Porous hypercrosslinked polymer-TiO2-graphene composite photocatalysts for visible-light-driven CO2 conversion

Significant efforts have been devoted to develop efficient visible-light-driven photocatalysts for the conversion of CO2 to chemical fuels. The photocatalytic efficiency for this transformation largely depends on CO2 adsorption and diffusion. However, the CO2 adsorption on the surface of photocatalysts is generally low due to their low specific surface area and the lack of matched pores. Here we report a well-defined porous hypercrosslinked polymer-TiO2-graphene composite structure with relatively high surface area i.e., 988 m2 g−1 and CO2 uptake capacity i.e., 12.87 wt%. This composite shows high photocatalytic performance especially for CH4 production, i.e., 27.62 μmol g−1 h−1, under mild reaction conditions without the use of sacrificial reagents or precious metal co-catalysts. The enhanced CO2 reactivity can be ascribed to their improved CO2 adsorption and diffusion, visible-light absorption, and photo-generated charge separation efficiency. This strategy provides new insights into the combination of microporous organic polymers with photocatalysts for solar-to-fuel conversion.

titania as well as further references such as the composite without titania, titania supported on graphene oxide-HCP composite etc. The results are presented in a systematic and sound manner.
Concerning the impact, the contribution is clearly important but justification for a publication in Nature Communications is not sufficiently provided. In addition, the hypothesis of the contribution is based on CO2 adsorption and diffusion length as crucial parameters to enhance catalytic activity. Following this line of argument, the major aspects to critically consider are: -CO2 adsorption on or close to the active site is important, as surface coverage presents the concentration subsequently determining the reaction rate to be achieved (r=k*surface coverage of reagents, considering a Langmuir Hinshelwood type of activation). In line, a higher surface coverage of the substrates enables reaching a higher reaction rate, while the intrinsic catalytic activity of the active site remains unaltered. Following this argument, the authors have to carefully proof their hypothesis that CO2 adsorption within the material indeed causes enhanced coverage on or close to the active sites, increasing reaction rate. Kinetic experiments varying partial pressure, temperature etc. and providing assessment of the influence on the reaction rate are indispensable to justify the hypothesis. In addition, a careful representation of the literature state of the art including rates achieve in previous contributions normalized to the active sites content of the different catalysts are essential.
-The second argument relates to diffusion length as important element of catalyst design. The provided data do not allow any conclusion of the role of surface diffusion within the overall system. Following the argument of a diffusion governed process, the rate of surface diffusion should be somewhat rate determining. Consequently, kinetic analysis following e.g. the temperature dependence are needed to support such a hypothesis. Diffusion limitation should cause a limited apparent activation energy according to the reduced temperature dependence of various types of diffusion compared to chemical reactions.
-As a minor point, the experimental data are not provided with sufficient information to understand the used experimental setup, e.g. in the related figure captions. Where the reactions carried out in a flow reactor or a batch system? What was the temperature? Partial pressure of CO2 and H2O were constant (was there water in the system)?
-Concerning CO2 adsorption experiments, data in presence of water appear important and water vapor sorption experiments would be complementary.
HOMO level is found to be enough for water oxidation.
The quantitative measurement of O2 evolution was conducted using an optical fiber oxygen sensor. The O2 evolution rate over HCP-TiO2-FG under visible-light irradiation was determined to be 1.6 μmol h -1 , while the O2 evolution over other photocatalysts was too low to be detectable (Supplementary Fig. S13). The electrons from the water oxidation are slightly higher than the total consumed electrons for the reduced products including CH4 and CO. To the best of our knowledge, this is the first example achieving the quantitative detection of the oxygen production during photocatalytic CO2 conversion in such a gas-solid reaction system. Moreover, the isotopic labeled H2 18 O vapors was used to verify the origin of the detected O2. The formation of 18 O2 suggests that the evolved O2 gas is derived from the photocatalytic water oxidation (Supplementary Fig. S16). Finally, we corrected the energy levels and proposed a mechanism for the overall CO2 conversion process over the HCP-TiO2-FG photocatalyst as shown in Fig. 4f.
In the revised manuscript, we have added the following text on page 9 and 11: "As a result of the relatively high photocatalytic performance of porous HCP-TiO2-FG, the O2 evolution can be measured to provide the evidence of the oxidation cycle offering a better insight of the mechanism that is seldom discussed in the literature 43 .The O2 evolution rate over HCP-TiO2-FG under visible-light irradiation was determined to be 1.6 μmol h -1 , while the O2 evolution over other photocatalysts was too low to be detectable (Supplementary Fig. S13). The electrons from the water oxidation are slightly higher than the total consumed electrons for the reduced products including CH4 and CO." "In an isotopically labeled experiment, the 13 CH4 and 13 CO signals at m/z = 17 and m/z = 29 appeared after the photocatalytic reaction. The results confirmed that the CO and CH4 products are indeed originating from the photocatalytic reduction of CO2 gas ( Supplementary Fig. S15). The isotopic labeled H2 18 O vapors led to the formation of 18 O2 (Supplementary Fig. S16), suggesting that the evolved O2 gas was derived from the photocatalytic water oxidation." "The HCP-FG showed that its highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) energy levels were located at −5.34 eV and −3.00 eV (vs. vacuum level) as calculated by optical absorption (Fig. 4d) and cyclic voltammetry (CV) measurement (Supplementary Fig. S25), which are more negative than the valence band (VB) and conduction band (CB) levels of TiO2 respectively. To further confirm, ultraviolet photoelectron spectroscopy (UPS) technique was employed to measure the HOMO location, (Supplementary Fig. S26), which was found to be very close to that of CV measurement. Based on the position of HOMO and LUMO energy levels, a tentative mechanism for the overall CO2 conversion process over the HCP-TiO2-FG photocatalyst is proposed and is shown in Fig. 4f." "The excited HCP-FG was recovered to its neutral state by oxidizing the absorbed water molecules to produce oxygen gas."   The work function (Φ) can be determined by the difference between the photon energy (21.2 eV) and the binding energy of the secondary cutoff edge. Φ = 21.20-16.86 = 4.34 eV The HOMO location is measured to be 1.10 eV below the Fermi level (E F ), corresponding to -5.44 eV vs vacuum level.
2. The reviewer wonders if the origin of CO production might be polymer degradation even though they showed good repeated property. The authors described that CO evolution increased in the absence of water, which implies that CO generation would be induced by partial oxidation of polymer or graphene. The reviewer recommends the authors to conduct isotope labeling experiment using carbon thirteen CO2 to show that the origin of their product would be CO2.
Response: This is a good suggestion. As suggested, we have conducted the isotope labeling experiment by using 13 CO2 as the substrate to detect the origin of the products.
The results indicate that the origin of CO and CH4 production is CO2 instead of polymer degradation (Supplementary Fig. S15).
We have now also included the corresponding description on page 9 in the manuscript.
"To verify the evolution of CO Fig. S15)." Figure S15. GC-MS spectra of 13 CH 4 (m/z=17) and 13 CO (m/z=29) after the photocatalytic reaction over . The isotopic labeled 13 CO 2 was used as the substrate in normal experiment condition.
3. The authors emphasized that their originality is the efficient adsorption of CO2, thus, they should show the reasonable experimental evidence like FT-IR to claim their efficient CO2 adsorption.

Response:
In most of the cases, FT-IR analysis was done to study the chemisorption of CO2 molecules on the catalysts. It is well established that the physical adsorption property of porous polymer materials can be evaluated by CO2 adsorption/desorption isotherms. The CO2 adsorption/desorption isotherms in Fig. 3e-f indicate the efficient adsorption of CO2 on HCP-TiO2-FG as 12.87 wt% at 1.00 bar and 273.15 K, which is more than 4-fold higher than that of TiO2 and TiO2-G.
Moreover, we have also conducted TGA-DSC analysis under CO2 atmosphere to get further evidence for CO2 adsorption. The results are shown in Supplementary Fig. S11.
The adsorption mode displays the pressure and temperature-dependent features with excellent recyclability for the repeated CO2 adsorption and desorption. Based on the literature reports (Energy Environ. Sci. 2014, 7, 3478;Chem. Soc. Rev. 2017, 46, 3322), the CO2 adsorption by such porous polymer materials is in consistent with the characteristics of physical adsorption, which cannot affect the signals in FT-IR spectra.
We have now also included corresponding description on page 6 in the manuscript: "It is well established that the CO2 uptake by porous polymer materials mainly results from its physical adsorption 34,35 . Such adsorption mode displays the pressure and temperature-dependent features with excellent recyclability for the repeated CO2 adsorption and desorption (Supplementary Fig. S11)." Figure S11. The temperature-dependent adsorption ability of porous HCP-TiO 2 -FG checked by TGA-DSC at CO 2 atmosphere.

Reviewer 2
This work by Wang et al. presents the fabrication of a porous hypercrosslinked polymer-TiO2-graphene (HCP-TiO2-FG) sandwiched structure for visible-light-driven CO2 conversion. The structure possesses relatively high surface area and CO2 uptake capacity, and shows relatively high photocatalytic performance for CO and CH4 production without the use of sacrificial reagents or precious metal co-catalysts. The authors argue that the superb catalytic activity is attributed to the improved CO2 adsorption and diffusion, visible-light absorption and photo-generated charge separation. Indeed, it is a topic of interest to the researchers in the related areas; however, the manuscript still cannot meet the level of Nature Communications. No deep understanding and important scientific issues have been given in this manuscript.
Several issues need to be clarified, and part of the conclusions cannot be supported by the current data.
Response: Thanks much for appreciating the quality of work presented in this manuscript. We believe that this work is novel and would have great impact in the field.
To date, there is no report on the combination of microporous organic polymers with photocatalytic materials for CO2 uptake and conversion. The photocatalytic performance is dramatically enhanced by the design and synthesis of porous HCP-TiO2-FG sandwiched photocatalysts. Based on the results obtained, we have presented the understanding of such superior structure from the aspects of improving CO2 adsorption and diffusion, visible-light absorption, and photogenerated charge separation efficiency. These factors are widely studied scientific issues in the related areas. In the revised version, we have conducted the kinetic experiments to provide further understanding. We also thank the reviewer's kind suggestions. We have addressed all the issues point-by-point in our response below.
Specific comments are listed below: 1. The authors have stated that this is the first example involving microporous organic polymers for CO2 conversion among numerous photocatalysts. This may overstate the significance of this work. As far as I know, conjugated microporous polymers have been used as materials for the capture and conversion of CO2 (Nature Communications, DOI: 10.1038/ncomms2960) and visible-light-driven conversion of CO2 (Green Chem., 2017, 19, 5777).
Response: While we appreciate the reviewer's comments, we want to emphasize that the present work is fundamentally different from those reports i.e., Nature Commun., 2013, DOI: 10.1038/ncomms2960 and Green Chem., 2017 The microporous polymers in the above-mentioned Nature Commun are developed as heterogeneous catalysts for the reaction of CO2 and propylene oxide for the formation of propylene carbonate. Indeed, there are many such reports on the use of microporous polymeric catalysts for chemical conversion of CO2 (Chem. Commun. 2015, 51, 11576;Adv. Mater. 2017, 29, 1700445;J. Mater. Chem. A 2018, 6, 374;et al). We also reported a metalporphyrin-based microporous polymer for catalyzing the reaction of CO2 with propylene oxide (J. Mater. Chem. A 2017, 5, 1509. Compared to the chemical conversion, the photocatalytic CO2 reduction is of great significance because it utilizes the abundant and sustainable solar energy to produce carbonaceous fuels.
Another report in the above-mentioned "Green Chem" journal is related to the photocatalytic conversion of CO2. However, the optimized pyrene-based polymer catalyst do not display porosity because it has extremely low surface area i.e., 23.9 m 2 g -1 . The photoreduction of CO2 depends on the chemical capture of CO2 molecules by a task-specific ionic liquid i.e., [P4444][p-2-O]. In addition, CO and H2 are detected as the main products from the photocatalytic system, whereas the desired CH4 production is absent. Thus the idea of using microporous organic polymers for CO2 uptake and photocatalytic conversion is missing in this work.
The previous studies were indeed focusing more on developing polymer materials for catalyzing CO2 conversion. These findings inspired us to incorporate the microporous organic polymers as CO2 capture materials into the photocatalytic system. As a result, we achieved high CO2 conversion efficiency with a rate of total consumed electron number (Re) as 264 μmol g -1 h -1 , including 83.7% selectivity for CH4 production and negligible side reaction of H2 production under visible-light irradiation.
To date, there is no report on the combination of microporous organic polymers with photocatalytic materials for CO2 uptake and conversion. We believe this strategy will be very helpful to overcome the constraint of deficient pore structure for semiconductor-based composites and to open a new pathway for the design and synthesis of well-defined porous materials with high CO2 uptake and photocatalytic conversion efficiency.
2. The authors have demonstrated that TiO2 crystals are supported on graphene sheets and encapsulated by ultrathin HCPs layer. As such, the TiO2 should be in the middle of the sandwiched structure. As stated in the manuscript, the photogenerated electrons migrate from HCP to TiO2 via their interfacial interaction. In this case, what is the role of the graphene? How can graphene improve the charge separation efficiency? The authors should clarify these issues in the manuscript.

Response:
The TiO2 crystals are located in the middle of the sandwiched structure because they are supported on graphene nanosheets and encapsulated by ultrathin HCPs layer ( Fig. 2 and Supplementary Fig. S1-4). It should be pointed out that the HCPs and graphene are not freestanding in the composite. Benefiting from the in-situ knitting strategy, ultrathin HCPs layers were integrated with the functionalized graphene nanosheets through the methylene linkers. The structures have been verified by XPS, FT-IR, and CP/MAS NMR characterizations (Fig. 3b-c and Supplementary Fig. S6-

S8).
As far as the role of the graphene is concerned, we have now added the EIS analysis of HCPs and HCPs-TiO2, which are compared with HCPs-FG and HCPs-TiO2-FG ( Supplementary Fig. S23). The covalent linking with graphene effectively improves the electronic conductivity of the HCPs and thus facilitates the electron transfer in the composite. The comparison of photocatalytic performance between HCP-TiO2 and HCP-TiO2-FG photocatalysts is also shown in Supplementary Fig. S24. The less efficient CH4 production over HCP-TiO2 photocatalyst can also reflect the influence of graphene on improving the charge separation efficiency.
The corresponding description has been revised on page 10 and 11 in the manuscript as: "The lower Ret of TiO2-FG than that of TiO2 indicates that FG modification favors the electronic conductivity due to its high electron mobility. Moreover, the covalent linking with graphene effectively improves the electronic conductivity of the HCPs and thus facilitates the electron transfer in the composite. The less efficient CH4 production over HCP-TiO2 photocatalyst can also reflect the influence of graphene on improving the charge separation efficiency (Supplementary Fig. S24). As a result, the porous sandwich structure possesses the improved efficiency in separating the photogenerated charge carriers."  3. In "Characterization of the resulting materials" part, the authors have stated that the percentage of the exposed {001} facets in the TiO2 crystal is calculated to be approximately 30%. What is the relationship between the {001} facets and the photocatalytic performance for CO2 conversion? If the {001} facets are more reactive, would it be better to use TiO2 nanosheets with dominant (001) facets?
Response: The exposed crystal facets have great impact on the photocatalytic performance of TiO2 crystals. In the studies of crystal facet engineering of anatase TiO2, both theoretical and experimental evidence demonstrates that the {001} facets are much more reactive but less thermodynamically stable than {101} facets due to higher average surface energy of the {001} facets than that of the {101} facets. As reported, the products with large percentage of {001} facets usually have a large size of micrometers or hundreds of nanometers and low surface area (Nature 2008, 453, 638;J. Am. Chem. Soc. 2009, 131, 4078;J. Am. Chem. Soc. 2009, 131, 3152;Chem. Commun. 2009, 29, 4381). When decreasing the particle size, the specific surface area increased, whereas the high-energy {001} facets tend to transform to the more thermodynamically stable {101} facets to reduce the high surface energy. For example, the TiO2 crystals with size of ~20 nm only exposed 9.6% {001} facets, but the photoactivity could be comparable to that of micro-sized TiO2 with dominant {001} facets (Nano Lett. 2009, 9, 2455. The TiO2 crystals with size of 30-85 nm only exposed 18% {001} facets, but they showed an increase in the specific surface area as 21 m 2 g -1 , and exhibited 5.6 times stronger photoactivity than microcrystals with 72% {001} facets (Chem. Commun. 2010, 46, 755).
Thus there should be a balance between the particle size and percentage of exposed {001} facets. The study on photocatalytic CO2 conversion indicated that the TiO2 crystal with 60% {001} facets (60 nm) showed 15% and 90% higher CO production than that with 92% {001} facets (150 nm) and 95% {101} facets (20 nm), respectively (ACS Catal. 2016, 6, 1097. Yu et al. systematically studied the influence of factors on the photocatalytic CO2 conversion including percentage of {001} facets, size, and surface area (J. Am. Chem. Soc. 2014, 136, 8839). The table below shows that the highest CH4 production was achieved by the TiO2 crystals with 58% {001} facets, size of 60 nm, and surface area of 45 m 2 g -1 (Table R1). 4. In thermogravimetric analysis (TGA, Figure S9), the authors have stated that the HCP-TiO2-FG composite structure exhibits excellent thermal stability compared to TiO2-G with resistance of degradation up to 400 °C. However, according to Figure S9, the thermal stability of TiO2-G is even better than HCP-TiO2-FG with resistance of degradation up to 400 °C.
Response: Sorry for this mistake. We have revised the sentence on page 6 as follows, "the HCP-TiO2-FG composite structure exhibited the excellent thermal stability comparable to TiO2-G with resistance to degradation up to 400 °C".

5.
For the photocatalytic test, the authors should provide the details for the optical density of light source and the illuminated area.

Response:
The details for the photocatalytic test are as follows: Under visible-light (λ≥420 nm) irradiation, the optical density of 300 W Xe lamp was measured to be 433 mW cm -2 and the illuminated area of photocatalyst is about 3.14 cm 2 .
In the revised manuscript, the following information is added, The optical density was measured to be 433 mW cm -2 and the illuminated area of photocatalyst was about 3.14 cm 2 ." In addition, to verify the origins of produced CO and CH4, the isotopic 13 CO2 must be used as a reactant to trace the carbon sources in the photocatalytic reaction.
Response: As suggested, to verify the origin of the products, we have conducted the isotopic labeling experiment using 13 CO2 as a reactant to trace the carbon sources in the photocatalytic reaction. The results are shown in Supplementary Fig. S15.
We have also included corresponding description on page 9 in the manuscript as follows:  Fig. S15)." Figure S15. GC-MS spectra of 13 CH 4 (m/z=17) and 13 CO (m/z=29) after the photocatalytic reaction over HCP-TiO 2 -FG under visible-light irradiation (λ≥420 nm). The isotopic labeled 13 CO 2 was used as the substrate in normal experiment condition.
6. According to Figure 4a and b, the HCP-FG material also exhibits a photocatalytic activity for CO and CH4 production. The authors have stated that the catalytic sites are located on TiO2. In this case, it remains elusive whether the catalytic sites are located on TiO2 or HCP-FG. The authors should clarify this point.

Response:
Yes, the HCP-FG material also exhibits a photocatalytic activity for CO and CH4 production. The porous property and photocatalytic performance of samples are shown in Fig. 3d-f, Fig. 4a-c, Fig. S10, Fig. S12  size distribution that calculated using DFT methods (slit pore models, differential pore volumes).
Time-dependent production of CH 4 (d) and CO (e) in photocatalytic CO 2 reduction with different catalysts under visible-light (λ≥420 nm). (f) Average efficiency of photocatalytic CO 2 conversion with different catalysts during 5 h of visible-light (λ≥420 nm) irradiation. 7. In Figure  8. According to the time-dependent production of CH4 and CO as shown in Figure S13, it remains unclear why the production of CH4 increases linearly while the production of CO has a stagnation effect.
Response: Yes, the CH4 production increased linearly with irradiation time, whereas the CO production was fast at initial irradiation time and then showed a sluggish increase during the photocatalytic reaction ( Fig. 4a-b and Fig. S19). In fact, the stagnation effect in CO production has also observed in photocatalytic CO2 reduction by many researchers (J. Am. Chem. Soc. 2018, 140, 38;J. Am. Chem. Soc. 2017, 139, 7217;J. Am. Chem. Soc. 2017, 139, 6538;Adv. Mater. 2016, 28, 6485;ACS Nano, 2015, 9, 2111. For example, Luo's group studied the mechanism of photoreduction of CO2 to CH4 on TiO2 surface by theoretical calculations. They proposed that CO was the initial product of CO2 photoreduction that could be further photo-reduced to CH3OH or CH4 (ACS Catal. 2016, 6, 2018. That is possibly why the CO production rate is fast at initial irradiation time and then shows a sluggish increase during the photocatalytic reaction. In contrast, the CH4 as the final product showed a steady increase. 9. In the electrochemical impedance spectra (EIS, Figure S15), HCP-FG has a much larger semi-circle diameter than HCP-TiO2-FG. Why the addition of TiO2 can dramatically reduce the electron-transfer resistance?
Response: In the EIS analysis ( Supplementary Fig. S23), the smaller arc in HCP-TiO2-FG sandwich structure than HCP-FG suggests that the formation of sandwich structure improves the electronic conductivity. According to the gas adsorptiondesorption analysis, it was found that the surface area and CO2 uptake of HCP-TiO2-FG were higher than those of HCP-FG (Supplementary Tab. S2). Thus we can infer that the TiO2 intercalation somewhat restricted the aggregation of HCPs layers on graphene nanosheets resulting in the reduced electron-transfer resistance.

In abstract, "…and highlights the importance of MOPs in combination with
photocatalysts for solar energy conversion.", there is no definition for the "MOPs" in the manuscript.

Response:
We are sorry for missing the definition of the "MOPs". In the revised manuscript, we have replaced it by "microporous organic polymers".

Reviewer 3:
The authors report on a novel catalyst for photocatalytic CO2 activation using visible light enabling unprecedented activity and CH4 selectivity. This high performance is achieved by tailoring the CO2 adsorption of the material together with short diffusion length to the active sites.
To achieve these features, the authors prepare a composite material composed of TiO2 anatase crystals with exposed {001} facets as active sites grown on graphene oxide.
The latter was functionalized by PhPh3 serving as anchoring groups enabling coating of a thin layer of hypercrosslinked polymer (HCP).
The material is comprehensively characterized. Its catalytic activity is reported compared to pure titania as well as further references such as the composite without titania, titania supported on graphene oxide-HCP composite etc. The results are presented in a systematic and sound manner.
Concerning the impact, the contribution is clearly important but justification for a publication in Nature Communications is not sufficiently provided. In addition, the hypothesis of the contribution is based on CO2 adsorption and diffusion length as crucial parameters to enhance catalytic activity.
Response: First, we would like to thank the reviewer to realize and appreciate the importance of work reported in this manuscript. We also thank the reviewer's overall comments and kind suggestions that helped us to improve the manuscript significantly.
We have now addressed all the reviewer's query point-wise in our response below.
Following this line of argument, the major aspects to critically consider are: -CO2 adsorption on or close to the active site is important, as surface coverage presents the concentration subsequently determining the reaction rate to be achieved (r=k*surface coverage of reagents, considering a Langmuir Hinshelwood type of activation). In line, a higher surface coverage of the substrates enables reaching a higher reaction rate, while the intrinsic catalytic activity of the active site remains unaltered.
Following this argument, the authors have to carefully proof their hypothesis that CO2 adsorption within the material indeed causes enhanced coverage on or close to the active sites, increasing reaction rate. Kinetic experiments varying partial pressure, temperature etc. and providing assessment of the influence on the reaction rate are indispensable to justify the hypothesis. In addition, a careful representation of the literature state of the art including rates achieve in previous contributions normalized to the active sites content of the different catalysts are essential.
Response: This suggestion inspired us to investigate the kinetic characteristics of the photocatalytic CO2 conversion which has never been discussed in the related literatures to the best of our knowledge. Generally, the reaction kinetics in a gas-solid system is studied by varying the partial pressure and temperature.
The photocatalytic reactions were carried out in a batch system under standard atmospheric pressure. Since the pressure in the photocatalytic reactor was settled as standard atmospheric pressure, we conducted the kinetic experiments by varying the partial pressure of CO2. The HCP-TiO2-FG photocatalyst exhibits a high CO2/N2 selectivity ratio of 25.8 calculated by the initial slopes of adsorption isotherms shown in Supplementary Fig. S20 (Adv. Mater. 2012, 24, 5703). The partial pressure of CO2 can be adjusted from 2.5% to 100% by varying the volume ratio of CO2 to N2. Since the kinetic model and reaction mechanism of photocatalytic CO2 conversion are ambiguous so far, the quantitative relationship between CO2 coverage and CH4 evolution rate is still unclear. Interestingly, it is observed that they show a similar trend of increase with CO2 proportion, e.g. both showed a dramatic increase at lower partial pressure and then displayed a sluggish increase at higher CO2 concentration ( Supplementary Fig. S21).
As far as temperature-dependent kinetics is concerned, the results and related discussion given below in our reponse would be helpful to address this comment.
In addition, the literature state of the art including rates normalized to the active sites content of the different catalysts are reviewed in the revised manuscript. Based on the discussions, we can deduce that the more efficient CH4 production over HCP-TiO2-FG should not result from the difference in the number of catalytic sites but mostly come from the higher surface coverage of CO2 on the active sites. Thus we can furthermore demonstrate the superiority of such porous sandwich structure towards the visible-lightdriven photocatalytic CO2 conversion.
On page 10, we have added the following content: The second argument relates to diffusion length as important element of catalyst design.
The provided data do not allow any conclusion of the role of surface diffusion within the overall system. Following the argument of a diffusion governed process, the rate of surface diffusion should be somewhat rate determining. Consequently, kinetic analysis following e.g. the temperature dependence are needed to support such a hypothesis.
Diffusion limitation should cause a limited apparent activation energy according to the reduced temperature dependence of various types of diffusion compared to chemical reactions.
Response: As suggested, we have conducted the kinetic experiments by varying the temperature on the photocatalyst surface. The temperature has a complicated influence on the rate of photocatalytic conversion from the aspects of adsorption, diffusion and photocatalytic processes.
The results are shown in Fig. 3e-f and Fig. R2-3. By elevating the temperature, the surface coverage of CO2 molecule on the catalyst surface was decreased due to the exothermic effect of adsorption process ( Fig. 3e-f), while the diffusion rate was increased as a result of the increased thermal motion of CO2 molecules (Fig. R2). For the photocatalytic process, it is well-known that the Gibbs free energy increases in the photocatalytic conversion of CO2 to CO and CH4. The increase of chemical potential originates from the energy of photons rather than heat, so the change in temperature by dozens of degrees would not cause a measurable difference in the photocatalytic reaction. That is why the activation energy is seldom discussed in the photocatalytic reaction in literatures. From our experience, the photocatalytic CO2 conversion efficiency over pure TiO2 photocatalyst under UV-light is indeed independent of temperature from 291 K to 353 K. Interestingly, we observed an obvious increase in CH4 evolution rate over porous HCP-TiO2-FG photocatalyst with temperature ( Fig.   R3).
Now we can conclude as follows: by increasing the temperature, the surface coverage of CO2 molecule declined, the diffusion rate speeded up, the photocatalytic reaction rate kept constant, and the overall reaction rate for CO2 conversion was increased.
Based on Arrhenius plot, the adsorption activation energy for CO2 adsorption is calculated to be 5.20 kJ mol -1 (Fig. R3a) using a microporous diffusion model (Ind.
Since the CH4 production increases linearly and possesses dominant electron consumption selectivity as 83.7%, we can adopt the pseudo-zero order model to estimate the rate constant for the overall reaction, obtaining apparent activation energy as 9.34 kJ mol -1 (Fig. R3b). By noting the temperature-dependent characteristics with activation energy values, we can conclude that the rate of surface diffusion is somewhat rate determining to the photocatalytic CO2 conversion.
In the present study, we aim to provide new insights into the design and synthesis of well-defined porous photocatalysts for CO2 uptake and conversion, and present an important first example towards solar-to-carbonaceous fuels conversion employing microporous organic polymers in combination with photocatalytic materials. The kinetic analysis is indeed very important to the study and applications of photocatalytic CO2 conversion but still there are many challenges that need to be addressed to unveiling the fundamental understanding of each reaction step. Therefore, we now intend to adopt the simple photocatalytic system such as pure TiO2 photocatalyst to probe the kinetics model and reaction mechanism of CO2 conversion, and then extend to complicated models involving adsorption, diffusion and photocatalytic processes. On page 9, we have added the following content: "Although the porous HCP-TiO2-FG also exhibits a high adsorption capacity towards water vapors, about 30 wt% at 90% humidity (Supplementary Fig. S17), the existence of water vapors brings a slight increase in CO2 uptake (Supplementary Fig. S18), presumably due to their affinity with the water molecules." Figure S17. The water adsorption of HCP-TiO 2 -FG at different humidity. Figure S18. CO 2 adsorption experiments, data in presence of water. q e is the equilibrium adsorption capacity at pure CO 2 atmosphere with 1 bar. q/q e represents fractional uptake at pure CO 2 atmosphere and mix atmosphere of 99% CO 2 +1% water vapors with 1 bar.

Reviewers' comments:
Reviewer #1 (Remarks to the Author): The authors reasonably answered to my comments and properly revised the paper. If the other reviewers also agree to accept, the revised version is publishable in Nature Communications.
Reviewer #2 (Remarks to the Author): Comments: I am pleased to see that the authors have made a big effort to improve the quality of this manuscript by performing new experiments, adding new discussion and completing a number of changes. Indeed, some comments have been partially addressed. However, the evidences are still insufficient to support their key conclusions. Although it is a topic of general interest, the current revised manuscript still cannot meet the level for publication in Nature Commun.
Specific comments are listed below: 1. In the isotopically labeled experiment, why the 13CH4 and 13CO signals are only at m/z = 17 and m/z = 29, respectively? Why there are no fragment signals at m/z = 13 and m/z = 16 for 13CH4 and 13CO?
2. As the authors stated that the catalytic sites are located on TiO2, the functions of the formed HCP-TiO2-FG sandwich structure remain not clear. In the sandwich structure, what are the TiO2 crystals directly contacting with (the porous hypercrosslinked polymer (HCP) or graphene)? If the TiO2 crystals are directly contacting with the HCP, the photogenerated electrons will migrate from HCP to TiO2 via their interfacial interaction. In this case, how can graphene improve the charge separation efficiency? If the TiO2 crystals are directly contacting with the graphene, the function of porous HCPs layers in enriching the adsorptive sites to achieve the high CO2 uptake may not work.
3. In the CV measurement for calculating the HOMO and LUMO energy levels of HCP-TiO2-FG catalyst in CH2Cl2, the reference electrode Ag/AgCl is not correct. The reference electrode Ag/AgCl is commonly suitable in aqueous solution. In organic solvent, the reference electrode should be Ag/Ag+ using the Fc/Fc+ couple as an internal standard.
4. The authors demonstrated that ultrathin HCPs layers were integrated with the functionalized graphene nanosheets through the methylene linkers. However, there is no direct evidence to conclude that HCPs layers were integrated with the graphene nanosheets through the methylene linkers. The current data can only support that there exists the methylene.
Reviewer #3 (Remarks to the Author): The authors were asked to provide further evidence for the governing nature of their two main Arguments, namely the superior CO2 adsorption as pre-requisite of high substrate concentration close to or on the active site and their hypothesis on the shortend diffusion lengths.
Indeed, a suitable kinetic analysis has been carried out illustrating that the investigated photocatalytic CO2 reduction is not under intrinsic kinetic control of the catalyst but the observed rates of CO and CH4 formation are rather determined by some transport effects.
The observed limited dependence of rate on temperature may hind towards diffusion control. I fully agree that kinetic analyses is a yet under-represented aspect in photo-catalysis. Though, the little rate dependence on temperature appears to point towards film diffusion effects. Hast the stirring speed or the main particle size been varied?
It remains unclear for me why the authors conclude on a reduced Diffusion length for the Optimum System, although they do not know which Diffusion is rate limiting: CO2 from bulk to the film, through the fild, in/on the porous material or Charge charrier diffusion?
The authors made an effort to identify the corresponding oxidation Reaction which has to balance the observed reduction reactions. Oxygen has been proven but quantitative Analysis is not yet available.
What about oxidation of the formed products, e.g. CO and CH4 to CO2 as reverse reactions of the attempted CO2 reduction. I suggest reference Experiments with these Substrates.
Overall, the contribution may become suitable for Nature Communications after providing further evidence and comments to the points made before.

Response to Reviewers' Comments
Many thanks for forwarding us the reviewers' comments that have helped us a lot to significantly improve the manuscript. Below are our responses (in BLUE colour) to the editor's queries. In the revised manuscript, more explanations on the HCP-TiO 2 -FG structure are presented in the response to Comment 4#. The diagram in Figure 1 has been updated to show the connections between HCPs and functionalized graphene. We are hoping that the structure is now much clear and clearly understandable.  According to the suggestion, the CV measurement using Ag/Ag + as reference electrode was also performed to ensure the identical electrolyte in the system. The CV curves were updated in Figure      the peak of C at 137 ppm is obviously enhanced, which indicates that the benzene ring in TiO 2 -FG has electrophilic substitution reaction with dichloromethane (DCM). It is noteworthy that the peaks belong to methylene at 37-39 ppm, which were commonly seen in HCPs materials and disappeared in Fig. R1a (Macromolecules 2011, 44, 2410; Sci. Adv. 2017, 3, e1602610). Instead, a new peak at 62.5 ppm was shown in Fig. R1b, which was corresponding to the methylene C peak of benzyl alcohol group (-CH 2 -OH). This indicates that the phenyl groups of TiO 2 -FG change into benzyl chloride group (-C 6 H 4 -CH 2 -Cl) in the reaction process firstly (Fig. R2a). However, due to the barrier effect of TiO 2 nanoparticles on graphene sheets, benzyl chloride groups can only exist in the form of original states instead of crosslinking with other benzene rings. With the quenching of dilute hydrochloric acid, the active benzyl chloride groups react with water and formed the benzyl alcohol groups (-C 6 H 4 -CH 2 -OH) (Fig. R2b). The model experiment indicates that the phenyl groups on TiO 2 -FG can react electrophilically with DCM to form benzyl chloride groups (-C 6 H 4 -CH 2 -Cl), which cannot undergo the further crosslinking reaction due to the blocking effect of TiO 2 nanoparticles. Meanwhile, the specific surface area of HO-CH 2 -TiO 2 -FG (128 m 2 g -1 ) obtained from the model experiment did not change compared with TiO 2 -FG, which also prove the above-mentioned conclusion.

Figure R2
Schematic diagram of reaction process in model experiment.
As co-monomer of synthesis for HCP-TiO 2 -FG, syn-PhPh 3 can be self-crosslinked to obtain SHCP-3a with a high specific surface area of 2525 m 2 g -1 by solvent knitting method (Sci. Adv. 2017, 3, e1602610). If TiO 2 -FG and SCHP-3a were mixed simply according to same mass ratio of co-monomers, the specific surface area and CO 2 uptake amount of the mixtures were lower than these of HCP-TiO 2 -FG, which suggest two monomers form a homogenous "sandwich" structure (Tab. R1).
Meanwhile, no freestanding SHCP-3a blocks were observed in SEM, TEM and scanning transmission electron microscopy (STEM) images (Fig. 2c-f and Supplementary Fig. S1d). This is because benzyl chloride group (-C 6 H 4 -CH 2 -Cl) formed has high reactivity for TiO 2 -FG and was linked to co-monomer syn-PhPh 3 , and the "sandwich" structure was formed in which the HCPs porous organic layers and TiO 2 -FG were linked by methylene. a Mass ratio of co-monomers for syn-PhPh 3 and TiO 2 -FG. b the sum of surface area for SHCP-3a and HCPs-TiO 2 -FG-X with mass ratio of co-monomers. c Surface area calculated from nitrogen adsorption at 77.3 K using Langmuir equation. d the sum of CO 2 uptake for SHCP-3a and HCPs-TiO 2 -FG-X with mass ratio of monomers at 1.00 bar and 273.15 K. e CO 2 uptake determined volumetrically using a Micromeritics ASAP 2020 M analyzer at 1.00 bar and 273.15 K.

Reviewer 3:
The authors were asked to provide further evidence for the governing nature of their two main Arguments, namely the superior CO 2 adsorption as pre-requisite of high substrate concentration close to or on the active site and their hypothesis on the shortend diffusion lengths. Indeed, a suitable kinetic analysis has been carried out illustrating that the investigated photocatalytic CO 2 reduction is not under intrinsic kinetic control of the catalyst but the observed rates of CO and CH 4 formation are rather determined by some transport effects. The observed limited dependence of rate on temperature may hind towards diffusion control. I fully agree that kinetic analyses is a yet under-represented aspect in photo-catalysis. Though, the little rate dependence on temperature appears to point towards film diffusion effects. Hast the stirring speed or the main particle size been varied? It remains unclear for me why the authors conclude on a reduced Diffusion length for the Optimum System, although they do not know which Diffusion is rate limiting: CO 2 from bulk to the film, through the fild, in/on the porous material or Charge charrier diffusion?
Response: Many thanks for appreciating our efforts on the kinetic analysis presented in the previous revision. The photocatalytic reactions were carried out in a gas-solid batch system without stirring the gas mixture. According to the suggestion, we studied the diffusion process by varying the stirring speed from zero to maximum. As shown in Supplementary Fig. S33, the increase in stirring speed greatly facilitates the photocatalytic conversion of CO 2 to CH 4 product, indicating that the diffusion plays a significant role in such a gas-solid reaction system. The related results and discussions have been added as a separate section "kinetic analysis" in the revised manuscript.
The particle size is another factor affecting the diffusion process, especially for internal diffusion. We would like to mention that the HCP-TiO 2 -FG composite produced through this strategy possesses the specific particle size. In the experience of optimizing synthesis conditions, TiO 2 crystals with larger or smaller size are inclined to form agglomerates, which cannot uniformly decorate on graphene or be fully wrapped by ultrathin HCPs layers. In the current system, we cannot achieve the varied particle size while keeping the designed structure with similar surface area, CO 2 uptake capacity and the number of active sites.
In order to reduce the diffusion length, we tried to construct the HCP-TiO 2 -FG sandwich structure with ultrathin HCP layers. The reduced diffusion length should be beneficial for the diffusion of CO 2 and photo-generated charges to the catalytically active sites, both of which may contribute to the improved CO 2 conversion efficiency.
Based on the gas diffusivity measurement, the diffusion coefficient of CO 2 in the HCP-TiO 2 -FG sandwich structure is calculated to be 1.8×10 -11 cm 2 s -1 at room temperature (Supplementary Fig. S32). Typical polymers for photoconversion application, the exciton diffusion and charge transfer dynamics of poly(3-hexyl thiophene) (P3HT) and phenyl-C61-buryric acid methyl ester (PCBM) have been studied, giving the diffusion coefficient of 1.8×10 -3 and 2.7×10 -4 cm 2 s -1 , respectively (Adv. Mater. 2008, 20, 3516; Nanoscale 2011, 3, 2280). The diffusion of charge carrier is much faster than that of the adsorbed gas by several orders of magnitude, implying that the gas diffusion should have more crucial influence at the rate limiting rather than the charge carriers.
In the present study, we aim to provide new insights into the design and synthesis of well-defined porous photocatalysts for CO 2 uptake and conversion. The designed sandwiched structure is somewhat complicated and makes it impossible to establish individual kinetic models. Although the kinetic mechanism cannot be clearly clarified in the current study, we indeed have made much progress on the kinetic analysis of photocatalytic CO 2 conversion that is seldom discussed in the literature. Inspired by the reviewers' suggestions, we have realized that the kinetic analysis is indeed very important for photocatalytic CO 2 conversion and there are yet many challenges that need to be addressed. In future study, we will use the simple photocatalytic system such as pure TiO 2 photocatalyst to probe the kinetics model and reaction mechanism of CO 2 conversion, and then extend to the complicated models involving adsorption, diffusion and photocatalytic processes.
On pages 12-13, all results and discussions on kinetic analysis are included and displayed as a separate section "kinetic analysis" in the revised manuscript.
"Kinetic analysis. The kinetics experiments were carried out to understand the contribution of CO 2 adsorption and diffusion to the enhancement of photocatalytic efficiency. The relationship between the CO 2 adsorption and CH 4 production can be explored by varying the surface coverage of CO 2 on the active sites. The partial pressure of CO 2 is adjusted in CO 2 /N 2 mixture because of a high CO 2 /N 2 selectivity ratio of 25.8 over the HCP-TiO 2 -FG photocatalyst (Supplementary Fig.   S29). Since the kinetic model and reaction mechanism of photocatalytic CO 2 conversion are ambiguous so far, the quantitative relationship between CO 2 coverage and CH 4 evolution rate is still unclear. Interestingly, it is observed that they show a similar trend of increase with CO 2 proportion, e.g. both of them dramatically increased at lower partial pressure and then displayed a slow increase at higher CO 2 concentration (Supplementary Fig. S30). Generally, the reaction rates that are normalized to the active sites allow the direct comparison of intrinsic reactivity on different catalysts [47][48][49] . For the catalytic system employing same catalyst, the reaction rate appears to be independent of the loading amount of catalyst after normalization to the same amount 50, 51 . In this regard, the porous HCP-TiO 2 -FG photocatalyst possesses equivalent catalytic active sites to TiO 2 /HCP-FG hybrid due to the same content of TiO 2 photocatalyst. That is, the more efficient CH 4 production over HCP-TiO 2 -FG should not result from the difference in the number of catalytic sites but mostly come from the higher surface coverage of CO 2 on the active sites.
The temperature has a complicated influence on the rate of photocatalytic conversion from the aspects of adsorption and diffusion. By increasing the temperature, the surface coverage of CO 2 molecules on the catalyst surface was decreased due to the exothermic effect of adsorption process ( Fig. 3e-f), while the diffusion rate was increased as a result of the increased thermal motion of CO 2 molecules (Supplementary Fig. S31). Based on Arrhenius plot, the adsorption activation energy for CO 2 adsorption is calculated to be 5.20 kJ mol -1 (Supplementary Fig. S32a) using a microporous diffusion model 52,53 . Since the CH 4 production increases linearly and possesses dominant electron consumption selectivity as 83.7%, we can use the pseudo-zero order model to estimate the rate constant for the overall reaction, obtaining apparent activation energy of 9.34 kJ mol -1 (Supplementary Fig. S32b). The diffusion process was further studied by varying the stirring speed. As shown in Supplementary Fig. S33, the increase of stirring speed greatly facilitates the photocatalytic conversion of CO 2 to CH 4 product. Combining the diffusion effect with pressure-/temperature-dependent characteristics, we can conclude that the photocatalytic CO 2 reduction over HCP-TiO 2 -FG is not under intrinsic kinetic control of the catalyst but the efficiency is rather determined by gas adsorption and diffusion. The elucidation of adsorption and diffusion that contributed to the photocatalytic reaction, and is seldom discussed in the literature, provides valuable information for understanding the relationship between the catalytic performance and structure properties. As a result, it clearly demonstrates the superiority of such porous sandwich structure towards the visible-light-driven photocatalytic CO 2 conversion. Further kinetic study is required to probe the kinetics model and reaction mechanism of photocatalytic CO 2 conversion."  Figure S30. Influence of the partial pressure of CO 2 on the CO 2 uptake (a) and CH 4 production rate (b). The photocatalytic reactions were carried out in a batch system under standard atmospheric pressure. The partial pressure of CO 2 can be adjusted from 2.5% to 100% by varying the volume ratio of CO 2 to N 2 . q e is the equilibrium adsorption capacity at pure CO 2 atmosphere with 1 bar. q/q e represents fractional uptake at different partial pressure of CO 2 . Figure S31. (a, b, c, d) (Supplementary Fig. S25), the amount of CO and CH 4 remained almost constant when the light was turned off. The extremely slow rate of reverse reaction indicates that the oxidation of CO and CH 4 to CO 2 is efficiently controlled over the HCP-TiO 2 -FG photocatalyst under such mild reaction conditions.
In the revised manuscript, we have added the following text on page 11:  Fig. S25)."

Figure S25. Changes in CH 4 and CO production over the HCP-TiO 2 -FG photocatalyst under dark conditions and visible-light (λ≥420 nm) irradiation and dark conditions.
Reviewers' comments: Reviewer #2 (Remarks to the Author): Comments: I am pleased again to see that the authors have made some efforts to improve the quality of this manuscript by performing new experiments, adding new discussion and completing a number of changes. Indeed, some comments have been partially addressed. However, the key issue in terms of structure-performance relationship is still ambiguous, which makes it insufficient to meet the level for publication in Nature Commun. For this reason, I have to recommend the rejection toward the publication in Nature Commun.
Specific comments are listed below: 1. By TEM images and EDX mapping, the authors demonstrated in the current revised manuscript that a distinct sandwich structure of HCP-TiO2-FG, in which the graphene surface and TiO2 crystals were covered by the HCPs layers and the TiO2 crystals were supported on the graphene sheets and encapsulated by the ultrathin HCPs layer, was formed. However, in the Response Letter, the authors pointed out that the HCPs and graphene existed as an integrated structure instead of freestanding parts in the composite. This seems to be very ambiguous.
2. According to Fig. 3d, the TiO2 intercalation somewhat restricted the aggregation of HCPs layers on graphene nanosheets and thus dramatically enlarged the specific surface area. As a result, the enhanced performance of HCP-TiO2-FG cannot support the argument that the catalytic sites are located on TiO2. As the HCP-FG material also exhibits a photocatalytic activity for CO and CH4 production, it remains elusive whether the catalytic sites are located on TiO2 or HCP-FG.
Reviewer #3 (Remarks to the Author): The careful revision of the authors is acknowledged. With regard to kinetic analysis and mass Transfer, all parameters reasonable well accessible by experiments were investigated and a careful discussion was added. The presented additional data appear comprehensive. I support acceptance.
Therefore we believe that the description of "sandwich structure" and "integrated structure" are not much contradictory. While responding to the comment 2# of Reviewer 2#, we used "integrated structure" in the response letter only to emphasize that the charge carriers can be moved throughout the whole HCP-FG structure.
Here we would like to explain the relationship between HCPs and graphene as HCPs being covalently linked on graphene through the methylene linkers. The TiO 2 crystals were intercalated into the HCPs layers and graphene sheets at discrete sites, as shown in Fig. 1. Therefore, the "HCP-TiO 2 -FG sandwich structure" is used to describe these sites with TiO 2 crystals loading. At the other sites without TiO 2 crystals, the HCPs are covalently linked on graphene that seems to be a "HCP-FG integrated structure". The model of covalent linking is now also presented in Fig. 1 according to the structural characterizations.
2. According to Fig. 3d, the TiO 2 intercalation somewhat restricted the aggregation of HCPs layers on graphene nanosheets and thus dramatically enlarged the specific surface area. As a result, the enhanced performance of HCP-TiO 2 -FG cannot support the argument that the catalytic sites are located on TiO 2 . As the HCP-FG material also exhibits a photocatalytic activity for CO and CH 4 production, it remains elusive whether the catalytic sites are located on TiO 2 or HCP-FG.

Response:
We have also explained the identification of catalytic sites in our previous response to R1 (response to the Comment 6# of Reviewer 2#). Given below is further explanation to clarify it.
Let's first compare the porous property and photocatalytic performance to clarify the location of catalytic sites. The results of HCP-FG, HCP-TiO 2 -FG, and TiO 2 /HCP-FG samples are taken from the manuscript and supplementary information, as shown in the Fig. R1 and Table R1 below. Obviously, the introduction of porous HCPs layers enriched the adsorptive sites to achieve the high CO 2 uptake and improved the visible light absorption. Thus the formation of well-defined HCP-TiO 2 -FG sandwich structure resulted in much higher photocatalytic CO 2 reduction rate. The HCP-FG material also exhibited broad visible-light absorption, high surface area and notable CO 2 uptake. However, its photocatalytic performance was far less than that of HCP-TiO 2 -FG, especially in the eight-electron reduction to CH 4 . These results show that the catalytic sites on TiO 2 are much more active for CO 2 reduction than those on HCP-FG. Besides, the catalytic sites can be further clarified by the comparison between HCP-FG and TiO 2 /HCP-FG hybrid. When TiO 2 crystals were supported on HCP-FG to form TiO 2 /HCP-FG hybrid, it is found that TiO 2 deposition blocked most of the adsorptive sites of HCP-FG and resulted in a dramatic decrease to less than one-third of CO 2 uptake. However, the CH 4 production over TiO 2 /HCP-FG was 7.4 times more than that over pristine HCP-FG. These results confirm that the TiO 2 crystals are introduced as catalytically active sites to facilitate the photocatalytic CO 2 conversion.
Second, the pathway of charge carriers transfer and separation was studied to clarify the location of catalytic sites. As shown in Fig. 4f, the lowest unoccupied molecular orbital (LUMO) level of HCP-FG is more negative than the conduction band (CB) level of TiO 2 . It is well-known that the polymer materials usually possess the excitons with high binding energy, which usually recombine at the excited states.
The photogenerated electrons of the excited HCP-FG can migrate to the CB of TiO 2 due to their matched energy levels (Type II heterojunction model: Semiconductors 1998, 32, 1; Angew. Chem., Int. Ed. 2012, 51, 10145;ACS Catal. 2014, 4, 3637;Energy Environ. Sci. 2015, 8, 731;Adv. Mater. 2017, 29, 1606198). Thus the photogenerated carriers of HCP-FG can be separated at the interface with TiO 2 , which largely reduced the recombination loss. Since the electrons are located on TiO 2 crystals, the adsorbed CO 2 molecules are more readily converted to CO than CH 4 at the catalytic sites of TiO 2 .
Based on the discussion above, it can be inferred that the CO 2 reduction better achieved at the catalytic sites on TiO 2 rather than those on HCP-FG. The identification of catalytic sites could better explain the difference in photocatalytic performance among HCP-FG, HCP-TiO 2 -FG and TiO 2 /HCP-FG.
We have also mentioned the above discussion at page 9 and 11 of the manuscript. Pore size distribution that calculated using DFT methods (slit pore models, differential pore volumes). Time-dependent production of CH 4 (d) and CO (e) in photocatalytic CO 2 reduction with different catalysts under visible-light (λ≥420 nm). (f) Average efficiency of photocatalytic CO 2 conversion with different catalysts during 5 h of visible-light (λ≥420 nm) irradiation. uptake determined volumetrically using a Micromeritics ASAP 2020 M analyzer at 1.00 bar and 273.15 K. c Pore volume calculated from nitrogen isotherm at P/P 0 =0.995, 77.3 K. d,e average of gas evolution rate (r) during 5 h of photocatalytic CO 2 reduction. f R e is the rate of total consumed electron number for the reduced product; R e = 8r(CH 4 )+2r(CO).

Reviewers' comments:
Reviewer #2 (Remarks to the Author): I still think that the key issue of structure-performance relationship remains ambiguous, which makes the manuscript insufficient to meet the level for publication in Nature Commun. If the authors could address the comments, I would recommend it for publication.
Specific comments are listed below: 1. The advantage of the "sandwich structure" is still not clear. As described by the authors, the catalytic sites are located on TiO2 while the adsorptive sites are on HCP, and the graphene improve the photogenerated electrons transfer from HCP to TiO2. In this case, the best structure may be TiO2-FG-HCP rather than HCP-TiO2-FG.
2. The authors demonstrated that the photogenerated carriers of HCP-FG can be separated at the interface with TiO2. However, according to the "sandwich structure", there is no interface between HCP-FG and TiO2. Instead, the interface should be formed between FG and TiO2 or HCP and TiO2. In addition, according to the "HCP-TiO2-FG sandwich structure", the TiO2 intercalation somewhat restricted the aggregation of HCPs layers on graphene nanosheets, but also blocked the charge transfer from HCP to TiO2 through graphene.
3. In the "sandwich structure", the catalytic sites are located on TiO2 while the adsorptive sites are on HCP, as stated by the authors. This requires a short diffusion length for the transfer of CO2 molecules from adsorptive sites to catalytic sites. However, there is no direct evidence to support this argument.
sites with TiO 2 crystals loading. Considering that this name caused confusion to the reviewer, we have replaced it by the "porous HCP-TiO 2 -FG composite". For better understanding, we have now revised the models in the diagram to show the interface between TiO 2 and HCP-FG (Fig. 1). As you will see, the photogenerated electron-hole pairs of the excited HCP-FG can move throughout the HCP-FG structure and then be separated at the interface with TiO 2 via their interfacial interaction, as depicted in Supplementary Fig. S31.  The proposed structure was previously demonstrated by TEM, STEM, HRTEM, SEM and AFM observations. In order to display the spatial distribution of HCP layers, we conducted further characterization by high-angle annular dark field (HAADF) mapping and three-dimensional rotating techniques. The results are now presented in Fig. 2f-i, Supplementary Fig. S6 and Supplementary Video, which could verify the above porous HCP-TiO 2 -FG composite structure. The detailed descriptions of the structure have been added in the morphology characterization section. We are hoping that the structure is now much clear and clearly understandable.
As suggested, another type of composite, HCP-FG supported TiO 2 (TiO 2 /HCP-FG), has already been prepared for comparison in this work. The HCP layers were hypercrosslinked on the functionalized graphene to form the HCP-FG structure with graphene surface rarely exposed, and then the HCP-FG was used as a supporting material for TiO 2 crystals growth during the solvothermal process. For better understanding of structure, we present the models in the diagram below (Fig. R1).
Model ( Table R1 below. Based on the above discussions, when TiO 2 was supported on HCP-FG instead of graphene, the TiO 2 deposition blocked most of the adsorptive sites of HCP-FG and resulted in a dramatic decrease in surface area, CO 2 uptake and photocatalytic efficiency. As a result, the designed HCP-TiO 2 -FG composite structure appears to be much superior to the TiO 2 /HCP-FG composite in our system. Of course, we expect that, in future, the synthesis strategy can be further improved to yield a much better structured combination of microporous organic polymers with photocatalysts.
We have also mentioned the above discussions in the revised manuscript and supplementary information.
On Pages 4, "The elemental mapping images in Fig. 2f- Fig. S14) (a-c and g-h). Therefore, the TiO 2 crystals on the graphene sheets were not exposed outside but encapsulated by the ultrathin HCPs layer."  Pore size distribution that calculated using DFT methods (slit pore models, differential pore volumes  2. The authors demonstrated that the photogenerated carriers of HCP-FG can be separated at the interface with TiO 2 . However, according to the "sandwich structure", there is no interface between HCP-FG and TiO 2 . Instead, the interface should be formed between FG and TiO 2 or HCP and TiO 2 . In addition, according to the "HCP-TiO 2 -FG sandwich structure", the TiO 2 intercalation somewhat restricted the aggregation of HCPs layers on graphene nanosheets, but also blocked the charge transfer from HCP to TiO 2 through graphene.

Response:
Since the proposed structure of HCP-TiO 2 -FG has been clarified in the above response, we are hoping that the interface between TiO 2 and HCP-FG could be clearly understandable now. As shown in Fig. 1, the TiO 2 -FG interface was first formed that served as "bridge" to favor the formation of interface between TiO 2 and HCP-FG. It should be noted that the HCP was covalently linked with FG for the formation of HCP-FG integrated structure instead of freestanding HCP blocks. Hence the interface we observed in Fig. 1-2 was actually the interface between TiO 2 and HCP-FG. In the absence of TiO 2 , the surface area and CO 2 uptake of HCP-FG were much lower than those of HCP-TiO 2 -FG (Supplementary Tab. S2). It was also found that the HCP-FG layers of TiO 2 /HCP-FG were much thicker than those of HCP-TiO 2 -FG ( Fig. 2 and Supplementary Fig. S13). The results implied that the TiO 2 intercalation somewhat restricted the aggregation of HCPs layers on graphene nanosheets. According to the charge separation pathway in Figure 4f, the photogenerated electron-hole pairs of the excited HCP-FG can move throughout the HCP-FG structure and then be separated at the interface with TiO 2 via their interfacial interaction. In the prerequisite of better understanding the structure, we can conclude that the graphene plays dual a role in the composite photocatalyst: (1) the interaction with TiO 2 serving as "bridge" to form well-defined porous HCP-TiO 2 -FG composite structure; (2) the covalent linking with HCPs improving the electronic conductivity of the HCPs and thus facilitating the electron transfer from HCP-FG to TiO 2 in the composite. For better understanding, here we would like to present a simple diagram illustrating the charge separation at the interface (Supplementary Fig. S31). We believe that given the above discussion, the query regarding "TiO 2 blocked the charge transfer from HCP to TiO 2 through graphene" is now better addressed.
Given below is the related explanation in the manuscript.
On Page 7, "The results implied the interaction of TiO 2 with graphene serving as "bridge" to the formation of well-defined HCP-TiO 2 -FG composite structure.
On Page 11, "the covalent linking with graphene effectively improves the electronic conductivity of the HCPs and thus facilitates the electron transfer in the composite.
The less efficient CH 4 production over HCP-TiO 2 photocatalyst can also reflect the influence of graphene on improving the charge separation efficiency ( Supplementary   Fig. S27) 3. In the "sandwich structure", the catalytic sites are located on TiO 2 while the adsorptive sites are on HCP, as stated by the authors. This requires a short diffusion length for the transfer of CO 2 molecules from adsorptive sites to catalytic sites.
However, there is no direct evidence to support this argument.

Response:
In the HCP-TiO 2 -FG composite, the HCP outer layers serve as adsorptive sites for CO 2 molecules and the TiO 2 crystals function as catalytic sites for CO 2 reduction. We agree that the diffusion of CO 2 molecules from the adsorptive sites to the catalytic sites plays an important role in CO 2 conversion.
(1) The evidence on the role of diffusion The kinetic study of CO 2 adsorption and diffusion provided a convincing evidence to clarify the significant role of diffusion in such a gas-solid reaction system. The related discussion is available at pages 12-13. More detailed descriptions can also be found in the response letter of R2 (response to the Comment of Reviewer #3). In the following peer-review process, the Reviewer #3 made comments on our revision as "The careful revision of the authors is acknowledged. With regard to kinetic analysis and mass transfer, all parameters reasonable well accessible by experiments were investigated and a careful discussion was added. The presented additional data appear comprehensive." For your convenience, we have pasted the previous response at the bottom as reference.
(2) The evidence on the effect of diffusion length Since the gas diffusion has a crucial influence on the rate limiting for CO 2 conversion, an appropriate diffusion length should be beneficial for the diffusion of CO 2 molecules from the adsorptive sites to the catalytically active sites. According to the structural analysis, the TiO 2 crystals were encapsulated by ultrathin HCP outer layers with a thickness of 3~8 nm (Fig. 2), indicating a short diffusion length for CO 2 molecules diffusing from HCP-FG to TiO 2 photocatalysts. For better understanding, we present the CO 2 diffusion model in the diagram below (Supplementary Fig. S15). Figure S15. Diagram of CO 2 diffusion from adsorptive sites to catalytic sites for conversion.
In addition, we have prepared the HCP-TiO 2 -FG composite with different thickness of HCP outer layers to study the effect of the diffusion length on the CO 2 uptake and conversion efficiency. The results suggested that there is an appropriate thickness of HCP layers which balanced the CO 2 adsorption and diffusion. In this work, we tried to construct the HCP-TiO 2 -FG sandwich structure with ultrathin HCP layers to reduce the diffusion length. Of course, we expect that, in the future, the synthesis strategy can be further improved to yield a much better structure with more efficient CO 2 adsorption and diffusion.
Given below is the related explanation in the manuscript.
On Page 9, "The model of CO 2 diffusion and conversion was presented in Supplementary Fig. S15." On Page 9-10, "the HCPs obtained by this strategy are comprised of ultrathin layers with a thickness of 3~8 nm wrapping around TiO 2 crystals (Fig. 2d-f) (Supplementary Fig. S16), however the size of TiO 2 particles was decreased accompanied by the thickening of the HCP layers (Supplementary Fig.   S17), which suggests that the HCP outer layers effectively suppress the growth of TiO 2 crystals. The distinct thickening of the outer layers was further verified from the characteristic morphology revealed in Fig. 1 Fig. S18a and Tab. S4), on the other hand, the diffusion length of CO 2 molecules also increased due to the thickening of the outer layer. As the CO 2 conversion efficiency increased initially and then decreased at higher amount (Supplementary Fig. S18b), there may be an appropriate thickness of HCP layers that balance the CO 2 adsorption and diffusion."   Reference:

Query on diffusion from Reviewer #3
The authors were asked to provide further evidence for the governing nature of their two main Arguments, namely the superior CO 2 adsorption as pre-requisite of high substrate concentration close to or on the active site and their hypothesis on the shortend diffusion lengths. Indeed, a suitable kinetic analysis has been carried out illustrating that the investigated photocatalytic CO 2 reduction is not under intrinsic kinetic control of the catalyst but the observed rates of CO and CH 4 formation are rather determined by some transport effects. The observed limited dependence of rate on temperature may hind towards diffusion control. I fully agree that kinetic analyses is a yet under-represented aspect in photo-catalysis. Though, the little rate dependence on temperature appears to point towards film diffusion effects. Hast the stirring speed or the main particle size been varied? It remains unclear for me why the authors conclude on a reduced Diffusion length for the Optimum System, although they do not know which Diffusion is rate limiting: CO 2 from bulk to the film, through the fild, in/on the porous material or Charge charrier diffusion?
Response: Many thanks for appreciating our efforts on the kinetic analysis presented in the previous revision. The photocatalytic reactions were carried out in a gas-solid batch system without stirring the gas mixture. According to the suggestion, we studied the diffusion process by varying the stirring speed from zero to maximum. As shown in Supplementary Fig. S40, the increase in stirring speed greatly facilitates the photocatalytic conversion of CO 2 to CH 4 product, indicating that the diffusion plays a significant role in such a gas-solid reaction system. The related results and discussions have been added as a separate section "kinetic analysis" in the revised manuscript.
The particle size is another factor affecting the diffusion process, especially for internal diffusion. We would like to mention that the HCP-TiO 2 -FG composite produced through this strategy possesses the specific particle size. In the experience of optimizing synthesis conditions, TiO 2 crystals with larger or smaller size are inclined to form agglomerates, which cannot uniformly decorate on graphene or be fully wrapped by ultrathin HCPs layers. In the current system, we cannot achieve the varied particle size while keeping the designed structure with similar surface area, CO 2 uptake capacity and the number of active sites.
In order to reduce the diffusion length, we tried to construct the HCP-TiO 2 -FG sandwich structure with ultrathin HCP layers. The reduced diffusion length should be beneficial for the diffusion of CO 2 and photo-generated charges to the catalytically active sites, both of which may contribute to the improved CO 2 conversion efficiency.
Based on the gas diffusivity measurement, the diffusion coefficient of CO 2 in the HCP-TiO 2 -FG sandwich structure is calculated to be 1.8×10 -11 cm 2 s -1 at room temperature ( Supplementary Fig. S39). Typical polymers for photoconversion application, the exciton diffusion and charge transfer dynamics of poly(3-hexyl thiophene) (P3HT) and phenyl-C61-buryric acid methyl ester (PCBM) have been studied, giving the diffusion coefficient of 1.8×10 -3 and 2.7×10 -4 cm 2 s -1 , respectively (Adv. Mater. 2008, 20, 3516;Nanoscale 2011Nanoscale , 3, 2280. The diffusion of charge carrier is much faster than that of the adsorbed gas by several orders of magnitude, implying that the gas diffusion should have more crucial influence at the rate limiting rather than the charge carriers.
In the present study, we aim to provide new insights into the design and synthesis of well-defined porous photocatalysts for CO 2 uptake and conversion. The designed sandwiched structure is somewhat complicated and makes it impossible to establish individual kinetic models. Although the kinetic mechanism cannot be clearly clarified in the current study, we indeed have made much progress on the kinetic analysis of photocatalytic CO 2 conversion that is seldom discussed in the literature. Inspired by the reviewers' suggestions, we have realized that the kinetic analysis is indeed very important for photocatalytic CO 2 conversion and there are yet many challenges that need to be addressed. In future study, we will use the simple photocatalytic system such as pure TiO 2 photocatalyst to probe the kinetics model and reaction mechanism of CO 2 conversion, and then extend to the complicated models involving adsorption, diffusion and photocatalytic processes.
On pages 12-13, all results and discussions on kinetic analysis are included and displayed as a separate section "kinetic analysis" in the revised manuscript.
"Kinetic analysis. The kinetics experiments were carried out to understand the contribution of CO 2 adsorption and diffusion to the enhancement of photocatalytic efficiency. The relationship between the CO 2 adsorption and CH 4 production can be explored by varying the surface coverage of CO 2 on the active sites. The partial pressure of CO 2 is adjusted in CO 2 /N 2 mixture because of a high CO 2 /N 2 selectivity ratio of 25.8 over the HCP-TiO 2 -FG photocatalyst (Supplementary Fig. S36). Since the kinetic model and reaction mechanism of photocatalytic CO 2 conversion are ambiguous so far, the quantitative relationship between CO 2 coverage and CH 4 evolution rate is still unclear. Interestingly, it is observed that they show a similar trend of increase with CO 2 proportion, e.g. both of them dramatically increased at lower partial pressure and then displayed a slow increase at higher CO 2 concentration (Supplementary Fig. S37). Generally, the reaction rates that are normalized to the active sites allow the direct comparison of intrinsic reactivity on different catalysts [47][48][49] . For the catalytic system employing same catalyst, the reaction rate appears to be independent of the loading amount of catalyst after normalization to the same amount 50,51 . In this regard, the porous HCP-TiO 2 -FG photocatalyst possesses equivalent catalytic active sites to TiO 2 /HCP-FG due to the same content of TiO 2 photocatalyst. That is, the more efficient CH 4 production over HCP-TiO 2 -FG should not result from the difference in the number of catalytic sites but mostly come from the higher surface coverage of CO 2 on the active sites.
The temperature has a complicated influence on the rate of photocatalytic conversion from the aspects of adsorption and diffusion. By increasing the temperature, the surface coverage of CO 2 molecules on the catalyst surface was decreased due to the exothermic effect of adsorption process (Fig. 3e-f), while the diffusion rate was increased as a result of the increased thermal motion of CO 2 molecules (Supplementary Fig. S38). Based on Arrhenius plot, the adsorption activation energy for CO 2 adsorption is calculated to be 5.20 kJ mol -1 (Supplementary Fig. S39a) (Supplementary Fig. S39b). The diffusion process was further studied by varying the stirring speed. As shown in Supplementary Fig. S40