Breaking through water-splitting bottlenecks over carbon nitride with fluorination

Graphitic carbon nitride has long been considered incapable of splitting water molecules into hydrogen and oxygen without adding small molecule organics despite the fact that the visible-light response and proper band structure fulfills the proper energy requirements to evolve oxygen. Herein, through in-situ observations of a collective C = O bonding, we identify the long-hidden bottleneck of photocatalytic overall water splitting on a single-phased g-C3N4 catalyst via fluorination. As carbon sites are occupied with surface fluorine atoms, intermediate C=O bonding is vastly minimized on the surface and an order-of-magnitude improved H2 evolution rate compared to the pristine g-C3N4 catalyst and continuous O2 evolution is achieved. Density functional theory calculations suggest an optimized oxygen evolution reaction pathway on neighboring N atoms by C–F interaction, which effectively avoids the excessively strong C-O interaction or weak N-O interaction on the pristine g-C3N4.

7、As we all know, there are two types of C atoms in CN. The authors are suggested to clarify which kind of C is modified by the F atom.
8、Can the author gives more direct and reliable evidence to prove the formation of C-F bond？ 9、Crystal defects including N vacancy (VN) in CN have been well studied, and their effects in photocatalysis have also been demonstrated. So, the authors are suggested to evaluate the possible effect of defects in this system. Reviewer #3: Remarks to the Author: This is a very interesting study aiming at exploring the critical issue of oxygen evolution reaction on graphitic carbon nitride. The authors claimed that the formation of C=O bonding during continuous photocatalytic overall waster splitting is the origin of carbon nitride failing to produce oxygen without the addition of small molecule organics. DRIFTA and NAP-XPS were used to support the claims, while fluorinated CN catalyst was designed to show that avoiding the formation of C=O can lead to the evolution of oxygen on carbon nitride without a sacrificing reagent. DFT calculations were also used to provide theoretical evidence. In general, the paper was well prepared, and can be considered for publication after some consideration of following issues.
(1) The fluorinated CN catalyst, i.e., F-CN, is a critical sample for the study, however the characterization is not sufficient. Also, AFM, photocurrent, EIS should be performed for all the samples in study. (2) The chemical state of fluorine on the sample should be carefully explained. The XPS F1s spectra (Fig S3(b)) only show very weak signals with a high s/n ratio. The other concern is how the fluorine evolves after the reaction, given the fact that oxygen and hydrogen peroxide would appear. (3) The stability of gases production, and the properties of the photocatalysts should be checked. (4) If just aiming at the production of oxygen, other methods like loading cocatalyst on carbon nitride can also play the same role. What is the necessity of fluorination? How is this compared to other methods? How can this approach inspire or help other scientific challenges in materials chemical or science?
Reviewer #4: Remarks to the Author: This manuscript by Wu et al. reports a study on the reasons leading to the overall water splitting inactivation on single-phase CN catalysts. The authors claimed that the formation of inert C=O intermediates by oxidizing surface carbon atoms during the OER is the bottleneck for overall water splitting on g-C3N4 catalyst. After analyzing isotopic-labelled in-situ diffuse reflection infrared Fourier transform spectroscopy and in-situ near-ambient pressure X-ray photoelectron spectroscopy, the formation of C=O intermediates is prevented via a simple surface fluorination strategy, and the F0.1-CN catalyst exhibits higher overall water splitting activity compared to the pristine CN. However, many mechanisms and modification strategies for g-C3N4 to improve the photocatalytic overall water splitting activity have been reported in previous literatures (Nat. Energy 6, 388(2021); Adv. Mater. 33, 2007479(2021); Chin. Chem. Lett. 32, 13(2021)). The novelty and significance of this work should be highlighted because the catalytic activity of F0.1-CN catalyst is general compared to the catalysts reported so far. And some main conclusions are not well supported by the present experimental/theoretical results. I consider that the manuscript cannot meet the high standards of Nat. Commun., so I cannot recommend the publication of this work on the Nat. Commun. catalytic mechanisms have been widely studied (Nat. Energy 6, 388(2021); Small 17, 2006851(2021); Adv. Mater. 33, 2007479(2021); Chin. Chem. Lett. 32, 13(2021)). Therefore, where are the advance and major breakthrough of the views mentioned by the author in relation to the reported results? 2. The F0.1-CN catalyst exhibits poor overall water splitting activity (H2 evolution rate of 174.77 μmol g-1 h-1) compared to the catalysts reported so far (H2 evolution rate > 500 μmol g-1 h-1, λ > 420nm; Small 17, 2006851(2021); Adv.Mater. 32, 1907296(2020)). Using isotopic-labelled in-situ diffuse reflection infrared Fourier transform spectroscopy and in-situ near-ambient pressure X-ray photoelectron spectroscopy to study mechanism is of little significance.
3. According to the signal-to-noise ratio of F1s spectra, the content of F should be very low for the F0.1-CN catalyst (< 0.1%). The authors believed that the improved OER activity is due to the C-F bond formation optimizing the OER pathway on adjacent N atoms. However, the number of F-modified N sites should be minimal. The coordination environment of most carbon sites in F0.1-CN is similar to that in pristine CN. Therefore, it is not convincing that F-CN does not generate C=O intermediates during the reaction. The authors should provide more experimental and theoretical results to explain this issue in detail. 4. In the in-situ NAP-XPS observations of N1s spectra, the peak area ratios of NHx, NC3, C-N=C change significantly for the pristine CN and F0.1-CN catalysts. This phenomenon should be explained in detail. In addition, the authors claimed that the N site in F-CN is the main OER active center. However, during continuous white light illumination, there is no peak positions change or new peaks appear on the N1s spectra for the F0.1-CN catalyst, which is unreasonable. 5. In the in-situ DRIFTS analysis, the authors claimed that the O source of C=O is from H2O. However, an obvious peak of the C=16O stretching vibration is still observed at 1725 cm-1 when H216O is replaced by 18O-labelled H218O. Furthermore, why does the bending vibration of H-O-H at 1645 cm-1 disappear in Fig. 1b-1d? More detailed analysis should be provided. 6. In the DFT calculations, the authors only compared the energy barrier of the surface N sites of F-CN and the surface C and N sites of pristine CN. Why did the authors not consider the C atom of the C-F bond in F-CN as the reaction site? 7. The authors should improve the accuracy of the expression. For example, what is the wavelength of white light?
Reviewer #5: Remarks to the Author: In this manuscript, the authors observed a collective C=O bonding during continuous photocatalytic overall water splitting on g-C3N4 catalyst by in-situ DRIFTS and NAP-XPS, and confirmed that the inert C=O bond directly hinders further OER steps, resulting in negligible O generation on CN. The F0.1-CN catalyst prepared based on this finding exhibited excellent overall water splitting activity with the order-of-magnitude improved H2 evolution rate compared to the pristine CN catalyst. This manuscript was interesting. However, I feel that there are still a number of uncertainties and obvious problems. The analysis of the data is not accurate enough. After careful evaluation, I think that this work does not meet the standard of Nature Communications.
Other comments: 1. In-situ NAP-XPS is the main method for the authors to prove the formation and disappearance of C=O bonds. This method is very interesting, but I think there are some issues with data analysis on NAP-XPS. For example, in the XPS spectra of CN under different illumination times, the fitted peak width of the O-H peak changed significantly. Actually, the total spectra at 0 min and 5 min of CN are close, but the C=O peak is formed due to the change in the peak width of the O-H peak during fitting. Authors should fit the XPS data under the same O-H peak width to determine the presence of C=O peak. 2. The consumption of e-and h+ during photocatalytic overall water splitting should be equal. However, in this work, the e-consumed by H2 production is not equal to the h+ consumed by O2 and H2O2 production. It is suggested that the authors calculate the amount of reduction and oxidation products in the photocatalytic water splitting process and analyze why the consumption ratio of e-and h+ is not 1:1. 3. The authors should point out the test conditions of Supplementary Figure 9, white light or AM1.5G? 4. The authors aim to illustrate the effect of F doping on enhanced OER performance, then supplemental separate OER tests may be more helpful in illustrating the authors' hypothesis. 5. The nanosheets of CN in Supplementary Figure 1 appear thinner than F-CN, but its BET surface is much smaller than that of F-CN. What is the reason for the increase in the specific surface area of F-CN? 6. It is difficult to illustrate the existence of F in Supplementary Figure 3. It is recommended that the authors re-test XPS and consider changing the sample preparation method to improve the accuracy.

Reply to reviewers' comments
To Reviewer 1: In this manuscript, the authors presented an interesting work on revealing the inhibiting effect on carbon nitride performance in photocatalytic water splitting to H2 and O2. The authors employed in-situ FTIR and XPS with isotopic detection for the identification of C=O formation in the process, which was suggested to play an important role for catalytic reaction. This work is new and can be published after revision. Some suggestions are listed below.
Response: Thanks for your great efforts in reviewing our manuscript. We appreciate your valuable comments and suggestions. Q1: It is believed that C=O formation on the surface of carbon nitride. Thus, it is suggested that the authors check the surface species on the catalysts (CN, CN-E, and F-CN) before and after reaction to see the difference and discuss the effect of C=O loading on their catalytic performance.
Response: Thanks for your valuable comment. We have carefully performed DRIFTS and XPS measurements on CN, CN-E, and F0.1-CN, before and after 5 hours of reaction. As shown in Fig. R1 and Fig. R2, IR and XPS spectra of samples before and after the reaction remain unchanged. The C=O accumulation was not observed on CN and CN-E samples after the reaction, which further suggests that C=O only formed during the OER as an intermediate species (only observable during the reaction) rather than be a stable surface group stoichiometrically produced in the reaction. We have added the additional data and discussion in the revised manuscript. Please see the Line 29 of Page 5.

Q2:
What is the mechanism of C=O involving in the catalytic reaction to reduce the performance?
Response: Thanks for your valuable comment. According to our in-situ DRIFTS ( Fig. 1a-1b) and NAP-XPS ( Fig. 1e) observations, C=O species collectively formed during continuous photocatalytic reaction on CN with pure water by oxidizing carbon sites. During this process, C=O is generated as the reaction intermediate of OER at the C site (*O). The subsequent OER reaction requires the dissociation of the C=O double bond to combine with the second water molecule (or OHion), forming the *OOH intermediate. Since the inert C=O double bond is thermodynamically difficult to dissociate, as a reaction intermediate, the C=O formation directly lowers the opportunity for OER. Q3: page 3, line 3 of the first paragraph. "F ion would slightly decrease the performance" is not correct. From Fig.1a, larger loading of F ion will significantly reduce the performance after F0.1.

To Reviewer 2:
The authors report that the C=O formation during the OER is the bottleneck of photocatalytic overall water splitting on CN, while this C=O bonding could be hindered through F modification. In-situ XPS and DRIFTS characterizations were also carried out. This assumption is innovative, but there are still several fundamental concerns need to be addressed.

Response:
Thanks for your great efforts in reviewing our manuscript. We appreciate your valuable comments and suggestions. Q1: In Fig. 3c, the rate-determining step of O2 evolution over CN(C) is the formation of *OH rather than the step from O* to *OOH. So, why the formation of C=O is the bottleneck of overall water splitting on CN-based catalysts? More illustration or an additional attention on the effect of C=O should be given out.
Response: Thanks for your valuable comment. As shown in Fig. 3c, during the OER over CN(C), the two steps of *→*OH and *O→*OOH have high-energy barriers, which would definitely lower the opportunity for the reaction. However, once the reaction starts with *OH formed, the subsequent formation of *O is easy after a low-barrier step of *OH→*O. Thus, the accumulation of *O (in the form of C=O) can be observed since the opportunity of the further reaction step is rather low with a high barrier (*O→*OOH), and once a small amount of *OOH is formed, it is quickly converted to O2 with no -COOH accumulation observed, which is consistent with our experimental observations. We have added this part of the discussion to the revised manuscript. Please see the Line 6 of Page 10.

Q2:
The authors proposed that the N site is the main OER center in F-CN. While N 1s XPS spectra indicates no obvious shift under continuous white light illumination. Why an oxidized N state did not form like C in CN?
Response: Thanks for your valuable comment. The unchanged N 1s state in F-CN indicates that the charge transfer at the N site was very fast with no obvious accumulation of intermediate states. This result further suggests that the N site as the main OER center facilitates the interfacial charge transfer after the surface F-modification.

Q3:
Since N is original present in F-CN and F modification is proposed to induce the change of reactive sites from C atoms in CN to N atoms adjacent to the C-F bond in F-CN, a fundamental discussion on the mechanism should be given out. The change of energy barrier from theoretical simulation is more like phenomenon other than underlaying comprehension.
Response: Thanks for your valuable comment. We have added a fundamental discussion on the mechanism in the DFT calculation part: Moreover, after the F atom occupies the C site, the C site is saturated, which becomes inert to the reactant or intermediate. As a result, the neighboring two-coordinated N site is the sole catalytic center on F-CN. The obtained reaction pathway shows that the surface N site in F-CN owns a much lower energy barrier (1.58 eV) than the surface C (2.25 eV) or N (2.86 eV) site in pristine CN (Fig. 3c), demonstrating that the F modification indeed can improve the OER activity on F-CN. According to the calculated charge density difference mappings, the improved OER activity on F-CN is attributed to the more local charge distribution between CN surface and *OH intermediate, which greatly promotes the formation of *OH intermediate (Fig. 3e) and also effectively avoids the excessively strong C-O interaction (Fig. 3d) or weak N-O interaction ( Supplementary Fig. 20b) in pristine CN. As a result, the F modification greatly decreases the formation energy of ratedetermining *OH. It should be noted that the F modification also significantly promote the formation of *OOH, which is also a high-barrier reaction step in OER (Fig. 3c). This implies that the excessively stable *O intermediate in the form of C=O bond on pristine CN is difficult to be further converted into *OOH. As a result, CN with an observable IR signal of C=O during reaction owns a lower activity than F-CN, which completely coincides with our experimental observations. Furthermore, the PDOS calculation provides an electronic-scale insight into the improved OER activity on F-CN. The obtained results show that F modification enables the N 2p states to move upward the Fermi level (Fig. 3f), which can be attributed to the transfer of partial electrons from N site to F site through C site. The more positive N 2p states promise the N site with higher oxidizing activity and more uncaptured orbits for bonding with *OH intermediate. Thus, the optimized bonding behavior between *OH and F-CN surface contributes to an improved OER activity. Hence, the N site in F-CN is the main OER center. Please see the Line 20 of Page 9 in the revised manuscript.

Q7:
As we all know, there are two types of C atoms in CN. The authors are suggested to clarify which kind of C is modified by the F atom.
Response: Thanks for your valuable comment. As shown in the structural configuration ( Fig. R2) of CN, there are two types of N (i.e., two coordinated pyridine N2c and three coordinated graphite N3c) and one type of C (i.e., three coordinated C3c) in CN. F modification was on C3c sites (blue circle).

Q8:
Can the author gives more direct and reliable evidence to prove the formation of C-F bond？ Response: Thanks for your valuable comment. We monitored C 1s and F 1s states on F-CN before and after hydrothermal treatment by XPS. Before hydrothermal treatment, F-ions were adsorbed on the CN surface, resulting in a distinct shift to the C 1s state (Fig. R3a), indicating the formation of C-F interaction. Moreover, after the hydrothermal treatment, the F 1s state slightly shifts towards higher binding energy, demonstrating the charge transfer from F atoms to C atoms, which solidly proves the strong C-F interaction. The additional experimental data was added to the Supplementary Information. Please see Supplementary Fig. 3.  Response: Thanks for your valuable comment. We quantified the C/N ratio and F ion content by using elemental analysis and ion chromatography. As shown in Table. R1, with the increase of F content, the C/N ratio was almost unchanged. C/N ratio in all samples was ~ 65%, demonstrating the existence of C defects rather than N defects. The unchanged C/N ratio excludes defect states as the major effect of enhanced performance on F-CN. Table. R1 was added to the Supplementary Information.

To Reviewer 3:
This is a very interesting study aiming at exploring the critical issue of oxygen evolution reaction on graphitic carbon nitride. The authors claimed that the formation of C=O bonding during continuous photocatalytic overall waster splitting is the origin of carbon nitride failing to produce oxygen without the addition of small molecule organics. DRIFTA and NAP-XPS were used to support the claims, while fluorinated CN catalyst was designed to show that avoiding the formation of C=O can lead to the evolution of oxygen on carbon nitride without a sacrificing reagent. DFT calculations were also used to provide theoretical evidence. In general, the paper was well prepared, and can be considered for publication after some consideration of following issues.
Response: Thanks for your great efforts in reviewing our manuscript. We appreciate your valuable comments and suggestions.

Q1:
The fluorinated CN catalyst, i.e., F-CN, is a critical sample for the study, however the characterization is not sufficient. Also, AFM, photocurrent, EIS should be performed for all the samples in study.

Q2:
The chemical state of fluorine on the sample should be carefully explained. The XPS F1s spectra (Fig S3(b)) only show very weak signals with a high s/n ratio. The other concern is how the fluorine evolves after the reaction, given the fact that oxygen and hydrogen peroxide would appear.
Response: Thanks for your valuable comment. We have re-tested the XPS F1s spectra on the F0.1-CN sample before and after the reaction. As shown in Fig. R4, the position of the F1s state was almost unchanged after the reaction. Please see the Line 3 of Page 8 in the revised manuscript.

Q3:
The stability of gases production, and the properties of the photocatalysts should be checked.

Response:
Thanks for your valuable comment. As shown in Fig. R5, we further extended the reaction time to 40 h, and the F0.1-CN can still maintain 70.35% of the initial efficiency on H2 and O2 production and continue to work.

Q4:
If just aiming at the production of oxygen, other methods like loading cocatalyst on carbon nitride can also play the same role. What is the necessity of fluorination? How is this compared to other methods? How can this approach inspire or help other scientific challenges in materials chemical or science?
Response: Thanks for your valuable comment. Compared with metal-based inorganic semiconductor catalysts, metal-free catalysts have irreplaceable advantages in cost-control and environmental friendliness. CN is essentially a metal-free inorganic catalyst, however, due to the limitation of its interfacial reaction energy, it cannot work without metal/metal oxide cocatalysts. Therefore, meticulously studying the reaction mechanism at the CN interface to guide a completely metal-free CN-based photocatalyst is of great significance in terms of both cost and environmental friendliness.

To Reviewer 4:
This manuscript by Wu et al. reports a study on the reasons leading to the overall water splitting inactivation on single-phase CN catalysts. The authors claimed that the formation of inert C=O intermediates by oxidizing surface carbon atoms during the OER is the bottleneck for overall water splitting on g-C3N4 catalyst. After analyzing isotopic-labelled in-situ diffuse reflection infrared Fourier transform spectroscopy and in-situ near-ambient pressure X-ray photoelectron spectroscopy, the formation of C=O intermediates is prevented via a simple surface fluorination strategy, and the F0.1-CN catalyst exhibits higher overall water splitting activity compared to the pristine CN. However, many mechanisms and modification strategies for g-C3N4 to improve the photocatalytic overall water splitting activity have been reported in previous literatures (Nat. Energy 6, 388(2021); Adv. Mater. 33, 2007479(2021); Chin. Chem. Lett. 32, 13 (2021)). The novelty and significance of this work should be highlighted because the catalytic activity of F0.1-CN catalyst is general compared to the catalysts reported so far. And some main conclusions are not well supported by the present experimental/theoretical results. I consider that the manuscript cannot meet the high standards of Nat. Commun., so I cannot recommend the publication of this work on the Nat. Commun.
Response: Thanks for your great efforts in reviewing our manuscript. We appreciate your valuable comments and suggestions.

Q1:
Up to now, the nonmetallic-elements modification of g-C3N4-based catalysts and the corresponding catalytic mechanisms have been widely studied (Nat.  (2021)). Therefore, where are the advance and major breakthrough of the views mentioned by the author in relation to the reported results?
Response: Thanks for your valuable comment. In this work, rather than reporting a world-champion CN-based catalyst for overall water splitting, we aimed to figure out a fundamental scientific question of why the pristine CN surface is inert to OER, which is a long-puzzled scientific challenge for years since the discovery of CN catalysts. Firstly, our in-situ DRIFTS and NAP-XPS provide valuable information at H2O/CN interface to generate an in-depth understanding of this issue. After effectively changing the reaction path of OER on the CN surface by surface fluorination, we observed the occurrence of OER on the classical CN catalyst (without the help of OER cocatalyst), which is completely different from previously reported non-metallic modification strategies by changing the band structure, facilitating the interfacial charge transfer, or enhancing the light-harvesting ability. Especially, all these reported non-metallic modification strategies without exception still require the participation of metal oxide OER cocatalysts in the overall water splitting, which does not essentially solve the bottleneck problem that CN cannot induce OER on the surface.
Secondly, on the basis of these works, we are aware of the important role of the interfacial reaction pathway, using interfacial observation to confirm C=O as the reaction bottleneck, and then through surface fluorination, without changing the band structure and light absorption range, achieved the overall water splitting on F-CN catalysts.
Thirdly, our work reveals the reason why pristine CN cannot directly split water, and the overall water splitting can be achieved through a simple surface atomic modification strategy, which relieves the dependence of CN catalysts on metalbased OER cocatalysts and further enhances understanding of the interfacial reaction mechanism on CN at the atomic scale basis. Based on this, we think the novelty of our work meets the standard of Nature Communications. Thank you again for your valuable comment.
Response: Thanks for your valuable comment. Our work realized the overall water splitting without metal-based OER cocatalysts, which reveals a long-hidden bottleneck of OER on the CN surface and provides a research basis for the future realization of completely metal-free CN catalysts. To reach the world-champion H2 production efficiency is not the goal of this work. Besides, the reported activity tests were not performed under a standard method. For example, the light intensity, irradiated area, mass of used catalyst can produce a significant effect on the activity. Hence, directly comparing the activities in different test methods cannot provide significant information on the performances of different photocatalysts. We have summarized the overall water splitting activities of recently reported CN-based catalysts, which are shown below in Table  R1. Compared with reported CN catalysts, the overall water splitting efficiency on F-CN (our work) is not low. Moreover, ours is the only catalyst without severely changing the chemical composition or structure of CN, which is of great scientific significance. Q3: According to the signal-to-noise ratio of F1s spectra, the content of F should be very low for the F0.1-CN catalyst (< 0.1%). The authors believed that the improved OER activity is due to the C-F bond formation optimizing the OER pathway on adjacent N atoms. However, the number of F-modified N sites should be minimal. The coordination environment of most carbon sites in F0.1-CN is similar to that in pristine CN. Therefore, it is not convincing that F-CN does not generate C=O intermediates during the reaction. The authors should provide more experimental and theoretical results to explain this issue in detail.
Response: Thanks for your valuable comment. We quantified the C/N ratio and F ion content by using elemental analysis and ion chromatography. The F content and calculated surface F coverage (the number of exposed C atoms on CN surface were calculated by the thickness of the sample, measured by AFM) were summarized in Table. R2. For the champion F0.1-CN sample, the surface F coverage is about 4 atom%. DFT calculations with varying F contents (with surface F coverage in the range of 1~8 atom%) were also conducted. As shown in Fig. R1, with surface F coverage increases from 1 atom% to 8 atom%, the barrier of OER on the adjacent N site (Fig. R1a) and the energy level of N 2p state (Fig. R1b) are almost unchanged, indicating that the adjacent N site of the C-F structure is only affected by the adjacent C-F hybridization rather than the concentration of F. The F coverage only affects the number of active N sites. Based on the above experimental and calculation results, we reason that once the active N site emerged, water splitting reaction has a higher chance of occurring at the active N site adjacent to C-F rather than at normal C or N sites, by which hinders the surface C=O formation. The additional experimental data and discussions were added to the revised manuscript. Please see the Line 6 of Page 10 and Line 21 of Page 10 in the revised manuscript.

Q4:
In the in-situ NAP-XPS observations of N1s spectra, the peak area ratios of NHx, NC3, C-N=C change significantly for the pristine CN and F0.  Fig. 1b-1d? More detailed analysis should be provided.
Response: Thanks for your valuable comment. First, we think the presence of C=O 16 peak in H2O 18 /CN system is due to the inevitable atomic exchange between the abundant surface -OH groups on CN surface and H2O 18 . Furthermore, the reason why the bending vibration of H-O-H at 1645 cm -1 was unobservable is that its position coincides with the positive C=O 18 peak.

Q6:
In the DFT calculations, the authors only compared the energy barrier of the surface N sites of F-CN and the surface C and N sites of pristine CN. Why did the authors not consider the C atom of the C-F bond in F-CN as the reaction site?
Response: Thanks for your valuable comment. As shown in the structural configuration (Fig. R2) of CN, there are two types of N (i.e., two coordinated N2c and three coordinated N3c) and one type of C (i.e., three coordinated C3c) in CN. F modification was on C3c sites (blue circle) to reach the maximum coordination (C4c), which is unable to bind with other reactant molecules.

To Reviewer 5:
In this manuscript, the authors observed a collective C=O bonding during continuous photocatalytic overall water splitting on g-C3N4 catalyst by in-situ DRIFTS and NAP-XPS, and confirmed that the inert C=O bond directly hinders further OER steps, resulting in negligible O generation on CN. The F0.1-CN catalyst prepared based on this finding exhibited excellent overall water splitting activity with the order-of-magnitude improved H2 evolution rate compared to the pristine CN catalyst. This manuscript was interesting. However, I feel that there are still a number of uncertainties and obvious problems. The analysis of the data is not accurate enough. After careful evaluation, I think that this work does not meet the standard of Nature Communications.
Response: Thanks for your great efforts in reviewing our manuscript. We appreciate your valuable comments and suggestions.

Q1:
In-situ NAP-XPS is the main method for the authors to prove the formation and disappearance of C=O bonds. This method is very interesting, but I think there are some issues with data analysis on NAP-XPS. For example, in the XPS spectra of CN under different illumination times, the fitted peak width of the O-H peak changed significantly. Actually, the total spectra at 0 min and 5 min of CN are close, but the C=O peak is formed due to the change in the peak width of the O-H peak during fitting. Authors should fit the XPS data under the same O-H peak width to determine the presence of C=O peak.
Response: Thanks for your valuable comment. We have reanalyzed the NAP-XPS data by fitting all spectra under the same peak width. As shown in Fig. R1, after the fitting, the C=O formation can be clearly observed. During 0-5 min, the change in O1s spectra (532.7 eV) was also observable. We have added the reanalyzed result to the revised manuscript. Please see Fig.  1e.

Q2:
The consumption of eand h + during photocatalytic overall water splitting should be equal. However, in this work, the econsumed by H2 production is not equal to the h + consumed by O2 and H2O2 production. It is suggested that the authors calculate the amount of reduction and oxidation products in the photocatalytic water splitting process and analyze why the consumption ratio of eand h + is not 1:1.
Response: Thanks for your valuable comment. Due to the inevitable generation of H2O2, the UV-induced conversion of O2 to ozone, and the presence of dissolved O2, it is reasonable that the H2/O2 ratio for the overall water splitting on semiconductor photocatalyst to be larger than a perfect stoichiometric ratio of 2:1 under white light. Similar phenomena were also observed in many other research works (Bai Y, et al, Angew. Chem. Int. Ed. 134, e202201299 (2022)). In our system, the main reason for the shortage of O2 production is the formation of H2O2, since CN is well-acknowledged as an efficient catalyst for H2O2 production (Nat. Commun. 12, 3701 (2021); ACS Catalysis 10, 14380-14389 (2020)). The Iion titration method we previously used is not sensitive enough and could only qualitatively analyze the presence of H2O2. To quantify the H2O2 production during the reaction, we have employed a more sensitive Ce 4+ back titration method (as reported in Adv. Mater. 34, e2107480 (2022).) (detailed methods see Supporting Information). As shown in Fig. R2, the H2O2 production rate on the champion F0.1-CN catalyst was determined to be 85.36 umol•g -1 •h -1 , almost identical to the short of H2:O2 ratio. The above discussion and additional experimental data were added to the revised manuscript. Please see the Line 32 of Page 7 in the revised manuscript.

Q5:
The nanosheets of CN in Supplementary Figure 1 appear thinner than F-CN, but its BET surface is much smaller than that of F-CN. What is the reason for the increase in the specific surface area of F-CN?
Response: Thanks for your valuable comment. TEM images (Supplementary Figure 1) can only distinguish local morphologies of the dispersed samples, which is difficult to intuitively identify the specific thickness of the sample. We thusly employed AFM to characterize the thickness of CN, CN-E, and F-CN samples. As shown in Fig. R4, CN-E and F-CN samples have a smaller layer thickness of ~ 4 nm, whereas the pristine CN layer is about ~5 nm. The increased BET surface area of F-CN compared to CN is due to the reduced layer thickness from the hydrothermal exfoliation. The additional experimental data was added to the Supplementary Information. Please see Supplementary Fig. 7. Q6: It is difficult to illustrate the existence of F in Supplementary Figure 3. It is recommended that the authors re-test XPS and consider changing the sample preparation method to improve the accuracy.
Response: Thanks for your valuable comment. We have retested XPS F1s spectra on CN, and F-CN samples before and after hydrothermal treatment. The existence of F is now observable (Fig. R5). The additional experimental data was added to the Supplementary Information. Please see Supplementary Fig. 3.

Reviewers' Comments:
Reviewer #1: Remarks to the Author: According to the responses, the authors made basic revisions based on the comments, but did not fundamentally solve the reviewer's doubts. Especially the reliability of in situ DRIFTS and NAP-XPS signals for identifying the surface C=O bonds. In my opinion, the revision of the paper does not reach the level of publication in this journal; unfortunately, I suggest the rejection of this manuscript.
1. Although the authors prove that the hydrophilicity of F-CN is greatly decreased with the increase of F content, this could not be an objective reason for the decreased OER activity. Otherwise, the authors need to cite references to demonstrate the relationship between hydrophilicity and OER performance.
2. The authors need to provide more direct evidence such as absorption spectra, to prove and reveal the coordination environment of C-F, not just through simulation calculation.
3. The authors still do not explain the reliability of in situ DRIFTS to identify C=O intermediate formation.
The in-situ test here cannot define as the real in situ conditions during the reaction, considering the pressure and temperature (150℃) discrepancy from the actual water splitting test. Also, the authors used a 420nm LED light source, which is totally different from the light source during the water splitting test ((≥300 nm ,1000 mW•cm-2)). Did the author add H2PtCl6•6H2O during the in situ DRIFTS test? If not, why? All of these will affect the analysis of the surface reaction mechanism.

The authors do not have a good explanation of what the merits of F-CN is.
Especially compared with some benchmark non-metallic element doped or complex modified catalysts (Nature Energy 6, 388-397 (2021); Advanced Energy Materials 7, (2017), Science 347, 970-974 (2015)). In addition, the authors used the UV-vis light irradiation with the high light intensity (1000 mW/cm2) and the high concentrated Pt cocatalyst for the reaction, which has no advantage. Therefore, from this point of view, the scientific importance of studying this material is not very significant.
5. The authors did not consider the effect of Pt cocatalyst on the F-CN. It was claimed that the adjacent N site of the C-F structure is responsible for the OER test; what is the reason for the largely enhanced HER half-reaction. 6. The stoichiometric ratio of hydrogen to oxygen production is far from two to one, which is crucial for studying fully decomposed water. The authors need to better illustrate this point by calculating Faraday efficiency or STH. In addition, the author can make a graph or bar chart with the hydrogen/oxygen ratio as the ordinate and the fluorine doping amount as the abscissa to study the influence of fluorine doping.
7. As far as I know, this journal has high requirements for repeatability of data, especially for statistics figures, such as Fig.2 a (the most important performance data I think); the error bars are usually needed and defined thoroughly. Figure 5 are inconsistent with the figure curves.

The captions in supplementary
Reviewer #2: Remarks to the Author: Though efforts have been taken by the authors, part of the comments have not been well responded: 1、 Q1 (" In Fig. 3c, the rate-determining step of O2 evolution over CN(C) is the formation of *OH rather than the step from O* to *OOH. So, why the formation of C=O is the bottleneck of overall water splitting on CN-based catalysts? More illustration or an additional attention on the effect of C=O should be given out.") is not properly responded. It seems that they authors do not make clear in the manuscript that what is "the formation of C=O". Is it the same with "the step from *OH to O*"? Even taking consideration of this point, some important even crucial conclusions are inconsistent with the DFT results. It can be seen from Fig. 3c Supplementary Fig. 9b, the authors should clearly note that the UV-vis absorption of the F-CN catalyst was slightly blue-shifted with high F content.
Reviewer #5: Remarks to the Author: In this manuscript, the authors used in-situ FTIR and NAP-XPS with isotopic detection to reveal that C=O formation is the inhibiting factor on carbon nitride for photocatalytic overall water splitting to form H2 and O2. A surface fluorination strategy was employed to improve the OER performance by occupying carbon-sites on CN with F-ions. This work is insightful and can be published after revision.
1) Why the authors choose fluorination strategy to inhibit the formation of C=O? There could be many other choices (such like sulfuration or phosphorization) to occupy the C sites of CN, but the authors seemed to go to fluorination without showing the underlying reason.
2) To illustrate the effect of F on enhancing the OER performance, the authors should show more detailed OER test rather than just the LSV. To understand why CN is inert to OER and why F-CN is conducive to OER, the OER kinetics analysis would be much more important and inspiring than simply knowing the onset potential.

Reply to reviewers' comments
To Reviewer 1: According to the responses, the authors made basic revisions based on the comments, but did not fundamentally solve the reviewer's doubts. Especially the reliability of in situ DRIFTS and NAP-XPS signals for identifying the surface C=O bonds. In my opinion, the revision of the paper does not reach the level of publication in this journal; unfortunately, I suggest the rejection of this manuscript.
Response: Thanks for your great efforts in reviewing our manuscript. We appreciate your valuable comments and suggestions.

Q1:
Although the authors prove that the hydrophilicity of F-CN is greatly decreased with the increase of F content, this could not be an objective reason for the decreased OER activity. Otherwise, the authors need to cite references to demonstrate the relationship between hydrophilicity and OER performance.
Response: Thanks for your valuable comment. The decreased photocatalytic activity of F-CN with high F content is not directly affected by the increased hydrophilicity. As we demonstrated in Fig. R1, hydrophilicity affects the sedimentation behavior of catalysts in water. With high F content, the F-CN catalyst can sediment very fast and be very difficult to form a uniform aqueous suspension. Such a fast sedimentation behavior of the catalyst will definitely render a lower light absorption efficiency than the well-suspended counterpart. In addition, we employed the in-situ UV-vis optical fiber spectroscopy to directly monitor the transmittance of white light (tungsten lamp, 5W) through different suspensions (0.3g/L). As shown in Fig. R2, with the increasing F content in F-CN catalysts (F0.01~F1), the transmittance of white light is significantly increased (30.03%~63.79%), demonstrating that more light is transmitted with higher F content, therefore reducing the photocatalytic performance. For the corresponding description in the revised manuscript see Page 7 Line 24.

Q2:
The authors need to provide more direct evidence such as absorption spectra, to prove and reveal the coordination environment of C-F, not just through simulation calculation.
Response: Thanks for your valuable comment. We performed the IR measurement on CN and F-CN samples as shown in Fig.  R3. Unfortunately, corresponding C-F vibrations were not observed due to the lower content of surface C-F in F-CN samples. We further employed XPS C1s and F1s spectra (Fig. R4) to monitor the strong C-F interaction in F-CN samples. Before the hydrothermal treatment, with the emerging of the F 1s state, a distinct shift of the C 1s state was observed (Fig. R4a), demonstrating the interaction between C and F atoms. Moreover, after the hydrothermal treatment, the F 1s state slightly shifts towards higher binding energy (Fig. R4b), demonstrating the charge transfer from F atoms after hydrothermal treatment, which solidly proves the strong C-F interaction in F-CN.

Q3:
The authors still do not explain the reliability of in situ DRIFTS to identify C=O intermediate formation. The in-situ test here cannot define as the real in situ conditions during the reaction, considering the pressure and temperature (150℃) discrepancy from the actual water splitting test. Also, the authors used a 420nm LED light source, which is totally different from the light source during the water splitting test ((≥300 nm ,1000 mW•cm-2)). Did the author add H2PtCl6•6H2O during the in situ DRIFTS test? If not, why? All of these will affect the analysis of the surface reaction mechanism.
Response: Thanks for your valuable comment. First, all in-situ DRIFTS experiments were at 25 o C under ambient pressure conditions rather than at 150 o C. As described in the experimental method that "Prior to isotopic experiments, pristine CN and F-CN samples were heated at 423 K for 30 min under flowing N2 to remove the remaining water.", the 423 K treatment was only employed before the isotopically labeled measurement of CN/F-CN samples to eliminate residue water. After that, samples were cooled down to room temperature and measured under identical conditions as other unlabeled experiments. A more detailed description of the experimental methods can be found in the revised Supplementary Information (See SI, Page 2, Line 10). Second, CN is a visible-light-driven catalyst. Therefore, we used the 420 nm LED lamp as the light source in the in-situ DRIFTS experiments. As requested, we further conducted the in-situ DRIFTS experiments under white light irradiation (Xe lamb, ≥300 nm, center intensity of 1000 mW•cm -2 ). As shown in Fig. R5, the experimental phenomenon is consistent with the use of a 420 nm lamp, indicating that the excitation light source does not affect the mechanism of the reaction on CN/F-CN.
Third, Pt was loaded on CN/F-CN to facilitate HER (2H + + 2e -→ H2) during the water splitting reaction, which has no IR absorption signals and should not affect the OER mechanism. Therefore, we did not consider Pt in DRIFTS experiments. As requested, we further conducted the in-situ DRIFTS experiments on CN/F-CN samples with the loading of Pt. As shown in Fig. R6, the experimental phenomenon is consistent with that on the pristine CN/F-CN samples, indicating that the Pt loading indeed does not affect the OER mechanism on CN/F-CN.
All these control experiments demonstrate the reliability of our in-situ DRIFTS results. Additional control experiments were added to the revised manuscript; Please see Page 5, Line 13.   (2015)). In addition, the authors used the UV-vis light irradiation with the high light intensity (1000 mW/cm2) and the high concentrated Pt cocatalyst for the reaction, which has no advantage. Therefore, from this point of view, the scientific importance of studying this material is not very significant.
Response: Thanks for your valuable comment. In this work, rather than reporting a world-champion CN-based catalyst for overall water splitting, we aimed to figure out a fundamental scientific question of why the pristine CN surface is inert to OER, which is a long-puzzled scientific challenge for years since the discovery of CN catalysts. Our work realized the overall water splitting without metal-based OER cocatalysts, which provides a research basis for the future realization of completely metal-free CN catalysts. Reaching the world-champion H2 production efficiency is not the goal of this work. Besides, the reported activity tests were not performed under a standard method. For example, the light intensity, irradiated area, and mass of the used catalyst can produce a significant effect on the activity. Hence, directly comparing the activities in different test methods cannot provide significant information on the performances of different photocatalysts. We have summarized the overall water-splitting activities of recently reported CN-based catalysts, which are shown below in Table R1. Compared with reported CN catalysts, the overall water splitting efficiency on F-CN (our work) is not low. Pt is the HER co-catalyst, which is almost necessary to all CN catalysts for water splitting. Moreover, ours is the only catalyst without severely changing the chemical composition or structure of CN, demonstrating the key role of surface reaction configurations, which is of great scientific significance.

Q5:
The authors did not consider the effect of Pt cocatalyst on the F-CN. It was claimed that the adjacent N site of the C-F structure is responsible for the OER test; what is the reason for the largely enhanced HER half-reaction.
Response: Thanks for your valuable comment. As HER cocatalyst, Pt loading should only affect HER half reaction rather than the CN catalyst itself. We monitored the XPS C 1s, N 1s, and F 1s spectra before and after Pt loading, and no changes were observed (Fig. R7 and Fig. R8), indicating that Pt loading hardly affects the surface C and N atoms in CN/F-CN. Furthermore, in our experiments, we observed enhanced water splitting efficiency on F-CN compared to CN and CN-E, which should be due to the enhanced OER rather than HER since the four electron-participated OER is the rate-determining step in water splitting. We further tested the HER efficiencies on CN, CN-E, and F-CN by using triethanolamine as the hole scavenger (Fig. R9). The result shows that the HER efficiency on F-CN is not significantly enhanced in comparison with CN and CN-E, demonstrating that the HER half-reaction is not severely affected by the F-modification. The enhanced water splitting efficiency stems from enhanced OER half-reaction. Please see Page 9, Line 11.

Q6:
The stoichiometric ratio of hydrogen to oxygen production is far from two to one, which is crucial for studying fully decomposed water. The authors need to better illustrate this point by calculating Faraday efficiency or STH. In addition, the author can make a graph or bar chart with the hydrogen/oxygen ratio as the ordinate and the fluorine doping amount as the abscissa to study the influence of fluorine doping.
to ozone, and the presence of dissolved O2, it is reasonable that the H2/O2 ratio for the overall water splitting on semiconductor photocatalyst to be larger than a perfect stoichiometric ratio of 2:1. Similar phenomena were also observed in many other research works (Bai Y, et al, Angew. Chem. Int. Ed. 134, e202201299 (2022)). In our system, the main reason for the shortage of O2 production is the formation of H2O2, since CN is well-acknowledged as an efficient catalyst for H2O2 production (Nat. Commun. 12, 3701 (2021); ACS Catalysis 10, 14380-14389 (2020)). To quantify the H2O2 production during the reaction, we have employed a Ce 4+ back titration method (as reported in Adv. Mater. 34, e2107480 (2022)) (detailed methods see Supporting Information). As shown in Fig. R10, the H2O2 production rate on the champion F0.1-CN catalyst was determined to be 85.36 umol•g -1 •h -1 under white light, almost identical to the short of H2:O2 ratio.
The STH of the whole reaction was also calculated based on the O2/H2O2 ratio (under AM1.5G simulated solar irradiation) on F0.1-CN. According to the following equation, the STH of F0.1-CN was determined to be 0.00195%.
Moreover, as requested, the bar chart with the H2/O2 ratio as functions of the F-content in F-CN samples was depicted as shown in Fig. R11. Result shows that the H2/O2 ratio was slightly increased with the increasing F-content in F-CN samples, but typically around 3~4. Corresponding descriptions and calculations were added to the revised manuscript. Please see Page 8, Lines 5-9.  Q7: As far as I know, this journal has high requirements for repeatability of data, especially for statistics figures, such as Fig.2 a (the most important performance data I think); the error bars are usually needed and defined thoroughly.
Response: Thanks for your valuable comment. As requested, error bars were added to Fig. 2a (Fig. R12) and the corresponding bar chart of H2/O2 ratios (Supplementary Fig. 17), which was defined by statistically repeating identical experimental results three times. Please see Fig. 2a and Supplementary Fig 17 in the revised manuscript.

To Reviewer 2:
Though efforts have been taken by the authors, part of the comments have not been well responded: Response: Thanks for your great efforts in reviewing our manuscript. We appreciate your valuable comments and suggestions. Q1: Q1 (" In Fig. 3c, the rate-determining step of O2 evolution over CN(C) is the formation of *OH rather than the step from O* to *OOH. So, why the formation of C=O is the bottleneck of overall water splitting on CN-based catalysts? More illustration or an additional attention on the effect of C=O should be given out.") is not properly responded. It seems that they authors do not make clear in the manuscript that what is "the formation of C=O". Is it the same with "the step from *OH to O*"? Even taking consideration of this point, some important even crucial conclusions are inconsistent with the DFT results. It can be seen from Fig. 3c that, the formation energy barrier of *OH is 2.25 eV and the energy barrier of O* to *OOH is about 2 eV while the corresponding value of the formation of O* or C=O is about 0.85, which indicates the ratedetermining step of O2 evolution over CN(C) is the formation of *OH. This conflicts with the authors conclusion that "formation of C=O intermediate is an important bottleneck for overall water splitting on single-phased CN" and even the title "Intermediate C=O Formation Is the Bottleneck of Overall Water Splitting on Carbon Nitride".
Response: Thanks for your valuable comment. Yes, we have made an error when describing the relationship between the C=O intermediate and OER barriers on CN. From the DFT calculations, the step of *O→*OOH (the further transition of C=O, with the break of the double bond) clearly has a larger barrier than the step of *OH→*O (the formation of C=O), which is essentially an important bottleneck of OER on single-phased CN catalysts with observable C=O accumulation during the reaction. We have modified the incorrect description in the revised manuscript; see Page 9 Line 15. We assumed that the transition of *O to *OOH is essentially the rate-determining step of OER on CN with C=O accumulation, Thus, the description of "the C=O formation is the bottleneck" is not technically accurate. The title is correspondingly changed to "Intermediate C=O Transition Is the Bottleneck of Overall Water Splitting on Carbon Nitride". Thank you again for your very useful suggestions.
Q2: The answer to Q2 (The authors proposed that the N site is the main OER center in F-CN. While N 1s XPS spectra indicates no obvious shift under continuous white light illumination. Why an oxidized N state did not form like C in CN?) is unconvincing. From the results of DFT, the energy barrier of CN(C) and F-CN(N) has a similar trend. So, the statement that "charge transfer at the N site was very fast with no obvious accumulation of intermediate states" lead to the unchanged N 1s state in F-CN is confusing. Please offer a more reasonable explanation.

Q3:
The theoretical O2 production of CN is very low and may exceed the minimum detection value of TCD. The original GC spectra of pristine CN photocatalytic water splitting (e.g., 5h results of Fig. 2b) should be given.

Response:
Thanks for your valuable comment. The original GC-TCD spectra of Fig. 2b were provided (Fig. R1), from which the O2 production differences between CN and F0.1-CN can be clearly observed. Q4: For Q4, we advise the authors to clarify the stability of *OH and the adsorption strength of *OH. Is it possible to use electrochemical methanol oxidation experiment to offer further evidence?
Response: Thanks for your valuable comment. Based on the calculated charge density difference mapping, we assume that the F modification in F-CN optimizes the bonding interaction between CN surface and *OH intermediate, which effectively avoids the excessively strong C-O interaction or weak N-O interaction. As carbon sites were occupied by F, leaving less C-OH coordination, therefore the N coordinated *OH on F-CN should be less stable than C coordinated *OH on CN. As suggested, we conducted the electrochemical methanol oxidation reaction (MOR) on CN and F-CN to mimic the break of *OH. As shown in Fig. R2 and Fig. R3, MOR potentials on both samples were lower than the corresponding OER potentials. Particularly, the MOR on F0.1-CN was slightly lower than that on CN, which coincides with the less stable *OH on F-CN. However, the MOR contains the break of both C-H and O-H, and the break of C-H is clearly the high-energy barrier step in MOR. Thus, we think the observed shift of MOR potential majorly reflects the reaction energy of C-H break rather than the absolute stability of *OH on different samples.  Q5: For Q7, please carefully check the reference of CN materials (e.g., Angew. Chem. Int. Ed. 2018, 57, 6848-6852) and get a basic understanding about its structure. Are you sure "there are two types of N one type of C"? Response: Thanks for your valuable comment. In our first response, we only considered the coordination number of C and N atoms and indicated that there are two types of N coordination (N3c and N2c) and one type of C coordination (C3c). If taking into account the neighboring chemical environments in CN structure, the C site can be further divided into two types, i.e., three-coordinated C3c 1 and three-coordinated C3c 2 (As depicted in Fig. R4). F atom should be able to connect with both types of C atoms, and we only considered C3c 1 in our original manuscript. Thus, we further conducted DFT calculation on F-CN with C3c 2 occupied. As shown in Fig. R5, the reaction energies of OER on F-CN with occupied C3c 1 /C3c 2 and its influence on the neighboring N site are almost identical. The distinguish of C sites and additional DFT calculation results were added to the revised manuscript. Please see Page 10 Line 12. Thank you again for your very useful suggestions.

To Reviewer 3:
The revision is satisfactory.
Response: Thanks for your great efforts in reviewing our manuscript.
In the revised manuscript NCOMMS-22-17621A, the authors highlighted the innovation of this work compared to previous reported similar works, and conducted a systematic comparison of overall water splitting activities with recently reported CNbased catalysts. The authors have added some experimental results (AFM, EIS and Quantification of H2O2) to support the structure and reaction mechanism of F-CN according to the reviewers' suggestions. The quality of the manuscript has been improved significantly, and now I recommend the publication of this work in Nature Communication. However, the related experimental results are still controversial. The authors should address the following issues.
Response: Thanks for your great efforts in reviewing our manuscript. We appreciate your valuable comments and suggestions.

Q1:
The authors claimed that no changes are found on the N1s spectra on both CN and F0.1-CN samples under the white light illumination. However, the peak area ratios of NHx, NC3, C-N=C change significantly. The authors believed that the hybridization of F and C atoms affects the electron orbital of N atoms. This is not sufficient to explain the changes in the in situ NAP-XPS. In addition, detailed results of the XPS fit parameters should be provided.
Response: Thanks for your valuable comment. As shown in Supplementary Fig. 4 (Fig. R1), the difference between CN and F0.1-CN at the initial before light-on is due to the F modification. After the light-on, the peak area ratios of N 1s spectra only changed from the initial to 5 min of light irradiation, which is possibly due to the change of light condition. After 5 min of irradiation, the peak area ratios were almost unchanged. Moreover, detailed fit parameters of NAP-XPS data were provided as shown in Table. R1.   (Fig. R2), which is consistent with the H2/O2 ratio under the identical reaction condition. Please see Page 8 Line 5.  Fig. R2. The photocatalytic H2O2 production profiles under AM1.5G irradiation on F0.1-CN determined by the back titration of Ce 4+ .

Q3:
In Supplementary Fig. 9b, the authors should clearly note that the UV-vis absorption of the F-CN catalyst was slightly blue-shifted with high F content.