Dipole field in nitrogen-enriched carbon nitride with external forces to boost the artificial photosynthesis of hydrogen peroxide

Artificial photosynthesis is a promising strategy for efficient hydrogen peroxide production, but the poor directional charge transfer from bulk to active sites restricts the overall photocatalytic efficiency. To address this, a new process of dipole field-driven spontaneous polarization in nitrogen-rich triazole-based carbon nitride (C3N5) to harness photogenerated charge kinetics for hydrogen peroxide production is constructed. Here, C3N5 achieves a hydrogen peroxide photosynthesis rate of 3809.5 µmol g−1 h−1 and a 2e− transfer selectivity of 92% under simulated sunlight and ultrasonic forces. This high performance is attributed to the introduction of rich nitrogen active sites of the triazole ring in C3N5, which brings a dipole field. This dipole field induces a spontaneous polarization field to accelerate a rapid directional electron transfer process to nitrogen active sites and therefore induces Pauling-type adsorption of oxygen through an indirect 2e− transfer pathway to form hydrogen peroxide. This innovative concept using a dipole field to harness the migration and transport of photogenerated carriers provides a new route to improve photosynthesis efficiency via structural engineering.

3. Clarify if the H2O2 production in C3N5 under N2 gas is due to oxygen pre-adsorbed on C3N5, and why the H2O2 production rate slightly decreases under air.Sonication of water can produce H2O2 through the sonolysis of H2O molecule.Please carry out more control experiments to confirm how much of H2O2 is generated from the sonolysis of water alone.4. In Figure 4e, the peak strength is reduced after releasing pressure compared to before applying pressure.Clarify how the pressure affects the material's structure.
5. Conduct scavenger tests with different concentrations to obtain more reliable quenching results.
6. Provide more details on the light-only experiment, such as whether it was conducted under stirring, and add a separate stirring experiment without light as a control.Also, clarify the differences between ultrasound and stirring.
7. Provide more details on how the authors conducted in-situ pressure-dependent PL spectra, such as how they achieved the high target pressure on quartz.
Reviewer #3 (Remarks to the Author): Dipole field in nitrogen-enriched carbon nitride Photosynthesis of H2O2 from O2 and water is an ideal way to convert and store solar energy as well as the production of useful chemicals, because it can make full use of the photogenerated electrons and holes.Li et al. reported the photocatalytic H2O2 production on nitrogen-rich C3N5.This present work has very fruitful characterizations and mechanism investigations.The structural engineering concept is novel and inspiring, and the results are rich and convincing.I think it can be published in Nature Communications after following issues being addressed.
1.There are conceptual problems in Figure 1c.There always exists a surface band bending for a semiconductor, whose direction depends on the semiconductor type (n or p) and the exposed environment [Chem. Rev. 2012, 112, 5520-5551].So the flat potential throughout the bulk and surface in (i) is not scientifically accurate.It is suggested to present a n-type case band bending, considering the n-type nature of the carbon nitride in this work.The concept demonstrated in (ii) is also problematic.Besides the band bending issue mentioned above, the transfer direction of charge carriers is not correct.As the energy level shown here generally represents the energy of electrons, the electrons should transfer from the high position to the low positon of the energy level lines, and holes migrate to the opposite direction.In terms of the effect of diople moment on the surface band bending, please refer to the paper [Angew.Chem. 2020, 132, 945-952].
2. In Figure 2f, "an increase in the average surface potential" is used to characterize the effect of light on the surface potential.Yet it is not rational because as we can see, the difference between the lowest potential and the highest potential with light on C3N5 is about 940-890=50 mV, even larger than the average increase of 30.61 mV.Actually, the CPD difference before and after light irradiation (ΔCPD) reflects the surface band bending extent [Chem. Soc. Rev. 2018, 47, 8238-8262].It is suggested to present ΔCPD upon distance, and compare the two ΔCPD curves of C3N5 and C3N4.Moreover, the huge variation of ΔCPD upon distance indicates the uneven spatial distribution of suface band bending, which may give more insights into photogenerated carriers' kinetics if it can be correlated with the local structures of C3N5 and C3N4.
3. Solid state NMR is the most efficient and direct proof to investigate the detailed structure of carbon nitride-based materials.The author also used this technology to confirm the molecular structure.However, the present resolution of 15N NMR spectra is not convincing enough to get the conclusion.More precise characterizations should be performed to clearly demonstrate the structure.4. In presence of alcohols as electron donors, some other carbon nitride-based catalysts show higher efficiency in H2O2 production, such as ACS Catal. 2020, 10, 24, 14380-14389, Nat. Commun., 2021, 12, 3701, Angew. Chem. Int. Ed., 2021, 60(48): 25546-25550, Chem Catal., 2022, 2(7): 1720-1733 etc.The Figure S20 should also include these activities in the latest publications.5. To evaluate the efficiency of the catalyst, activity under monochromatic light should be performed to calculate the apparent quantum yield.
6.The in-situ pressure-dependent UV-vis absorption spectra characterization is very impressive.It may be inspiring to future works on the pressure-induced band gap modulation.It is suggested to provide a detailed picture of the set-up for this measurement to make it clearer to the readers.

Minor problems:
The pictures for adsorption configuration in Figure 6a is fuzzy.Clear and distinct pictures should be provided."The surface plane O2 is adsorbed in a Pauling-type manner at the N active sites in C3N5" should be "On the surface plane, O2 is adsorbed ..." 1.The authors don't mention in the manuscript any aspect regarding the synthesis and its rationale, just briefly in the SI, a triazole-based molecule is employed as starting monomer.Such molecules have been employed previously for the synthesis of carbon nitride observing a higher release of ammonia and nitrogen gas (by means of GC-MS) rather than the formation of a C3N5 (J.Mater. Chem. A 2017, 5, 8394).The differences in characterization between the so called C3N4 and C3N5 are less than minimal, for instance it is pretty much impossible to distinguish between both N1s spectra other than the y-scale is not normalized, while in previous C3N5 reports the difference was very substantial (J.Am.Chem. Soc. 2019, 141, 5415-5436).In the case of the solid state NMR, 3 out of the 5 chemical shifts for 15 N cant be distinguished from the background signal.
Author response: We appreciate the reviewer's critical comments.I am sorry that the previous version did not mention about any synthesis and its principles.In the revised version, we further demonstrate our structure using organic elemental analysis (UNICUBE-Elementar), Nuclear magnetic resonance (NMR, Bruker ADVANCE III 400 and 600 for solid-state NMR spectra and JNM-ECZ400R for liquid-state NMR spectra), Matrix-assisted laser desorption/ionization-time of flight mass spectrometry (MALDI-TOF-MS, Bruker Autoflex Speed TOF/TOF), liquid chromatography time-of-flight mass spectrometer (LC-TOF-MS, AB Sciex Triple TOF 5600), near-edge X-ray absorption finestructure (NEXAFS) at the beamline BL14W1 station of the Beijing Synchrotron Radiation Facility, China and Raman spectrometer (LabRAM HR Evolution spectrometer).
Indeed, the triazole-based molecule (3-amino-1,2,4-triazole, 3-AT) has been used for the synthesis of carbon nitride in a previous report [J. Mater. Chem. A 2017, 5, 8394].The synthesis method described in this report is that 3-AT is treated with hydrochloric 2/34 acid, ammonium persulfate and alkali, after which the resulting mixture is subjected to high temperature heat treatment with N2.However, our synthesis method is quite different from this report.We used a one-step pyrolysis method using air for the high temperature synthesis of triazole-based C3N5 backbone.Our methods are shown below: About 2.0 g of 3-amino-1,2,4-triazole powders was put into an Al2O3 crucible, and the crucible was covered with an Al2O3 cover to keep a half-cover state.The crucible was then heated to 500°C in a muffle oven at a rate of 5°C/min, kept at 500°C for 3 h, and then cooled down to room temperature in the muffle oven.
(Chemical structure of C3N5 in our work and previous work) Also, we are not at all the same structure as the C3N5 report [J.Am. Chem. Soc. 2019, 141, 5415-5436] you mentioned before, the specific structure is shown in above figure, therefore, their structural characterization is also completely different.And they are also synthesized in a completely different process, which can be shown in previous report.Therefore, it is not appropriate to compare our work with the above two works.
(Table S3: in the revised Supplementary Information) Here, carbon nitride structures at different temperatures including 200, 300 and 400°C using 3-AT (viz.CN-200°C, CN-300°C and CN-400°C) were also synthesized for comparison.First, we evaluate the elemental ratio of C to N for all samples using organic elemental analysis.The CHN results of C3N5 (CN-500°C) confirm that the average weight percentages of C and N are 33.1% and 63.1%, respectively (Table S3).The average atomic rate of C/N is thus determined to be 0.611, which is very close to the theorical value (0.60), confirming the successful synthesis of C3N5.Specifically, for 3-AT and CN-200°C, the elemental proportion of CNH has barely changed, which may be due to the fact that the boiling point of 3-AT is 244.9°C,implying that the chemical composition has not changed.For CN-300°C, CN-400°C and C3N5 (CN-500 o C), with the increase of synthesis temperature, the percentage weight of N and H decreases gradually, this may be due to the possible release of ammonia and hydrogen during the synthesis process.The total mass fraction of 3-AT exceeded 100%, due to the water absorption of the sample.
(Figure S7: in the revised Supplementary Information)  MALDI-TOF-MS was used to determine the chemical structure of intermediates.Figure S7a presents the whole MALDI-TOF-MS spectra of the intermediate of CN prepared at 400°C.Since the amino groups on the carbon nitride are easily charged with positive protons, we tested the intermediates in positive ion mode.The peaks with m/z of 64, 71, 96, 106, 127,192, 196 and 211 Da are observed in the MALDI-TOF MS spectra of CN-400°C.Therefore, some possible molecular structures of the intermediates in the CN polymerization are listed in Figure S7b.Based on this, the above m/z can correspond to the eight positively ionized products of the four molecules as above.This implies the possible existence of these four intermediates.Other excess peak may be due to the effect of incomplete polymerization of 3-AT and other impurities.
(Figure S8: in the revised Supplementary Information) LC-TOF-MS was used to determine the chemical structure of intermediates.Figure S8a presents the whole LC-TOF-MS spectra of the intermediate of CN prepared at 400°C.We also tested the intermediates in positive ion mode, because the amino groups on the carbon nitride are easily charged with positive protons.The peaks with m/z of 43, 64, 71, 96, 106, 127,192, 196 and 211 Da are observed in the LC-TOF-MS spectra of CN-400 o C. Therefore, the positively ionization mode of these possible intermediates is shown in are listed in Figure S8b.These results are also very consistent with the results of MALDI-TOF-MS.This also demonstrates the possible existence of these four intermediates.
(Figure S5: in the revised Supplementary Information)

6/34
Near-edge X-ray absorption finestructure (NEXAFS) analysis was also used to explore the chemical bonds of triazole-based C3N5.The C K-edge NEXAFS spectra (Figure S5) of C3N5 and C3N4 show characteristic excitations including 1s→π*out of plane C=C at ~285.2 eV and 1s→π*N-C=N at 288.0 eV [Nat. Commun. 2014, 5, 3783].Compared with C3N4, C3N5 displays a blue shift of 1s→π*N-C=N excitation, which is attributed to increased N-C=N bond strength upon the formation of triazole moiety.The N K-edge NEXAFS spectra (Figure 2c) of two samples show similar response at 399.3 and 402.5 eV, attributing to the 1s→π*N-C=N and π*C-N resonance.Note that, relative to C3N4, triazolebased C3N5 displays a red shift of 1s→π*N-C=N excitation and a new peak at 401.1 eV, which are attributed to the formation of triazole moiety and 1s→  *heterocyclic N-N of triazole, respectively [Nat.Commun. 2014, 5, 3783].These experimental findings provide strong evidence of the coexistence of triazole and triazine moieties in the tetrazole derived C3N5 materials.
The chemical structure of the intermediates was further demonstrated by NMR spectroscopy (Figure S9).The 15 N spectra of 3-AT and CN-200 o C were tested with liquid NMR analysis (JNM-ECZ400R with 5mm Royal probe) dissolved by DMSO.The 13 C solidstate NMR spectra of all samples and 15 N solid-state NMR spectra of CN-300 o C, CN-400 o C, C3N5 (CN-500 o C) and C3N4 were acquired on Bruker ADVANCE III 400 and 600 equipped with a 4 mm double resonance MAS NMR probe using the cross-polarization magic-angle spinning (CPMAS). 13C spectra were referenced to TMS (δ( 13 C) = 0.00 ppm) by setting the high frequency 13 C peak of solid adamantane to 38.56 ppm. 15N spectra were referenced to nitromethane δ( 15 N) = 0.00 ppm by setting the isotropic peak of a glycine sample (98 % 15 N) to −347.6 ppm.
Based on the above results, we proposed the possible synthesis process.(Figure S12: in the revised Supplementary Information) Our modification to the manuscript: The corresponding description was added to the revised manuscript (Page 3) and Supplementary Information (Text S3, Figures 2c, 2d, S5 and S7-12 and Table S3).
2. The prepared material does present defects and a substantially narrower band gap which could be contributing to the enhanced photocatalytic performance.Claiming that such performance is higher than the state of the art, however, is quite misleading as the performance strongly depends also on reactor design so one could just make such claim if other catalysts have been tested under sonication and illumination with the same intensity (and the reactor design shown here is quite unconventional).
Author response: We appreciate the reviewer's professional comments.Your suggestion is very valuable.Indeed, the performance of the catalyst depends greatly on the reactor design.Here, we apply both ultrasonic and light for catalytic H2O2 production, and compared the performance of related reports presented in Figure S30 and Table S4.Since there are few reports on the production of H2O2 by applying both external force and light, I summarized the H2O2 production performance reported carbon nitride photocatalysts, all piezoelectric catalysts and all piezoelectric photocatalysts.Furthermore, we also added photocatalysts that previously lacked carbon nitride (Figure S30 and Table S4).However, it is very difficult to normalize the same intensity, especially for ultrasonic conversion efficiency, and most reports do not give the light intensity and frequency and power of ultrasound.Therefore, we have 10/34 further summarized the detailed reaction conditions including light intensity, ultrasound power and frequency, and rotational speed in all reference lists (Table S4) for further comparison.Overall, H2O2 yield we reported exceeds that of most catalysts.
(Figure S30 and Table S4: in the revised Supplementary Information)  Our modification to the manuscript: The corresponding description was added to Supplementary Information (Figure S30 and Table S4).
3. Additionally, the 2e -selectivity has been assessed electrochemically by means of an RRDE which substantially different to a photocatalytic scenario, one depending on surface area and adsorption desorption and the other one depending more heavily in energy band position.Average quantum yield for instance would be more useful.
Author response: We appreciate the reviewer's professional comments.
(Figure S25: in the revised Supplementary Information) Our modification to the manuscript: The corresponding description was added to the revised manuscript (Page 5) and Supplementary Information (Text S7 and Figure S25).Summary Comments: Following this reasons i don't think this manuscript can be considered for publication in Nature Communications and i recommend the authors to either reword the discussion acknowledging the ambiguity of the structure, or try to elucidate further the structure by means of mass spectrometry or elemental analysis, which is very useful to prove the structure of melem analogues (RSC Adv. 2021, 11, 38862) Author response: We highly appreciate the comments from the reviewer, and they are all considered in corrected manuscript.In the revised version, we further demonstrate our structure using organic elemental analysis, NMR, MALDI-TOF-MS, LC-TOF-MS, NEXAFS and Raman spectrometer.
Thank you very much again for your kind and appropriate comments.We are sure that these comments help improve the quality of our manuscript significantly.

14/34
Reviewer #2 (Remarks to the Author): Summary Comments: The authors have demonstrated the importance of the internal dipole field in nitrogen-rich carbon nitride photocatalysts and its impact on charge separation and H2O2 production.This work is of significant contribution to the field and deserves to be published in this journal after revisions addressing the following comments and questions: Author response: We highly appreciate the comments from the reviewer, and they are all considered in corrected manuscript.
1. Clarify if the polarization induced by the dipole field is triggered by an external force, such as pressure or ultrasonication, or if it is a physical property of nitrogen-rich carbon nitride.
Author response: We appreciate the reviewer's critical comments.As the reviewer questioned, the formation of dipole fields is an inherent property of nitrogen-rich carbon nitride, which originates from the dipole induced by the material's own asymmetric structure.This dipole is defined as a pair of opposite charges "q" and "-q" separated by a distance "d".The direction of the dipole moment (p) in space is from negative charge "-q" to positive charge "q".This dipole moment with an electron cloud distribution forms a dipole field.It is worth noting that this dipole field can be adjusted by applying an external force such as pressure or ultrasonication to enhance the field strength, thereby promoting the separation efficiency of photogenerated carriers (Figure 1b).Here, polymerization of the triazole and triazine framework to form a nitrogen-rich carbon nitride (viz.C3N5) leads to asymmetry of the structure, and a dipole moment is generated by the noncoincidence of the positively and negatively charged centers with a value of 2.80 D for a single unit (Figure 1d).When the number of units increases to 6, the dipole moment is enhanced to 22.22 D (Figure 1e).This strong dipole moment means that dipole field-driven spontaneous polarization in C3N5 can be used to harness photogenerated charge separation kinetics.
(Figures 1b, d and e: in the revised manuscript) 15/34 2. In Figure 3b, C3N4 produces H2O2 only with sonication despite lacking an internal dipole field.Recent reports suggest that C3N4 nanosheets exhibit in-plane piezoelectricity.Were the samples sonicated to form nanosheets before conducting photosynthesis?
Author response: We appreciate the reviewer's professional comments.As you mentioned, C3N4 can produce hydrogen peroxide under ultrasound without a dipole field, which can be due to following reasons: C3N4 has been shown to own in-plane piezoelectric response [J.Mater. Chem. A, 2019, 7, 20383;Adv. Mater. 2021, 33, 2101751], and its piezoelectric effect is due to the polarization field formed by the uneven distribution of positive and negative charge centers under external force, rather than its own structurally induced dipole field.Prior to photosynthesis, our samples were subjected by ultrasound treatment for 30 min to form nanosheets.

Our modification to the manuscript:
The corresponding description was added to revised manuscript (Page 8).
3. Clarify if the H2O2 production in C3N5 under N2 gas is due to oxygen pre-adsorbed on C3N5, and why the H2O2 production rate slightly decreases under air.Sonication of water can produce H2O2 through the sonolysis of H2O molecule.Please carry out more control experiments to confirm how much of H2O2 is generated from the sonolysis of water alone.

Author response:
We appreciate the reviewer's professional comments.
(1) Clarify if the H2O2 production in C3N5 under N2 gas is due to oxygen pre-adsorbed on C3N5.
Based on your suggestion, we re-performed the N2 atmosphere experiment, further checked the gas tightness, and extended the N2 blast time to 20 min.The new experimental results demonstrated that almost no H2O2 was generated in the N2 atmosphere (Figure 3d), and the results were roughly the same as the acoustic yield of water's sonification (Figure S23).These demonstrate that this may have been due to the poor gas tightness previously, which led to the entry of air and caused the elevated yield.
(2) why the H2O2 production rate slightly decreases under air.
As the following equation shows, the O2 concentration is crucial for the synthesis of H2O2, and a higher O2 concentration facilitates the positive proceeding of reaction and the synthesis of H2O2.Therefore, when a sealed reactor is filled with O2, it can show a higher H2O2 yield compared to an open air atmosphere.
(3) Sonication of water can produce H2O2 through the sonolysis of H2O molecule.Please carry out more control experiments to confirm how much of H2O2 is generated from the sonolysis of water alone.
Based on your suggestion, we conducted a direct ultrasonic experiment of pure water, and the result show that H2O2 of 21.3 μM was produced after 1h ultrasound.
(Figure 3d: in the revised manuscript)  Our modification to the manuscript: The corresponding description was added to revised manuscript (Page 5, Figure 3d) and Supplementary Information (Figure S23).4e, the peak strength is reduced after releasing pressure compared to before applying pressure.Clarify how the pressure affects the material's structure.

In Figure
Author response: We appreciate the reviewer's professional comments.To monitor the charge migration process in photoexcited C3N5 under an external force, in situ pressure-dependent PL spectroscopy was performed.With gradually increasing pressure from ambient to 15 GPa, both C3N4 and C3N5 display similar changes from initial blue to green and finally to colorless at high pressure (Figures S48 and 49).These results are consistent with the chromaticity diagram of the Commission Internationale de I'Eclairage (CIE) (Figures S50).As shown in Figures S51 and 4e, as the pressure 17/34 increases, the PL peak positions of the two samples show an obvious redshift, and the peak intensity of C3N5 is significantly lower than that of C3N4.When the pressure is released to 0 GPa, the peak position recovers, whereas the peak intensity decreases because the destruction of the structure under applied pressure weakens the luminescence of the material [Nanoscale 2020, 12, 12300-12307].These results confirm that the inhibition efficiencies for charge recombination in photoexcited C3N4 and C3N5 increase with increasing external pressure, and the charge separation efficiency for C3N5 is higher than that for C3N4.Therefore, the dipole field in C3N5 improves the charge migration behavior.
Recent reports have investigated the potential mechanism of these anomalous transitions in PL behavior via in situ high pressure XRD [Nanoscale 2020, 12, 12300-12307].As the pressure increases, the layer stacking order of the carbon nitride material decreases while the layer interactions increase.In this case, the tri-s-triazine of carbon nitride shifting to the porous position of the nearest neighboring layer with an obvious drop in volume and the electron interactions are enhanced, especially, at the positions with larger electronic density where lone pair electrons of nitrogen occur in two samples under high pressure.This should further affect the PL emission related to the lone pair electrons of nitrogen.Hence, we observed that as two samples transforms into a less compressible state, and exhibit a more significant decrease in intensity under pressure.

Our modification to the manuscript:
The corresponding description was added to Supplementary Information (Note of Figure S51).

Conduct scavenger tests with different concentrations to obtain more reliable quenching results.
Author response: We appreciate the reviewer's professional comments.Based on your suggestion, 1, 2 and 5 mM of scavengers were added to the initial solution in photocatalytic H2O2 production with Us.TBA for *OH have little effect on the production of H2O2.It is clearly visible, p-BQ for *O2 − ) mainly contribute to H2O2 production, and with the increase of p-BQ, the production of H2O2 gradually decreases.
(Figure S56: in the revised Supplementary Information)

18/34
Our modification to the manuscript: The corresponding description was added to Supplementary Information (Figure S56).
6. Provide more details on the light-only experiment, such as whether it was conducted under stirring, and add a separate stirring experiment without light as a control.Also, clarify the differences between ultrasound and stirring.
Author response: We appreciate the reviewer's professional comments.Based on your suggestion, we added a separate stirring experiment without light as a control.In Figure S24, C3N4 and C3N5 produce almost no H2O2 with only stirring.20 mg of catalyst was added to 40 mL of pure water containing ethanol (10 vol%) at pH=3.The catalyst was dispersed by ultrasonication for 10 min, and air was bubbled through the solution for 10 min.The reactor was kept at 25 ± 0.5°C with cooling circulating water and was irradiated at λ ≥ 420 nm using a 300 W Xe lamp (PLS-SXE300D, Beijing Perfectlight Technology Co., Ltd) with a light intensity of 100 mW cm -2 , and simultaneously subjected to ultrasonication by an ultrasonic cleaner (40 kHz, 100 W, Jielimei, Kunshan, China).The light-only experiments were placed under a xenon lamp with stirring.
In catalytic reactions, stirring is often used for macroscopic mixing of liquids.However, it is difficult to achieve adequate mixing and generate enough force to excite the material for piezoelectric effects.While ultrasound provides microscopic mixing of liquid materials and induces an ultrasonic cavitation effect by applying a periodic force that causes the rupture of cavitation bubbles.This leads to a high pressure of up to 10 8 Pa at the non-homogeneous catalyst/water interface, which provides sufficient energy for electron excitation.Moreover, ultrasound can directly excite pure water through cavitation to produce H2O2.These conclusions are also supported by the results presented in Figure S23.
(Figure S24: in the revised Supplementary Information) Our modification to the manuscript: The corresponding description was added to revised manuscript (Page 5) and Supplementary Information (Figure S24).
19/34 7. Provide more details on how the authors conducted in-situ pressure-dependent PL spectra, such as how they achieved the high target pressure on quartz.
Author response: We appreciate the reviewer's critical comments.High-pressure experiments were performed using a symmetric diamond anvil cell (DAC).A pair of ultra-low fluorescence diamonds with anvil surface of 400 µm diameter was used to generate pressure for the in situ high-pressure PL experiments.A T301 stainless steel gasket pre-indented to the thickness of 45 µm was laser drilled to obtain a 150 µm diameter hole for loading the sample and a ruby ball.The ruby fluorescence technique was adopted for the pressure calibration in connection with the shift of the R1 line of the ruby fluorescence.The silicon oil was applied as pressure-transmitting medium around the sample.All the high-pressure experiments were conducted at room temperature.The high-pressure PL measurement were performed by using a combined home-made optical measurement system.The pressure-dependent PL spectra were measured by a semiconductor laser with an excitation wavelength of 355 nm, and recorded with an optical fiber spectrometer (Ocean Optics, QE65000).Note that the effects of different excitation laser intensities and luminous fluxes on the obtained PL intensity of carbon nitride were avoided by fixing all the parameters during each high-pressure PL experiment.
(Figure S61: in the revised Supplementary Information) Our modification to the manuscript: The corresponding description was added to revised manuscript (Page 8-9) and Supplementary Information (Figure S61).
8. Evaluate the electron transport capacity of C 3N5 and C3N4 using photocurrent and impedance measurements.
Author response: We appreciate the reviewer's professional comments.To reveal carriers migration of C3N5, the transient piezoelectric current response is depicted in Figure S47, manifesting obvious and repeatable piezo-photo current signals with on/off of applying ultrasound or visible light.As shown in Figure S47, the current intensity (j) of C3N5 and C3N4 in various scenarios at 60 min follows the sequence C3N5/Us/Vis > C3N4/Us/Vis > C3N5/Vis > C3N4/Vis > C3N5/Us > C3N4/Us.The C3N5 under both ultrasound and visible light displays the highest migration rate of carriers, and 20/34 the result is matched well with H2O2 production.The electrochemical impedance spectroscopy (EIS) further explains a high-efficient carriers transfer of C3N5.The arc diameter of C3N5 is the smallest than that of C3N4.The smaller diameter is due to the lower charge transfer resistance and the faster mobility of electrons, indicating that the dipole field contributes to highest charges transfer efficiency of C3N5.These results fully demonstrate that the dipole field of triazole-based C3N5 can rapidly improve the separation rate of carriers.(Figure S47: in the revised Supplementary Information) Our modification to the manuscript: The corresponding description was added to revised manuscript (Page 6) and Supplementary Information (Figure S47).9. Include more references related to polarization-induced exciton dissociation in polymeric photocatalysts, such as Chem Catal 2, 1734-1747; Chem Catalysis 2 (7), 1517-1519.
Author response: We appreciate the reviewer's critical comments.Based on your suggestions, I have added corresponding references.
Our modification to the manuscript: The corresponding references has been added in the revised manuscript (Ref.7 and 11).
Thank you very much again for your kind and appropriate comments.We are sure that these comments help improve the quality of our manuscript significantly.

21/34
Reviewer #3 (Remarks to the Author): Summary Comments: Dipole field in nitrogen-enriched carbon nitride Photosynthesis of H2O2 from O2 and water is an ideal way to convert and store solar energy as well as the production of useful chemicals, because it can make full use of the photogenerated electrons and holes.Li et al. reported the photocatalytic H2O2 production on nitrogenrich C3N5.This present work has very fruitful characterizations and mechanism investigations.The structural engineering concept is novel and inspiring, and the results are rich and convincing.I think it can be published in Nature Communications after following issues being addressed.
Author response: We highly appreciate the positive comments from the reviewer, and they are all considered in corrected manuscript.
1.There are conceptual problems in Figure 1c.There always exists a surface band bending for a semiconductor, whose direction depends on the semiconductor type (n or p) and the exposed environment [Chem. Rev. 2012, 112, 5520-5551].So the flat potential throughout the bulk and surface in (i) is not scientifically accurate.It is suggested to present a n-type case band bending, considering the n-type nature of the carbon nitride in this work.The concept demonstrated in (ii) is also problematic.
Besides the band bending issue mentioned above, the transfer direction of charge carriers is not correct.As the energy level shown here generally represents the energy of electrons, the electrons should transfer from the high position to the low positon of the energy level lines, and holes migrate to the opposite direction.In terms of the effect of diople moment on the surface band bending, please refer to the paper [Angew.Chem. 2020, 132, 945-952].
Author response: We thank you for the reviewer's constructive comment.Your suggestion will have a great significance to improve the quality of my work.Based on your suggestion, we have reproposed an n-type case band bending model and the transfer direction of charge carriers [Chem. Rev. 2012, 112, 5520-5551].It is worth noting that polar materials with dipole moments have a profound effect on surface energy band bending.The degree of upward band bending is more pronounced for polar materials accompanied by a dipole moment than for surfaces of nonpolar materials.Compared to nonpolar materials, the surfaces of polar materials with dipole moments exhibit more significant upward band bending [Angew.Chem. 2020, 132, 945-952;Chem. Soc. Rev. 2018, 47, 8238-8262].
(Figure 1c: in the revised manuscript) 22/34 Our modification to the manuscript: The corresponding description has been added in the revised manuscript (Page 2 and Figure 1c).
2. In Figure 2f, "an increase in the average surface potential" is used to characterize the effect of light on the surface potential.Yet it is not rational because as we can see, the difference between the lowest potential and the highest potential with light on C3N5 is about 940-890=50 mV, even larger than the average increase of 30.61 mV.Actually, the CPD difference before and after light irradiation (ΔCPD) reflects the surface band bending extent [Chem. Soc. Rev. 2018, 47, 8238-8262].It is suggested to present ΔCPD upon distance, and compare the two ΔCPD curves of C3N5 and C3N4.Moreover, the huge variation of ΔCPD upon distance indicates the uneven spatial distribution of suface band bending, which may give more insights into photogenerated carriers' kinetics if it can be correlated with the local structures of C3N5 and C3N4.
Author response: We thank you for the reviewer's constructive comment.Based on your valuable suggestions, we reselected the appropriate areas and presented the CPD according to the distance.To further identify surface charge modulation, the surface piezoelectric potential distribution with the surface morphologies of C3N5 and C3N4 was evaluated by Kelvin probe force microscopy (KPFM) (Figures 2f and S15).Upon illumination, we can observe that the KPFM images become brighter for n-type carbon nitride [Chem. Soc. Rev. 2018, 47, 8238-8262].The results agree well with the CPD increases for n-type semiconductors.The range of the contact potential difference (CPD) for C3N5 under dark condition is approximately 859∼888 mV, which is apparently higher than the range of 607∼635 mV for C3N4; that for C3N5 under light condition is approximately 908∼946 mV, which is also higher than the range of 620∼649 mV for C3N4.Moreover, C3N5 with light irradiation exhibits an increase in the average surface potential of ≈ 44.42 mV relative to that without light, while C3N4 with light shows an increase of ≈ 10.41 mV (Figure 2f).This is because the spontaneous polarization induced by the dipole field of C3N5 amplifies the directional charge transfer upon light irradiation.As previously reported, compared to nonpolar materials, the surfaces of polar materials with dipole moments exhibit more significant upward band bending [Chem.Soc. Rev. 2018, 47, 8238-8262], which effectively inhibits charge recombination upon irradiation with modulated light, the overall surface potential of the whole material is increased (Figure S15).Furthermore, the huge variation of ΔCPD upon distance indicates the uneven spatial distribution of surface band bending, which may be correlated with the local polarization structures of C3N5 and C3N4.Your suggestion will have a great significance to improve the quality of my work.It is a valuable and far-reaching proposal.However, due to time constraints, we will conduct study on your suggestion in the follow-up work, especially the role of relationship between ΔCPD and local structures of material.I truly believe that your opinion will significantly improve the quality of our future work.Our modification to the manuscript: The corresponding description has been added in the revised manuscript (Page 4, Figure 2f) and Supplementary Information (Figure S15).
3. Solid state NMR is the most efficient and direct proof to investigate the detailed structure of carbon nitride-based materials.The author also used this technology to confirm the molecular structure.However, the present resolution of 15 N NMR spectra is not convincing enough to get the conclusion.More precise characterizations should be performed to clearly demonstrate the structure.MALDI-TOF-MS was used to determine the chemical structure of intermediates.Figure S7a presents the whole MALDI-TOF-MS spectra of the intermediate of CN prepared at 400°C.Since the amino groups on the carbon nitride are easily charged with positive protons, we tested the intermediates in positive ion mode.The peaks with m/z of 64, 71, 96, 106, 127,192, 196 and 211 Da are observed in the MALDI-TOF MS spectra of CN-400°C.Therefore, some possible molecular structures of the intermediates in the CN polymerization are listed in Figure S7b.Based on this, the above m/z can correspond to the eight positively ionized products of the four molecules as above.This implies the possible existence of these four intermediates.Other excess peak may be due to the effect of incomplete polymerization of 3-AT and other impurities.LC-TOF-MS was used to determine the chemical structure of intermediates.Figure S8a presents the whole LC-TOF-MS spectra of the intermediate of CN prepared at 400°C.We also tested the intermediates in positive ion mode, because the amino groups on the carbon nitride are easily charged with positive protons.The peaks with m/z of 43, 64, 71, 96, 106, 127,192, 196 and 211  which are attributed to the formation of triazole moiety and 1s→  *heterocyclic N-N of triazole, respectively [Nat. Commun. 2014, 5, 3783].These experimental findings provide strong evidence of the coexistence of triazole and triazine moieties in the tetrazole derived C3N5 materials.
The chemical structure of the intermediates was further demonstrated by NMR spectroscopy (Figure S9).The 15 N spectra of 3-AT and CN-200 o C were tested with liquid NMR analysis (JNM-ECZ400R with 5mm Royal probe) dissolved by DMSO.The 13 C solidstate NMR spectra of all samples and 15 N solid-state NMR spectra of CN-300 o C, CN-400 o C, C3N5 (CN-500 o C) and C3N4 were acquired on Bruker ADVANCE III 400 and 600 equipped with a 4 mm double resonance MAS NMR probe using the cross-polarization magic-angle spinning (CPMAS). 13C spectra were referenced to TMS (δ( 13 C) = 0.00 ppm) by setting the high frequency 13 C peak of solid adamantane to 38.56 ppm. 15N spectra were referenced to nitromethane δ( 15 N) = 0.00 ppm by setting the isotropic peak of a glycine sample (98 % 15 N) to −347.6 ppm.

Figure S7 :
Figure S7: MALDI-TOF-MS spectra (a) of CN-400 o C; Major species (b) constituting the precipitate formed from CN-400 o C. The solid sample was adopted for testing.

Figure S8 :
Figure S8: LC-TOF-MS spectra (a) of H 2O and CN-400 o C; Major species (b) constituting the precipitate formed from CN-400 o C. DMSO was used as a solvent for CN-400 o C, and DMSO was used as a control sample.

Figure S9 :
Figure S9: 13 C and 15 N NMR spectra (a and b) of 3-AT, CN-200 o C, CN-300 o C, CN-400 o C, C3N5 (CN-500 o C) and C3N4 and possible structural model representation (c).The 15 N spectra of 3-AT and CN-200 o C were tested with liquid sample dissolved by DMSO, and the other samples were tested with solid samples.

Figure S10 :
Figure S10: Structural model and theoretical correspondence of 13 C NMR chemical shift produced from ChemDraw.(Figure S11: in the revised Supplementary Information)

Figure S11 :
Figure S11: Micro Raman spectra of C3N4 and C3N5.The circles and squares indicate Raman modes of triazine and triazole rings, respectively.524nm shows the signal of Si, since the substrate is made of silicon.

Figure S12 :
Figure S12: The possible synthetic steps of C3N5.

Figure S30 :
Figure S30: H2O2 production rates for C3N5 in this work compared with reported work.

Figure S25 :
Figure S25: Apparent quantum yield (AQY) of H2O2 production at specific wavelengths superimposed with its UV-Vis absorption curve.

Figure 1 :
Figure 1: b, Dipole moment and its electron cloud distribution, and dipole field and its change with external forces.d, Structural unit and dipole moments of C3N4 and C3N5 with positive and negative charge centers.e, Dipole moments of C3N4 and C3N5 with different structural unit numbers.

Figure 3d :
Figure 3d: Effect of dissolved oxygen on H2O2 production for C3N5/Us/Vis in 1 h.(Figure S23: in the revised Supplementary Information)

Figure
Figure S23: a, Time profiles of H2O2 production via pure water with Us. b, Corresponding histograms of the H2O2 yield at 60 min.

Figure S24 :
Figure S24: Time profiles of H2O2 production by stirring-only with C3N4 and C3N5.

Figure
Figure S47: a, Transient current density-time curves of C3N4 and C3N5 with on-off cycles of US, Vis, and US/Vis at a potential of -0.5 V in 0.5 M Na2SO4 solution, b, Nyquist plots of C3N4 and C3N5.

(
Figure 2f: Surface potential from KPFM images of C3N4 and C3N5 with (w/) and without (w/o) light irradiation.(Figures S15: in the revised Supplementary Information)

Figure S15 :
Figure S15: Surface morphologies and corresponding KPFM potential images of C3N4 (a) and C3N5 (b).i, AFM 3D topography images.ii, AFM 2D topography images.iii, iv, contact potential difference (CPD) of C3N4/C3N5 with and without light.The arrow is the selected surface potential area.

Figure S7 .
Figure S7.MALDI-TOF-MS spectra (a) of CN-400 o C; Major species (b) constituting the precipitate formed from CN-400 o C. The solid sample was adopted for testing.MALDI-TOF-MS was used to determine the chemical structure of intermediates.FigureS7apresents the whole MALDI-TOF-MS spectra of the intermediate of CN prepared at 400°C.Since the amino groups on the carbon nitride are easily charged with positive protons, we tested the intermediates in positive ion mode.The peaks with m/z of64, 71, 96, 106, 127,192, 196 and 211  Da are observed in the MALDI-TOF MS spectra of CN-400°C.Therefore, some possible molecular structures of the intermediates in the CN polymerization are listed in FigureS7b.Based on this, the above m/z can correspond to the eight positively ionized products of the four molecules as above.This implies the possible existence of these four intermediates.Other excess peak may be due to the effect of incomplete polymerization of 3-AT and other impurities.
Figure S8.LC-TOF-MS spectra (a) of H2O and CN-400 o C; Major species (b) constituting the precipitate formed from CN-400 o C. DMSO was used as a solvent for CN-400 o C, and DMSO was used as a control sample.
Figure S5.C K-edge NEXAFS spectra of C3N4 and C3N5.(Figure 2c: in the revised manuscript)

15N
Figure S9. 13C and 15 N NMR spectra (a and b) of 3-AT, CN-200 o C, CN-300 o C, CN-400 o C, C3N5 (CN-500 o C) and C3N4 and possible structural model representation (c).The 15 N spectra of 3-AT and CN-200 o C were tested with liquid sample dissolved by DMSO, and the other samples were tested with solid samples.(Figure S10: in the revised Supplementary Information)

Table S4 .
H2O2 production rates for C3N5 in this work compared with representative recently reported work.