Reviewers' Comments:
Reviewer #1: Remarks to the Author: The authors have correctly responded to almost all the questions previously raised. The only questions that are only partially clarified are: "5) Indication that pH 9 is near neutral pH is erroneous and should be avoided. Neutral pH is 7 -7.5; 9 is an alkaline pH value." the response relies on the fact this concept is related to the B-L acid-base concept and that in the field of RFB alkaline pH is close to 14; despite the fact that the near-neutral is indicated, the readers of Nature may consider this has an odd statement. The near-neutral could be maintained but a clear sentence related to the explanation given "Actually, it is commonly found in RFB publications that pH 9 or even pH 12 is classified as the nearneutral region" should be included.
Question 7 "In the Supplementary information section when the battery lifetime is predicted an energy storage capacity value is given of 2kWh for an active surface area of 1m^2. How the value of energy storage capacity is obtained should be better elucidated." is not totally convincing:"The energy storage capacity of 2 kWh is one of our assumptions/targets in the lifetime estimation calculations, instead of any experimental value. Based on the materials utilization ratio and capacity decay rate as measured in our laboratory-scale cells, we calculated how much electrolyte would be needed to achieve an energy storage capacity of 2 kWh as well as how long the battery stack would remain in service before reaching the loss of 20% of the total capacity." But the article is of quality and worth of publication in Nat. Comm.
Reviewer #2: Remarks to the Author: I have reviewed the responses to the reviewers' comments and have the following additional comments: 1. The authors have made a sincere attempt to address the reviewers ' comments. I appreciate this.
2. Regarding reviewer 2 comment on conductivity -the issue was not K ion vs. proton. The issue was more to do with hte measurement being performed in KOH (as opposed to in water with the K ion/counter-ion form of the polymer). Under these circumstances, the value seemed quite low. An easy way to clear this up would be to report the conductivity measured with a the blank cell (no membrane, in KOH) alongside the measurement with the membrane in the cell.
3. The response to the issue of Donnan exclusion is only partly satisfactory -the data is conclusive in that it is size exclusion. The authors acknowledge this in their response but continue to mention in the paper that Donaan exclusion may contribute. I do not think there is any contribution from Donnan exclusion whatsover in this system. Hence, I request the authors to amend the manuscript to remove references to Donnan exclusion as a possible mechanism as I feel this would be scientifically inaccurate.
4. The authors have acknowledged that the cycle time (and not just number of cycles( should be reported, and merely stating 1000s of cycles is disingenuous. I am glad to see that total duration is also now mentioned. However, I caution the authors that the reader usually picks up on cycles and hence, it would not be appropriate to in any way, in the manuscript, claim longevity. I agree that multiple cycles with short duration is a good "accelerated test" and this should be clearly stated. No attempt should be made to sensationalize the "number of cycles without capacity fade" as it looks like each cycle is on the order of a few minutes at most. Do note: regardless of chemistry, the only (narrow) market potential for RFBs will be for long-duration (multi-hour cycles) storage. Any other operational manifestation is not likely to be useful. I feel that if the above comments are properly addressed, this work can move towards publication.
Reviewer #3: Remarks to the Author: The authors well addressed all concerns raised by the reviewers, and the manuscript is now in the final stage for acceptance. But, a few minor issues should still be addressed.
1. I agree that the pH condition where the experiments were conducted in this study can be regarded as near-neutral pH. But, to avoid confusion and thus to help readers from broad fields understand better, the authors are suggested to add exact pH value in the parentheses. For example, nearneutral (pH 9). Those can be added in the abstract part and the main text as well. 2. In the discussion part, the authors compared the performance of their new membrane in an organic RFB with the previously reported membranes. As the authors already concluded in this study, its superior performance was attributed to the exceptionally low crossover of active materials through the new membrane. Although the crossover significantly depends on the size of the active materials, most of the active materials used in the previous studies have smaller sizes, for example, 2,6-DHAQ (see Table 1). For a fair comparison and showing the versatility of the new membrane, the authors are suggested to conduct an ion permeation test or cycle stability test of RFB using 2,6-DHAQ.
We are grateful to all reviewers again for their time and additional valuable comments. Below are the pointto-point response.

Reviewer #1 (Remarks to the Author):
The authors have correctly responded to almost all the questions previously raised. The only questions that are only partially clarified are: "5) Indication that pH 9 is near neutral pH is erroneous and should be avoided. Neutral pH is 7 -7.5; 9 is an alkaline pH value." the response relies on the fact this concept is related to the B-L acid-base concept and that in the field of RFB alkaline pH is close to 14; despite the fact that the near-neutral is indicated, the readers of Nature may consider this has an odd statement. The near-neutral could be maintained but a clear sentence related to the explanation given "Actually, it is commonly found in RFB publications that pH 9 or even pH 12 is classified as the near-neutral region" should be included. Response: We appreciate the reviewer's positive comments and valuable suggestion. To make to the definition of near-neutral condition clearer, we have added the following explanation in the revised manuscript: (Manuscript, page 1, abstract) Importantly, these cation exchange membranes demonstrate ultrahigh conductivity for aqueous salt electrolytes at near-neutral pH (pH = 9), in which both a redox-active anthraquinone and ferrocyanide show long-term stability.
(Manuscript, page 3) In this work, we report new ion-exchange membranes with subnanometer ion transport pathways derived from PIMs and demonstrate their exceptional performance in aqueous organic RFBs operated at near neutral pH conditions (pH = 9, Fig. 1a).
(Manuscript, page 3) Importantly, the membranes demonstrate high conductivity for aqueous electrolytes at near-neutral pH of 9, in which both redox-active anthraquinone and ferrocyanide show long-term stability.
(Manuscript, page 11) In contrast to conventional aqueous organic RFB chemistries operated with highly alkaline solutions (e.g., pH = 14), electrolyte solutions with lower pH (i.e. pH range 9-12), which are commonly termed near-neutral pH in aqueous RFBs, are proven to be beneficial for minimizing the degradation of both redox-active molecules and polymer membranes.
(Manuscript, page 18) In summary, we have demonstrated a new generation of ion-exchange membranes that provide exceptional performance in aqueous organic RFBs operated at near neutral pH conditions (pH = 9). Question 7 "In the Supplementary information section when the battery lifetime is predicted an energy storage capacity value is given of 2kWh for an active surface area of 1m^2. How the value of energy storage capacity is obtained should be better elucidated." is not totally convincing: "The energy storage capacity of 2 kWh is one of our assumptions/targets in the lifetime estimation calculations, instead of any experimental value. Based on the materials utilization ratio and capacity decay rate as measured in our laboratory-scale cells, we calculated how much electrolyte would be needed to achieve an energy storage capacity of 2 kWh as well as how long the battery stack would remain in service before reaching the loss of 20% of the total capacity." But the article is of quality and worth of publication in Nat. Comm. Response: We have added the more detailed explanation on the calculation of energy storage capacity in the revised supporting information.
(Supplementary information, page 47) As demonstrated in this work (Figs. 4 and 5), the typical power density of organic redox flow batteries is 1 kW/m 2 (100 mW cm -2 ) at 0.1 A cm -2 . Given a small flow battery cell with an active surface area of 1 m 2 and rated power of 1 kW, and energy storage for 2 h, the energy storage capacity is 2 kWh. The current is equal to the area multiplied by the current density: 0.1 A cm -2 × 1 m 2 =1000 A; The charge is equal to current multiplied by time: 2 h × 1000 A= 7200 s × 1000 C s -1 =7.2 x 10 6 C. Moles of electrons = 7.2×10 6 C / 96,485 C mol -1 = 74.6 moles of electrons.

Reviewer #2 (Remarks to the Author):
I have reviewed the responses to the reviewers' comments and have the following additional comments: 1. The authors have made a sincere attempt to address the reviewers ' comments. I appreciate this.
2. Regarding reviewer 2 comment on conductivity -the issue was not K ion vs. proton. The issue was more to do with the measurement being performed in KOH (as opposed to in water with the K ion/counter-ion form of the polymer). Under these circumstances, the value seemed quite low. An easy way to clear this up would be to report the conductivity measured with a blank cell (no membrane, in KOH) alongside the measurement with the membrane in the cell. Response: We agree with the reviewer's concern about the deviation of real membranes conductivity caused by the blank cell filled with aqueous electrolyte. Indeed, we have taken this factor into the consideration and the resistance of blank cells had been subtracted to evaluate the real ionic conductivity of sPIM-SBF membranes as reported in the manuscript. In detail, after the EIS measurement of each membrane-loaded coin cell (Type 2032), the membrane was removed and the otherwise-identical cell without any membrane was reassembled to obtain the blank resistance. The resistance of all blank cells was similar with values of around 0.100 Ω cm 2 based on individual 12 measurements.
For apparent ionic conductivity measurement in 1M aqueous KCl, sPIM-SBF membranes with sufficient IEC values (i.e., except sPIM-SBF-0.53) show low area specific resistance (ASR) ranging from 0.244 to 0.698 Ω cm 2 , which are superior to Nafion 115 membrane and even comparable to those of blank coin cells (Table R1). And these low ASR values leads to ultrahigh ionic conductivity with values in the range of 0.01-0.04 S cm -1 for sPIM-SBF membranes, representing the highest level for potassium ion conduction as compared to commercial benchmark membranes, recently reported state-of-the-art ion exchange membranes and ion sieving membranes [1][2][3][4][5][6][7][8] . To help readers clearly identify the low resistance and high ionic conductivity of our sPIM-SBF membranes, we have added the membrane resistance values in the revised supplementary information: Table R1 | Area specific resistance (ASR) of blank and membrane-loaded coin cells, and the derived real ionic conductivity (σ) of sPIM-SBF and Nafion membranes.

Polymer
Area specific resistance (Ω cm 2 ) Apparent ionic conductivity (10 -3 S cm -1 ) c Blank cells a Membrane loaded cells b Membranes