Reversible modulation of interlayer stacking in 2D copper-organic frameworks for tailoring porosity and photocatalytic activity

The properties of two-dimensional covalent organic frameworks (2D COFs), including porosity, catalytic activity as well as electronic and optical properties, are greatly affected by their interlayer stacking structures. However, the precise control of their interlayer stacking mode, especially in a reversible fashion, is a long-standing and challenging pursuit. Herein, we prepare three 2D copper-organic frameworks, namely JNM-n (n = 7, 8, and 9). Interestingly, the reversible interlayer sliding between eclipsed AA stacking (i.e., JNM-7-AA and JNM-8-AA) and staggered ABC stacking (i.e., JNM-7-ABC and JNM-8-ABC) can be achieved through environmental stimulation, which endows reversible encapsulation and release of lipase. Importantly, JNM-7-AA and JNM-8-AA exhibit a broader light absorption range, higher charge-separation efficiency, and higher photocatalytic activity for sensitizing O2 to 1O2 and O2•− than their ABC stacking isostructures. Consequently, JNM-8-AA deliver significantly enhanced photocatalytic activities for oxidative cross-coupling reactions compared to JNM-8-ABC and other reported homogeneous and heterogeneous catalysts.

1. Figure 1: the bond between Cu and N should be coordination bond, so please use dash line instead of solid line.2. Can the authors explain why JNM-9-ABC has a ABC stacking directly instead of having AA stacking?3. Figure 2; the colors in Figure 2 are very difficult to distinguish.I advise to change the colors of these plots to make them clearer.4. Figure 3: HRTEM measurements are much more important than SEM and low-resolution TEM, but in Figure 3, the authors only present the TEM images with crystal lattice in insets, which is very confusing.Please provide HR TEM data and present the electron diffraction mode. 5. Figure 4: how do the authors rule out the possibility that the topology of COFs changed rather than the change of layer stacking?6.The N2 sorption capacity of these materials are extremely low for all COFs, which might be resulted from the low crystallinity of these materials.Please improve the quality of these COFs.7. Since the N2 sorption is so low, is it accurate to determine the pore size distribution vias such low sorption isotherms?8.In Figure 4d, the difference of 2.26 and 1.2 is 1, and the difference of 2.95 and 2.26 is 0.7.Is it hard to say which is closer based on so minor difference.This is the same for JNM-8.Another question is that why does the pore size not go back to the original size?Is this because of partial transformation?If so, why are there no two peaks for the recovered samples?Please explain this.9.The DFT calculation is very unreasonable.The DMF molecule in AA-stacking model is so close to the skeleton, while the DMF molecule in ABC-stacking model is adjacent to the skeleton.It is very apparent that closer distance will definitely result in higher energy levels.Additionally, the calculation does not make too much sense as these results are based on theoretical models.All DFT researchers know that ABC stacking will lead to higher energy levels than AA stacking regardless the materials, so the conclusion is very obvious and not surprising.10.SEM EDS is not as accurate as TEM.Since the authors can get TEM measurements, why not use TEM EDS for elemental mapping?11.Please provide elemental analysis data for these COFs in this work.12.What is the possible mechanism for the claimed layer stacking?Is there any proof for this mechanism?Please clarify.
13.I noticed that this manuscript has many typos or grammar issues.Please check through very carefully and improve the readability of this work.
Reviewer #2 (Remarks to the Author): The authors have prepared three 2D COFs composed from Cu-CTU and three di-amine linkers with different length.Interestingly, the interlay stacking structure of JNM-7 and JNM-8 can be reversibly modulated, leading to the reversible encapsulation and release of lipase.Due to the alteration of stacking, AA stacking structure exhibited better photocatalytic activity for oxidative cross-coupling reaction.Obviously, it is very hard to achieve reversible structure transformation and obtain permanent AA and ABC structure with the identical components.Thus, these results are of great importance and deserve publication in such a reputed journal as Nat.Comm.Moreover, the work is solid, the materials are thoroughly characterized, and structural transformation mechanism is well investigated.I think it is a very good addition to COF and MOF area and am happy to see it published after addressing some points: 1.The ABC stacking with R3 space group should be trigonal, not hexagonal.2. For the reversible interlayer structure transformation, did author try other solvent instead of DMF? 3. From the BET analysis, it seems the as synthesized JNM-7-AA have a smaller pore at 1.20 nm, did this can be attributed to ABC stacking model? 4. The For elemental analysis of JNMs, the calculated and observed ratio are not in good agreement with each other.It's better to re-check.
Reviewer #3 (Remarks to the Author): In this paper, the authors reported the synthesis of a series of CuOFs, and their reversible structural transformation triggered by acid or heat.Due to the modulation of interlayer stacking model, the porosity and photocatalytic performance can be tuned.The materials and properties were convincingly and systematically studied including crystal structure, porosity, stability, host-guest chemistry and catalytic performance.In addition, the authors extensively investigated the reversible structural transformation and its mechanism.This paper is interesting and would be a valuable contribution to the field.Therefore, this reviewer recommends its publication in Nat.Comm., after minor revisions after the following comments are addressed.
1.The space group "P6/M" should be changed to "P6/m".2. Graphics and texts are inconsistent.In Figure 5a, it shows that LP@JNM-8-AA was heated in DMF to release the LP.But in manuscripts, "…..after centrifugation, the collected powder of LP@ JNM-8-AA was added to phosphate buffer solution and the resulting mixture was heated at 80°C for 6 h.Lipase …..(Figs.5b, and 5c).".Please check and revise.3.In the Supplementary Table 12, did the author calculate the TOF based on one Cu ion or one Cu-CTU (three Cu ions)? 4. In the Supplementary Figure 35, it should be the signal of 1O2 instead of O2•−. 5.The theoretical and experimental values of the elemental analysis of JNM-7-AA, JNM-8-AA and JNM-9-ABC are quite different.
significantly decreased.The only difference in this manuscript is that the authors studied this problem based on two copper-containing COFs.Finally, there are several confusing parts in this manuscript which significantly decrease the soundness of the conclusion and need to be addressed.Therefore, I do not recommend it for publication in Nature Communications.It can be transferred to other journals after addressing the following concerns.
Author's response: We thank this reviewer for reviewing our manuscript and for raising important questions to improve our work.
Therefore, compare to their work, we can obtain the stable AA and ABC structure with identical components in a reversible manner (See Fig. R1b), allowing us to further reveal the interlayer stacking induced properties change such as adsorption and photocatalysis.
Overall, we would like to emphasize the importance and novelty of our work as follows: 1.Although Zhao's group has reported the guest-triggered reversible interlayer shifting, the resulting quasi-AA or -AB stacking structure are not stable when the guests were removed.In addition, no ABC stacking structure can be achieved.JNM-7 and JNM-8 are the first reported examples that can exhibited reversible interlayer shifting and the AA or ABC stacking isomer can preserve in the absence of guests.
To further clarify this point, a schematic demonstration is shown in the Fig. R1. 2. Owing to reversible interlayer shifting, the reversible encapsulation and release of enzyme can be achieved, which have never been achieved before in the field of COFs.We believed that these unique properties would be applied in the drug delivery systems.
3. Although photocatalytic application of COFs has been extensively studied in the area, in our work, we intent to systematically study the relationship between stacking structure (i.e., AA and ABC isomers) and photocatalytic activities.Importantly, we have been found that AA stacking structure features broader light absorption range, higher charge-separation efficiency and higher photocatalytic activity for CDC reaction than their ABC stacking isostructure.These observations are different from the reported COFs (ACS Appl. Mater. Interfaces 2021, 13, 29471−29481), indicating the introduction of Cu-CTU might be bring new functions into COFs.Such understanding could be significantly important for designing highly efficient COF-based photocatalysts.Moreover, JNM-7-AA delivered a high TOF for CDC reaction, which is much faster than many other reported homogeneous and heterogeneous catalysts.
1. Figure 1: the bond between Cu and N should be coordination bond, so please use dash line instead of solid line.
2. Can the authors explain why JNM-9-ABC has a ABC stacking directly instead of having AA stacking?
Author's response: For mesoporous systems, due to their high surface area, it is expected to have large interfacial energy, leading to a natural trend of minimizing their free energy by closing energetically unfavorable pores (ref. Macromol. rapid comm. 2004(ref. Macromol. rapid comm. , 25, 1487(ref. Macromol. rapid comm. -1490)).Therefore, in the field of COFs, the increases of the length of linker did not always produce the large pore size (See ref. Chem. Sci., 2019, 10, 4293).Moreover, although the theoretical calculation can be used to reveal the more energetically favorable stacking models, the change of synthetic condition often alters the interlayer Specifically, change the solvent from n-butyl alcohol to dioxane, the AA stacking structure will be obtained instead of AB stacking structure.Zhang's group (See ref. Small 2023, 2303684) have been prepared NKCOF-11-AA and NKCOF-11-ABC by alteration of synthetic method, although the total stacking energy calculation indicated that AA stacking modes (140.18kcal mol −1 ) was more energetically favorable than that of ABC stacking (83.04 kcal mol −1 ).Overall, it is still very hard to precisely predict the interlayer stacking structure in 2D COFs, since the DFT calculation only can be conducted without consideration of solvent and catalysts.
Nevertheless, we also conducted the DFT calculation to evaluated the total stacking energy of JNM-9.As shown in Table R1, the total stacking energy of AA (105.81 kcal mol −1 ) and ABC stacking (60.95 kcal mol −1 ) was much higher than AB stacking (39.13 kcal mol −1 ).Therefore, without consideration of solvent (i.e., DMF, o-DCB and n-BtOH) and catalysts (i.e., TFA), the AA stacking structure is the most energetically favorable.But, we have tried several synthetic conditions and AA stacking structure was not obtained.
3. Figure 2; the colors in Figure 2 are very difficult to distinguish.I advise to change the colors of these plots to make them clearer.
Author's action: We thank this reviewer for pointing out this issue, and we have changed the colors of these plots to make them easier distinguish in Figure 2. 4. Figure 3: HRTEM measurements are much more important than SEM and low-resolution TEM, but in Figure 3, the authors only present the TEM images with crystal lattice in insets, which is very confusing.Please provide HR TEM data and present the electron diffraction mode.
Author's response: We thank this reviewer's suggestion and we revised Figure 3 as shown here as Fig. R2.The high-resolution transmission electron microscopy (HR-TEM) and Fourier transform (FFT) of JNM-7-AA, JNM-8-AA, and JNM-9-ABC demonstrated that the well-ordered lattice fringe with dspacing of 3.10, 3.70, and 3.10 nm, corresponding to the lattice planes of ( 100), ( 100) and (110), respectively (Figs. 3b,3d,and 3f).This result is in good agreement with their refined PXRD pattern.Author's response: It is well-known that the honeycomb nets can be obtained by imine condensation between monomers with C3 and C2 symmetry (See Chem. Rev., 2020, 120, 8814−8933).More importantly, the PXRD pattern of JNMs exhibited characteristic peaks that are similar to 2D honeycomb structure.Recently, a 3D COF have been prepared from 2D honeycomb COF through inclined interpenetration (see J. Am.Chem. Soc. 2023, 145, 13537), however, the PXRD of 3D COF featured much more complicated peaks compared to 2D COF.In addition, the BET analysis also revealed narrow pore size distribution of JNMs that similar to 2D honeycomb structure.Combining all these evidences, we believe it is the alteration of interlayer stacking rather than topology change.
6.The N 2 sorption capacity of these materials are extremely low for all COFs, which might be resulted from the low crystallinity of these materials.Please improve the quality of these COFs.8.In Figure 4d, the difference of 2.26 and 1.2 is 1, and the difference of 2.95 and 2.26 is 0.7.Is it hard to say which is closer based on so minor difference.This is the same for JNM-8.Another question is that why does the pore size not go back to the original size?Is this because of partial transformation?
If so, why are there no two peaks for the recovered samples?Please explain this.
Author's response: Comment 6 and 8 are highly related, therefore, we reply to these comments together.
Thank you very much for your comments.The BET surface areas of porous materials are not only related to their crystallinity but also highly affected by the activation processes.In addition, in our work, due to the incorporation of Cu-CTU, the density of CuOF will be higher compared to corresponding COF with similar pore size.During the structure transformation, the heating and magnetic stirring are required, which will reduce the crystallinity and particle size of samples (Supplementary Fig. 24.) and might block the pore of JNMs.Therefore, the relatively low BET surface areas and smaller pore size distribution were observed.
To improve the quality of samples, we carefully washed the samples by soxhlet extraction with various solvent, and then, the sample were activated by supercritical carbon dioxide before the N 2 sorption measurements.According to this reviewer's suggestion, we have retired the N 2 sorption experiment and the results have been shown in the Fig. R3.It clearly showed that the BET surface areas of all CuOF are largely improved.More importantly, their pore size distributions are more reasonable.
For instance, JNM-7-AA (pristine) and JNM-7-AA (regenerated) exhibited closer pore size distribution (3.4 and 3.0 nm), which is much more difference with the JNM-7-ABC with a pore size distribution of 1.4 nm.Same for JNM-8.After improve the quality of sample, the pore size of JNM-7 and JNM-8 can almost go back to the original size (3.4 and 3.0 nm for JNM-7-AA pristine and regenerated samples, and 4.0 and 3.9 nm for JNM-8-AA pristine and regenerated samples).7. Since the N 2 sorption is so low, is it accurate to determine the pore size distribution via such low sorption isotherms?
Author's response: Thank you very much for your comment.Taking the JNM-7-ABC as an example to explain how we determine the pore size distribution.JNM-7-ABC showed a Type-I microporous sorption isotherm, and the BET surface area was calculated to be 157.9m 2 g −1 by using the BEL-Master software (Fig. R4).To accurately determine the pore size distribution of JNM-7-ABC, we chose the nonlinear density functional theory (NLDFT) slit-pore model.As shown in Fig. R4, the experimental N 2 adsorption isotherm of JNM-7-ABC was in good agreement with the adsorption isotherms simulated by the NLDFT slit-pore model, demonstrating the reliability of the pore size distribution.The calculated pore size of JNM-7-ABC is centered at 1.4 nm, which was close to the simulated pore size (1.3 nm) for the eclipsed ABC stacking model.apparent that closer distance will definitely result in higher energy levels.Additionally, the calculation does not make too much sense as these results are based on theoretical models.All DFT researchers know that ABC stacking will lead to higher energy levels than AA stacking regardless the materials, so the conclusion is very obvious and not surprising.
Author's response: Due to the larger pore size in AA-stacking, after optimization, the DMF molecule is far away to the frameworks with a distance larger than 10 Å, while it is close to the skeleton in ABC stacking model with distance ranged from 2.297 to 5.672 Å.As the reviewer suggested, the closer distance will lead to higher energy levels.Therefore, the DFT calculation suggests that the energy input is necessary for the interlayer structural transformation process, which is consistent with the experimental observations.Specifically, JNM-7-AA in DMF cannot transfer to JNM-7-ABC without heating and stirring.
Notably, the AA stacking is NOT always more energetically favorable than ABC stacking.For instance, Cui and co-authors have found that the total crystal stacking energy of ABC stacking (116.52 kcal mol −1 ) for COF1-i Pr is much higher than those of the AA (44.67 kcal mol −1 ) and AB (70.33 kcal mol −1 ) stacking, and the PXRD of COF1-i Pr confirmed it had ABC stacking model (See ref. J. Am. Chem. Soc. 2018, 140, 16124−16133).Therefore, it should be beneficial to conducted the DFT calculation for further understanding the structural transformation.Author's action: We have provided the elements analysis data and revised in manuscript and pages 4-5 of Supplementary.
12. What is the possible mechanism for the claimed layer stacking?Is there any proof for this mechanism?Please clarify.
Nevertheless, in our case, we assumed that the DMF molecules have disturbed the Cu-Cu interactions between layers, leading to the AA stacking transfer to ABC stacking.To further support our assumption,  2020, 26, 4350 -4377, Dalton Trans., 2011, 40, 10742-10750).These results implied that the solvent   13.I noticed that this manuscript has many typos or grammar issues.Please check through very carefully and improve the readability of this work.
Author's action: We thank this reviewer's suggestion, and we have carefully checked and polished the English.
The authors have prepared three 2D COFs composed from Cu-CTU and three di-amine linkers with different length.Interestingly, the interlay stacking structure of JNM-7 and JNM-8 can be reversibly modulated, leading to the reversible encapsulation and release of lipase.Due to the alteration of stacking, AA stacking structure exhibited better photocatalytic activity for oxidative cross-coupling reaction.
Obviously, it is very hard to achieve reversible structure transformation and obtain permanent AA and ABC structure with the identical components.Thus, these results are of great importance and deserve publication in such a reputed journal as Nat.Comm.Moreover, the work is solid, the materials are thoroughly characterized, and structural transformation mechanism is well investigated.I think it is a very good addition to COF and MOF area and am happy to see it published after addressing some points: Author's response: We thank this reviewer for finding our manuscript worthy of publishing in Nature Communications and for raising some questions to improve our work.
1.The ABC stacking with R3 space group should be trigonal, not hexagonal.
Author's action: We thank this reviewer for pointing out this issue, and we have changed ABC tacking with hexagonal R3 space group to trigonal R3 space group.
2. For the reversible interlayer structure transformation, did author try other solvent instead of DMF?
Author's response: We thank this reviewer for pointing out this issue.Please see the response for Reviewer 1's Comment 12.
3. From the BET analysis, it seems the as synthesized JNM-7-AA have a smaller pore at 1.20 nm, did this can be attributed to ABC stacking model?
Author's response: We thank this reviewer for pointing out this issue, and we simulated the JNM-7-ABC structure and measured its pore size to be ∼1.40 nm, which is similar to the experimentally measured pore size (∼1.30nm) after the JNM-7-AA layer-stacking structure conversion.This further demonstrates the success of the layer stacking transformation.
4. The For elemental analysis of JNMs, the calculated and observed ratio are not in good agreement with each other.It's better to re-check.
Author's action: We have retried elements analysis and provided in manuscript and SI.
In this paper, the authors reported the synthesis of a series of CuOFs, and their reversible structural transformation triggered by acid or heat.Due to the modulation of interlayer stacking model, the porosity and photocatalytic performance can be tuned.The materials and properties were convincingly and systematically studied including crystal structure, porosity, stability, host-guest chemistry and catalytic performance.In addition, the authors extensively investigated the reversible structural transformation and its mechanism.This paper is interesting and would be a valuable contribution to the field.Therefore, this reviewer recommends its publication in Nat.Comm., after minor revisions after the following comments are addressed.
Author's response: We also thank this reviewer for supporting our manuscripts published on Nature Communication.
Author's action: We thank this reviewer for pointing out this issue, and we have changed the space group "P6/M" to "P6/m in the manuscript and Supplementary.
2. Graphics and texts are inconsistent.In Figure 5a, it shows that LP@JNM-8-AA was heated in DMF to release the LP.But in manuscripts, "…..after centrifugation, the collected powder of LP@ JNM-8-AA was added to phosphate buffer solution and the resulting mixture was heated at 80°C for 6 h.
Author's action: We thank this reviewer for pointing out this issue, and we have modified Figure 5a and "…the collected powder of LP@ JNM-8-AA was added to phosphate buffer solution and the resulting mixture was heated at 80°C for 6 h…" have changed to "…the collected powder of LP@ JNM-8-AA was added to phosphate buffer solution and DMF solution and the resulted mixture was heated at 80 °C for 6 h…" in the manuscript.
3. In the Supplementary Table 12, did the author calculate the TOF based on one Cu ion or one Cu-CTU (three Cu ions)?
Author's action: The TOF based on one Cu-CTU (three Cu ions) and have explained in Supplementary

Fig. R1 .
Fig. R1.The schematic demonstration for tuning the interlayer stacking a) previous works, and b) this work.

Fig. R4 .
Fig. R4.(a) N 2 adsorption and desorption isotherms of JNM-7-ABC at 77 K. (b) BET plot for surface area calculation.(c) The experimental and simulated N 2 adsorption isotherm of JNM-7-ABC.(d) Calculated pore-size distribution of JNM-7-ABC based on the adsorption branch of the N 2 isotherm at 77 K.
10. SEM EDS is not as accurate as TEM.Since the authors can get TEM measurements, why not use TEM EDS for elemental mapping?Author's action: We thank this reviewer's suggestion, and we retested EDS during TEM measurements as shown in Fig. R5-Fig.R7.These Figures have been added into SI as Supplementary Fig. 11-13.
we have tired various solvent such as DMF, dimethylacetamide (DMA), H 2 O, dioxane, and CH 3 OH under same condition (i.e., 80 °C for 8 h).As shown in Fig.R8.We have found that DMF and DMA can trigger the structure transformation of JNM-7 and JNM-8 from AA to ABC stacking model.However, other solvent like dioxane, and CH 3 OH cannot induce structure change.In addition, the immersion of JNM-7 into H 2 O only reduced the crystallinity, while can partially trigger the structure transformation of JNM-8 from AA to ABC stacking model.It is well known DMF, and DMA have strong coordinating ability toward transition metals than alcohols and esters (See ref.Chem.Eur.J. ability might trigger the structure transformation.We further tried stronger coordinating solvent like 1,8-diazabicyclo(5.4.0)undec-7-ene (DBU).As shown in Fig.R9, the addition of DBU could increase the crystallinity of JNM-7-ABC after the structural transformation.These experimental results further supported that coordinating solvent can disturbed the Cu-Cu interactions between layers, leading to the AA stacking transfer to ABC stacking.

Table R1 .
The space group, cell parameters, total energy, and total crystal stacking energy per layer of the possible structures of JNM-9 under different states.

Table 12 .
4. In the Supplementary Figure35, it should be the signal of 1 O 2 instead of O 2•− .Author's action: We have changed O 2 •− to 1 O 2 in the Supplementary Figure35.