Photocatalytic CO2 reduction to syngas using metallosalen covalent organic frameworks

Metallosalen-covalent organic frameworks have recently gained attention in photocatalysis. However, their use in CO2 photoreduction is yet to be reported. Moreover, facile preparation of metallosalen-covalent organic frameworks with good crystallinity remains considerably challenging. Herein, we report a series of metallosalen-covalent organic frameworks produced via a one-step synthesis strategy that does not require vacuum evacuation. Metallosalen-covalent organic frameworks possessing controllable coordination environments of mononuclear and binuclear metal sites are obtained and act as photocatalysts for tunable syngas production from CO2. Metallosalen-covalent organic frameworks obtained via one-step synthesis exhibit higher crystallinity and catalytic activities than those obtained from two-step synthesis. The optimal framework material containing cobalt and triazine achieves a syngas production rate of 19.7 mmol g−1 h−1 (11:8 H2/CO), outperforming previously reported porous crystalline materials. This study provides a facile strategy for producing metallosalen-covalent organic frameworks of high quality and can accelerate their exploration in various applications.

structures?The 2theta positions of the major peaks seem to be a bit too high if the COF structures are actually as drawn in figure 2a-c. 5.In principle, the aldehyde groups can directly react with the tridentate amine ligands (TAPB/TAPT), which is a competition reaction of them reacting with ethylenediamine.If so, COF can also be produced.From the characterizations in this manuscript, it cannot rule out the possibility of these two reactions happening at the same time.6.From the nitrogen sorption isotherms and pore size distributions, the pores of these M-COFs seem to be highly distributed, indicating the disordered structures.7. The influence of synthetic methods is frequently emphasized in this manuscript.However, all the contrasts between M-COF-1 to M-COF-2 can be attributed to their crystallinity difference (as shown in Figure S5).Amorphous counterparts should be put into comparisons.8. What is the stability of these COFs under photocatalytic conditions?
Reviewer #3 (Remarks to the Author): This manuscript described one-step synthesis of three M-COFs, which were further used as photocatalysts.Characterizations have been made to analyze the M-COFs and to evaluate their performance for CO2 photoreduction.The topic is interesting, but the authors failed to demonstrate enough novelty and impact which can demand the publication in Nature Communications.The monomers and reactions are well-known.The crystallinities of the products do not reach the state of the art in this field.In addition to some flaws and claims without clear support, there are critical concerns about the modelled structures of the M-COFs, i.e., the layer stacking.Therefore, I cannot recommend its publication in Nature Communications.The detailed comments and questions are listed as following.
1.In the abstract, the authors identified low crystallinities of COFs are the bottleneck of the field.While the statement is true, the high crystallinity of the COFs claimed by the authors in this study is misleading.The COFs only show less than ten peaks in the PXRD pattern, which is far from high crystallinity in the state of the art of M-COFs.See e.g., Nat. Commun. 2020, 11, 1434. 2. The unit cell parameters and space groups in Figure 2 are inconsistent to those in Tables S1 and S2 and in the text.3. The Pawley fitting shown in Figure 2 have serious issues.(i) There is no space group of P62m.(ii) For Figure 1d, there should be more Bragg positions before 5 degree for the space group P-62m and the given unit cell.(iii) Why the c-axis of ZnZn-TAPB-COF-1 is 19.99 Å while the structure model looks having the same layer stacking as the others?(iv) Why there is no peak at the first several Bragg positions?4. In modelling the structures of the COFs, the authors only considered the AA eclipsed stacking, which could cause misleading.How about other common layer stacking behaviors, such as AA inclined and AB staggered stacking? 5. To justify the structural models, I suggest to include a figure comparing experimental PXRD patterns with those calculated from the structural models.6.How large are the pores measured from the structural models and how good do they agree with the N2 sorption results?The pore size distributions in Figures S16-21 shows two pore sizes.One of c.a. 1.5 nm, and another of c.a. 3.0 nm.The authors need to describe where they come from.7. The authors mentioned the yield of ZnZn-TATP-COF-1 on line 130, and conclude the one step synthesis as having high yield.This is misleading.The authors need to show the yield of other one step made COFs and compare the yields to two step synthesis.8. On line 150, the authors claimed that one step synthesis show higher BET surface area than those synthesized by two steps.This is not true.Table S3 shows that half are the opposite.9.The Zn-O bonding distance should be in the range of 1.9-2.1 Å for standard zinc oxides.The EXAFS results shown in Figure 3 show a distance of c.a. 1.6 Å, which has a significant deviation.10.The section of morphological control of the COFs reads abruptly.Do the morphologies related to the focus of this manuscript, i.e., performance of CO2 reduction?11.It is difficult to access the structural models from Tables S1 and S2.I suggest to upload cif files as supplementary information, even though they are modelled structures.12.The authors need to check the convention of writing space groups.For example, there is no space group should be written as P6/M.Reviewer #4 (Remarks to the Author): The synthesis of metallo-covalent organic frameworks (M-COFs) with high crystalline is highly desirable for their application in photocatalysis.This manuscript reported a very easy one-pot synthesis of highly crystalline M-COFs under ambient conditions and further demonstrated their application for syngas production from CO2.The authors clearly characterized the structures of M-COFs with XRD, TEM, and EXAFS.Meanwhile, the authors demonstrated the generality of the methods for synthesizing other M-COFs.In photocatalysis, this M-COF disclosed a high syngas production rate.In overall, this work represents the new development in the synthesis and application of M-COFs.Considering the high novelty and significance, this Reviewer recommends the acceptance of this manuscript for publication in Nature Communications.There are some issues that should be addressed before publication.Response: We thank the reviewer for this valuable comment.The key difference between MOFs and M-COFs is that metal atoms act as the indispensable nodes in MOFs, whereas metal atoms are dispensable in M-COFs.In this work, COFs are generated instead of MOFs based on the following two pieces of evidence.First, we synthesized the COFs without metal salts (N-TAPB-COF-1 and NN-TAPB-COF-1, which were counterparts to Zn-TAPB-COF-1 and ZnZn-TAPB-COF-1, respectively) and obtained crystal products with similar peak positions with their counterparts (Please see Figure R2).This proves that metal atoms are not indispensable nodes in the crystalline products.Second, the peaks around 65 ppm in 13C-NMR spectra of the M-COFs confirm the C-N from ethylenediamine, which act as necessary linker in the Salen-COFs (Please see Figures S13-S16) [R1-R2].
As suggested by the reviewer, we have provided the simulated PXRD patterns of the MOF counterparts to Zn-TAPB-COF-1 and ZnZn-TAPB-COF-1 (Figure R3), demonstrating that the products are M(salen)-COFs.Materials Studio is the software which is used to simulate the structures.We have also added the AB stacking structures for comparison (Please see Figure 2 in the revised Manuscript).The peak positions and relative intensities in experimental line all coincided better with AA stacking than AB stacking.Moreover, the pore size (~3.5 nm) detected by N2 adsorption-desorption isotherms of the M(salen)-COFs were entirely consistent with AA stacking (3.7-3.8 nm) (Please see Figure 2

Comment 2:
The regularity of BET data listed in Table S3 is not good, such as Zn-TAPB-COF-1 836 m 2 g -1 , Zn-TAPB-COF-2 338 m 2 g -1 , Zn-TAPT-COF-1 319 m 2 g -1 and Zn-TAPT-COF-2 1616 m 2 g -1 .Since this paper does not study the influence of synthesis strategy on the morphology and structure of M-COF, and the support of proving that the synthesis is M-COF is insufficient, how about the repeatability of one-step synthesis strategy?It is recommended to supplement supporting data.Response: We thank the reviewer for pointing this out.We have retreated the COFs with solvent soaking and supercritical CO2 (scCO2) drying to remove the impurities in pores and maintain the mesopores of the COFs.The detailed descriptions have been added in the Methods section in the revised manuscript (Please see pages 17-19 in the revised Manuscript).After solvent soaking and scCO2 drying, the regularity of BET data of M(salen)-COFs was greatly improved (Please see Table S5 and Figures S18-S23 in the revised Supporting Information).Moreover, the repeatability of onestep synthesis strategy was also characterized with Zn-TAPB-COF-1 as an example, showing similar crystallinity, BET surface area and pore volume between two batches of samples (Please see Figure R4) Comment 3: Please explain why Zn-TAPT-COF-1 has the lowest specific surface area but the best catalytic performance in Table S3.Response: As we can see from Figure 4b-f, Co-TAPT-COF-1 shows the best catalytic performance.In CO2RR, the formation of *COOH intermediate from the hydrogenation of CO2 is the rate- ) Pore diameter (nm) 3.5 nm c determining step.Firstly, the excellent performance of Co-TAPT-COF-1 can be mainly attributed to the lowest reaction energy barrier of CO2RR on Salen-Co site (0.75 eV) (Please see Table S10 in the revised Supporting Information).Secondly, the Co-TAPT-COF-1 with nanotube morphology and higher crystallinity possesses more efficient electron transfer in comparison with Co-TAPB-COF-1 and Co-TAPT-COF-2 (Please see Figure 4i in the revised Manuscript and Figure S48 in the revised Supporting Information).Thirdly, the nanotube morphology of Co-TAPT-COF-1 endows it higher CO2 uptake capacity than Co-TAPT-COF-2 (Please see Figure S49 in the revised Supporting Information).All in all, the active sites, efficient electron transfer and high CO2 uptake capacity jointly bring the best catalytic performance of Co-TAPT-COF-1.The influence of specific surface area on the catalytic performance is of secondary importance (Please see Table S5 in the revised Supporting Information).
Comment 4: Why does the synthesis route greatly affect the photocatalytic activity?
Response: The synthesis routes greatly affected the crystallinity and morphology of the M(salen)-COFs (Please see pages 6-11 in the revised Manuscript, Figure S7 and Figures S26-S33 in the revised Supporting Information).The M(salen)-COFs-1 obtained in one-step synthesis route showed higher crystallinities and larger-scale morphology than M(salen)-COFs-2 obtained from two-step synthesis.Steady-state photoluminescence spectra and nanosecond transient absorption spectra all indicated that M(salen)-COFs-1 with higher crystallinities exhibited longer promoted separation of electron-hole pairs, faster electron transfer and longer excited state lifetime.The above promoted properties greatly benefited photocatalytic activity (Please see pages 15-16 in the revised Manuscript).Moreover, the different morphologies of M(salen)-COFs from different synthetic routes also affected their CO2 adsorption capacity and mass transfer efficiency (Please see Figure S49 in the revised Supporting Information).To make it more clear, we have rewritten the descriptions in the manuscript (Please see pages 15-16 in the revised Manuscript).
Comment 5: The model for calculating the free energy of CO2 conversion to CO is too simple.It is just a metal ligand, and does not include the covalent structure in COF.It is recommended to redesign the calculation model to explain where the active center of the catalyst is.
Response: We thank the reviewer for pointing this out.We have redesigned the calculation models with more ligands (Figure R5 or Figure S42 in the revised Supporting Information).Free energy diagrams for CO2 conversion to CO were calculated for different M(salen)-COFs (Please see Figure R6 and Figure S43 in Supporting Information).The calculated results indicated that Co-Salen site in Co-TAPT-COF showed the lowest rate-determining step (RDS) free energy.The RDS free energy on Zn-TAPT-COF and ZnZn-TAPT-COF were all higher than Co-TAPT-COF, confirming that the photocatalytic activities improved with lower RDS free energy (Please see Table S10 in the revised Supporting Information).Comment 7: It is suggested to supplement the circulating experimental data of the catalyst for CO2 reduction to explain the chemical stability of the catalyst.

Response:
We thank the editor for pointing this out.We have carried out the recycling experiments with Co-TAPT-COF-1 (stored for about one year), suggesting that there was no significant decrease in the catalytic performance over time (Please see Figure R7 or Figure S40 in Supporting Information).The stability of the Co-TAPT-COF-1 after photocatalysis was also assessed with PXRD patterns, FT-IR spectra and N2 adsorption-desorption isotherms.The above characterizations proved that the crystallinity and covalent C=N linkage maintained after photocatalysis, while the BET surface area and pore volume decreased slightly (Please see Figure R8 or Figure S41 in Supporting Information).

Reviewer #2:
This manuscript describes the synthesis of COFs for CO2 reduction.One big problem is that the PXRD patterns of COFs are inconsistent with their simulated or proposed structures.Another key issue is the experimentally observed pore size is far different from the proposed structure.The 'onepot' method has been reported previously for a similar coordination system and is not novel.The catalytic activity and selectivity of the COFs are not impressive compared to previously reported other framework catalysts.Eye-catching phrases 'Metallo-COFs' and 'single atom' should be used carefully and avoided; 'metallosalen COFs' is more accurate for this case.I cannot suggest the publication in Nat Commun.
Response: We thank the reviewer very much for the valuable comments.We have learned a lot from your points and this manuscript has been greatly improved based on your valuable comments and suggestions.
We have retreated the obtained M(salen)-COFs with solvent soaking and supercritical CO2 (scCO2) drying to remove the impurities in pores and maintain the mesopores of the M(salen)-COFs.After retreatment, the major peaks corresponding to (100) lattice plane of the M(salen)-COFs arose obviously and then the experimental PXRD patterns agreed well with their simulated AA stacking structures (Please see Figure R9 or Figure 2 in the revised Manuscript).Moreover, the pore sizes are also consistent with AA stacking structures after the retreatment (Please see Figures S5-S6 and Figures S18-S23 in the revised Supporting Information).
In this manuscript, M(salen)-COFs are used for photocatalytic CO2 reduction to syngas for the first time, which will enrich the catalysts types for CO2 photoreduction.The syngas production rate outperforms reported COFs.We are sorry that the title might mislead you and the "one-pot" is overemphasized.We have deleted the "One-step synthesis" in the title and changed the descriptions in the abstract to weaken this point (Please see page 1 in the revised Manuscript).
We have changed the phrases "metallo-COFs" into "metallosalen-COFs (M(salen)-COFs)" (Please see the revised Manuscript and Supporting Information).We have changed the "singleatomic metal site and dual-atomic metal sites" to "mononuclear metal sites and binuclear metal sites", respectively (Please see the revised Manuscript).

Comment 2:
The synthesis routes indicate that in figure 1 are quite confusing.Could you use a clearer layout or arrows to better convey the differences between these two methods?Response: We thank the reviewer for this valuable suggestion.We have redrawn Figure 1 to better convey the differences between these two methods (Please see page 4 in the revised Manuscript).

Comment 3:
In the legends of figure 2d-f, the black curve should be "Pawley-refined" instead of "simulated".Response: We thank the reviewer for pointing this out.We have rearranged the PXRD patterns to make it clear in Figure 2 (Please see Figure R9 or Figure 2 in page 4 of the revised Manuscript).
Comment 4: Could you show the exact peak position of the PXRD patterns or provide cif files of the simulated structures?The 2theta positions of the major peaks seem to be a bit too high if the COF structures are actually as drawn in figure 2a-c.Response: As suggested by the reviewer, we have retreated the M(salen)-COFs samples with scCO2 and retested the PXRD patterns of them.The exact peak positions of the PXRD patterns have been shown in Figure 2 and the major peaks with smaller 2theta positions corresponding to (100) lattice planes of the COFs arose obviously.The cif files of the simulated structures have been also provided in the revised files.
Comment 5: In principle, the aldehyde groups can directly react with the tridentate amine ligands (TAPB/TAPT), which is a competition reaction of them reacting with ethylenediamine.If so, COF can also be produced.From the characterizations in this manuscript, it cannot rule out the possibility of these two reactions happening at the same time.Response: We thank the reviewer for pointing this out.We have constructed the structure of COFs without ethylenediamine (TFB-TAPB-COF) [R6] .The simulated PXRD pattern of TFB-TAPB-COF is shown in Figure R10, which is totally different from the experimental PXRD of ZnZn-TAPB-1.While the experimental PXRD of ZnZn-TAPB-1 is consistent with the simulated AA stacking PXRD of ZnZn-TAPB-1, proving that ethylenediamine also react with the aldehyde groups.Moreover, the peaks around 65 ppm in 13C-NMR spectra of the M(salen)-COFs confirm the C-N from ethylenediamine, which act as necessary linker in the Salen-COFs (Please see Figures S13-S16) [R1-R2] .Comment 6: From the nitrogen sorption isotherms and pore size distributions, the pores of these M-COFs seem to be highly distributed, indicating the disordered structures.Response: The highly distributed pores originate the residual impurities in the pores.We have retreated the COFs with solvent soaking and supercritical CO2 (scCO2) drying to remove the impurities in pores and maintain the mesopores of the M(salen)-COFs.The detailed descriptions have been added in the Methods section in the revised manuscript (Please see pages 17-19 in the revised Manuscript).After solvent soaking and scCO2 drying, the pore size distributions of the M(salen)-COFs-1 centred around 3.5 nm, agreeing well with the theoretical values in AA stacking structures (Please see Figures S5-S6 and Figures S18-S23 in the revised Supporting Information).
Comment 7: The influence of synthetic methods is frequently emphasized in this manuscript.However, all the contrasts between M-COF-1 to M-COF-2 can be attributed to their crystallinity difference (as shown in Figure S5).Amorphous counterparts should be put into comparisons.Response: As suggested by the reviewer, we synthesized amorphous counterpart to Co-TAPT-COF-1 and used it as photocatalyst for CO2 reduction.The photocatalytic activities, PXRD pattern and FT-IR spectrum of the amorphous counterpart were shown in Figure R11, indicating worse performance than the Co-TAPT-COF-1 and Co-TAPT-COF-2.This confirmed that the crystallinity of M(salen)-COFs significantly affected their performance.Comment 8: What is the stability of these COFs under photocatalytic conditions?Response: We thank the reviewer for pointing this out.We have assessed the stability of the Co-TAPT-COF-1 after photocatalysis with PXRD patterns, FT-IR spectra and N2 adsorption-desorption isotherms.The above characterizations proved that the crystallinity and covalent C=N linkage maintained after photocatalysis, while the BET surface area and pore volume decreased slightly (Please see Figure R8 or Figure S41 in Supporting Information).

Reviewer #3:
This manuscript described one-step synthesis of three M-COFs, which were further used as photocatalysts.Characterizations have been made to analyze the M-COFs and to evaluate their performance for CO2 photoreduction.The topic is interesting, but the authors failed to demonstrate enough novelty and impact which can demand the publication in Nature Communications.The monomers and reactions are well-known.The crystallinities of the products do not reach the state of the art in this field.In addition to some flaws and claims without clear support, there are critical concerns about the modelled structures of the M-COFs, i.e., the layer stacking.Therefore, I cannot recommend its publication in Nature Communications.The detailed comments and questions are listed as following.Response: We thank the reviewer very much for the valuable comments on our work.The crystallinities of the products have been significantly improved after solvent soaking and scCO2 drying.The major peaks corresponding to (100) lattice plane of the COFs have arisen obviously (Please see Figure R9 or Figure 2 in the revised Manuscript).For the modelled structures, we have provided the simulated AA stacking and AB stacking PXRD patterns of the COFs (Please see Figure R9 or Figure 2 in the revised Manuscript).The peak positions and relative intensities of experimental line all coincided better with AA stacking than ABstacking.Moreover, the pore size detected by N2 adsorption-desorption isotherms of the M(salen)-COFs were entirely consistent with AA stacking (Please see Figures S5-S6 and Figures S18-S23).We have made detailed modifications to the manuscript and responded to each question according to your comments.The specific contents are as follows: Comment 1: In the abstract, the authors identified low crystallinities of COFs are the bottleneck of the field.While the statement is true, the high crystallinity of the COFs claimed by the authors in this study is misleading.The COFs only show less than ten peaks in the PXRD pattern, which is far from high crystallinity in the state of the art of M-COFs.See e.g., Nat. Commun. 2020, 11, 1434.Response: We thank the reviewer for pointing this out.We have deleted "high crystallinity" in the revised Manuscript.Related reference (Nat.Commun. 2020Commun. , 11, 1434) ) has been cited in the revised Manuscript (Please see page 4 in the revised Manuscript).

Comment 2:
The unit cell parameters and space groups in Figure 2 are inconsistent to those in Tables S1 and S2 and in the text.Response: We thank the reviewer for pointing this out.We have modified the cell parameters and space groups in Figure 2 and Tables S2-S4 (Please see Figure 2 in the revised Manuscript and Tables S2-S4 in the revised Supporting Information).

Comment 3:
The Pawley fitting shown in Figure 2 have serious issues.(i) There is no space group of P62m.(ii) For Figure 1d, there should be more Bragg positions before 5 degree for the space group P-62m and the given unit cell.(iii) Why the c-axis of ZnZn-TAPB-COF-1 is 19.99 Å while the structure model looks having the same layer stacking as the others?(iv) Why there is no peak at the first several Bragg positions?Response: We thank the reviewer very much for pointing these issues out.The manuscript has been improved greatly based on your comments.(i) We have corrected the space group (Please see Figure 2 in the revised Manuscript and Tables S2-S4 in the revised Supporting Information).(ii) We have retreated the COFs with solvent soaking and supercritical CO2 (scCO2) drying to remove the impurities in pores and maintain the mesopores of the COFs.After retreatment, the major peaks before 5 degree corresponding to (100) and ( 200) lattice planes of the COFs arose obviously (Please see Figure R9 or Figure 2 in the revised Manuscript).(iii) The unit cell parameters of ZnZn-TAPB-COF-1 have been modified (Please see Figure R9 or Figure 2 in the revised Manuscript and Table S3 in the revised Supporting Information).(iv) We have retreated the COFs with solvent soaking and scCO2 drying to remove the impurities in pores and maintain the mesopores of the COFs.After retreatment, the major peaks before 5 degree corresponding to (100) and ( 200) lattice planes of the COFs arose obviously (Please see Figure R9 or Figure 2 in the revised Manuscript).

Comment 4:
In modelling the structures of the COFs, the authors only considered the AA eclipsed stacking, which could cause misleading.How about other common layer stacking behaviors, such as AA inclined and AB staggered stacking?Response: We thank the reviewer for this meaningful comment.We have added the simulated AA inclined and AB stacking structures for comparison (Please see Figure R12 and Figure 2   Response: We have retreated the COFs with solvent soaking and supercritical CO2 (scCO2) drying to remove the impurities in pores and maintain the mesopores of the COFs.After retreatment, the pore sizes distributions were more centralized.The M(salen)-COFs-1 synthesized in one-step route showed high crystallinities and their pore sizes all centred around 3.5 nm, agreeing well with the theoretical pore sizes in AA stacking structures (Please see Figure 2 in the revised Manuscript, Figures  in the revised Supporting Information).For M(salen)-COFs-2 synthesized in two-step route, their crystallinities were much weaker, indicating partial pore collapse.Hence the pore sizes of the M(salen)-COFs-2 centred between 1.1 nm and 1.4 nm due to partial pore collapse.
Comment 7: The authors mentioned the yield of ZnZn-TATP-COF-1 on line 130, and conclude the one step synthesis as having high yield.This is misleading.The authors need to show the yield of other one step made COFs and compare the yields to two step synthesis.Response: As suggested by the reviewer, we have added yield of the COFs in the Methods section in the manuscript (Please see pages 17-19 in the revised Manuscript).The yields of the COFs are also listed for comparison (Please see Table R1 or Table S1 in the revised Supporting Information).Comment 8: On line 150, the authors claimed that one step synthesis show higher BET surface area than those synthesized by two steps.This is not true.Table S3 shows that half are the opposite.
Response: We thank the reviewer for pointing this out.We have deleted the description "the M(salen)-COFs synthesized by one-step method show higher SBET than those synthesized by twostep methods in general" (Please see page 7 in the revised Manuscript).
Comment 9: The Zn-O bonding distance should be in the range of 1.9-2.1 Å for standard zinc oxides.The EXAFS results shown in Figure 3 show a distance of c.a. 1.6 Å, which has a significant deviation.

Response:
The peak positions in FT-EXAFS spectra are normally 0.3-0.5 Å smaller than the true bonding distances, as the data are given without phase correction [R7-R9] .This deviation results from the phase shift (δj(K)) in the formula below, which is used in the fourier transform.To make it clear, we have added "without phase correction" in the revised manuscript (Please see page 9 in the revised Manuscript).
() = ∑    0 2   () −2  /()  −2 2   I have carefully reviewed the revised version of the manuscript and Point-by-point Response to the Comments.Basically met the revision requirements and agreed to be published.
Reviewer #2 (Remarks to the Author): The author did revise certain minor issues in this manuscript.However, some major problems are still not addressed.The novelty of this method and the catalytic performance are still not impressive enough.Therefore, I cannot suggest publication in Nat Commun.
Reviewer #3 (Remarks to the Author): I appreciate the authors' efforts on revising the manuscript, which has been improved.However, concerns still remain on the proposed structural models of the COFs.2. The authors need to describe in the manuscript how it was found that "The peak positions and relative intensities of experimental line all coincided better with AA stacking than AA inclined".Due to the low number of peaks in the experimental pattern, it is difficult to distinguish them.
3. Only 1D fringes were observed by TEM imaging in Figure 2.This is not convincing to support the proposed eclipsed AA-stacking models, which have honeycomb structures.The 1D fringes actually agree better with AA inclined models or AA serrated models, which do not have 6-fold or 3-fold symmetry.I suggest the authors to investigate the details of the TEM images, as well as their FFT, which can provide important structural information.In addition, the author need to mention from which direction the images are recorded.4. As 001 reflection is not obvious from PXRD pattern, how did the authors determine the inter-layer distance (c-axis) of the models?Why does Zn-TAPB-COF-1 have a much longer c-axis than the others? 5.The authors need to report corrected EXAFS results because they can give indication not only about Zn-O distance, but also Zn-Zn distance, which is related to the inter-layer distance of the COFs (see my comment above).
Reviewer #4 (Remarks to the Author): In this version, the authors have properly responded the comments from the reviewers and revised the manuscript well.This reviewer suggests the acceptance of this manuscript for publication now.
Reviewer #5 (Remarks to the Author): The design, synthesis and structural characterization of this work are interesting.The X-ray absorption spectroscopy results are convincing.However, the photophysical experiments and the subsequent discussions raise concerns.Using Ru(bpy)3 in photocatalytic experiments amplifies these issues further.1) Firstly, no experimental details of photoluminescence and nanosecond transient absorption measurements were provided.
2) Furthermore, the results derived from the steady-state photoluminescence and the nanosecond absorption spectroscopy do not corroborate the proposed photocatalytic performance.First of all, the authors did not report what the photoluminescence is originated from.Is it due to COFs or the metallic segment of the structure?Which unit is the emission quencher?
3) There is significant issue concerning the nanosecond transient absorption spectra.The authors only displayed the kinetics at a wavelength of 500 nm.Without clarification about what this 500 nm signal denotes, the use of the lifetime to substantiate the charge separation efficiency can be misleading.This issue is further complicated when Ru(bpy)3 is employed as a photosensitizer.This is because, in this context, both the lifetime of Ru(bpy)3 and the charge transfer process from Ru(bpy)3 to M-COF become crucial factors influencing the photocatalytic efficiency.In another word, the excited state dynamics from COF may not be important.
symmetry were more plausible modes for the synthesized M(salen)-COFs.We have

Figure R1 .Comment 1 :
Figure R1.PXRD pattern of Ni-TAPB-COF-1.Comment 1: Whether the three metal complexes in Fig. 1b can also generate MOFs.Fig. 2 only shows the structural model (a-c), experiment (red) and Pawley refined (black) PXRD pattern (d-f) of Zn-TAPB-COF-1, ZnZn-TAPB-COF-1 and Co-TAPB-COF-1.It is suggested to provide PXRD diagrams of MOFs for comparison, indicating that M-COFs are generated.At the same time, explain what software is used to simulate the parallel AA-stacking structure, supplement the misplaced ABstacking structure for comparison, and indicate that the AA stack is not the AB stack.It is recommended that the experimental line and analog line in the PXRD diagram should be separated and not overlapped so that the black line cannot be seen clearly.Response: We thank the reviewer for this valuable comment.The key difference between MOFs and M-COFs is that metal atoms act as the indispensable nodes in MOFs, whereas metal atoms are dispensable in M-COFs.In this work, COFs are generated instead of MOFs based on the following two pieces of evidence.First, we synthesized the COFs without metal salts (N-TAPB-COF-1 and NN-TAPB-COF-1, which were counterparts to Zn-TAPB-COF-1 and ZnZn-TAPB-COF-1, respectively) and obtained crystal products with similar peak positions with their counterparts (Please see FigureR2).This proves that metal atoms are not indispensable nodes in the crystalline in the revised Manuscript, Figures S5-S6 and Figures S18-S23 in the revised Supporting Information).In addition, the experimental line and analog line in the PXRD diagram have been separated.

Figure R6 (
Figure R6 (Figure S43 in Supporting Information).Calculated free energy diagrams for CO2 reduction to CO on Zn-TAPT-COF, ZnZn-TAPT-COF and Co-TAPT-COF.Comment 6: The three catalysts in Fig. S3a have good absorption between 400-600 nm.Why do you add photosensitizer [Ru (bpy) 3] Cl2 • 6H2O.Response: The [Ru(bpy)3]Cl2•6H2O which can offer a large absorption coefficient for visible light functions as the photosensitizer in this work.In photocatalytic CO2 reduction, the photosensitizer can harvest light and transfer the photoinduced electrons to the photocatalysts[R3-R5] .As the M(salen)-COFs are heterogeneous catalysts and the light intensity decreases with distance from the solution surface, the light absorption by M(salen)-COFs catalysts are limited.So we add homogeneous photosensitizer [Ru(bpy)3] Cl2•6H2O to enhance the visible light harvest.

Figure R7 (
Figure R7 (Figure S40 in Supporting Information).Stability test of Co-TAPT-COF-1 (stored for about one year) in five recycles for the photocatalytic CO2 reduction reactions.

Figure R9 .Comment 1 :
Figure R9.(a-c) 3D AA stacking structures, (d-f) TEM images and (g-i) experimental and simulated PXRD patterns for AA stacking and AB stacking modes (inset: simulated structure) of Zn-TAPB-COF-1, ZnZn-TAPB-COF-1 and Co-TAPT-COF-1, respectively.Comment 1: In figure 1c, X should be CH or N. Response: We thank the reviewer for pointing this out.We have changed the description in Figure 1 (Please see page 4 in the revised Manuscript).

Figure R11 .
Figure R11.Time-dependent production rate of (a) CO and (a) H2 with 10 mg amorphous counterpart to Co-TAPT-COF-1 as catalyst.PXRD pattern and FT-IR spectrum of the amorphous counterpart to Co-TAPT-COF-1.
in the revised Manuscript).The peak positions and relative intensities of experimental line all coincided better with AA stacking than AA inclined or AB stacking.Moreover, the pore size of M-COF-1 detected by N2 adsorption-desorption isotherms of the M(salen)-COFs were entirely consistent with AA stacking (Please see Figures S5-S6 and Figures S18-S23).

Figure R12 .
Figure R12.Simulated AA stacking, AA inclined, AB stacking PXRD patterns and experimental PXRD patterns for (a) Zn-TAPB-COF-1, (b) ZnZn-TAPB-COF-1, (c) Co-TAPT-COF-1.Comment 5: To justify the structural models, I suggest to include a figure comparing experimental PXRD patterns with those calculated from the structural models.Response: As suggested by the reviewer, we have added the simulated PXRD patterns in Figure 2 in the revised Manuscript.Comment 6: How large are the pores measured from the structural models and how good do they agree with the N2 sorption results?The pore size distributions in Figures S16-21 shows two pore sizes.One of c.a. 1.5 nm, and another of c.a. 3.0 nm.The authors need to describe where they come from.Response: We have retreated the COFs with solvent soaking and supercritical CO2 (scCO2) drying to remove the impurities in pores and maintain the mesopores of the COFs.After retreatment, the pore sizes distributions were more centralized.The M(salen)-COFs-1 synthesized in one-step route showed high crystallinities and their pore sizes all centred around 3.5 nm, agreeing well with the theoretical pore sizes in AA stacking structures (Please see Figure2in the revised Manuscript, Figures in the revised Supporting Information).For M(salen)-COFs-2 synthesized in two-step route, their crystallinities were much weaker, indicating partial pore collapse.Hence the pore sizes of the M(salen)-COFs-2 centred between 1.1 nm and 1.4 nm due to partial pore collapse.

103
in the revised Manuscript and Figures S5-S6 in the revised Supporting Information).104Detaileddescriptions have been improved as "Simulated eclipsed AA-stacking, serrated
While the AA eclipsed model can be distinguished from AB staggered model by PXRD and N2 sorption results, the authors failed to describe why the AA eclipsed model can be distinguished from AA inclined model and AA serrated model.The detailed comments and questions are listed as following.1.The authors only presented the discussion of AA eclipsed model and AB staggered model in the manuscript.As different stacking behaviors can greatly affect the properties of 2D-COFs (see ACSAppl.Nano Mater.2022, 5, 10, 14377), the authors need to describe how the AA eclipsed model can be distinguished from other stacking models, i.e., AA inclined and AA serrated models in the main text.