Polyamide membranes with nanoscale ordered structures for fast permeation and highly selective ion-ion separation

Fast permeation and effective solute-solute separation provide the opportunities for sustainable water treatment, but they are hindered by ineffective membranes. We present here the construction of a nanofiltration membrane with fast permeation, high rejection, and precise Cl-/SO42- separation by spatial and temporal control of interfacial polymerization via graphitic carbon nitride (g-C3N4). The g-C3N4 nanosheet binds preferentially with piperazine and tiles the water-hexane interface as revealed by molecular dynamics studies, thus lowering the diffusion rate of PIP by one order of magnitude and restricting its diffusion pathways towards the hexane phase. As a result, membranes with nanoscale ordered hollow structure are created. Transport mechanism across the structure is clarified using computational fluid dynamics simulation. Increased surface area, lower thickness, and a hollow ordered structure are identified as the key contributors to the water permeance of 105 L m2·h−1·bar−1 with a Na2SO4 rejection of 99.4% and a Cl-/SO42- selectivity of 130, which is superior to state-of-the-art NF membranes. Our approach for tuning the membrane microstructure enables the development of ultra-permeability and excellent selectivity for ion-ion separation, water purification, desalination, and organics removal.


Responses to the Reviewer #1:
This paper reported the fabrication of ultrahigh flux NF membrane by incorporation of g-C3N4 in PIP aqueous solution during interfacial polymerization. Comparing with other method reported, this paper did show major advantage in flux increase, as shown in trade-off curve. Therefore I recommend for publication with addressing following problems.

Response:
Thank you for the insightful comments on our paper and helpful suggestions.

Comments 1:
When author check the stability test, why using methylene blue dye instead of salt?

Response:
Thank you for the comment. We used methylene blue dye solution to carry out the long term test was to demonstrate the photo-catalytic cleaning property of our membrane. The fouling behavior of methylene blue is more obvious than salt to investigate membrane performance. We have also done the stability test with a NaSO4 solution. As shown in the Supplementary Figure 19 (which was also given below), in a 72h test salt rejection and water flux were barely changed. It showed that the membrane was stable for use. Figure S19. The stability test of permeance and Na2SO4 rejection for PA-g-C3N4 membrane. 2

Comments 2:
In the preparation of free-standing nanofilm, TMC concentration (0.4%) used for IP interfacial polymerization is abnormally high, why? As normally TMC concentration is used less than 0.2%.

Response:
Thank you for the comment. We agree that normally TMC concentration is less than 0.2 wt%. In this study, we studied different TMC concentrations at 0.135 wt% g-C3N4 to investigate their effects on the interfacial polymerization. The surface morphologies are given below. It can be seen that with the increase of TMC concentration till 0.4 wt%, the nanoscale structure becomes more ordered which increases the surface area and roughness. When TMC concentration reaches 0.5 wt%, the surface becomes less regular, likely due to the extensive crosslinking that limit the formation of nanoscale hollow structure. And the permeance of this membrane at 0.4 wt% was the highest among all 5 concentrations (see Supplementary Table 6) under high Na2SO4 rejection of 99%. The corresponding results have been added into the Supplementary Figure 11 and Supplementary Table   6 in the revised supporting information.

Comments 3:
In the experimental section for interfacial polymerization, with g-C3N4 case, what the concentration for each reagents and additives? Please give it in detail.

Response:
Thank you for the comment. We have provided the experimental details in the Methods session of the manuscript. Revisions are as follows: "Preparation of freestanding nanofilms. Firstly, 2.0g PIP, 0.5g trimethylamine (TEA) and 0.135 wt% g-C3N4 were dissolved in 100 mL water. Secondly, 0.16 mL phosphoric acid was added to adjust the solution pH to 9.
Next, a 0.4 wt% TMC n-hexane solution was prepared. Then, the PIP solution was poured into a petri dish. After that, the TMC organic solution was gently added on top of the aqueous solution to induce interfacial polymerization. After 5 min, a freestanding film was formed that was taken out of the petri dish, washed by DI water for tree times and stored in DI water.
Preparation of PA and PA-g-C3N4 membranes. The PA and PA-g-C3N4 NF composite membranes were prepared by IP on top of the polyether sulfone (PES, MWCO 100 kDa) ultrafiltration (UF) membranes. The aqueous solution was prepared with 0.2 to 2.0 g of PIP and 0.5mL TEA in 100 mL water, and its pH was tuned to 9 by adding 0.16 mL phosphoric acid. To prepare the PA-g-C3N4 composite membrane, g-C3N4 nanosheets were added at the concentrations of 0.075 to 0.145 wt%. The organic solution was prepared by dissolving 0.1 to 0.5 wt% TMC in n-hexane. Then, the PES membrane was dipped into the aqueous phase for 5 min. After being taken out, the excess solution on its surface was removed using a rubbery roller and the support was left at room temperature until the surface appeared dull and dry. Next, the membrane was immersed in the organic phase for 1 min. After that, the membrane was heated in an oven at 80 o C for 5 min, and stored in DI water before use.
Three replicate membranes for each experiment were fabricated and examined to get a reproducible performance.
Membrane performance test. The permeation performance of the membrane was measured on the cross-flow filtration equipment. The effective area of the membrane was 26 cm 2 . The solution temperature was maintained at 25°C by a heat exchanger. In order to achieve steady state, the membranes were pre-pressurized for 2 hours under 6 bar. The flow rate was 1.5 L·min -1 . The concentrations of salts (including single NaCl, Na2SO4) and dyes in feed solutions were 2000 ppm and 100 ppm, respectively.
The water permeance was calculated based on Eq. 1 .

4
(1) where L is the water permeance (L m -2 h -1 bar -1 ), V (m 3 ) is the volume of permeate collected over Δt, A is the effective membrane area (m 2 ), Δt and Δp represent the filtration time (h) and the transmembrane pressure (bar).
( 2) where R is the salt rejection (%),Cp and Cf represent the concentration of permeate and feed solutions, respectively. The salt concentration was quatified by conductivity measurement, and dye solutions were measured by UV-Vis.
In addition, mixed salt tests were performed in a similar way using the mixed solutions of NaCl and Na2SO4, where the total concentrations were kept constant at 2000 ppm while the weight ratios of Cl -:SO4 2were varied.

Comments 4:
What is the chemical nature of PIP/ g-C3N4 interactions, also the nature of hexane/g-C3N4 as mentioned in the paper.

Response:
Thank you for the comment. We investigated the nature of PIP/ g-C3N4 and hexane/g-C3N4 interactions by MD simulations. We used the LAMMPS package to calculate the interaction energy, including Van der Waals force, hydrogen bonding and electrostatic interaction energy between g-C3N4 and other molecules. The results in Fig. 2c provides an overall comparison of such interactions between g-C3N4 and water, hexane and PIP. To gain further sights, The radial distribution functions (RDF, g(r)) of one representative atom in g-C3N4, named as N1, around different atoms in PIP, water, and n-hexane at the start (Fig. 2d) and end (Fig. 2e) of the simulation were computed and given in Fig. 2d. It is noticed that the peak height for N1-H2O is reduced and N1 becomes closely surrounded by the nitrogen atoms in PIP at the end of simulation, likely through hydrogen bonding.
A similar trend is observed with C atom in g-C3N4 (see Supplementary Figure 10). It suggests that the hydrogen bonding between g-C3N4 and PIP may contribute to strong interactions. We have included such discussions in the manuscript.
For the interaction between g-C3N4 and hexane, the nitrogen and carbon in g-C3N4 is found to interact closely with the carbon in hexane (Fig. 2d). It is difficult to determine the exact nature of interactions using the current methods, but based on our understanding, we think it's van der Waals force.

Comments 5:
How is the hollow strand morphology formed? Please using the data to explain the chemical nature and kinetics.

Response:
Thank you for your comment. In this work, based on MD simulations, we provided explanations on the hollow strand morphology in the revised manuscript as follows: "The emergence of ordered structure might be attributed to the retarded diffusion of PIP in the presence of g-C3N4, similar to the local activation and lateral inhibition phenomenon reported in earlier studies 23,35,36 . However, it is interesting that the number of arched channels increases and channel size decreases as g-C3N4 content rises until reaching 0.135 wt%. This is well explained by the tiling effect of nanosheets. PIP diffuses into the hexane phase through the gap between adjacent nanosheets, reacts with TMC at the other side, forming a hollow structure above the nanosheets where the PIP concentration is lean. With more nanosheets at the interface, the gap density is higher and its length is shortened. Consequently, the density and size of arched channels are changed. 0.135 wt% of g-C3N4 is found to give the most uniform structure with a hollow channel. Further increment in nanosheet content may lead to extensive stacking of nanosheets and disturb the ordering of structure." 6

Responses to the Reviewer #2:
This manuscript describes the addition of graphitic carbon nitride (g-C3N4) to interfacial polymerization solutions to enhance the permeability of nanofiltration membranes. Many papers have examined the effects of nanoparticles on interfacial polymerization, but this work shows uniquely high water permeabilities along with nearly complete Na2SO4 rejection. The authors propose that high permeability is due to a low membrane thickness and high surface area that result from g-C3N4 limiting piperazine diffusion to specific regions during interfacial polymerization. The permeation rate combined with the high Na2SO4 rejection is remarkable. Thus, I recommend publication after some significant changes. These include clearly specifying the interfacial polymerization procedure so others can repeat the work, discussing whether the technique is reproducible for multiple batches of membranes, and clarifying parts of the manuscript. Specific comments follow.

Response:
Thank you for your positive evaluation of our work and the insightful comments. Following your suggestions, we have provided details of the interfacial polymerization process so that others can repeat the work and revised our manuscript for better clarity.

Comments 1:
The authors should note what the "g" stands for in "g-C3N4".

Response:
Thank you for the comment. We are sorry for forgetting to note the "g" stands for in "g-C3N4".
The "g" for in "g-C3N4" has been noted in the revised manuscript as follow: "graphitic carbon nitride (g-C3N4) nanosheet."

Comments 2:
The manuscript mentions Cl -/SO4 2selectivity in the introduction, but I never saw any data on chloride permeation. Did the authors do an experiment with NaCl rejection or some mixed-salt experiment? It is not clear how they determined Cl-/SO4 2selectivity. Was the NaCl rejection around 20% as the selectivity values suggest? Do the authors know why the control and g-C3N4 membranes show such high selectivity compared to most NF membranes? 7

Response:
Thank you for the comment. We did experiment with NaCl rejection and some mixed-salt experiment and measured the rejection to Cland SO4 2in the revised manuscript. Mixed salt tests were performed using the mixed solutions of NaCl and Na2SO4, the weight ratios of Cl -:SO4 2 in 2000 ppm mixed solution were varied. The Cland SO4 2concentrations were measured by ion chromotography. The Cl -/SO4 2selectivity was calculated based on Eq. (3). Details of the experiment were given in the Methods session of the revised manuscript.
The results from mixed salt test were provided in Supplementary Figure 18. The Clrejection is around 20%, and the rejection to SO4 2is > 99% in all tests. The Cl -/SO4 2selectivity is 130. The high selectivity is attributed to the negative surface charge on membranes and the suitable pore size.  We explained in the manuscript as follows: " Fig. 4d shows that the MWCO of PA-g-C3N4 membrane is 472 Da, which is slightly larger than that of PA membrane and is consistent with earlier discussions. A narrow pore size distribution with pore radius of 0.364 nm is achieved, which lies right between the sizes of hydrated Clions (0.332 nm) and SO4 2ions (0.379 nm) (Fig. 4e). It is also revealed that the membrane surfaces are highly negatively charged. The membrane is thus able to achieve high selectivity of 130 for Clover SO4 2-

Comments 3:
The resolution in some figures is not high enough.

Response:
Thank you for the comment. We apologize for the problem with figure quality, and has now increased the resolutions of our figures in the revised manuscript.

Comments 4:
The authors should probably reduce the number of significant figures in the permeability value 105.2. The 0.2 is not significant. Also, the manuscript should state how many replicate membranes were examined and if the reproducibility holds for several different batches of interfacial polymerization. This is vital to demonstrate.

Response:
Many thanks for your kind comments. According to your suggestion, we have reduced the number of significant figures in the permeability value 105.2 to 105. Also, three replicate membranes for each experiment were examined and the data were consistent.
The details have been added into the Methods session of the revised manuscript as follow: "Three replicate membranes for each experiment were fabricated and examined to get a reproducible performance."

Comments 5:
I am not sure why the manuscript refers to piperazine as an activator. The trimesoyl chloride is already activated, and piperazine is a reactant. Also, what is the inhibitor to which they refer. I looked up reference 29 but did not see this terminology.

Response:
9 Thank you for your comment. We used the terminology of activator and inhibitor based on other studies in literature; and we apologize for the wrong reference used, which should be reference 23 instead. However, we do agree that the concepts here are blur and hence removed relevant discussions in the revised manuscript. The discussions are revised as 'When g-C3N4 was added into the PIP solution, the monomer diffusivity decreases to 10 -11 m 2 s -1 , one order of magnitude lower than its original value. By real-time online optical monitoring (see Supplementary Figure 8), it is observed that the spreading area of the (PIP+g-C3N4)/TMC system (Fig.1b) is smaller than that of the PIP/TMC system within the same time (Fig.1c), which confirms that the mixing of spreading and diffusion within the (PIP+g-C3N4)/TMC system is slower than that with the PIP/TMC system.
Hence, the diffusion rates of the amino monomers may be adjusted by tuning the g-C3N4 to reach an ideal difference in the diffusion of monomers in both the aqueous and organic phases'.

Comments 6:
The authors discuss that g-C3N4 slows piperazine diffusion through interactions with g-C3N4.
What is the ratio of g-C3N4 to piperazine in the solution? Is most of the piperazine adsorbed on g-C3N4 or is the effect of g-C3N4 primarily at the interface?

Response:
Thank you for the question. The mass ratio of g-C3N4 to piperazine in the water solution is 1:2.5.
As is mentioned in your question, based on our MD simulations, the slowed diffusion of PIP in water might be due to two effects: 1) The interaction (or adsorption) of PIP with g-C3N4; since the quantity of g-C3N4 in water is 2.5 times less (and roughly 2.5 times less in molar ratio) than PIP, the majority of PIP molecules are free. However, even if only part of the PIP are adsorbed onto g-C3N4, they form large complex in water, increase the viscosity of the solution, and may then hinder the diffusion of other PIP molecules through the spatial and viscous effects; hence, the hindrance effect caused by PIP adsorption onto g-C3N4 could be substantial; 2) Nanosheets tend to tile at the interface, which causes direct spatial barrier for PIP diffusion. It is difficult to quantify the relative importance of each effect, as MD only gives qualitative explanations. We tend to believe that both effects play important roles with tiling effects being even more substantial given the low concentration of g-C3N4 in the solution.

Comments 7:
I don't understand Figures 2g and 2h. In Figure 2g, what are the meanings of the colors. In Figure   2h, the two plots do not look that different, and I am not sure of the meaning of the x-axis.

Response:
Thank you for your question. In Fig 2g,  On the other hand, the diffusion is more likely to be prohibited and the interface plane would be closer to the water phase (to the right-hand side) when g-C3N4 is present. We have clarified the information in the caption of Figs. 2h and 2g as "Time dependence of the position of four pieces of g-C3N4 nanosheets along z-axis; and h the PIP number density along z axis in the simulation systems. Z-axis starts from the hexane phase to the water phase as is captured in 2a."

Comments 8:
The TEM images in Figure 3 are difficult to compare.

Response:
Many thanks for your kind comments. We have edited the TEM images in Figure 3 in the revised manuscript.
The corresponding TEM images in Figure 3 is given below: membrane. g-i C1s XPS spectra for g-C3N4, PA, and PA-g-C3N4 membrane respectively. j, k N1s XPS spectra for PA and PA-g-C3N4 membrane respectively. l S parameter as a function of positron energy for PA and PA-g-C3N4 membrane. The concentration of g-C3N4 for membrane preparation was 0.135 wt%.

Comments 9:
Figures 4a, 4d, and 4e need labels on the axes. I am not sure how the authors are determining the pore-size distribution in Figure 4e.

Response:
Many thanks for your kind comments. We are sorry for the missing labels. We have now added the labels on the axes of Figs. 4a, 4d and 4e in the revised manuscript and given below:

13
In addition, we have added the methods for measurement of pore-size distribution in section 3.1 of the revised supporting information.
The corresponding descriptions are also given below: "By calculating the Stokes radius of the spherical solute of interest and neglecting the effect of steric and hydrodynamic interactions between solute and pore space on solute rejection, the pore size distribution of the NF membrane can be expressed as the following probability density function.
where μp is the average pore size of the composite membrane, μp is the Stokes radius of the spherical solute corresponding to a rejection rate of 50.0%, σp is the ratio between the Stokes radius with a rejection of 84.13% to the Stokes radius with a rejection of 50.0%, rp is the Stokes radius of the solutes."

Comments 10:
The photocatalytic cleaning raises questions. How will you get light into a membrane module?
Are reactive species diffusing through the membrane from the underlying g-C3N4 to react with methylene blue? Did the authors perform any control experiments to verify that light is required for the membrane regeneration? Does the same procedure without light but with rinsing give a similar result? I am not sure that this part of the manuscript is necessary, but if included it should contain control experiments.

Response:
Many thanks for your comments. We have done additional experiment to answer the questions and further confirm our explanations: 1) In our experiment, after the membrane was removed from the membrane module, light was used to irradiate the membrane to perform the photocatalytic cleaning. It is a good question how it could be possible to get light into a large membrane module in the real applications. Most flat sheet membrane modules for NF today are spiral wound and it is difficult to dissemble the module for cleaning. In this case, we may adopt a different module configuration, such as plat and frame, which is convenient for photocatalytic cleaning. The photocatalytic cleaning might be cost effective and 14 environmental friendly in the future.
2) For you second question, are reactive species diffusing through the membrane from the underlying g-C3N4 to react with methylene blue? Yes, we think reactive species diffuse through the membrane from the underlying g-C3N4 to react with methylene blue. The reaction between g-C3N4 and organic species has been studied in literature and the mechanism is illustrated in Supplementary  3) For you third question, did the authors perform any control experiments to verify that light is required for the membrane regeneration? Does the same procedure without light but with rinsing give a similar result? Based on your suggestions, we have conducted control experiment to verify the role of light. Specifically, we applied different cleaning methods to the fouled PA-g-C3N4 membrane, including soaking, rinsing without light, and rinsing with light. The SEM images of membrane surfaces after cleaning as captured below show that foulants were removed more efficiently with light irritation than the other two cleaning methods. The existence of light is crucial for membrane cleaning. The corresponding results has been added to the Supplementary Figure 21 of the revised supporting information.

Comments 11:
The procedure for membrane preparation does not gives the amounts of materials dissolved in solutions. Are concentrations the same as for the free-standing film? The free-standing film does not mention phosphoric acid, but membrane fabrication does. The reader needs sufficient detail to 16 reproduce the procedure. What was the effective area of the membranes in the permeation tests?
How long did the permeation occur to achieve steady state?

Response:
Thanks a lot for your kind comments. The compositions of the aqueous phase and organic phase were the same for preparing the free-standing films and the composite membranes. Phosphoric acid was added in both cases. The effective area of the membranes in the permeation tests was 26 cm 2 .
In order to achieve steady state, the membranes were pre-pressurized for 2 hours under 6 bar.
According to your suggestions, we have added the protocol for membrane preparation including concentrations, detailed procedure, the effective area of the membranes in the permeation tests and stabilization time into the Methods session of the revised manuscript.
The corresponding descriptions are also given below: "Preparation of freestanding nanofilms. Firstly, 2.0g PIP, 0.5g trimethylamine (TEA) and 0.135 wt% g-C3N4 were dissolved in 100 mL water. Secondly, 0.16 mL phosphoric acid was added to adjust the solution pH to 9. Next, a 0.4 wt% TMC n-hexane solution was prepared. Then, the PIP solution was poured into a petri dish. After that, the TMC organic solution was gently added on top of the aqueous solution to induce interfacial polymerization. After 5 min, a freestanding film was formed that was taken out of the petri dish, washed by DI water for tree times and stored in DI water.
Preparation of PA and PA-g-C3N4 membranes. The PA and PA-g-C3N4 NF composite membranes were prepared by IP on top of the polyether sulfone (PES, MWCO 100 kDa) ultrafiltration (UF) membranes. The aqueous solution was prepared with 0.2 to 2.0 g of PIP and 0.5mL TEA in 100 mL water, and its pH was tuned to 9 by adding 0.16 mL phosphoric acid. To prepare the PA-g-C3N4 composite membrane, g-C3N4 nanosheets were added at the concentrations of 0.075 to 0.145 wt%. The organic solution was prepared by dissolving 0.1 to 0.5 wt% TMC in nhexane. Then, the PES membrane was dipped into the aqueous phase for 5 min. After being taken out, the excess solution on its surface was removed using a rubbery roller and the support was left at room temperature until the surface appeared dull and dry. Next, the membrane was immersed in the organic phase for 1 min. After that, the membrane was heated in an oven at 80 o C for 5 min, and stored in DI water before use. Three replicate membranes for each experiment were fabricated and examined to get a reproducible performance.' "Membrane performance test. The permeation performance of the membrane was measured on the cross-flow filtration equipment. The effective area of the membrane was 26 cm 2 . The solution temperature was maintained at 25°C by a heat exchanger. In order to achieve steady state, the membranes were pre-pressurized for 2 hours under 6 bar. The flow rate was 1.5 L·min -1 . The concentrations of salts (including single NaCl, Na2SO4) and dyes in feed solutions were 2000 ppm and 100 ppm, respectively."

Comments 12:
Can the authors give uncertainties in Table S6 and note how many different membranes were used in each case? These and other experiments should employ replicate membranes, not just multiple experiments on the same membranes.

Response:
Many thanks for your kind comments. Three membranes were used in each case. The experiments were conducted with replicate membranes. According to your advice, we have added the uncertainties in original Table S6 (now Table S7) of the revised supporting information.

Comments 13:
If concentration polarization is not large, the membrane rejection will increase with flow rate. Is this a possible reason why the membranes show such high Na2SO4 rejection along with high permeability?

Response:
Thank you for the comment. We agree that membrane rejection will increase with flow rate due to reduced concentration polarization. We tested the salt rejection at different flow rates; as shown in Figure R1, the rejection increases only slightly at higher flow rate.
Our tests were done in a round cell (with dimensions given below) at the feed flow rate of 1.5 L min -1 . This is high compared to real applications; but is comparable or higher to some extent than literature work with which we compared our performances with in Fig. 4. Since rejection only decreases slightly at lower flow rate, we may conclude that the high selectivity at high permeance is mainly resulted from the unique membrane structure. Figure R1. Effect of flow rate on Na2SO4 rejection. Figure R2. Illustration of the NF testing cell.
The corresponding results has been added into the Supplementary Figure 20 of the revised supporting information.

Comments 14:
The manuscript suggests that slow spreading "confirms that the diffusion rate of the (PIP+g-C3N4)/TMC system is slower than that of the PIP/TMC system." Spreading and diffusion are not the same. Perhaps the term mixing, rather than diffusion, is appropriate. This seems more of a mixing and spreading phenomenon.

Response:
Many thanks for your suggestion. We agree with you that spreading is not the same as diffusion.
However, mixing seems to be improper since PIP reacts with TMC rather than mixing with it. The optical experiment captured the formation of the polyamide layer, which is related to the movement of PIP into the hexane phase. In addition, we do have NMR data to show the changes in the diffusion rate of PIP. We have revised the relevant statements to "which confirms that the mobility of PIP within the (PIP+g-C3N4)/TMC system is lower than that with the PIP/TMC system" in the revised manuscript.

Responses to the Reviewer #3:
Present work demonstrates fast permeation and selective ion separation by regulation of interfacial polymerization (IP). g-C3N4 is applied for controlled diffusion of PIP by one order of magnitude, thereby forming a nanoscale-ordered hollow cone structure. MD simulation explains the role of g-C3N4 in the IP process to understand the role of membrane microstructure in the nanofiltration process. Although I cannot recommend this manuscript for publication in Nature Communications due to lack of novelty, let me advise the following points for the authors' consideration.

Response:
Thank you for the valuable comments and suggestions. We have revised our manuscript accordingly, and a point-to-point reply is provided below. We hope that we have clarified the novelty of our work, and addressed your other concerns.

Comments 1:
Incomplete literature survey: The authors need to survey previously reported papers such as "Synthesis and characterization of g-C3N4 nanosheet modified polyamide nanofiltration membranes with good permeation and antifouling properties" and "g-C3N4 nanofibers network reinforced polyamide nanofiltration membrane for fast desalination". I suggest mentioning of these works in the introduction section with emphasizing the novelty of the present work in their respect.

Response:
Thank you for the valuable suggestion. The two papers you mentioned are very helpful.
According to your kind advice, we have cited these papers and highlighted the novelty of our work by compared with these papers in our Introduction session.
The novelties of our works mainly include:1) We provide the way to incorporate the g-C3N4 in the IP process to spatially and temporarily modulate the reaction which leads to a nano-scale ordered morphology; 2) The mechanism of the g-C3N4 in IP process and the transport mechanism of water molecules are systematically studied using molecular simulation and CFD; 3) Extraordinary separation capacities, in particular, high permeance along with high ion-ion selectivity for value-added separations, have been achieved.
Comparing to reference 1 ("Synthesis and characterization of g-C3N4 nanosheet modified polyamide nanofiltration membranes with good permeation and antifouling properties"), there are four major differences: 1) The membranes in ref. 1 after the addition of g-C3N4 look rough on the surface, and no nanoscale ordered structure is observed; as a result, the overall membrane 20 performances after addition of g-C3N4 are much lower than our membrane; this might be due to the different way g-C3N4 was prepared and different IP protocol that was employed; 2) In ref.1, they did not explain the underlying mechanisms, while MD is employed in our work to provide fundamental understanding of the role of g-C3N4 in interfacial polymerization and CFD is used to probe into the transport phenomena; 3) Extraordinary separation capacities, in particular, high permeance along with high ion-ion selectivity for value-added separations, have been achieved in our work; 4) Self cleaning properties were not reported in ref. 1 but in our work. These main differences are the most important features that make our work unique among existing work.
Second, the work in reference 2 ("g-C3N4 nanofibers network reinforced polyamide nanofiltration membrane for fast desalination") is essentially different from ours. g-C3N4 was used as the interlayer, neither nanoscale ordered structure nor outstanding selectivity/permeance (comparable to our work) was observed in ref.
On the other hand, in our work, g-C3N4 was mixed with the PIP solution to tune the diffusion and interface spreading of PIP; as a result, nanoscale ordered structure and outstanding performances are achieved.
Again, we greatly appreciate the comment by the reviewer, and we have revised our manuscript as follows: 'Notably, g-C3N4 has been used in prior studies as nanofillers 32 or interlayer 33 during interfacial polymerization. However, this is the first time g-C3N4 was used for spatial and temporal modulation of the reaction to get a nano-scale ordered morphology and extraordinary separation capacities, and the first time the role of g-C3N4 was systematically studied using molecular simulation. Hence, this research demonstrates a unique approach to overcoming the trade-off between permeability and selectivity, providing insight into the design of ultrapermeable and highly selective NF membranes in water purification, desalination and resource recovery.'

Comments 2:
The section titled "Ultrafast permeation and ion-ion separation" requires more discussion. For example, the authors studied separation performance of PA-g-C3N4 as a function of g-C3N4 concentration and failed to explain the reason of decreased flux beyond 0.145 % of g-C3N4.

Response:
Many thanks for your kind comments. In our experiment we found that beyond 0.145 % of g-C3N4, the order of hollow nanoscale surface decreases and correspondingly water transport pathways decreases, which leads to flux decline.  Table 7 gives the permeance comparison of the PA-g-C3N4 membranes with different g-C3N4 concentrations. With the increase of g-C3N4 concentration, the permeance gradually increases, reaching the maximum at 0.135 wt% g-C3N4. At 0.145 wt% g-C3N4, the permeance decreases again, which is consistent with the less regular nanoscale structure captured in Supplementary Figure 12."

Comments 3:
In page 9, line 3, the authors mention that the Cl -/SO4 2selectivity reaches 130 without a technical detail. Experiments carried out for the selectivity measurement need be provided.

Response:
22 Thank you for the comment. In our original submission, we measured the rejections to NaCl and Na2SO4 individually and calculate the selectivity based on Eq. (3). In the revised manuscript, we have done additional experiment by measuring the ion rejections with mixed salt solutions.
Specifically, mixed salt tests were performed using the mixed solutions of NaCl and Na2SO4, where the total concentrations were kept constant at 2000 ppm while the ratios of were varied. The Cland SO4 2concentrations were measured by ion chromotography. The Cl -/SO4 2selectivity was calculated based on Eq.
(3). Details of the experiment were given in the Methods session of the revised manuscript.
The results from mixed salt test were provided in Supplementary Figure 18. The Clrejection is around 20%, and the rejection to SO4 2is > 99% in all tests. The Cl -/SO4 2selectivity can reach up to 130. The high selectivity is attributed to the negative surface charge on membranes and the suitable pore size. Fig. 4f compares our membranes with other membranes reported in literature.
The highlight of our work is boosting the membrane permeance while maintaining high selectivity. We explained in the manuscript as follows: " Fig. 4d shows that the MWCO of PA-g-C3N4 membrane is 472 Da, which is slightly larger than that of PA membrane and is consistent with earlier discussions. A narrow pore size distribution with pore radius of 0.364 nm is achieved, which lies right between the sizes of hydrated Clions (0.332 nm) and SO4 2ions (0.379 nm) (Fig. 4e). It is also revealed earlier that the membrane surfaces are highly negatively charged. The membrane is thus able to achieve high selectivity of 130 for Clover SO4 2- (Fig. 4f) through both size sieving and Donnan exclusion effects."

Comments 4:
The authors presented the self-cleaning property of PA-g-C3N4 as one of the highlights of the work whereas they do not demonstrate or explain the mechanism.

Response:
Thanks a lot for your comment. The reaction between g-C3N4 and organic species has been studied in literature (refs, 37, 38). According to your kind advice, we have demonstrated the mechanism figure as the following. As shown in the figure below, the related mechanism is that electrons (e -) and holes are generated by visible light with g-C3N4. The generated ecan react with dissolved oxygen (O2) to form superoxide radical anion O2 -. Then h + and O2degrade the MB to CO2 and H2O, membrane surface becomes clean again. Figure S22. The mechanism of photocatalytic degradation of methylene blue on the PA-g-C3N4 membrane.
The corresponding results has been added into the Supplementary Figure 22 of the revised supporting information, the corresponding descriptions have been added into the revised manuscript as the following.

Comments 5:
Self-cleaning properties of PA and PA-g-C3N4 need be compared and emphasized. At the same time, the authors may want to consider visual demonstration of the anti-fouling behavior of PA and PA-g-C3N4 via SEM images of PA and PA-g-C3N4 before and after the fouling tests.

Response:
Many thanks for your kind comment. According to your kind advice, self-cleaning properties of PA and PA-g-C3N4 has been compared and shown in Supplementary Figure 20. The results show that self-cleaning properties of PA-g-C3N4 membrane is better than that of PA membrane. we have done the anti-fouling behaviour of PA and PA-g-C3N4 via SEM images of PA and PA-g-C3N4 before and after the fouling tests with light irritation for each round. It can be seen that the anti-fouling behaviour of PA-g-C3N4 membrane was better than that of PA membrane. The corresponding results has been added into the revised supporting information. Second-round photocatalytic g-C3N4 membrane. (b5) Third-round photocatalytic g-C3N4 membrane.