Intermolecular cascaded π-conjugation channels for electron delivery powering CO2 photoreduction

Photoreduction of CO2 to fuels offers a promising strategy for managing the global carbon balance using renewable solar energy. But the decisive process of oriented photogenerated electron delivery presents a considerable challenge. Here, we report the construction of intermolecular cascaded π-conjugation channels for powering CO2 photoreduction by modifying both intramolecular and intermolecular conjugation of conjugated polymers (CPs). This coordination of dual conjugation is firstly proved by theoretical calculations and transient spectroscopies, showcasing alkynyl-removed CPs blocking the delocalization of electrons and in turn delivering the localized electrons through the intermolecular cascaded channels to active sites. Therefore, the optimized CPs (N-CP-D) exhibiting CO evolution activity of 2247 μmol g−1 h−1 and revealing a remarkable enhancement of 138-times compared to unmodified CPs (N-CP-A).

I am supportive of this work as this is very topical area of research and a range of interesting measurements (such as TRMC and TAS) are used to gain understanding and the observed AQYs are high, which is also important in this field. Having said this there are a couple of flaws which I believe should be addressed before this can be accepted: -I strongly disagree with the premise of 'molecular architecture' or design. I actually think it makes the manuscript weaker as there is a risk that other important factors have not been considered, the data set is simply to small to draw strong conclusions, in particular in photocatalysis where many different processes are taking place. I am also a little unsure about the notion of improving alkynyl materials: As far as I can tell these have not been previously reported and based on what has been reported for hydrogen production from water, I would not have expected them to act as particularly good photocatalysts for CO2 reduction. I would recommend focusing on reporting these materials and dropping the design aspect (including the title), as it is questionable in this field -Transient absorption spectroscopy can be very insightful to understand the dynamics in these materials. However, it would be much more useful to perform the measurement in the presence of scavenger and/or CO2 to understand what is happening under photocatalysis conditions. A few other more minor comments and questions: 1) At first I was very excited about these results only to see later that TEOA was used. This is ok at this stage, but the manuscript needs to state clearer in the main-text that a scavenger was used.
2) Net-like materials are usually referred to as conjugated microporous polymers in the materials community.
3) Comparing rates to other reports makes little sense as this ignores the fact that set-ups are very different and light sources vary (statement: 'highest CO evolution efficiency' and in a few other places). It would be much better to compare AQYs, which are already in the manuscript, to other reports. 4) Particle size and dispersion in the photocatalysis mixture are currently not considered. Some reports indicate that this might be important and should be measured and added. 5) Important experimental details are missing: What was the pressure of the photolysis experiment? This needs to be added to figure captions and also mentioned in the main-text. How much of the Co complex was used? I was a little unsure at times and it needs to be stated clearer. 6) CO2 absorption of a dry powder can be very different compared to a measurement under wet conditions. This has been reported for conjugated microporous polymers and should be measured if it is considered to be important. 7) Selectivity is only briefly discussed in the main-text but omitted from Fig.3. I am a little unsure how important it is right now as the area is in its infancy, but it seems to be standard practise in the field and should be added. 8) Were the TR-PL measurements in Fig.4 performed in the presence of a scavenger?
Reviewer #3 (Remarks to the Author): The manuscript of J.H. Ye et al. reports on the original construction of intermolecular cascaded πconjugation channels for powering CO2 photoreduction by modifying both intramolecular and intermolecular conjugation of conjugated polymers. This study is of great interest and reports a rather novel approach. Detailed experimental analysis and theoretical calculations were conducted to verify the proposed approach and conclusion. This work might interest those in pursuit of better photogenerated electron delivery from the viewpoint of the molecular architecture. Though I inclined to recommend the publication with Nat. Commnun., there are also some important revisions need to be made: 1. Could you please give the loading content of the Co (II) bipyridine complexes cocatalyst on the CPs as this may be critical for the photocatalytic efficiency? 2. As the author showed in Table S4, there are some residual palladium and copper metals detected in the CPs. Please explain if these metals, especially the Pd, will have any activity in the photocatalytic process. 3. In this study, the LUMO and HOMO were obtained by hybrid density-functional-theory-based firstprinciples calculations. The experimental verifications for these values by UPS or Mott-Schottky plot should be conducted. 4. Authors mentioned that "The conductivity transients and calculated charge mobilities for CPs are displayed in Fig. 2a, in which the L-CP-A, with a linear structure and alkynyl group, exhibits charge mobility (μtot) of 0.32 cm2 V-1 s-1 (φ Σμ= 7.4×10-5 V-1 s-1). As expected, the charge mobility of L-CP-D in the absence of alkynyl decreased to a much lower value of 0.15 cm2 V-1 s-1(φ Σμ= 3.4×10-5 V-1 s-1)." (Page 6). However, there is no corresponding results shown in Fig.2a. Please check this carefully. 5. Authors claimed that "Besides, the net-like CPs (N-CP-A and N-CP-D) possess more cocatalyst absorption sites of phenyl in the units, thus resulting in enhanced performance of net-like CPs (N-CP-A and N-CP-D) than those of linear CPs (L-CP-A and L-CP-D)" (Page 9). As Fig.4a shows, the N-CPs exhibited greater CO2 adsorption than the L-CPs. Does the CO2 adsorption have any effects on the CO2 photoreduction efficiency except for the cocatalyst absorption sites number? This may be a result of multi-factors. 6. There is a clerical error in SI (page 46). "As shown in Figure S4" should be "As shown in Figure  S34".
Then, I would highly suggest these minor revisions to be thoroughly addressed before re-submission.

Reviewers' comments:
Reviewer #1 (Remarks to the Author): This manuscript developed the construction of intermolecular cascaded π-conjugation channels by modifying both intramolecular and intermolecular conjugation to optimize the photogenerated electron delivery and improve the performance of CO 2 photoreduction. The highest CO evolution activity of 2247 μmol g -1 h -1 among all reported conjugated polymers has been achieved. These results may be interesting and inspire scientists in the field to further improve the performance of CO 2 photoconversion. However, there are some issues on data analyses about structure and properties. Therefore, revision on this draft is necessary before acceptance for publication.

Response:
The authors thank the reviewer for the valuable comments. These comments are very helpful for improving the quality and value of this article.
1. The discussion of the band structure is too superficial. Please added the detailed investigation in the manuscript, such as Mott-Schottky test and UPS spectra.
Response: Based on previous reports, the energy band structures of these CPs were investigated via cyclic voltammetry measurement, which was regarded as the one of most suitable way to determine the LUMO and HOMO levels of these conjugated polymers (Adv. Mater. 2015, 27, 6265;Angew. Chem. Int. Ed. 2015, 54, 13594;Angew. Chem. Int. Ed. 2016, 55, 9783;Angew. Chem. Int. Ed. 2016, 55, 9202;). Combining with the bandgap energy from diffuse reflectance spectrum (DRS), the energy levels of these CPs were preciously positioned and showed in the Supplementary Table S1. According to the reviewer's suggestion, we also conducted the Mott-Schottky test ( Figure R1) and UPS spectra ( Figure R2). to further explore the band structures of these CPs that are coincidence well with our cyclic voltammetry (CV) results..  Table R1, which showed high accordance with the energy levels determined by cyclic voltammetry measurement.    Table R2, which showed high accordance with the energy levels determined by cyclic voltammetry measurement and Mott-Schottky test.  In summary, we have revised the discussion of the band structure in the Manuscript as follow: "Cyclic voltammetry (CV) measurements were also conducted, the HOMO position can be determined by the irreversibility of the oxidation peaks due to the irreversible oxidation process of the CPs at the impressed voltage ( Supplementary Fig. 10) revealed different energy levels within the CPs (Supplementary Table 1) 39 . In addition, their energy levels were further investigated by the UPS (Ultraviolet Photoelectron Spectroscopy) ( Supplementary Fig. 11) and Mott-Schottky test ( Supplementary Fig. 12), which showed a high accordance with the energy levels determined by CV measurement (Supplementary Table 2)." 2. Page 8, the cyclic-voltammetry method and EPR measurement cannot demonstrate that the electrons on LUMO of CPs can have the ability to produce Co (I) from Co (II), because of solvent effect. The test conditions for EPR measurement should be provided. More convincing evidence for the existence of Co (I) should be provided, e.g. in-situ XPS.

Response:
We thank the reviewer for this comment. We obtained the N-CP-D powder with Co complexes by a natural adsorption process and then place it into the EPR sample tube. We used a degassing device to make sure that tube is oxygen-free and seal the sample tube by sintering the nozzle. After these careful pre-treatments, we tested the EPR spectra of Co before and after light illumination. Through the above mentioned method, we cannot find a peak that can be assigned to the Co (I) because of its low-spin state of the Co (I). Only a weak peak assigned to Co (II) were observed in the EPR measurement. However, based on the previous reports (J. Mol. Catal. A-Chem. 2013, 193, 27;Angew. Chem. Int. Ed. 2016, 55, 14310;ChemSusChem 2019, 12, 4493), the in-situ generated Co (I) are widely recognized as the major active spices for CO 2 reduction.
According to reviewer's suggestion, we also conducted the in-situ XPS measurement to explore the existence of Co (I) ( Figure R3). Figure R3. In-situ XPS spectra of N-CP-D with Co complexes (a) and the corresponding multi peaks separation spectra (b). Figure R3, compared to the N-CP-D without irradiation, the binding energy of Co 2p in N-CP-D with irradiation shifted to a lower binding energy. From the further multi peaks separation process, two new peaks appear Co 2p 3/2 and Co 2p 1/2 that show lower binding energy than that of Co 2+ , clearly revealing the photo-induced generation of Co + in N-CP-D.

As shown in
We have added the following sentences for the EPR measurement and the description of the photo-induced generation of Co + both in Supplementary information and Manuscript.
In Manuscript: "We also demonstrated that photo-excited electrons on the LUMO of CPs in the present work do have the ability to reduce Co (II) bipyridine complexes to Co (I) bipyridine complexes via the cyclic-voltammetry spectrum ( Supplementary Fig. 19). In addition, the weakened EPR signal ( Supplementary Fig. 20, see Methods for experimental details) 22 indicates the weakening of high-spin-state Co (II) upon visible-light irradiation that maybe reduced to low-spin-state Co (I).
In order to get more convincing evidence for the existence of the low-spin-state of Co (I), the in-situ XPS ( Supplementary Fig. 21) were performed over N-CP-D. It showed a lower binding energy than that of Co 2+ , which clearly revealed the existence of photo-induced low-spin-state Co + in this system." In Supplementary information: "EPR measurement. N-CP-D powder with Co complexes was obtained by an adsorption process and then place it into the EPR sample tube. A degassing device was used to make the tube oxygen-free and seal the sample tube by sintering the nozzle. After these pre-treatments, the sample was tested before and after irradiation with an electron spin resonance (JES-FA200, JEOL, Japan) spectrometer." 3. According to supplementary Table S4, residual palladium (0.661 wt.%) retains in CPs matrix.
Whether the residual Pd affect the photocatalytic CO 2 reduction activity? There is no explanation about it in the manuscript.
Response: Indeed, there are trace amounts of residual palladium retains in all CPs. Despite it is inevitable that trace amounts of palladium remained in the CPs, all CPs without adding Co complexes as cocatalyst showed negligible photocatalytic CO 2 reduction activity. After adding cocatalyst, the CP-A series still exhibited low activities in CO 2 reduction, while the CP-D series showed a great enhancement, revealing the enhanced activities are attributed to the intermodular electron transfer between CP-D series and Co complexes.
We have added the following sentences for the explanation that the residual Pd has negligible effect on the photocatalytic CO 2 reduction activity in Manuscript.
"All CPs showed quite low activity in the absence of cocatalyst, as well as the CP-D series in the presence of isolated cobalt chloride or dipyridyl ( Supplementary Fig. 22), implying that the residual Pd haa negligible effect on the photocatalytic CO 2 reduction activity (Supplementary Table 5)." 4. In Fig. 4a, CO 2 absorption-desorption isotherms curves were employed to demonstrate that the capacity for CO 2 adsorption was not the determining factor for the CO 2 photoreduction. However, the chemisorption of CO 2 is a more important factor in CO 2 photoreduction. The chemisorption of CO 2 , e.g. CO 2 -FTIR, should be added.

Response:
We thank the reviewer for this comment. The CO 2 absorption of a dry powder is very different compared to measurement under wet conditions based on the previous report (J. Am. Chem. Soc. 2012, 134, 10741). The chemisorption of CO 2 is a more important factor than CO 2 adsorption in CO 2 photoreduction. We therefore employed the in-situ infrared technology to investigate the adsorption and chemisorption of CO 2 over CPs in a solvent-containing environment ( Figure R4). Figure R4. In-situ FT-IR spectra of N-CP-D with Co complexes for the adsorption of CO 2 (a) and the chemisorption of CO 2 (b) in a solvent-containing environment.
As shown in Figure R4, although the CP-A series exhibited greater CO 2 adsorption than the CPs-D series, the L-CP-D and N-CP-D showed an enhanced intensity both in the region of CO 2 adsorption (a) and the region of CO 2 chemisorption (b) than L-CP-A and N-CP-A. It suggests that the CO 2 adsorption capacity of CPs-D series is stronger than that of CPs-A series under the solvent-containing environment, which provides favorable conditions for the subsequent CO 2 reduction reaction.
We have revised the following sentence in Manuscript for the explanation of the CO 2 adsorption.
"Nevertheless, the CO 2 absorption of a dry powder is very different from the wet conditions 50 , the in-situ FT-IR indicates L-CP-D and N-CP-D exhibit enhancement both in CO 2 adsorption and CO 2 chemisorption than L-CP-A and N-CP-A ( Supplementary Fig. 30). It means that the CO 2 adsorption capacity of CPs-D series is stronger than that of CPs-A series under the solvent-containing environment, which provides favorable conditions for the subsequent CO 2 reduction reaction."  Table S6, the author tried to highlight the fact that the photoexcited electrons can be fast delivered to CO 2 from catalyst. Nevertheless, the short lifetime (τ1) was also shortened in CO 2 atmosphere. The average lifetime and experimental details of TRPL should be provided.

Response:
We thank the reviewer for this comment. We have added the average lifetime in Table   R3 and the experimental details of TR-PL in the Supplementary Information.
"TR-PL measurement. Photoluminescence spectra decay curves were obtained by using a Hamamatsu instrument (Hamamatsu, Japan) with a 1kHz Ti: sapphire regenerative amplifier (Solstice, Spectra-Physics) was used as an excitation light source of an optical parametric amplifier (OPA) (TOPAS prime, NIR-UV-Vis. LIGHT CONVERSION Inc.). The CPs were dispersed in the solvent of acetonitrile and water solution with scavenger of triethanolamine. The argon was firstly pumped into a sealed cell for 40 min by using a pipe inlet to exclude the dissolved oxygen and other atmosphere. After the above pre-treatment process, the TR-PL signals were recorded and these samples are named L-CP-A in Ar, L-CP-D in Ar, N-CP-A in Ar and N-CP-D in Ar, respectively. As a comparison, the CO 2 was subsequently pumped into this sealed cell for 40 min by using a pipe inlet. The TR-PL signals were also recorded and these samples are named L-CP-A in CO 2 , L-CP-D in CO 2 , N-CP-A in CO 2 and N-CP-D in CO 2 , respectively." As shown in Table R3, it was indeed mentioned by the reviewer, the short lifetime (τ1) was also shortened in a CO 2 atmosphere. It can be mainly attributed to a large number of photogenerated electrons which can be rapidly delivered to CO 2 via the cocatalyst and only a small number of electrons with a short lifetime involved in the detectable recombination process of TR-PL under CO 2 atmosphere.
We can find that the average lifetime (τ av ) of L-CP-D and N-CP-D also showed an obvious decrease in CO 2 atmosphere than that in an argon atmosphere, while the average lifetimes of L-CP-A and N-CP-A exhibited no obvious change either in CO 2 or in argon. We thus attribute these differences to the fact that electrons can be fast delivered to CO 2 via the cocatalyst in L-CP-D and N-CP-D. Table R3. Parameters of the time-resolved photoluminescence decay curves according to a biexponential decay in different atmosphere.
We also revised the description of TR-PL in the Manuscript as follows.
"For both N-CP-D (Fig. 4c) and L-CP-D (inset of Fig. 4c), the average lifetime of electrons (Supplementary Table 8) in a CO 2 atmosphere was significantly shortened compared to that in an argon atmosphere, which can be mainly attributed to the large amount of photogenerated electrons can be fast delivered to CO 2 via the cocatalyst and only a small number of electrons with short lifetime involved into the detectable recombination process under the CO 2 atmosphere ( Supplementary Fig. 34), which can be further confirmed by the TA spectra in different atmosphere ( Supplementary Fig. 35)." 6. The decomposition of the CP should be considered on the conversion of CO 2 to CO.

Response:
We thank the reviewer for this comment. To avoid possible decomposition of CPs by the photo-induced thermal effects to CO, we kept the system temperature during the reaction from being too high through a circulating cooling system. By using this condition, the CPs form decomposition by the photo-induced thermal effects can be effectively avoided as indicated by the isotopic experiment.
From the isotopic experiment, the ratio of generated 13 CO to 12 CO was found to be also the same as the ratio of 13 CO 2 to 12 CO 2 in the source gas, indicating that almost all produced 13 CO indeed originated from 13 CO 2 in the CO 2 photoreduction over the CPs while not the decomposition product. In addition, the CO 2 reduction experiment of CPs without adding Co complexes shows negligible activity in CO generation, which indicates the decomposition over CPs during the CO 2 reduction is ngeligible.
7. The photocatalytic stability should be carried out and the mechanism of the apparent degradation of the photocatalytic activity should be discussed.

Response:
We have carried out the photocatalytic stability showed in Figure 3c, from which there is only a slight decrease in the fifth cycle and the photocatalytic activity remains to be 91% of the original cycle. We attribute this slight decrease to the fresh Co complexes which are not added to the system in the photocatalytic stability measurement.
To better clarify this point, we also added the discussion about the slight decrease of the photocatalytic activity in the Manuscript as follows.
"Moreover, the stability test over L-CP-D and N-CP-D indicates that after 5 cycles, the CO evolution was still high (maintaining 91% to the original cycle without adding fresh Co complexes to the system) as compared to the initial values (Fig. 3c), suggesting the adequate stability of these CPs ( Supplementary Fig. 25)." Reviewer #2 (Remarks to the Author): I am supportive of this work as this is very topical area of research and a range of interesting measurements (such as TRMC and TAS) are used to gain understanding and the observed AQYs are high, which is also important in this field.

Response:
The authors thank the reviewer for the valuable comments. These comments are very helpful for improving the quality and value of this article.
Having said this there are a couple of flaws which I believe should be addressed before this can be accepted: -I strongly disagree with the premise of 'molecular architecture' or design. I actually think it makes the manuscript weaker as there is a risk that other important factors have not been considered, the data set is simply too small to draw strong conclusions, in particular in photocatalysis where many different processes are taking place. I am also a little unsure about the notion of improving alkynyl materials: As far as I can tell these have not been previously reported and based on what has been reported for hydrogen production from water, I would not have expected them to act as particularly good photocatalysts for CO 2 reduction. I would recommend focusing on reporting these materials and dropping the design aspect (including the title), as it is questionable in this field.
Response: Thank you very much for your helpful suggestion. According to your insightful suggestion, we have revised the title into "Intermolecular Cascaded π-Conjugation Channels for Electron Delivery Powering CO 2 Photoreduction", which weakened the design aspect in the title.
Besides, we agree that photocatalysis is a very complicated reaction where many different processes are taking place. So, we do our best to weaken the concept of molecular architecture in the Manuscript and highlight that intermolecular electron delivery is one of the critical factors for CO 2 reduction, which has never been reported in previous research.
About the notion of improving alkynyl materials, the alkynyl materials are reported as the candidates for hydrogen production from water better than materials without alkynyl (J. Catal. 2017, 350, 64). But, the materials without alkynyl showed significant enhancement in CO 2 reduction, which acts as good photocatalysts for CO 2 reduction. We attribute this difference in two similar photocatalytic reactions to the different kinds of cocatalyst. In hydrogen production reaction, the Pt loaded on the surface of materials as the cocatalyst and the photogenerated electrons need to be transferred to cocatalyst across the materials. However, the Co complexes act as the cocatalyst in the CO 2 reduction reaction, which is the independent molecule to the materials.
The absence of alkynyl in the material concentrates the photogenerated electron to promote the intermolecular electron delivery.
-Transient absorption spectroscopy can be very insightful to understand the dynamics in these materials. However, it would be much more useful to perform the measurement in the presence of scavenger and/or CO 2 to understand what is happening under photocatalysis conditions.

Response:
We thank the reviewer for this comment. We have conducted the transient absorption spectroscopy over N-CP-D in a solvent-containing environment in the presence of scavenger. We have compared the different dynamics between the atmosphere of CO 2 and argon in Figure R5. As shown in Figure R5, in first 40 ps there is no difference between the kinetics in CO 2 and argon because the main step in this relatively short temporal domain is the charge carrier generation and trapping (Chem. Rev. 1995, 95, 69). After these steps, the trapped electrons were transferred to the absorbed target molecules in the temporal domain of 50-500 ps. Compare to the inert gas molecule of argon, the CO 2 molecule could accept the photogenerated electron thus showing a decrease in the temporal domain of 50-500 ps. This phenomenon is consistent with the results shown by TR-PL (Figure 4c).
We have revised the description in the Manuscript as follows.
"For both N-CP-D (Fig. 4c) and L-CP-D (inset of Fig. 4c), the average lifetime of electrons (Supplementary Table 8) in a CO 2 atmosphere was significantly shortened compared to that in an argon atmosphere, which can be mainly attributed to the large amount of photogenerated electrons can be fast delivered to CO 2 via the cocatalyst and only a small number of electrons with short lifetime involved into the detectable recombination process under the CO 2 atmosphere ( Supplementary Fig. 34), which can be further confirmed by the TA spectra in different atmosphere ( Supplementary Fig. 35)." A few other more minor comments and questions: 1) At first, I was very excited about these results only to see later that TEOA was used. This is ok at this stage, but the manuscript needs to state clearer in the main-text that a scavenger was used.

Response:
We have added the description of TEOA as the scavenger in the main text as follows.
"To study if the electron delivery from the CPs to cocatalyst has critical effects on CO 2 photoreduction properties, the evaluation of CO 2 photoreduction activities (see Methods for experimental details) were carried out in a closed gas circulation system by using CPs as the catalyst and 5 μmol Co (II) bipyridine complexes as cocatalyst. The acetonitrile/water (7:3) mixture with triethanolamine (TEOA) as sacrificial agent was also added." 2) Net-like materials are usually referred to as conjugated microporous polymers in the materials community.

Response:
We have added the description of the Net-like materials which are usually referred to as conjugated microporous polymers when we first mentioned Net-like materials in this revision.
"To validate the above strategy, four goal-oriented materials including linear and net-like (Net-like materials are usually referred to as conjugated microporous polymers) CPs with simple structure 24,25 , but different π-conjugations were built by using Suzuki-Miyaura coupling instead of  Chem. Soc. Rev. 42, 8012-8031 (2013).
3) Comparing rates to other reports makes little sense as this ignores the fact that set-ups are very different and light sources vary (statement: 'highest CO evolution efficiency' and in a few other places). It would be much better to compare AQYs, which are already in the manuscript, to other reports.

Response:
We thank the reviewer for this comment. We have added the comparison of AQYs with other similar CO 2 reduction systems in Table R4. We also mentioned in the Manuscript as follows.
"With the addition of appropriate cocatalyst (Supplementary Table 6), the CO selectivities of L-CP-D and N-CP-D was measured to be 86% and 82% (Supplementary Fig. 23) and achieved an apparent quantum yield (AQY) as high as 3.39 % and 1.23 % at 400 nm, respectively ( Supplementary Fig. 24), which is considered higher than that in most reports until now (Supplementary Table 4)." 4) Particle size and dispersion in the photocatalysis mixture are currently not considered. Some reports indicate that this might be important and should be measured and added.

Response:
We thank the reviewer for this comment. From the TEM and SEM images, the particle sizes are different, and the net-like CPs are larger than the linear CPs. However, as shown in Figure R6, the dispersion of them in the solvent of acetonitrile/water (7:3) mixture is similar without adding Co complexes. After adding Co complexes, the CPs-A series still maintain the dispersion, while the CPs-D series stack together due to the intermolecular π-π stacking with the Co complexes. Although there are differences in dispersion among CPs, there are still good contacts between the materials and the solvent in the stirring condition. We have added this information to the Supplementary Information. Figure R6. The dispersion of CPs in in the solvent of acetonitrile/water (7:3) mixture with or without adding Co complexes at different times. 5) Important experimental details are missing: What was the pressure of the photolysis experiment?
This needs to be added to figure captions and also mentioned in the main-text. How much of the Co complex was used? I was a little unsure at times and it needs to be stated clearer.

Response:
We thank the reviewer for this comment. The pressure of the photocatalytic experiments is about 80 kPa, and we have compared the different amounts (0, 1, 2, 5, 10 μmol) of the Co (II) bipyridine complexes as cocatalyst in Figure S20 of Supplementary  "To study if the electron delivery from the CPs to cocatalyst has critical effects on CO 2 photoreduction properties, the evaluation of CO 2 photoreduction activities (see Methods for experimental details) were carried out in a closed gas circulation system by using CPs as the catalyst and 5 μmol Co (II) bipyridine complexes as cocatalyst. The acetonitrile/water (7:3) mixture with triethanolamine (TEOA) as sacrificial agent was also added." 6) CO 2 absorption of a dry powder can be very different compared to a measurement under wet conditions. This has been reported for conjugated microporous polymers and should be measured if it is considered to be important.

Response:
We thank the reviewer for this comment. The CO 2 absorption of a dry powder is very different compared to measurement under wet conditions based on the previous report (J. Am. Chem. Soc. 2012, 134, 10741). The chemisorption of CO 2 is a more important factor than CO 2 adsorption in CO 2 photoreduction. We therefore employed the in-situ infrared technology to investigate the adsorption and chemisorption of CO 2 over CPs in a solvent-containing environment ( Figure R4). Figure R4. In-situ FT-IR spectra of N-CP-D with Co complexes for the adsorption of CO 2 (a) and the chemisorption of CO 2 (b) in a solvent-containing environment.
As shown in Figure R4, although the CP-A series exhibited greater CO 2 adsorption than the CPs-D series, the L-CP-D and N-CP-D showed an enhanced intensity both in the region of CO 2 adsorption (a) and the region of CO 2 chemisorption (b) than L-CP-A and N-CP-A. It suggests that the CO 2 adsorption capacity of CPs-D series is stronger than that of CPs-A series under the solvent-containing environment, which provides favorable conditions for the subsequent CO 2 reduction reaction.
We have revised the following sentence in Manuscript for the explanation of the CO 2 adsorption.
"Nevertheless, the CO 2 absorption of a dry powder is very different from the wet conditions 50 , the in-situ FT-IR indicates L-CP-D and N-CP-D exhibit enhancement both in CO 2 adsorption and CO 2 chemisorption than L-CP-A and N-CP-A ( Supplementary Fig. 30). It means that the CO 2 adsorption capacity of CPs-D series is stronger than that of CPs-A series under the solvent-containing environment, which provides favorable conditions for the subsequent CO 2 reduction reaction." 7) Selectivity is only briefly discussed in the main-text but omitted from Fig.3. I am a little unsure how important it is right now as the area is in its infancy, but it seems to be standard practise in the field and should be added.

Response:
We have discussed the selectivity in the Supplementary Information based on the Figure S20. Based on the report with a similar system the obtained selectivity from the CPs-D series is not the highest one because the CPs could generate H 2 . However, after adding with Co complexes, the photogenerated electrons were oriented transferred to the cocatalyst and the selectivity of CO increased with the amount of cocatalyst and eventually reached 82% and 86% for N-CP-D and L-CP-D, respectively, which is considered to be above the current level of reports.
In addition, we also added the comparison of the selectivity of similar system for CO 2 reduction in Table S4 and revised the description in the Manuscript as follows.
"With the addition of appropriate cocatalyst (Supplementary Table 6), the CO selectivities of L-CP-D and N-CP-D was measured to be 86% and 82% (Supplementary Fig. 23) and achieved an apparent quantum yield (AQY) as high as 3.39 % and 1.23 % at 400 nm, respectively (Supplementary Fig. 24), which is considered higher than that in most reports until now (Supplementary Table 4)." 8) Were the TR-PL measurements in Fig.4 performed in the presence of a scavenger?

Response:
We thank the reviewer for this comment. We have added the experimental details of TR-PL in the Supplementary Information, which also list as follows.
"TR-PL measurement. Photoluminescence spectra decay curves were obtained by using a The authors thank the reviewer for the valuable comments. These comments are very helpful for improving the quality and value of this article.
1. Could you please give the loading content of the Co (II) bipyridine complexes cocatalyst on the CPs as this may be critical for the photocatalytic efficiency?
Response: We thank the reviewer for this comment. We also believe that the content of the Co (II) bipyridine complexes cocatalyst on the CPs is critical for the photocatalytic efficiency. As mentioned in Supplementary information, the production of H 2 decreased with an increasing amount of cocatalyst, while the production of CO gradually increased. The increase in CO production over L-CP-D was not apparent after adding more than 1 μmol of cocatalyst, and this phenomenon appeared in the N-CP-D system after adding more than 5 μmol of cocatalyst. With the addition of 10 μmol of cocatalyst, the CO evolution selectivity of L-CP-D and N-CP-D obtained through our experiment reached 86% and 82%, respectively.
In addition, we also employed ICP-AES measurement to determine the loading content of the Co species on the CPs (Table R5) when adding with 5 μmol of Co (II) bipyridine complexes cocatalyst. The loading content of the Co species on the CPs is due to the difference in the specific surface area. Moreover, the CO 2 photoredcution activity over N-CP-D (11.37 μmol h -1 ) is much higher than that over N-CP-A (0.08 μmol h -1 ). Despite the different loading amount of Co (II) bipyridine complexes on N-CP-D and N-CP-A, the enhancement (~138 folds) of activities over N-CP-D is much higher than the enhanced loading amount (~2.33 folds) of Co complexes. It demonstrates that the main reason of enhanced activities is attribute to the intermolecular conjugate interactions between N-CP-D and Co complexes. We have added the following sentences for the explanation of the loading content of Co species in Supplementary information.
In addition, the ICP-AES measurement was employed to determine the loading amount of the Co species on the CPs (Table S6) when adding with 5 μmol of Co (II) bipyridine complexes cocatalyst. The main reason of improved activities over CPs-D series is attribute to the enhanced intermolecular conjugate interactions.
2. As the author showed in Table S4, there are some residual palladium and copper metals detected in the CPs. Please explain if these metals, especially the Pd, will have any activity in the photocatalytic process.
Response: Indeed, there are trace amounts of residual palladium retains in all CPs. Despite it is inevitable that trace amounts of palladium remained in the CPs, all CPs without adding Co complexes as cocatalyst showed negligible photocatalytic CO 2 reduction activity. After adding cocatalyst, the CP-A series still exhibited low activities in CO 2 reduction, while the CP-D series showed a great enhancement, revealing the enhanced activities are attributed to the intermodular electron transfer between CP-D series and Co complexes.
We have added the following sentences for the explanation that the residual Pd has negligible effect on the photocatalytic CO 2 reduction activity in Manuscript.
All CPs showed quite low activity in the absence of cocatalyst, as well as the CP-D series in the presence of isolated cobalt chloride or dipyridyl ( Supplementary Fig. S22), implying that the residual Pd has negligible effect on the photocatalytic CO 2 reduction activity.
3. In this study, the LUMO and HOMO were obtained by hybrid density-functional-theory-based  Figure R1) and UPS spectra ( Figure R2). to further explore the band structures of these CPs that are coincidence well with our cyclic voltammetry (CV) results.. Figure R1. Mott-Schottky plots for CPs in 0.1 M Na 2 SO 4 at pH=7 by using Ag/AgCl with saturated KCl as reference electrode.
As shown in Figure R1.  Table R1, which showed high accordance with the energy levels determined by cyclic voltammetry measurement.    Table R2, which showed high accordance with the energy levels determined by cyclic voltammetry measurement and Mott-Schottky test.  In summary, we have revised the discussion of the band structure in the Manuscript as follow: "Cyclic voltammetry (CV) measurements were also conducted, the HOMO position can be determined by the irreversibility of the oxidation peaks due to the irreversible oxidation process of the CPs at the impressed voltage ( Supplementary Fig. 10) revealed different energy levels within the CPs (Supplementary Table 1) 39 . In addition, their energy levels were further investigated by the UPS (Ultraviolet Photoelectron Spectroscopy) ( Supplementary Fig. 11) and Mott-Schottky test ( Supplementary Fig. 12), which showed a high accordance with the energy levels determined by CV measurement (Supplementary Table 2)." 4. Authors mentioned that "The conductivity transients and calculated charge mobilities for CPs are displayed in Fig. 2a, in which the L-CP-A, with a linear structure and alkynyl group, exhibits charge mobility (μtot) of 0.32 cm 2 V -1 s -1 (φ Σμ= 7.4×10 -5 V -1 s -1 ). As expected, the charge mobility of L-CP-D in the absence of alkynyl decreased to a much lower value of 0.15 cm 2 V -1 s -1 (φ Σμ= 3.4×10 -5 V -1 s -1 )." (Page 6). However, there is no corresponding results shown in Fig.2a.
Please check this carefully.

Response:
We are sorry for this careless mistake. We have mistaken the graphic symbol of the inset in Figure 2a. Actually, the graphic symbol in the inset in Figure 2a is L-CP-D and L-CP-A.
We have revised it in this revision. shows, the N-CPs exhibited greater CO 2 adsorption than the L-CPs. Does the CO 2 adsorption have any effects on the CO 2 photoreduction efficiency except for the cocatalyst absorption sites number?
This may be a result of multi-factors.

Response:
We thank the reviewer for this comment. The CO 2 absorption of a dry powder is very different compared to measurement under wet conditions based on the previous report (J. Am. Chem. Soc. 2012, 134, 10741). The chemisorption of CO 2 is a more important factor than CO 2 adsorption in CO 2 photoreduction. We therefore employed the in-situ infrared technology to investigate the adsorption and chemisorption of CO 2 over CPs in a solvent-containing environment ( Figure R4). Figure R4. In-situ FT-IR spectra of N-CP-D with Co complexes for the adsorption of CO 2 (a) and the chemisorption of CO 2 (b) in a solvent-containing environment.
As shown in Figure R4, although the CP-A series exhibited greater CO 2 adsorption than the CPs-D series, the L-CP-D and N-CP-D showed an enhanced intensity both in the region of CO 2 adsorption (a) and the region of CO 2 chemisorption (b) than L-CP-A and N-CP-A. It suggests that the CO 2 adsorption capacity of CPs-D series is stronger than that of CPs-A series under the solvent-containing environment, which provides favorable conditions for the subsequent CO 2 reduction reaction.
We have revised the following sentence in Manuscript for the explanation of the CO 2 adsorption.
"Nevertheless, the CO 2 absorption of a dry powder is very different from the wet conditions 50 , the in-situ FT-IR indicates L-CP-D and N-CP-D exhibit enhancement both in CO 2 adsorption and CO 2 chemisorption than L-CP-A and N-CP-A ( Supplementary Fig. 30). It means that the CO 2 adsorption capacity of CPs-D series is stronger than that of CPs-A series under the solvent-containing environment, which provides favorable conditions for the subsequent CO 2 reduction reaction." 6. There is a clerical error in SI (page 46). "As shown in Figure S4" should be "As shown in Figure S34".