Mechanistic analysis of multiple processes controlling solar-driven H2O2 synthesis using engineered polymeric carbon nitride

Solar-driven hydrogen peroxide (H2O2) production presents unique merits of sustainability and environmental friendliness. Herein, efficient solar-driven H2O2 production through dioxygen reduction is achieved by employing polymeric carbon nitride framework with sodium cyanaminate moiety, affording a H2O2 production rate of 18.7 μmol h −1 mg−1 and an apparent quantum yield of 27.6% at 380 nm. The overall photocatalytic transformation process is systematically analyzed, and some previously unknown structural features and interactions are substantiated via experimental and theoretical methods. The structural features of cyanamino group and pyridinic nitrogen-coordinated soidum in the framework promote photon absorption, alter the energy landscape of the framework and improve charge separation efficiency, enhance surface adsorption of dioxygen, and create selective 2e− oxygen reduction reaction surface-active sites. Particularly, an electronic coupling interaction between O2 and surface, which boosts the population and prolongs the lifetime of the active shallow-trapped electrons, is experimentally substantiated.


Reviewer #5 (Remarks to the Author):
This is a detailed investigation (mostly experimental, with some calculations) on the photocatalytic evolution of H2O2 using polymeric carbon nitride (PCN) based photocatalysts that have been treated with NaSCN in order to add sodium cyanaminate moieties. The paper shows an enhancement of H2O2 evolution over non-treated PCN materials. The enhancement is quantified by a table in the SI, showing an approximately sixfold increase in H2O2 production rate over the past attempts show, a list which is restricted to different varieties of PCN.
I have a variety of questions and criticisms of the paper. Some are fundamental and some are technical.
1) I do not see a clearly quantified reason why this study reflects a high-profile advance (Nature Communications) rather than a contribution to the chemical literature that is, while valuable, essentially a technical report. It is not clear that the reported sixfold increase will make a difference between what is an interesting technical observation and a true advance that would (as indicated in the introduction) sway industry away from its conventional processes. How does the process compare to other processes in the literature? The substances used/synthesized here are not new, so there is no new understanding of the fundamental processes and/or components of the system either.
2) The mechanistic interpretation of the results, while approached for multiple angles, is, ultimately, not stringent. I recognize that this is a difficulty that anyone working on PCN materials must face. The atomic structure of the actual materials in question, especially the nature of active sites, is simply not understood in detail -nor is it possible to determine this atomic structure with certainty by any existing experimental or computational techniques, to the best of my understanding. Any model must therefore remain qualitative and speculative, except for the overall observations that can be extracted directly from experiment or from theory. But even given these difficulties, some questions remain inherently open based on the present paper.
2a) The paper assumes that O2 is being converted to H2O2 -but no evidence is given that it is indeed O2, and not something else, that is being converted. What is the source of this O2? What is its partial pressure / concentration? There is glycerol in this system. Why is it not glycerol that reacts and, in the process, releases H2O2? There are other interesting ingredients in the system (KOH, HClO4, perhaps others) and the role of the oxygen-containing anions is not discussed.
So, evidence that it is indeed O2 that is being converted, and not something else, should be provided.
But in any case, I missed what is the O2 source and how it is controlled. In a quantitative chemical study, it would be essential to control, characterize and report the reagents appearing on either side of the reaction. 2b) Structural assumptions made in the discussion of the material. The paper is careful in its text descriptions of PCN derived materials. In particular, it is good to see that a variety of melon, rather than a hypothetical H-free C3N4 material, forms the foundation of at least the computational understanding.
Nevertheless, Figure 1a shows a depiction of a hypothetical structure of PCN-NaCA-n that is, in my view, unsubstantiated in this paper and has been subject to an extensive discussion in the literature. The crux is that the C3N4 like structure shown here likely cannot be made and, to my knowledge, there is no firm evidence in the literature that it ever has been made.
Among the references cited (indirectly) on this point: Refs 37, 38, and 47, which are used as evidence, are early and predate this debate. In these papers, the existence of a fully C3N4 like condensate was essentially a plausible assumption, but later debunked. Ref. 35 is a review and Ref 36 also presents no structural evidence. Ref. 35 states specifically: "Unfortunately, the crystal structure of heptazine-based CN is still not very clear." Furthermore, there is at least one reference (Chemistry of Materials 29 (10), 4445-4453) that shows from simple thermodynamic considerations that the C3N4 like structure shown Figure 2a (right) cannot arise under plausible thermodynamic conditions. In short, I think the unsubstantiated structural hypothesis of heptazine based C3N4 condensation (Fig 2a right)  2c) Similar for "the improved polymerization degree increases layer buckling, ..." in the discussion ... there is no evidence for this in the paper.
2d) I agree with much of the qualitative discussion of the experimental results (trapped charges, charge migration). One thing that remains unclear to me is the SPV spectrum in Fig 7b. The abrupt onset at 0.8 mu-s is not clearly explained and the explanation in terms of charge diffusion from bulk to surface is not clear to me at all. The sharp onset does not look like diffusion, which is a gradual process. It is difficult to understand why there is some time constant here with no signal at all. 2e) I believe that the computational data shown are inconclusive and not well substantiated. I should prequel this by saying that I understand well that the models itself need to remain conceptual since no experimental structural evidence regarding the actual reaction site or atomic geometry exists at all, nor is it clear which statistical ensemble of reaction sites should be considered. So the conceptual models themselves are certainly necessary and the existence of the PCN sites shown is not implausible.
However, even then, there are several unsubstantiated assumptions and technical omissions in the theory, which raise doubts: -How plausible is the presence of Na+ in the site shown? This is an equilibrium with H2O, or so I understand, and Na+ is rather soluble. What is the expected concentration of Na+ in such a site, compared to Na+ in solution? There are multiple other ionic species around. A thermodynamic analysis of the probability of Na+ occupying the sites in question should be provided. (The presence of Na+ is a critical prerequisite for the energetics claimed and so this assumption should be subtantiated.) -What is the spin state of the adsorbed O2? This would be important.
-What is the spin state of any other intermediates considered? -How were the O2 adsorption sites determined? There are many possible adsorption sites -was a search for other possible adsorption sites performed?
-What are the charges found on individual species in these simulations? are they physical? If these overall simulation cells are electrically neutral, the presence of different ionic moieties still implies a significant role of charge transfer. However, the PBE functional used here (or any GGA) is known to describe charge transfer unphysically.
-In particular, the drastic increase of the O2 adsorption energy in the presence of Na+ / NCNremains unexplained from a fudamental point of view. Given that the O2 molecule bonds so strongly, some significant charge transfer is likely. But this charge transfer could be entirely an artifact of the simulation -again, charge transfer is simply not described correctly by GGAs. A much more exhaustive analysis of the simulations would be necessary to clarify these issues.
-Is 2 nm of "vacuum" really enough to electrostatically decouple different sites with strong dipolar characteristics from one another? 2 nm is not much at all. There should be some electrostatic interaction between the supercells considered.
-There is no "van der Waals interaction" in the PBE functional (or in any GGA). This is very well known. Given that the bonding here has van der Waals character at least in parts, their absence is a technical error that could invalidate the simulations altogether.
- Figure 8 shows "free energies". And indeed, given that the energies of several reagents are partial pressure / concentration dependent, free energies should be used.
However, the authors say nothing regarding partial pressure of O2 during the reaction, either in experiment or in the theory part. The correct way to couple the free energies of reference gas phases into such simulations is well known, e.g., Physical Review B 65 (3), 035406. Did the authors do this?
If not, the numbers should be corrected and the appropriate analysis should be provided.
-Without availability of the geometries used in the computations (all computational steps) the simulations will not be reproducible. Since numerous public repositories are now available to deposit such information, all pertinent input files should be made available. R3 / 60 (2) In the revised manuscript, we employed ICP-OES to analyze the overall sodium content in the sample; and the weight percentage of sodium in PCN-NaCA-2 is 6.9 %.

Revisions:
Revisions in the manuscript on page 3: The sodium content of PCN-NaCA-2 was determined to be 6.9 wt.% by inductively coupled plasma atomic emission spectroscopy (ICP OES).

Comments:
"4. Continuing with XPS, in the N1s spectrum, I really believe that what is fitted as just one peak in the 400.1-eV region should be definitely fitted with more than 1, as the shape of a peak doesn't properly fit the spectrum (top spectrum of the d graph). Can this N1S spectra be revised?"

Responses:
Following the reviewer's suggestions, N 1s peaks in the spectra were deconvoluted into three peaks (398.6, 400.0, and 401.2 eV), and the fitting curve matches much better with the original N 1s signals (Figures S1c and S1d). The interpretation of the XPS spectra is revised as well in the supporting information.

Revisions:
Revised Figure S1b in Supplementary Information on page S8:

R4 / 60
Comments: "5. Porosity measurements are highly recommended to increase value to the SEM and TEM images observed in the Supplementary Information document."

Responses:
The samples were characterized with nitrogen physisorption. As shown in Figure S4, PCN has pores with sizes of 2 − 6 nm, and PCN-NaCA-1 presents sharper size distribution curve centered at 3.5 nm. and PCN-NaCA-2 has similar pore size distribution to PCN, but much lower BET surface area, e.g., 83.2 m 2 /g for PCN versus 11.9 m 2 /g for PCN-NaCA-2. PCN-NaCA-3 shows wide pore size distribution and BET surface area of 18.1 m 2 /g. The surface area and pore structure are not decisive factors for the photocatalytic H 2 O 2 production performance, as PCN-NaCA-2 shows the lowest BET surface area, but highest photocatalytic activity among all the photocatalysts. Figure S4 in Supplementary Information on page S12:  (Table S1), and pore size distribution ( Figure S4) information have been added in the revised supporting information. From the SEM/TEM characterizations, the morphology differences in micrometer scale is not obvious. By HRTEM, PCN-NaCA-2 is the only sample that could be observed with clear layer stacking structures in nanometer scales.

Revisions:
(1) Figure S3 and notes in Supplementary Information on page S11: Notes: SEM images shows that these samples have similar morphology; and due to the poor crystallinity of these samples, layer-stackings structures are not observed by HRTEM.

Responses:
It is in Figure S1, panels g and h show the EDS mapping images of sodium and nitrogen elements, respectively, in sample PCN-NaCA-2. (1) There are technologies for observing the molecules on a surface with atomic-level roughness, such as scanning tunneling microscope (STM). While on the surface of nanoparticles, such as the case here, observing a surface-adsorbed dioxygen molecule is rather challenging, to the best of our knowledge. We are unable to clearly describe the details on the surface of a nanoparticle in atomic-level at the current stage.
(2) In this work, the samples were prepared, stored, and handled in air, i.e., chemical/physical R7 / 60 modification of the surface by oxygen, if any, has already happened before the spectroscopy measurement. We discovered the significant differences in transient absorption spectra under vacuum and oxygen atmosphere. These spectroscopic observations experimentally verify the impact of dioxygen on the excitation and charges trapping process of the photocatalyst, which is recently described in an theoretical simulation investigation (Nat. Commun. 12, 320 (2021)).

Comments:
"9. Can the author clarify which useful information or conclusion is extracted from the transient photovoltage (TPV) characterization? It seems that it was included in the article because it was done more than because its relevant to the project."

Responses:
(1) The amount and the life-time of charge carriers are critical to an efficient photocatalytic oxygen reduction reaction. We thus employed TPV for monitoring the charges recombination behavior dynamically. It is found that the photovoltage signal can last for 2.5 ms, which is much longer than that of PCN (0.6 ms). Accordingly, the depletion of charge carriers by recombination is significantly attenuated by construction of the sodium cyanaminate moiety on polymeric carbon nitride framework.
This result is conducive for understanding the rationale behind the superior performance of PCN-NaCA-2 in photocatalytic H 2 O 2 production.
(2) In the revised the manuscript, we have added more discussions on the TPV results as well as the implications to the photocatalytic performance.

Revisions:
(1) Revisions in the manuscript on page 13: The dynamics behavior of the charge carriers is further investigated by transient photovoltage (TPV) characterization. For both PCN and PCN-NaCA-2, there are two photovoltage peaks (peak a and peak b, Figure 7b), which are, respectively, attributed to drift and diffusion of the photo-induced charges. 83,84 PCN-NaCA-2 presents a remarkably strong and negative peak b lasting for 2.5 ms before decaying to zero, speaking for the diffusion of large number of photo-induced electrons to the surface after photons excitation event. The photovoltage (PV) characterizations demonstrate, obviously, that the depletion of the charge carriers by recombination is significantly attenuated by the construction of the sodium cyanaminate moiety on polymeric carbon nitride framework. (2) Revised Figure 7b in the manuscript on Page 13: (1) We apologies for missing the information on O 2 -TPD measurement. In the revised supporting information, there is detailed descriptions on O 2 -TPD measurement.

R8 / 60
(2) Inspired by the reviewer's comments, we conducted TGA-IR-MS analysis on the sample under oxygen atmosphere to check the possibility of other changes on the surface. As shown in Figure S31, there is a small weight loss of 2.1 % in the TG curve. The infrared spectra observe the release of H 2 O during heating, and the mass spectra also confirms that water is the only species evolved during heating. The evolved H 2 O comes from the surface adsorption, since the sample was stored in air and did not pre-treated before measurement.

R9 / 60
(3) The samples were prepared by polymerization reaction under high temperatures (e.g., 400 o C or 600 o C, please refer to METHODs in the revised manuscript). In the O 2 -TPD measurement, the sample was pretreated at 300 o C in helium flow to remove the surface adsorbates. After the adsorption equilibrium with a O 2 (2 %)/He pulse, the physically adsorbed oxygen was removed by He flow; the TPD measurement then started with the temperature program, and desorbed oxygen was monitored by thermal conductive detector (TCD).

Revisions:
(1) Revisions in Supplementary  (2) Figure S31 in Supplementary  Notes: There is a small weight loss of 2.1 % in the TG curve. The infrared spectra observe the release of H 2 O during heating, and the mass spectra also confirms that water is the only species evolved during heating. The evolved H 2 O comes from the surface adsorption, since the sample was stored in air and did not pre-treated before measurement. These results indicate that pretreatment in helium flow at 300 o C is enough for removing the surface adsorbed molecules in O 2 -TPD. Responses: (1) For an in-depth understanding of critical factors governing the solar conversion efficiency, a series of techniques were employed for a step-by-step analysis of the complicated processes involved in the photocatalytic H 2 O 2 production, e.g., photons absorption, charge separation and diffusion, and surface reactions.
(2) In the revised manuscript, we have added more necessary explanations to clarify the reasons and conclusions.

Revisions:
(1) Revisions in manuscript on page 10: As the non-emissive trapped electrons have the potential for participating chemical conversion reaction on the surface. We then focus on the status of the non-emissive trapped electrons and their interaction with the surface adsorbed dioxygen on PCN and PCN-NaCA-2.
(2) Revisions in manuscript on page 12: For initiating oxygen reduction reactions, the photo-induced charge carriers must diffuse to the surface active-sites. The amount of the photo-induced charges and the charges diffusion properties are of great importance, and thus be investigated by surface photovoltage (SPV) spectroscopy.
(3) Revisions in manuscript on page 13: The photovoltage (PV) characterizations demonstrate, obviously, that the depletion of the charge carriers by recombination is significantly attenuated by the construction of the sodium cyanaminate moiety on polymeric carbon nitride framework.
(4) Revisions in manuscript on page 14: It is difficult to further reveal the reaction mechanism by monitoring the transient intermediates on the surface via experimental methodologies. However, computational quantum chemistry offers a platform that makes the prediction of the surface-structure dependent reaction mechanism feasible. 89 Density functional theory (DFT) calculation of the ORR reaction mechanism (both 2e − and 4e − ORR pathways) is therefore performed for further understanding the rationale behind the superior performance of PCN-NaCA-2 for H 2 O 2 production.

Responses:
We appreciate the reviewer for these valuable comments. We have carefully considered the comments and made intensive revisions accordingly, as shown in the revised manuscript and Supplementary Information.

Comments:
The authors claim that their sodium cyanaminate treatment creates p-type and n-type regions. I am somewhat skeptical that this cyanaminate treatment will indeed engender sufficiently high doping densities to achieve true n and p regions in the traditional sense. I find it instead more likely that the resulting material that the authors have obtained comprises a larger distributions of sheet sizes are particle sizes that may lead to a distribution in the energy landscape that causes additional charge localization following photoexcitation."

Responses:
(1) We appreciate the reviewer's very constructive comments. We fully understand the reviewer's skepticism on the description of conductivity-type in the manuscript; and we are very interested in the proposed mechanism by the reviewer on the impact of sheet/particle sizes on the distribution of energy landscape. We, indeed, would like to explain the data from the reviewer's perspective, however, the theoretical or experimental basis is not enough for supporting such a novel proposal so far.

R13 / 60
(2) In this work, the sodium content in PCN-NaCA-2 was determined to be 6.9 wt.%, stating that the amount of sodium cyanaminate moiety in the polymeric carbon nitride framework is not negligible.
We measured the M-S curve in dark condition; and most importantly, the correlation between the slope of M-S curve and the conductivity-type is well-established. We therefore explain the experimental data accordingly.
(3) We will keep the reviewer's mechanism in mind, and further consider it in the future work based on available experimental/theoretical data. Meanwhile, we introduced this idea partially in the revised manuscript.

Revisions:
(1) Revisions in the Abstract: The structural features of cyanamino group and pyridinic nitrogen-coordinated soidum in PCN-NaCA promote photon absorption, alter the energy landscape of the framework and improve charge separation efficiency, enhance surface adsorption of dioxygen, and creates highly selective 2e − ORR surface-active sites.
(2) Revisions in the manuscript on Page 3: Introduction of sodium cyanaminate moiety by molten-salt treatment creates the electron-withdrawing cyanamino-goup and coordinative interaction between sodium and pyridinic nitrogen, and thus alters the distribution of the energy landscape of the carbon nitride framework.
(3) Revisions in the manuscript on Page 13: The p-type conductivity might be attributed to the strong electron withdrawing property of the cyanamino-group and the coordinative interaction between sodium and pyridinic nitrogen in the framework. These structural features could lead to a unique energy landscape that may promote the charge separation under illumination. Moreover, three data points are generally insufficient to support the assertion that there are no two-photon effects contributing to the observed optical signal when using such short laser pulses with high peak powers."

Responses
(1) Considering the reviewer's comments, the plots is in linear scale in the revised Figure 6.
(2) We agree with the reviewer that two-photon excitation could happen under high power density, and more data points are usually necessary. However, in this case, as shown in Figure S27, the data plots significantly deviate from a quadratic scaling as would be expected for two-photon excitation, indicating that two-photon excitation is not relevant to the power densities used in our experiment.

Revisions:
(1) Revised Figure 6 in the manuscript on page 11: (2) Figure S27 and notes in Supplementary Information on page S28: Notes: the data plots significantly deviate from a quadratic scaling as would be expected for two-photon excitation, indicating that two-photon excitation is not relevant to the power densities used in our experiment.

Comments:
"The authors say that the surface photovoltage signal is proportional to the amount of trapped charge.
However, this doesn't seem correct. It seems like the photovoltage signal will likely either be quadratic or logarithmic with charge density depending on what the energy profile of the density of tail states looks like.
In any event, I believe that this assertion warrants further discussion."

Responses:
(1) We appreciated this constructive comment, and apologize for our wrong description on the SPV signal intensity.
(2) The structure of the polymeric carbon nitride is complicated, and the description of the excitation behaviour by mathematical models is so far unavailable. At the current stage, we can only roughly analyse the differences in charge carrier amount and diffusion properties between samples by comparing the intensities of their SPV signals.
(3) We revise the description on the intensity of SPV signal as shown below.

Revisions:
Revisions in the manuscript on page 12: The intensity of SPV signal is positively related with the amount of photo-induced charges as well as

Reviewer #3 (Remarks to the Author):
Comments: "In this paper, the authors describe the utilization of modified carbon nitride as photocatalyst for oxygen reduction to hydrogen peroxide. The results are encouraging and the authors seriously attempt to reveal the reaction mechanism. The weak points are the materials synthesis and final structure, stability studies (including reasons for possible instability of the photocatalyst), lack of other and more sustainable hole scavengers, and other points as state below. Overall, I think that the work has the potential to be published in Nature Comm after rigorous revision."

Responses:
We acknowledge the reviewer for the constructive comments and the manuscript is improved after an intensive revision based on the reviewer's suggestions. (1) We analyzed the evolved gases during the materials synthesis process on thermal gravimetric analyzer-gas chromatograph-mass spectrometer (TGA-GCMS). The major components of the evolved gases were identified. We thus proposed a conversion mechanism based on the results and the related organic conversion reactions in literatures.
(2) The results are discussed in the revised manuscript, and Figures S6-S8 are added in the Supplementary Information.

Revisions:
(1) Revisions in the manuscript on page 3: By thermal gravimetric analysis-gas chromatography mass spectrometry (TGA-GC-MS), the major components of the evolved gases during synthesis were identified (Figures S5-S7); and the mechanism of the conversion of amino group to cyanaminate moiety was proposed accordingly ( Figure S8). (4) Figures S7 in Supplementary Information on pages S14:  Figure S7. The mass spectra of every retention peaks in the chromatogram in Figure S6.

Responses:
(1) In the revised manuscript, we discussed the status of the glycerol production in the growing bio-diesel industry. The glycerol by-product is surplus, and the utilization of this biomass derivative is important.
(2) We choose glycerol as the electron/proton donor also for the following advantages: (1) glycerol oxidation by photo-induced holes presents fast reaction kinetics, efficiently supplying electrons/protons to the selective oxygen reduction reaction; (2) glycerol and its oxidation products are non-toxic and easily bio-degradable, and we propose the utilization of produced H 2 O 2 aqueous R21 / 60 solution containing glycerol oxidation intermediates directly in the environmental remediation, such as the reductive elimination of the highly toxic hexavalent Cr ( Figure S20).
(3) In anaerobic reaction (photocatalytic H 2 evolution), PCN-NaCA-2 shows a protons reduction performance that is similar to PCN ( Figure S25); while in the aerobic reaction with bisphenol A (BPA) as the electron/protons donor, both the H 2 O 2 production rate and the BPA degradation rate on PCN-NaCA-2 are around five times of that on PCN ( Figure S24); in the presence of glycerol as the electron/proton donor, the H 2 O 2 production performance on PCN-NaCA-2 is more than 10 times of that on PCN. In the flow-reactor under sufficient photons irradiation and with glycerol as the proton/electron donor, PCN-NaCA-2 presents superior H 2 O 2 production activity, which is 24.6 times of that on PCN. These experiments indicate that the potential of the unique surface-active sites for 2e − ORR on PCN-NaCA-2 could be fully developed on condition that sufficient electrons/protons are available.

Revisions:
(1) Revisions in the manuscript on page 5: Developing a proper process of consuming and valorizing glycerol well matches the market need. (2) Revisions in the manuscript on page 7: The photocatalytically produced H 2 O 2 aqueous solution is an enviornmental benign and efficient reductant for elimination of the highly-toxic hexavalent Cr ( Figure S20).
(3) Revisions in the manuscript on pages 7-8: The distinction between the photocatalytic performances of PCN and PCN-NaCA-2 varies with the reaction conditions. In the flow-reactor under sufficient photons irradiation and with glycerol as the proton/electron donor, PCN-NaCA-2 presents superior H 2 O 2 production activity, which is 24.6 times of that on PCN. In the photocatalytic H 2 O 2 production coupled with BPA (Bisphenol A) degradation, the initial H 2 O 2 production rate and BPA degradation reaction rate constant on PCN-NaCA-2 are 5 times of those on PCN ( Figure S24); While in the anaerobic H 2 evolution reaction, PCN-NaCA-2 and PCN exhibit very close photocatalytic activity in terms of H 2 production rate ( Figure S25). It is proposed that PCN-NaCA-2 possesses unique surface-active sites for efficient outputting of the photo-induced electrons to surface adsorbed dioxygen, and its potential for H 2 O 2 production can be maximized in the presence of sufficient supply of electron/proton ( Figure S25).
(4) Figure S20 and notes in Supplementary  Responses: (1) The content of Na in PCN-NaCA-2 is 6.9 wt. % by ICP-OES analysis, while when it was dispersed in the 50 mL solution, the sodium ion concentration in solution is only 4.2 ppm (sodium distribution: in catalyst / in solution = 85 / 15).
(2) The chemical environment of nitrogen and sodium in PCN-NaCA-2 was investigated with x-ray adsorption spectroscopy (XAS). It is experimentally and theoretically evidenced that the sodium is coordinatively interacting with four pyridinic nitrogen from two adjacent tri-s-triazine units. All this  indicates that the sodium cation is chemically bonded within the framework, rather than just physically adsorbed on the surface. This part is discussed in the revised manuscript.

Revisions:
(1) Revisions in the manuscript on pages 3 − 4: The sodium content of PCN-NaCA-2 was determined to be 6.  46,47 indicating no interaction between these nitrogen atoms and sodium. Furthermore, the chemical environment of the sodium was characterized with X-ray absorption spectrometer ( Figure S10). Sodium in PCN-NaCA-2 presents a K-edge absorption profile that is different from that of NaSCN and NaSCN/PCN, but is very similar to the profile of sodium diformylamine. The XAS results are indicating that the sodium ion is strongly interacting with PCN-NaCA-2 framework via the coordination with pyridinic nitrogen, which matches well with the theoretical simulation. (2) Revised Figure 2 in the manscript on page 4: (3) Figure S10 in Supplementary Information on page S16: Comments: "4. A comparison to PCN after heat treatment at 400 C should be given as well. In addition, how can the authors exclude the creation of N vacancies after post-heating to 400 C (as shown many times before)?
The latter may be the reason for the absorption shift and enhanced catalytic activity."

Responses:
(1) PCN was prepared under 600 o C in nitrogen atmosphere. Additional 3 hours heat treatment under 400 o C in nitrogen flow is unable to obviously change the surface properties.
(2) Following the reviewer's suggestions, we treated PCN under 400 o C for 3 hours (PCN-400). As shown in Figure 1a, PCN-400 and PCN were employed as the control samples, and they exhibited the same performance in the photocatalytic H 2 O 2 generation.

Revisions:
(1) Revisions in the manuscript on page 5: Responses: (1) Long-term photocatalytic reaction was conducted. As shown in Figure S17, H 2 O 2 concentration increases with the irradiation time for 10 hours, and H 2 O 2 concentration reached 11.1 mM. The photocatalyst was separated for post-characterization.
(2) The cycle performance of PCN-NaCA-2 in the photocatalytic H 2 O 2 production reaction was examined.
As shown in Figure S16, the 8 th recycled photocatalyst still exhibited good photocatalytic performance that is 88.1 % of initial run.
(3) The sodium content of the photocatalyst after 10 hours reaction was determined to be 1.8 wt. % by ICP-OES.
(4) PCN-NaCA-2 after 10 h of photocatalytic reaction was collected and characterized by XPS. The changes in N1s and C1s signal was discussed in the revised supporting information.

Revisions:
(1) Revisions in the manuscript on pages 5-6: PCN-NaCA-2 shows stable performance for recycling, and at 8th run, the photocatalytic performance is 88.1% of the initial run ( Figure S16). Long-term running stability of PCN-NaCA-2 was examined in a batch reaction, and H 2 O 2 concentration reaches a remarkable value of 11.1 mM after 10 h R28 / 60 irradiation ( Figure S17). The characterizations of the recycled photocatalyst indicate that long-term photocatalytic reaction leads the loss of the amine moiety as well as sodium ion in the framework ( Figure S18).
(2) Figure S16 in Supplementary Information on page S20: Figure S16. (3) Figure S17 in Supplementary Information on page S21: Figure S17. Long term running performance of PCN-NaCA-2 in photocatalytic H 2 O 2 production. Reaction conditions: 20 mg photocatalyst was dispersed in 50 mL aqueous solution with glycerol contents of 3.5 wt. % was charged in the photoreactor and irradiated with solar simulator.
(4) Figure S18 and notes in Supplementary  Notes: for N1s signal, the deconvoluted peaks at 400.1 eV decreases as compared to that of as-prepared PCN-NaCA-2, which might result from the changes in the amine moiety of PCN-NaCA-2 after long-term running under irradiation; for C1s signal, there is no obvious changes except that the increased peak intensity at 284.9 eV, which could be attributed to the surface deposited adventitious carbon after the reactions involving organics. The sodium content after long-term irradiation was determined to be 1.8 wt.% by ICP-OES.

Responses:
We appreciate the reviewer for the suggestions. We agree with the reviewer and revised the discussions on absorption spectra.

Revisions:
Revisions in the manuscript on page 9: The absorbance at 450 -500 nm might result from the excitation to the defects states below the conduction band.

R30 / 60
Comments: "7. How can they avoid the oxidation of Glycerol by hydroxyl radicals? It may inhibit that reaction after a short time as the peroxide will be consumed."

Responses:
(1) As shown in Figure S15., the photocatalytic H 2 O 2 decomposition reaction was found to be very slow.
(2) Glycerol plays the role of electron/proton donor for the H 2 O 2 production, and what we are expecting is the efficiency of H 2 O 2 production; the oxidative degradation of the glycerol is not concerned in this work.
(3) Moreover, the selectivity of electron/proton towards the formation of H 2 O 2 production is determined to be >93% ( Figure S23), i.e., in this photo-redox reaction, O 2 to H 2 O 2 is predominant in the reduction side and glycerol oxidation by hole is the majority in the oxidation side.

Revisions:
(1) Revisions in the manuscript on page 5: The hydrogen peroxide decomposes slowly on PCN-NaCA-2 and PCN ( Figure S15), which contributes positively to the H 2 O 2 accumulation in the reaction system.

Responses:
The degradation mechanism of the glycerol in the photocatalytic reaction was studied via intermediates analysis. Based on various identified intermediates, we proposed the mechanism of the glycerol degradation ( Figure S22).

Revisions:
(1) Revisions in the manuscript on page 7: The biomass-derived glycerol serves as the electron/proton donor for O 2 to H 2 O 2 conversion. The glycerol degration process was investigated via reaction intermediates identification. By gas chromatogaphy-mass spectrometry (GC-MS), four major intermediates, e.g., dihydroxylacetone, glyceraldyhyde, glyceric acid, and glycolic acid, are identified. Based on these major imtermediates, the glycerol degradation process by hole oxidation is proposed accordingly ( Figure S22).
(2) Figure S22 in Supplementary Information on page S25: R32 / 60 Figure S22. Scheme of the proposed reaction mechanism of dioxygen reduction and glycerol degradation. The glycerol oxidation reaction mechanism was proposed based on the GC-MS identified intermediates, e.g., dihydroxylacetone, glyceraldyhyde, glyceric acid, and glycolic acid.

Comments:
"9. A summary of the outcome of the TAS and SPV will be useful for the reader (also as an illustration)."

Responses:
In this work, fs-TAS shows the dynamics of the trapped electrons; and SPV measurements reveal the R33 / 60 information on the amount of charge carriers reaching the surface. They are consistent in demonstrating that the construction of the sodium cyanaminate moiety significantly improves the charge separation efficiency. Considering the constructive comments, in the revised manuscript, there is a short summary on these results.

Revisions:
Revisions in the manuscript on page 12: As shown in Figure 7a "10. The title is slightly misleading -in this paper the authors don't really identify the factor for…"

Responses:
A few words in the title has been revised.

Revisions:
Revision on the title of the manuscript:

Responses:
We appreciate the reviewer for the following constructive suggestions, and we have carefully designed experiments and addresses all the concerns, and made detailed revisions on the manuscript and Supplementary Information.

Comments:
"1. Fig.S1e the NMR of PCN-NaAC-2 did not show any obvious differences compared with that of the PCN sample. Can the author enlarge the specific part to show the peak at 171 ppm?"

Responses:
The shoulder peak at 171 ppm has been enlarged in Figure S1.

Revisions:
Revised Figure S1e  For instance, how much of the raw glycerol is oxidized to glyceraldehyde after the photochemical reaction?
Can the concentration of glycerol be lower considering that 0.38M is quite high?"

Responses:
(1) As shown in the glycerol oxidation reaction mechanism analysis, there are various intermediates, which are including 2e/2H + , 4e/4H + , and more electrons/protons transfer reaction products. We thus examined the selectivity of proton/electron towards O 2 to H 2 O 2 conversion with 4-methylbenzyl alcohol as the electron/proton donor, due to the accuracy in quantifying the 2e − oxidation product (4-methylbenzaldehyde). The selectivity of electron/proton towards the formation of H 2 O 2 production is determined to be >93% ( Figure S23), i.e., in this photo-redox reaction, alcohol oxidation is supplying proton/electron to the conversion of O 2 to H 2 O 2 .
(2) The impact of the glycerol concentration on the hydrogen peroxide generation performance was examined and discussed in the Supporting Information. As shown in Figure S11, the H 2 O 2 production performance increases with the glycerol concentration; the impact becomes mild when the glycerol concentration is higher than 3.5 wt. % (0.38 M). For a detailed mechanistic investigation on the photocatalysts, we evaluated the catalytic performance in the presence of sufficient electron/proton R36 / 60 donor (0.38 M / 3.5 wt. % glycerol).

Revisions:
(1) Revisions in the manuscript at page 7: As there are multiple intermediates/products in glycerol oxidation reaction, selective oxidation of 4-methylbenzyl alcohol to 4-methylbenzaldehyde was employed for further examining the H 2 O 2 selectivity. As shown in Figure S23, H 2 O 2 production selectivity (in terms of the molar ratio beween

4-Methylbenzaldehyde was quantified by GCMS with calibration curve and 1,4-dicyanobenzene as the internal standards.
(3) Figure S11 in Supplementary Information on page S17:

Comments:
"3. Following the previous question, the use of glycerol as sacrifice agent would increase the cost and bring impurities to the final H2O2 production. Glycerol itself is a useful chemical. Do the authors think that the produced H2O2 will be more valuable? Besides, how to purify the H2O2 from the final mixture for further usage?"

Responses:
(1) In the rising biodiesel industry, especially in Europe Union, glycerol is the byproduct and its yield accounts for 10 wt.% of the biodiesel production (https://www.ebb-eu.org/stats.php). Although glycerol is a versatile chemical, the increasing production and limited consumption makes glycerol surplus. To develop the utilization of glycerol is quite important. with 10 mM H 2 O 2 in the reductive elimination of Cr(VI). As shown in Figure S20, the mixture is capable of eliminate Cr(VI) efficiently.

Revisions:
(1) Revision in the manuscript on page 5: Developing a proper process of consuming and valorizing glycerol well matches the market need. (2) Revision in the manuscript on page 7: The photocatalytically produced H 2 O 2 aqueous solution is an enviornmental benign and efficient reductant for elimination of the highly-toxic hexavalent Cr ( Figure S20).
(3) Figure S20 and notes in Supplementary Information on page S23:

Figure S20. Reductive elimination of Cr(VI) by photocatalytically generated hydrogen peroxide. Reaction conditions: 1.8 mL H 2 O 2 solution with concentration of 10 mmol produced by PCN-NaCA-2 photocatalysis was charged into a silica cuvette. 200 uL of Cr(VI) solution with concentration of 400 ppm was added in the cuvette and mixed by magnetic stirrer. The concentration of Cr(VI) was determined by the absorbance of the solution at 340 nm with a calibration curve.
Notes: Reductive conversion of the highly toxic hexavalent Cr to less toxic trivalent Cr is one of the most important approaches for treating hexavalent Cr contamination. Hydrogen peroxide is an efficient and environmental-benign reductant for hexavalent Cr conversion: Responses: In the revised manuscript, we analyzed the photocatalytic H 2 O 2 decomposition behavior on PCN and PCN-NaCA-2. As shown in Figure S15, 2 hours light irradiation in the absence of dioxygen led to less than 9 % decrease in the H 2 O 2 concentration for both PCN and PCN-NaCA-2, and PCN-NaCA-2 exhibited slightly faster degradation rate than PCN.

Revisions:
(1) Revisions in the manuscript on page 6: The hydrogen peroxide decomposes slowly on PCN-NaCA-2 and PCN ( Figure S15), which contribute positively to the H 2 O 2 accumulation in the reaction system.
(2) Figure S15 in Supplementary Information on page S20:  (1) By employing EPR technique, we have qualitatively confirmed the presence of superoxide radicals ( Figure S21a).
(2) To evaluate the contribution of the superoxide route to the H 2 O 2 formation, superoxide dismutase (SOD) was used for accelerating the superoxide radicals dismutation reaction for H 2 O 2 production.
As shown in Figure S21b., the addition of SOD does not obviously change H 2 O 2 production rate, i.e., superoxide dismutation reaction contributes negligible to the overall H 2 O 2 production, and one-step 2e transfer is the predominant pathway.

Revisions:
(1) Revisions in the manuscript on page 7: There are two pathways for H 2 O 2 production, e.g., one-step two electrons transfer pathway and superoxide radicals involved pathway. For analyzing the H 2 O 2 formation mechanism, the contribution of each pathway in the reaction system was evaluated. As shown in the electron spin resonance (2) Figure S21 in Supplementary  PCN-NaCA-2 shows optimum photocatalytic H 2 O 2 performance in weak basic conditions, and 2e − ORR selectivity measurements with RRDE was also conducted in the NaOH solution. We thus performed the theoretical simulation in the alkaline condition.

Responses:
(1) We appreciate the reviewer for very detailed and constructive comments; and the revisions based on these comments substantially improve the manuscript.
(2) After an intensive revision, we would like to briefly emphasize the background and the significance of this work.
There is no doubt that solar energy conversion is of great importance for a sustainable energy future. (3) In this work, we are dedicated to revealing the rationale behind molecular structure-efficiency relationship in a superior photocatalytic H 2 O 2 production system. All the events, from photons excitation step to the final H 2 O 2 molecule formation step, are systematically examined, during which some previously unknown structural features and interactions are experimentally substantiated.
 As described in the reviewer's comments, the detailed structures of various polymeric carbon nitride materials are so far controversial, due to the lack of substantiated structural information.
In this work, the sodium in the carbon nitride framework, was evidenced to be interacting with four pyridinic nitrogen from two adjacent tri-s-triazine units by both XANES characterizations and theoretical simulations, which greatly promotes our understanding on the nature of polymeric carbon nitride materials. Meanwhile, the outcomes are very informative for rational design of materials for efficient solar-to-chemical conversion processes.

Revisions:
(1) Revisions in the Abstract: The overall photocatalytic transformation process is systematically analyzed, and some previously

R45 / 60
Comments: "2) The mechanistic interpretation of the results, while approached for multiple angles, is, ultimately, not stringent. I recognize that this is a difficulty that anyone working on PCN materials must face. The atomic structure of the actual materials in question, especially the nature of active sites, is simply not understood in detail -nor is it possible to determine this atomic structure with certainty by any existing experimental or computational techniques, to the best of my understanding. Any model must therefore remain qualitative and speculative, except for the overall observations that can be extracted directly from experiment or from theory.

But even given these difficulties, some questions remain inherently open based on the present paper."
Responses: (1) We appreciate the reviewer for understanding the challenges in this work.
(2) After considering the comments from all the reviewers, we conducted intensive experimental and theoretical simulation works. The revised manuscript now reveals more substantiated structural details.

Comments:
"2a) The paper assumes that O2 is being converted to H2O2 -but no evidence is given that it is indeed O2, and not something else, that is being converted. What is the source of this O2? What is its partial pressure / concentration? There is glycerol in this system. Why is it not glycerol that reacts and, in the process, releases H2O2? There are other interesting ingredients in the system (KOH, HClO4, perhaps others) and the role of the oxygen-containing anions is not discussed.
So, evidence that it is indeed O2 that is being converted, and not something else, should be provided.

But in any case, I missed what is the O2 source and how it is controlled. In a quantitative chemical study,
it would be essential to control, characterize and report the reagents appearing on either side of the reaction."

Responses:
(1) Oxygen was bubbled in the photocatalyst suspension, and the dissolved oxygen is converted to H 2 O 2 .
By controlling the atmosphere, O 2 is proven to be the oxygen source of H 2 O 2 production.
(2) It is experimentally evidence that the electrons/protons from alcohol to O 2 reduction for H 2 O 2

R46 / 60
production present a selectivity of >93%, stating that the alcohol is the electron/proton donor for the O 2 reduction to produce H 2 O 2 .

Revisions:
(1) Revisions in the manuscript on page 5: As expected, the oxygen partial pressure directly impacts the H 2 O 2 production performance; air atmosphere lowers the H 2 O 2 production, and in the nitrogen atmosphere, trace H 2 O 2 is produced from the residual dissolved oxygen in water after nitrogen flushing ( Figure S14).
(2) Figure S14 in Supplementary Information on page S19: Figure S14. Comparison of the photocatalytic H 2 O 2 production performance in O 2 , air, and N 2 atmosphere. Reaction conditions, the same to that in figure S13, except that the atmosphere is controlled. Reaction conditions: 10 mg photocatalyst, 50 mL aqueous solution with glycerol concentration of 3.5 wt.% was mixed by ultrasonication and irradiated by solar simulator for 45 min.

Comments:
"2b) Structural assumptions made in the discussion of the material. The paper is careful in its text descriptions of PCN derived materials. In particular, it is good to see that a variety of melon, rather than a hypothetical H-free C3N4 material, forms the foundation of at least the computational understanding." Nevertheless, Figure 1a shows a depiction of a hypothetical structure of PCN-NaCA-n that is, in my view, unsubstantiated in this paper and has been subject to an extensive discussion in the literature. The crux is that the C3N4 like structure shown here likely cannot be made and, to my knowledge, there is no firm evidence in the literature that it ever has been made.

Responses:
We appreciate the reviewer for the constructive comments; and we have revised the scheme of the chemical structure of PCN-NaCA-n, which is now similar to PCN in linear melon structure.

Revisions:
Revised Figure 2a in the manuscript on page 4:
there is no evidence for this in the paper."

Responses:
Following the abovementioned revision on the proposed structure of PCN-NaCA-2, this description has been removed from the manuscript.

R48 / 60
Comments: "2d) I agree with much of the qualitative discussion of the experimental results (trapped charges, charge migration). One thing that remains unclear to me is the SPV spectrum in Fig 7b. The abrupt onset at 0.8 mu-s is not clearly explained and the explanation in terms of charge diffusion from bulk to surface is not clear to me at all. The sharp onset does not look like diffusion, which is a gradual process. It is difficult to understand why there is some time constant here with no signal at all."

Responses:
(1) We acknowledge the reviewer for this comment and apologies for the unclearly description of the SPV/TPV experiments and data interpretation. We added some marks/notes on the TPV spectra as well as more data interpretation in the revised manuscript for promoting the reader's understanding on the data.
(2) There are two peaks in the TPV spectra for both PCN and PCN-NaCA-2, the first peak (peak a in Figure 7b) at 0.8 μs rising sharply and is resulting from the drift of the charge carriers. The second peak (Peak b in Figure 7b) comes from the diffusion of the charge carriers.
(3) We have described the details of the equipment for collecting the TPV data in Supplementary   Information. Due to the resolution of the equipment, the retardation time is unable to be monitored.
The time constant before the peaks conveys no information related with the photo-physic process of the samples. For avoiding misunderstanding, we added a note in Figure 7b, indicating the photovoltage signal appears as soon as the incidence of laser pulse.

Revisions:
(1) Revised Figure 7b in the manuscript on page 13: Responses: (1) By ICP-OES (Inductively coupled plasma atomic emission spectroscopy) analysis, PCN-NaCA-2 has a sodium content of 6.9 wt.%; and in the equilibrated aqueous suspension with PCN-NaCA-2 concentration of 0.2g/L, nearly 90 % of the sodium is distributed in PCN-NaCA-2, indicating a strong interaction between sodium and the carbon nitride framework.
(2) The position of sodium is screened for a reasonable configuration of PCN-NaCA. Three possible positions for sodium are proposed, and the optimization processes affords the same final configuration, in which the sodium interacts with four pyridinic nitrogen from two adjacent tri-s-triazine units.
(3) The chemical environment of sodium and nitrogen in the framework was further characterized with XANES (X-ray absorption near edge structure). In nitrogen K-edge XANES measurement, the blue shift and intensity enhancement of the π* resonances at 399.3 eV (pyridinic N) speaks for the charge transfer and interaction between pyridinic N and Na. The Na K-edge XANES data demonstrates that the chemical environment of sodium is different from the surface adsorbed sodium ion, but presents some similarities to the sodium in sodium diformylamine with chemical bond between sodium and nitrogen.
(4) The experimental results match well with the outcomes in theoretical simulation.

Revisions:
(1) Revisions in the manuscript on page 3: The sodium content of PCN-NaCA-2 was determined to be 6. there is interaction between the sodium ion and four pyridinic nitrogen atoms of two adjacent heptazine units (Figures 2b, S9). Nitrogen K-edge X-ray near-edge structure (XANES) measurements were thereafter used to investigate the interaction (Figure 2c). PCN shows typical π* resonance at 399.3 eV corresponding to pyridinic N of the tri-s-triazine moiety; 42,43 most importantly, blue shift and enhancement of the pyridinic N peak is observed, demonstrating the presence of coordination interaction and charge transfer from pyridinic nitrogen to sodium. 44,45 Meanwhile, there is no shift for the other π* resonances (401.2 eV, amino; 402.3 eV, graphitic), 46,47 indicating no interaction between these nitrogen atoms and sodium. Furthermore, the chemical environment of the sodium was characterized with X-ray absorption spectrometer ( Figure S10). Sodium in PCN-NaCA-2 presents a K-edge absorption profile that is different from that of NaSCN and NaSCN/PCN, but is very similar to the profile of sodium diformylamine. The XAS results are indicating that the sodium ion is strongly interacting with PCN-NaCA-2 framework via the coordination with pyridinic nitrogen, which matches well with the theoretical simulation.

Revisions:
Computational methods in Supplementary Information on page S5-S7.

R54 / 60
Comments: "-How were the O2 adsorption sites determined? There are many possible adsorption sites -was a search for other possible adsorption sites performed?"

Responses:
(1) Sodium cyanaminate moiety plays an essential role for the superior performance in the photocatalytic H 2 O 2 production in the experimental investigations; and this work mainly focus on the mechanism behind the superior performance initiated by sodium cyanaminate moiety.
(2) In addition to the systematic spectroscopy investigations, the theoretical simulation is employed for further understanding the role of cyanaminate moiety in ORR. The sites closing to sodium and cyanamino-group (Sites 1 -4 in Figure S32a) was thus screened for O 2 adsorption.
(3) It was found that O 2 is unable to adsorb on sites 2 and 4, but the adsorption is feasible on sites 1 and Notes ： Sodium cyanaminate moiety plays an essential role for the superior performance in the photocatalytic H 2 O 2 production in the experimental investigations; and this work mainly focus on the mechanism behind the superior performance initiated by sodium cyanaminate moiety. The theoretical simulation is employed for further understanding the role of cyanaminate moiety in ORR. The sites closing to sodium and cyanamino-group (Sites 1 -4) was thus screened for O 2 adsorption. It was found that O 2 is unable to adsorb on sites 2 and 4, but the adsorption is feasible on sites 1 and 3 with E ad (adsorption R55 / 60 energy) of 0.32 eV and 0.39 eV, respectively. O 2 adsorption on site 3 was selected as the optimum initial structure for the following calculation.

Comments:
"-What are the charges found on individual species in these simulations? are they physical?
If these overall simulation cells are electrically neutral, the presence of different ionic moieties still implies a significant role of charge transfer. However, the PBE functional used here (or any GGA) is known to describe charge transfer unphysically."

Responses:
(1) The Mulliken charge population values of ORR intermediates on PCN and PCN-NaCA based on different functional were summarized below. It was found that the charge population based on GGA-PBE and B3LYP is consistent with each other (Table S4).
(2) It was found that on an electrically neutral cell (of both PCN and PCN-NaCA), ORR process is unlikely to happen, because the initial step of ORR (O 2 adsorption) is energetically unfavorable ( Figure S33). When the cell is negatively charged with one additional electron, ORR on PCN and PCN-NaCA can proceed with surmountable barriers. This is consistent with recently published work from Prof. Qiao's group (J. Am. Chem. Soc. 133, 20116 (2011)).

Responses:
(1) Based on the improved configurations which are closer to the experimental results, the new DFT calculations states that the adsorption energy of O 2 on PCN-NaCA is lower than that on PCN; meanwhile, the interaction energy of O 2 on PCN-NaCA is higher than that on PCN (Table S5). Most importantly, the charge transfer between adsorbate and substrate is consistent with the interaction energy, i.e., the higher the interaction energy is, the more the charge transfer.
(2) For double-checking the charge transfer results, hybrid functional B3LYP was employed as well. As shown in Table S4, the charge transfer between ORR intermediates and substrate based on hybrid functional B3LYP is consistent with that based on GGA-PBE.

Responses:
Considering the suggestions, the adsorption energy values of O 2 and OOH on PCN-NaCA with different vacuum values were checked (Table S3). It was found that the variation of the adsorption energy is negligible, indicating that a vacuum with 2 nm is enough to eliminate the electrostatic interaction between the adjacent supercells. Notes: It was found that the variation of the adsorption energy is negligible, indicating that a vacuum with 2 nm is enough to eliminate the electrostatic interaction between the adjacent supercells.

Comments:
"-There is no "van der Waals interaction" in the PBE functional (or in any GGA). This is very well known.
Given that the bonding here has van der Waals character at least in parts, their absence is a technical error that could invalidate the simulations altogether."

Responses:
Thanks for the comment. Van der Waals interaction was considered in the revised manuscript. The GGA + vdW approach within Grimme scheme was adopted to describe the vdW interaction.

Revisions:
Computation methods in page 6 in Supplementary Information.

Comments:
"- Figure 8 shows "free energies". And where ∆E is the difference in the total energy, ∆E ZPE and ∆S are the differences in the zero-point energy and the change of entropy, T is the temperature (T = 298.15 K in this work), ∆G U and ∆G pH are the contributions from the electrode potential (U) and pH value.
(2) The CHE model defines that the chemical potential of a proton/electron in solution is equal to one half of the chemical potential of one gas-phase H 2 .

Response:
We are so grateful for the reviewer's effort on improving this work. As shown in the revised Supplementary Information (Figure S27), the data plots deviate from a quadratic scaling significantly, stating that two-photon excitation is not relevant to the power densities used in our experiments. We tend to believe that the deviation is too significant, and the additional data plots are thus not necessary.

Reviewer #3 (Remarks to the Author):
The authors revised the paper carefully, addressing the critical concerns, and therefore I recommend accepting the paper.

Reviewer #4 (Remarks to the Author):
All my concerns have been addressed in the revisions. It can be accepted for publication

Reviewer #5 (Remarks to the Author):
Comments: "The authors provided a very comprehensive response to several reviewers. I believe that this is commendable. I apologize to the authors for a review that focuses on a few and partial points only. Most of them concern the theory part."

Response:
We appreciate the reviewer's great effort in reviewing this manuscript. The suggestions are quite valuable for improving the manuscript.

Comments:
"1) The revised Fig 1a is much more precise (great) but the Na+ is still shown in a position that is inconsistent with the theoretical model. Should this depiction be consistent with the theory?"

Response:
The position of sodium ion has been revised, and is now consistent with that in the theoretical model.

Comments:
"2) The revised theory details are very helpful. In particular, use of the COSMO model and of charged unit cells are key details that could be critical. (1). We are sorry to miss a clarification on this point in the previous revision. To However, there is no obvious change in the adsorption energy of OOH on PCN-NaCA observed with the increase of the cell dimensions (Table S4).
We, therefore, tend to believe that the effect of background charge on the COSMO model is negligible in this case. This clarification part has been 5 / 8 added in the revised Supplementary Information. We also find similar method reported in literature (Electrochimica Acta 2020, 354, 136620).
(3). The theoretical simulation is employed for investigating the differences in the ORR steps on PCN and PCN-NaCA; and the reaction mechanisms on PCN and PCN-NaCA are compared under the same conditions, such as the charged unit cell and the presence of neutralizing positive background. In the presence of the above-mentioned clarification, the discussions on the distinctions in the reaction mechanism on PCN and PCN-NaCA are now more reasonable in the revised manuscript.

Revisions:
(1) Revisions in the experimental part in Supplementary Information on page S7.
To check the possible interaction between the positive charge background and the COSMO model, we examined the adsorption energy of OOH on PCN-NaCA with a "cell-extrapolation" method using increasingly larger cells; 17 there is no obvious change in the adsorption energy of OOH on PCN-NaCA observed with the increase of the cell dimensions (Table S4). Reference: [17] Chan, K. & Nørskov, J. K. Electrochemical barriers made simple. J. Phys.
(2) Table S4 in Supplementary Information on page S33. "3) What is the partial charge associated with Na in the overall negative calculations?"

Response:
The Mulliken charge population of the negatively charged PCN-NaCA cell is added in the Supplementary Information (Figure S9b). The charge of Na is 0.838 e.
Revisions: Figure S9 in Supplementary Information on page 16. Comments: "4) The authors do not include O2 partial pressure in the computations but their experimental results now indicate a key impact of O2 partial pressure. Can this difference be rationalized? Shouldn't the O2 partial pressure be accounted for in the simulations?" Response: (1). We may not clearly explain this experiment in the previous revision. The experimental comparison of the H 2 O 2 production performance under different atmosphere was for supporting that the H 2 O 2 production comes from the dioxygen reduction reaction, rather than the other reactions.
(2 Meanwhile, the electrons come from excitation of the photocatalyst by photons: ℎ + PC + ℎ The electrons generation rate is: Therefore, H 2 O 2 production rate is determined both by the concentration of dissolved oxygen and the electrons generation rate. With sufficient photo-induced electrons supply, H 2 O 2 production rate is determined by the concentration of dissolved oxygen; and H 2 O 2 production rate will be controlled by electrons generation rate when the oxygen supply is in excess, which is desirable for mechanistic studies on the photocatalytic system.
In the experimental part, the concentration of dissolved oxygen in 1 atm O 2 atmosphere is roughly 5 times of that in air atmosphere; however, the H 2 O 2 production rate of the reaction system under 1 atm O 2 is only 1.6 times 8 / 8 of that in air atmosphere, which states that the dissolved O 2 is an excess reagent under the reaction conditions in this project.
The impact of the oxygen partial pressure on the H 2 O 2 production performance in the experimental part could be understood via the discussion on the reaction kinetics. We made revision in the manuscript on this point.

Revision:
Revision in manuscript on page 5.
As dioxygen is the source for H 2 O 2 production, the oxygen partial pressure directly impacts the H 2 O 2 production performance.
I reread the manuscript and thank the reviewers for their answers and additions. The work presented here is still impressively comprehensive and should be accepted. --- The supercell size dimension test (Table S4) hopefully settles the potential problem of a charged supercell in practice.
---Regarding the O2 partial pressure. What the authors are doing in the experimental part is consistent with what I meant to say for the theory part.
Specifically, the O2 partial pressure, if considered, would enter the calculation as an O2 chemical potential (dependent on temperature and pressure for instance via ideal gas laws). This would show up as a pressure/temperature dependent ideal gas term in Delta S for O2, on page S6, line 180.
My understanding is now that the authors likely did not consider this term. In any event, I suspect that this term would only shift the energy zero in Figure 8, relative to all other terms.
If that is so, a short comment might be good (and all that is needed).
---Very minor comment: I think some text segments (also in the SI) can still benefit from some polishing for correct English.
I did not pay close attention, but, e.g.: Comments: I reread the manuscript and thank the reviewers for their answers and additions.
The work presented here is still impressively comprehensive and should be accepted.

--
The supercell size dimension test (Table S4)  My understanding is now that the authors likely did not consider this term. In any event, I suspect that this term would only shift the energy zero in Figure 8, relative to all other terms.
If that is so, a short comment might be good (and all that is needed).
---Very minor comment: I think some text segments (also in the SI) can still benefit from some polishing for correct English. Other typo (line 162): "imtermediates"

Response:
We are grateful to the reviewer's effort in improving this manuscript.
We have double checked the manuscript and all the typo and grammatical error have been corrected, which are highlighted in blue in the revised manuscript.