A class of organic cages featuring twin cavities

A variety of organic cages with different geometries have been developed during the last decade, most of them exhibiting a single cavity. In contrast, the number of organic cages featuring a pair of cavities remains scarce. These structures may pave the way towards novel porous materials with emergent properties and functions.We herein report on rational design of a three-dimensional hexaformyl precursor 1, which exhibits two types of conformers, i.e. Conformer-1 and -2, with different cleft positions and sizes. Aided by molecular dynamics simulations, we select two triamino conformation capturers (denoted CC). Small-sized CC-1 selectively capture Conformer-1 by matching its cleft size, while the large-sized CC-2 is able to match and capture both conformers. This strategy allows the formation of three compounds with twin cavities, which we coin diphane. The self-assembly of diphane units results in superstructures with tunable proton conductivity, which reaches up to 1.37×10-5 S cm-1.

In the manuscript titled "A Size-Matching Strategy to Differentiate Flexible Conformers for the Discovery of Novel Cages with Twin Cavities" Professor Zhang and co-workers present a strategy by which conformationally flexible conformers of hour-glass-shaped molecule 1 can be distinguished by covalently locking the energetically favored conformer(s) via imine formation between formyl functionalities of 1 and two different tripodal amines. The presented strategy is elegant yet simple and, although imine formation chemistry is used in similar fashion in the many facets of supramolecular chemistry to bring about (the desired) minimum energy structure of a system, to my knowledge no prior work exists where targeted identification of conformers is presented in similar manner. The validity of the applied approach is highlighted by the 1H NMR spectra of the resulting "diphanes" in which the two different viable conformers are well resolved. Further validation is presented in the form of by X-ray crystallographic and computational investigations. My only minor critique of the presented strategy relates to its restriction by the self-correcting nature of DCC, which limits its applicability to a broader group of potentially interesting conformationally flexible compounds, and perhaps, therefore, attractiveness to even wider audience.
My main criticism is related to some aspects of the X-ray crystallography part of the work. These are discussed in detail below and should be addressed before further considering of the publication of the work.
There are no major problems in the presented single crystal X-ray analyses, which are an important part regarding the main arguments and findings of the work. There are some minor issues that need resolving and commenting; these are listed at the end of this report. There are more significant issues, however, in the powder X-ray diffraction (PXRD) part of the work, which affect, for example, how the structure-property correlation regarding proton conductivities is interpreted.
Firstly, there is a significant lack of details in the description of the PXRD sample preparation which makes the reproduction of the work difficult. Are PXRD measurements carried out using the 'as synthesized' bulk samples, with or without the addition of TFA, or ground single crystals? These details should be presented in clear fashion in the experimental part.
Secondly, visual comparison between the experimental PXRD patterns and the respective simulated patterns, corresponding to the single crystal X-ray diffraction data, suggests that the experimental and simulated patterns are not similar. If this is indeed the case -i.e. the respective powder samples and the single crystals of endo- [1,2,4]diphane, endo- [1,2,5]diphane and exo- [1,2,5]diphane are not structurally similar -there is an issue in using the single crystal structures to discuss the solid state molecular organization of the diphanes in the context of proton conductivity, assuming that the proton conductivity pellet samples were made using the bulk samples, not single crystals. In other words, the current interpretation on the correlation between the proton conductivities and differences in molecular packing, including TFA mobility, of the diphanes is not merited without further evidence on the structural similarity of the crystal structures and the proton conductivity samples. To clarify: -Details of the PXRD sample preparation should be included in experimental section.
-The PXRD patterns that are presented should be of the same bulk material used in the preparation of the proton conductivity samples.
-The simulated and experimental PXRD patterns should be analyzed in terms of their similarity, preferably by Rietveld or Pawley fitting.
-If there is no evidence on structural correlation between the experimental and simulated patterns, the use of single crystals structures to draw conclusions on the structural properties of the bulk, including the possible TFA transportation in the channels of the solid material, is not warranted.
I would also like to ask the authors to provide further proof on the claim that transportation of TFA molecules is a significant factor in the different proton conductivities of the synthesized diphanes. In this context, I suggest reexamining the methodological part as well as the discussion in the paper Nat. Commun. 7, 12750 (2016), where the authors describe only the disordered chloride anions and water molecules to be mobile. None of the TFA anions show any signs of disorder in the crystal structures reported in the present work.
List of specific comments regarding X-ray analyses that need resolving: -N-H proton missing from atom N48_1 in the structure refinement of exo- [1,2,5]diphane -Typographical error in cif of endo- [1,2,4]diphane describing the color of the crystal -Incorrect and missing parameters (absorption correction and moiety formula) in cif file of exo- [1,2,5]diphane. Also, many of the parameters in Table do not correspond to those reported in the cif. Please check and change accordingly.
-Lacking or incorrect single crystal X-ray instrument details in cif of exo- [1,2,5]diphane the device type is assigned as Bruker Apex-II, whereas in the ESI Bruker D8 Venture is given.
-Single crystal X-ray instrumentation should be described in more detail. The text should give the full details of the (two?) instrument(s) with used X-ray source(s) (Cu, Ga?) as well as the detector(s).
-Each of the crystal structures seems to contain voids in the crystal lattice with unresolved electron density. This should be mentioned/discussed and details of the Olex2 Mask/Platon SQUEEZE procedure, which has been used to treat this, should be given in the supplementary information.
-According to the X-ray analyses, the strained endo- [1,2,4]diphane is the only diphane structure which does not seem to have the tertiary amines protonated, perhaps due to the steric shielding of the amine moiety caused by rigid strained conformation. I believe this is an interesting point and worth discussing in the text.
General remarks: -Spelling mistake in Abstract, line 2 "fileds" -> "fields" -Spelling mistake on page 4, in the beginning of Results: "As briefly mentioned above, the geometry optimizations (DFT, B3LYP/6-31G*) showed that molecule…" -In page 4, third row from the bottom, "spectroscopies" should be changed to "spectroscopy" -In page 4, sentence "Its single crystals suitable for X-ray crystallography were obtained by slow evaporation of solvent from its chloroform solution, but only Conformer-1 could be identified" would benefit from rephrasing. I believe that the authors wish to say that in the crystal structure of 1 all molecules adopt the conformation corresponding to Conformer-1. However, the sentence can also be understood that both conformers existed in the crystal structure, but only Conformer-1 could be identified. Therefore, please clarify the sentence. -In page 6, please rephrase the sentence "The single crystals suitable for X-ray analysis were obtained by slow evaporation of its THF solution with additional trifluoroacetic acid (TFA) for better solubility. It crystallized into triclinic… ", in order make clear to the reader what "its" and "it" refer to -In page 7, please change "The twisting of these three chains are…" to "The twisting of these three chains is…" -Spelling mistake in caption of Fig. 4, "trancated" should read "truncated" -In page 13, please add "as" to the sentence "These molecules with unique shapes and configurations can also be as used novel supramolecular synthons…" Reviewer #2 (Remarks to the Author): The paper by Zhenyu Yang et al. describes the reaction of a hexaaldehyde substrate with two triamines to form various imine double-cages. The hexaaldehyde is conformationally semi-labile -it is partially rigidified by biaryl-type of connection of aromatic rings but it retains low-barrier rotations about single bonds. The unique feature of this hexaaldehyde is the possibility of forming two types of double clefts with the 3-fold symmetry that substantially differ by dimensions (among the unlimited number of other conformations). The authors reacted this hexaladehyde with two aliphatic triamines and obtained imines by a reversible type of reaction. The short triamine formed only one type of a double-cage utilizing smaller clefts, while the longer triamine was able to produce two types of double-cages utilizing either small or larger clefts. For the obtained double-cages, interesting proton transporting properties have been reported, that rely on the high charge and porosity in the solidstate.
Despite the fact that I think that the "dual-purpose" hexaaldehyde is interesting, and resulting doublecages may exhibit unique properties, I can't agree with the "methodological" interpretation of the results that the authors provided. The authors gave a new name ("Conformation capturing") to the well-known and intuitive phenomenon. It is quite typical that conformationally labile compounds, especially those with low activation energy barriers can easily adjust and form the products that thermodynamically most stable. It is also straightforward, that for a "small" partner, that does not fit to the large cleft only one type of product is formed while for the larger but still flexible partner two types of product are possible and formed. The idea of multidimensional reactivity that depends on the conditions (the reaction partner, template, or environment) and relies on size-match and conformational shape-adjustment constitutes the core of dynamic covalent chemistry. Therefore, the results presented in this paper are not "a novel yet simple size-matching strategy to identify and distinguish the interconverting conformational" a new strategy to "capture" conformers or to probe conformational space that "had remained unexplored" as the authors claim, but the typical and intuitive behavior of conformationally labile molecules participating in reversible reactions.
I think that the compounds that the authors synthesized are indeed interesting and create many new possibilities that can, in the future, become suitable for publication in Nat. Commun., but, in this interpretation, the authors missed the point. Instead of focusing on the unique properties of the double cage products, which are way more interesting (but require more work and more examples), they described rather typical and intuitive behavior, presenting it as a new methodology. Therefore, I recommend the rejection of the paper in the current form. The results may become suitable for publication in the future, with the focal point changed and with more results about the obtained double cages.
Additional remarks: PXRD patterns -experimental and simulated are very different. Most likely more polymorphic forms are present in the samples.
It is quite typical that for the conformationally labile compound only one of the possible conformers is found in the crystal phase. So depriving the value of X-ray analysis by saying "SC-XRD, only able to identify Conformer-1 in solid-state" is not OK.
The discussion part does not contain a discussion but a summary-so either some discussion should be added or the part should be called summary.
In this manuscript, Z. Yang et al. provides a novel strategy to differentiate and identify rapid interconverting conformational isomers which would be otherwise difficult to be detected by standard characterization techniques. The simple but interesting approximation consists of the idea to catch conformational isomers of a targeted system (precursor) by reacting them through dynamic covalent chemistry with molecular units (capturers) of different size. The strategy is demonstrated for a three-dimensional hexaformyl model precursor with an extremely-rich conformational space (i.e., infinite conformers), which can be broadly classified into two distinct type of conformers. The outcome of the reported strategy (final species) not only allows the identification of two elusive conformers of the precursor but also provides new molecular cages with potential functionalities and applications (e.g., favorable proton conduction).
Within my limited knowledge about the literature of molecular cages, the proposed synthetic method (key result of the study) looks quite simple and relatively easy to be used for the design of novel 3D molecular cages. It is difficult, however, to me to assess the significance and broad scope of the proposed approach for a general scientific community in the frontier of chemistry, materials chemistry and chemical biology. Nevertheless, the relevance of the synthetic procedure for the design of novel and exotic molecular cages can be strengthened and highlighted with a more extended introduction (only a few sentences) emphasizing the advantages of this new approach with respect to the previous ones to obtain organic molecular cages.
The study is very well-organized, well-written and can be easily followed with the main idea clear along the manuscript. I have really enjoyed the paper. Nonetheless, a criticism, related with the previous one (significance), would be to provide a sufficient context with more comparisons with previous works to make clear the relevance of the study.
Concerning validity of the study, the experimental findings seem to be solid and support the conclusions drawn. Nevertheless, I believe that the theoretical part, which is quite relevant in the study, can be improved. In this respect, I would suggest the following points to be considered: 1) The experimental analysis (H NMR) about the conformational interconversion of 1 is performed in solution. However, DFT calculations seem to be performed in gas phase. That effect may have an effect on the energy difference between conformer-1 and -2. Therefore, I would include solvent effects in the calculations (a continuum model can be enough for this study).
2) van der Waals (vdW) intramolecular interactions can also have an effect on the energy difference between conformer-1 and -2. Plain B3LYP functional cannot capture vdW interactions and should be augmented with external approximations (the most practical would be the Grimme's D3 approach). I would re-do the calculations at least at the B3LYP-D3/6-31G* level to check that intramolecular vdW interactions are not playing an important effect.
3) As the energy difference between conformer-1 and -2 is quite small even in the range of error of the DFT method, I would check that the energy order is preserved with another density functional. In this sense, extra calculations with the M06-2X, which can describe vdW interactions at short range, can be a good alternative to check that the theoretical results are robust. Fig. 2a, the authors write: "Fast interconversion of Conformer-1 and -2 with extremely low isomerization barrier." However, the authors have not computed the isomerization barrier, they provide an energy difference between both conformers. An estimate of the low isomerization barriers can provide by calculating the torsional potential through: i) the rotation of the carbon-carbon bond of the central benzene ring and the subsequent sp3 carbon atom and ii) the rotation around the inter-ring carboncarbon bond of the two aryl rings within a peripheral arm.

4) In the caption of
Then the author say: "Free energy of each isomer was determined by geometry optimization (DFT, B3LYP/6-31G*)". I was wondering if the energy provided is actually a free energy difference computed from vibrational calculations or is simply the energy difference calculated between the optimized conformer 1 and 2. If it is obtained by the latter approach is not a free energy estimate.

5)
In the Supplementary Information (DFT calculations section), the author define the strain energy as the energy difference between the EC unit found in the cage and that obtained after optimization. The abbreviation EC is not defined.
6) In the Supplementary Information, there is a section devoted to Molecular dynamics simulations. However, the authors do not call to this section during the discussion in the main text which is weird. On the other hand, I do not understand the histograms in Supplementary Figure 4, is it possible that they are interchanged? If it is the case, it would have more sense to me. Nevertheless, I think that Supplementary Figure 4 deserves a brief discussion at least in the Supplementary Information.

7)
Apart from the strain energies evaluated for the four diphane cages, the relative difference between the two types of cages (exo and endo) should be provided to see if they can correlate with the experimental yield obtained for the final cages.
Typo in the caption of Fig. 4 (second line). "truncated" should be substituted by "truncated" As a future suggestion (not for this paper), it would be great if you are able to have control and break the final molecular cage by external stimuli but isolating the conformational isomers.
In the manuscript titled "A Size-Matching Strategy to Differentiate Flexible Conformers for the Discovery of Novel Cages with Twin Cavities" Professor Zhang and co-workers present a strategy by which conformationally flexible conformers of hour-glass-shaped molecule 1 can be distinguished by covalently locking the energetically favored conformer(s) via imine formation between formyl functionalities of 1 and two different tripodal amines. The presented strategy is elegant yet simple and, although imine formation chemistry is used in similar fashion in the many facets of supramolecular chemistry to bring about (the desired) minimum energy structure of a system, to my knowledge no prior work exists where targeted identification of conformers is presented in similar manner. The validity of the applied approach is highlighted by the 1 H NMR spectra of the resulting "diphanes" in which the two different viable conformers are well resolved. Further validation is presented in the form of by X-ray crystallographic and computational investigations. My only minor critique of the presented strategy relates to its restriction by the self-correcting nature of DCC, which limits its applicability to a broader group of potentially interesting conformationally flexible compounds, and perhaps, therefore, attractiveness to even wider audience.
Answer: First of all, we would like to thank the Reviewer for carefully reviewing our manuscript and appreciating our research work. As mentioned by the Reviewer, self-correcting DCC is a useful strategy for the preparation of organic cages, which we fully agree. On the other hand, we rather use DCC as a synthetic tool to discover interesting structures related to the flexible conformations of a precursor, including but not limited to the three diphanes reported in the present work.
For example, by employing the size-matching strategy, the precursor molecule 2 reacted with large-sized CC-2 to selectively yield monofunctional Cage-2 with a single cavity and one pendant -OH (Fig. C1a), which is presented as a by-product in the original manuscript ( Fig. 5 and Fig. C1a); on the other hand, the reaction between molecule 2 and small-sized CC-1 selectively produced bifunctional Cage-3 with two -OH at both ends of the molecule (Fig. C1a). We are currently polymerizing these two cages (as monomers) to form lateral and main-chain polycages, respectively, which will be reported in due course.
Taking R-endo- [1,2,4]diphane for instance, it also self-assembled into superstructures with four hierarchical levels (Fig. C3). It is worth noting that chiral channels were formed in the tertiary structures, which allows us to explore the induced chirality and spontaneous polarization of confined water. It endows the superstructure with interesting properties such as non-linear optics and ferroelectrics, which are under investigation in our lab.
In a word, the strategy reported in the current study provides a series of interesting molecular platform for the exploration of novel hierarchical superstructures, which allows us to search for emergent properties of these superstructures. Figure C3. Hierarchical self-assembly of R-endo- [1,2,4]diphane during crystallization. a) Side view of its crystal structure (primary structure), b) supramolecular dimer (secondary structure) self-assembled by two R-endo- [1,2,4]diphane molecules, c) supramolecular 32-Helix self-organized by supramolecular dimer, forming a chiral intermolecular channel as shown in blue, and d) the super lattice (quaternary structure) formed by an array of 32-Helices. Hydrogen atoms and solvents are omitted for clarity, and the crystal belongs to trigonal space group P3221.
Ø In the powder X-ray diffraction (PXRD) part of the work. 2) As the structural analysis of each bulk material was ascertained, we set out to revise the correlation between molecular packing and the proton conductivity of the three diphanes (Supplementary Table 7 and Fig. 18 Table 7  Ø List of specific comments regarding X-ray analyses that need resolving: Q1: N-H proton missing from atom N48_1 in the structure refinement of exo-[1,2,5]diphane.

A1
: N-H proton of N48_1 has been revised in the structure refinement of exo- [1,2,5]diphane and the revised CIF has been redeposited at CCDC.

A2:
The typographical error word "coloueless" describing the color of the crystal in CIF of endo- [1,2,4]diphane has been revised as "colourless", and the revised CIF has also been redeposited at CCDC.

Q3:
Incorrect and missing parameters (absorption correction and moiety formula) in cif file of exo- [1,2,5]diphane. Also, many of the parameters in Table do not correspond to those reported in the cif.
Please check and change accordingly.
i. The empirical formula of Molecule 1 "C90H62Cl12O6" has been revised as "C90H62Cl12O6".
In supplementary table 6, crystal data for Cage-2.
ii. The line of "Reflections collected/unique, 79008/19768 [Rint = 0.1267]" has been revised as "Reflections collected, 79008". After which, a new row was added and its content is "Independent Q4: Lacking or incorrect single crystal X-ray instrument details in cif of exo-[1,2,5]diphane the device type is assigned as Bruker Apex-II, whereas in the ESI Bruker D8 Venture is given.
information has also been mentioned in the Techniques part of the Revised Supplementary Information.

Q7:
According to the X-ray analyses, the strained endo- [1,2,4]diphane is the only diphane structure which does not seem to have the tertiary amines protonated, perhaps due to the steric shielding of the amine moiety caused by rigid strained conformation. I believe this is an interesting point and worth discussing in the text.

A7:
Thanks for the reviewer's careful inspection of the protonated crystal structure. In Page 7 of the Revised Manuscript, we added "In addition, the X-ray structure of endo- [1,2,4]diphane showed that the tertiary amines on the aliphatic chains were not protonated, even with the presence of TFA. This is presumably due to the steric hindrance of the irregularly stretched amine moiety caused by the rigid strain conformation, as well as the electrostatic repulsion of the neighbouring TFA molecules." We also added Supplementary Fig. 12 to illustrate this interesting point.
3) In page 4, third row from the bottom, "spectroscopies" should be changed to "spectroscopy". 4) In page 4, sentence "Its single crystals suitable for X-ray crystallography were obtained by slow evaporation of solvent from its chloroform solution, but only Conformer-1 could be identified" would benefit from rephrasing. I believe that the authors wish to say that in the crystal structure of 1 all molecules adopt the conformation corresponding to Conformer-1. However, the sentence can also be understood that both conformers existed in the crystal structure, but only Conformer-1 could be identified. Therefore, please clarify the sentence.

5)
In page 6, please rephrase the sentence "The single crystals suitable for X-ray analysis were obtained by slow evaporation of its THF solution with additional trifluoroacetic acid (TFA) for better solubility. It crystallized into triclinic… ", in order make clear to the reader what "its" and "it" refer to.
6) In page 7, please change "The twisting of these three chains are…" to "The twisting of these three chains is…". 7) Spelling mistake in caption of Fig. 4, "trancated" should read "truncated". 8) In page 13, please add "as" to the sentence "These molecules with unique shapes and configurations can also be as used novel supramolecular synthons…".
Answer: Thanks for the reviewer's comments.
1) In Page 1, the second line of the abstract, the word "fileds" has been revised as "fields".
2) In Page 4，in the beginning of Results, the sentence "As briefly mentioned above, the geometry optimizatoins (DFT, B3LYP/6-31G*) showed molecule 1 exists infinite forms of conformers via all possible C-C single bond rotations." has been revised as "As briefly mentioned above, the geometry optimizatoins (DFT, B3LYP/6-31G*) showed that molecule 1 exists infinite forms of conformers via all possible C-C single bond rotations" in the Revised Manuscript.
3) In Page 5, the fourth line of the second paragraph, the word "spectroscopies" has been revised as "spectroscopy". 4) In Page 5, the sentence "Its single crystals suitable for X-ray crystallography were obtained by slow evaporation of solvent from its chloroform solution, but only Conformer-1 could be identified" has been revised as "Its single crystals suitable for X-ray crystallography were obtained by slow evaporation of solvent from its chloroform solution, but molecule 1 only exists the conformation corresponding to Conformer-1 in the crystal structure (Fig. 2c), which is common for the conformationally labile compounds that often crystallize into the most efficient molecular packing. 26  First of all, we would like to thank the Reviewer for carefully reviewing our manuscript and appreciating our research work.

Q1:
Despite the fact that I think that the "dual-purpose" hexaaldehyde is interesting, and resulting doublecages may exhibit unique properties, I can't agree with the "methodological" interpretation of the results that the authors provided. The authors gave a new name ("Conformation capturing") to the well-known and intuitive phenomenon. It is quite typical that conformationally labile compounds, especially those with low activation energy barriers can easily adjust and form the products that thermodynamically most stable. It is also straightforward, that for a "small" partner, that does not fit to the large cleft only one type of product is formed while for the larger but still flexible partner two types of product are possible and formed. The idea of multidimensional reactivity that depends on the conditions (the reaction partner, template, or environment) and relies on size-match and conformational shape-adjustment constitutes the core of dynamic covalent chemistry. Therefore, the results presented in this paper are not "a novel yet simple size-matching strategy to identify and distinguish the interconverting conformational" a new strategy to "capture" conformers or to probe conformational space that "had remained unexplored" as the authors claim, but the typical and intuitive behavior of conformationally labile molecules participating in reversible reactions.

A1:
Thanks for the reviewer's comments. We agree with the reviewer's opinion on "It is quite typical that conformationally labile compounds, especially those with low activation energy barriers can easily adjust and form the products that thermodynamically most stable". However, as we calculated the conformation energy of the two conformers of molecule 1 (Supplementary Fig. 9 and Table 1), we found that their energy difference is negligible (ca. 0.1 kJ mol -1 ). It therefore indicates their fast interconversion, which makes it hard to say which conformation is thermodynamically more stable to pack into the corresponding crystal.
The reviewer stated that the conformation capturing is a well-known and intuitive phenomenon, and the comment itself is indeed correct but not in our case. First of all, we think the wording "large/small cleft" is somehow misleading, which has been revised as "As the formyl groups are permanently moving, the region with the probability of containing formyl moieties forms a circular ring within each cleft, and the range of the ring is different for Conformer-1 and -2 (middle column, Fig. 1)." (Page 3 in the Revised Manuscript, highlighted in yellow). Secondly, this is not intuitive as the reviewer thought to be. As shown in Fig. 1  The rational of this design has been carefully explained in the Revised Supplementary Information, along with the revised Supplementary Fig. 10, give as follows: As the formyl groups are permanently moving, the region with the probability of containing formyl moieties form a circular ring within each cleft, and the range of the cleft of Conformer-1 and -2 is different. The region with formyl motion is similar to that of electron cloud surrounding an atomic nucleus, which spans from 3.48 to 16.52 Å for Conformer-1 and 6.76 to 12.28 Å for Conformer-2, respectively.
We also calculated the sizes of CC-1 and -2, which are 3.766 and 4.770 Å, respectively. Taken into consideration of the reactive distance between an amine and an aldehyde (esteemed to be 2.911 Å S22 ), the effective sizes of CC-1 and -2 were calculated to be 6.677 and 7.681 Å, respectively. It therefore means that both CC-1 and -2 can easily cover the "formyl cloud" of Conformer-1, producing both endo- Indeed, we used the size-matching strategy to discover interesting structures related to the flexible conformations of a precursor, including but not limited to the three diphanes reported in the present work.
For example, by employing the size-matching strategy, the precursor molecule 2 reacted with large-sized CC-2 to selectively yield monofunctional Cage-2 with a single cavity and one pendant -OH (Fig. C1a), which is already presented as a by-product in the original manuscript ( Fig. 5 and Fig. C1a); on the other hand, the reaction between molecule 2 and small-sized CC-1 selectively produced bifunctional Cage-3 with two -OH at both ends of the molecule (Fig. C1a). We are currently polymerizing these two cages (as monomer) to form lateral and main-chain polycages, respectively, which will be reported in due course.
Besides, we found that Cage-2 (not Cage-1) self-assembled into superstructures with distinct four hierarchical levels (Fig. C2). The tertiary structure is supramolecular 21-helix, and an array of these helices formed the quaternary superstructure.
Taking R-endo- [1,2,4]diphane for instance, it also self-assembled into superstructures with four hierarchical levels (Fig. C3). It is worth noting that chiral channels were formed in the tertiary structures, which allows us to explore the induced chirality and spontaneous polarization of confined water. It endows the superstructure with interesting properties such as non-linear optics and ferroelectrics, which are under investigation in our lab.

b) The synthesis of R-endo-[1,2,4]diphane and S-endo-[1,2,4]diphane.
In a word, the strategy reported in the current manuscript provides a series of interesting molecular platform for the exploration of novel hierarchical superstructures, which allows us to search for emergent properties of these superstructures.  28 Q4: It is quite typical that for the conformationally labile compound only one of the possible conformers is found in the crystal phase. So depriving the value of X-ray analysis by saying "SC-XRD, only able to identify Conformer-1 in solid-state" is not OK.

A4:
Thanks for the reviewer's comment, and the reviewer is indeed correct. In the original manuscript, we have already put a similar statement that "X-ray crystallography is a powerful method to unambiguously determine the absolute conformation of a molecule, but it often only distinguishes the conformer that provides the most efficient molecular packing.", which agrees well with the reviewer's opinion. In order to avoid the misleadingness, we revised the previous statement in Page 4 in the Original Manuscript to "Its single crystals suitable for X-ray crystallography were obtained by slow evaporation of solvent from its chloroform solution, and molecule 1 only exists the conformation corresponding to Conformer-1 in the crystal structure (Fig. 2c), which is common for the conformationally labile compounds that often crystallize into the most efficient molecular packing.", which has been highlighted in yellow in the Revised Manuscript (Line 10, Page 5).

Q5:
The discussion part does not contain a discussion but a summary-so either some discussion should be added or the part should be called summary.

2,4]diphane', endo-[1,2,5]diphane' and exo-[1,2,5]diphane'.
In the calculation part of the manuscript (Fig. 1), the results were carried out by molecular dynamics simulations based on the formation of imine form of diphanes. In this section, we investigated the design principle of the size-matching strategy of diphanes construction, which should rely on reactive distance between formyl and amino groups for the effective formation of imine bonds. As we illustrated in the Supplementary Fig. 10 however, this has been largely overlooked".
3. Concerning validity of the study, the experimental findings seem to be solid and support the conclusions drawn. Nevertheless, I believe that the theoretical part, which is quite relevant in the study, can be improved. In this respect, I would suggest the following points to be considered: 3) As the energy difference between conformer-1 and -2 is quite small even in the range of error of the DFT method, I would check that the energy order is preserved with another density functional.
In this sense, extra calculations with the M06-2X, which can describe vdW interactions at short range, can be a good alternative to check that the theoretical results are robust.
Answer: Thanks for the reviewer's comment. We recalculated the free energy of each isomer (conformer-1 and conformer-2) with another M06-2X-D3 hybrid functional to check whether the energy order is preserved. As shown in the Supplementary Table 1 in the Revised Supplementary Information, although the energy order is different for B3LYP-D3 and M062X-D3 functionals, the overall deviation between them is quite small. Thus, the results reported in this paper are all calculated by B3LYP-D3 method in chloroform unless the specific instructions.  (2) The energy of each conformers provided in Fig. 2a is free energy, which was recalculated by geometry optimization using DFT calculation (B3LYP-D3 in PCM with chloroform as solvent).

Supplementary
In addition, we revised the caption of  Table 1 and Supplementary Fig.   9)" to call the molecular dynamic simulations. As suggested by the Reviewer, we also discussed the rational of this paper with Supplementary Fig. 10 in the Revised Supplementary Information.
Please refer to it for detailed discussion. 7) Apart from the strain energies evaluated for the four diphane cages, the relative difference between the two types of cages (exo and endo) should be provided to see if they can correlate with the experimental yield obtained for the final cages.
Answer: Thanks for the reviewer's comments. The Reviewer is indeed correct. For example, as we already put it in the original manuscript, "We also noticed that the conversion rates of endoand exo-[1,2,5]diphane' were different, as the former was generated faster than the latter, leading to a product distribution of 75% endo-and 25% exo-[1,2,5]diphane', respectively". This is in line with the strain energy of the diphanes. Fig. 4 (second line) "truncated" should be substituted by "truncated".

8) Typo in the caption of
Answer: Thanks for the reviewer's comments. In Page 8 of the Revised Manuscript, the word "trancated" in the caption of Fig. 4 (second line) has been revised as "truncated". 9) As a future suggestion (not for this paper), it would be great if you are able to have control and break the final molecular cage by external stimuli but isolating the conformational isomers.

Answer: Thanks for the reviewer's comments. Capture conformers in solution in a controllable
and reversible way is our ultimate goal and we are making efforts to achieve it in the near future.  Fig. 11).  Fig. 7, the sentence "The comparison of molecular structure of three diphanes" has been revised as "The comparison of molecular structure of three diphanes with the presence of TFA" in the Revised Manuscript (Page 12, the first line of the caption in Fig. 7).
37. In Page 11 of the Original Manuscript, line 8 of the caption in Fig. 7, "at 303 and 343 K" has been revised as "at 308 and 338 K" in the Revised Manuscript (Page 12, line 9 of the caption in Fig. 7). i.e. sandglass-shaped structure (Fig. 7a and c). However, their self-assembled superstructures are remarkably different. endo- [1,2,4]Diphane molecules are closely packed, and it provides no clear transportation channels for protons hopping (Fig. 7b), where TFA molecules are trapped within the intrinsic and extrinsic cavities of the diphanes. In contrast, endo-[1,2,5]diphanes self-assemble into a crystalline phase with ordered channels (highlighted with red column) that allow the transportation

REVIEWER COMMENTS
Reviewer #1 (Remarks to the Author): I am pleased to have read the revised version of the manuscript by Professor Zhang and co-workers which has gone through a substantial edit and as a result has been significantly improved. The PXRD part of the new version of the manuscript has been thoroughly revised and I am glad to see that a correlation between the simulated and recorded powder data now exists. The peak widths of the experimental patterns are rather large considering that single crystals were used in data collection. This could be explained, however, by partial desolvation of the crystals if the capillaries used in data collection were loaded with dried crystals and not together with mother liquor. This small detail, i.e. were the capillaries loaded with or without crystallization solvent, is an important information and should be added in the experimental detail in the supporting information to avoid confusion.
In addition, the following minor errors have been introduced into the SI material during the revision process, which should be addressed: 1) The given densities in single crystal X-ray diffraction data tables (supplementary tables 2-6) should be in g/cm<sup>3</sup> (not in M g/cm<sup>3</sup>).
2) Check the graph legend in Supplementary Figure 14, which has the descriptions of the different graphs in wrong order.
In overall, the X-ray structure determination and crystallographic analyses are expertly done and after the above minor corrections have been made, I am pleased to support the publication of the manuscript in terms of the quality of its crystallographic part.
Reviewer #2 (Remarks to the Author): Despite numerous arguments adduced by the authors, I am still not convinced that the idea of "conformation capturing" represents a novel concept (see below for detailed discussion) and constitutes the main value of the paper. However, because I don't neglect experimental facts, but only their interpretation, and I really appreciate the intriguing structures of the cages and the huge amount of experimental work on their properties, I can recommend the paper for publication. However, I think that the title and the abstract should be modified according to the current content.a 1. The authors mention the energy difference for 1 as one of the arguments in the discussion (: "However, as we calculated the conformation energy of the two conformers of molecule 1 (Supplementary Fig. 9 and Table 1), we found that their energy difference is negligible (ca. 0.1 kJ mol-1)." For a dynamic system, the relative stability of the products matters not of the substrates. Not to mention that for identification of conformers it is the barrier of interconversion, not the relative stability that matters. 2. Size match -the authors argue that the results are not intuitive because the average size of the cleft does not agree with the observed selectivity. However, this is not the average size, but the minimum size is the factor that matters and, in this context, the results are intuitive. The authors themselves admit this " " as the maximally stretched amines of CC-1 154 still cannot reach the three formyl groups within each cleft" 3. The sentence "conversion rates of endo-and exo- [1,2,5]diphane' were different, as the former was generated faster than 211 the latter, leading to a product distribution of 75% endo-and 25% exo-[1,2,5]diphane', respectively." is not precise, because it is not the kinetics but thermodynamics that determines the final distribution. It should be corrected.
Reviewer #3 (Remarks to the Author): Although the authors have satisfactorily addressed most of my previous comments, there are some theoretical points that need a better explanation for a total recommendation of the manuscript in Nature Communications. These points are: 1) According to my suggestions, authors have performed calculations including solvent effects and considering two density functionals with Grimme's dispersion corrections (B3LYP-D3 and M062X-D3). Nevertheless, I do not see why the authors have decided to preserve the results at the B3LYP-D3 level since the energy order using total energies or free energies is similar at the M062X level. A justification should be provided. In Table 1, the units are missing (I guess kJ/mol).
2) The computed torsion potential for the rotation of the carbon-carbon bond of the central benzene ring and the subsequent sp3 carbon atom does not seem to suggest an easy rotation (according to the energetics) and the simple cartoon used in Figure 1, middle-left cannot be appropriate. A clarification of this should be provided. The type of torsion potential, either rigid or fully relaxed should be indicated in the text or supplementary information. Finally and for consistency with other energies, the units in the torsion graph should be in kJ/mol.
3) The reply to my previous comment (comment 6th in the previous letter) related to MD calculations is elusive unsatisfactory. First, the sentences added explain the type of DFT calculations that have been performed over the two Conformers-1 and -2. A call for the MD calculations is appropriate when Figure 1 is explained because, based on those calculations, the cartoon in Figure 1 (middle-left) is created. Second, I keep thinking that the histograms in Supplementary Figure 10 have been able to be interchanged. If we compare the cartoon for Conformer-2 with its histogram there are possible events with distances lower than 0.4 nm, which does not match with the cartoon (distances between 0.6 -1.3 nm). This should be clarified if there are no mistakes. 4) Again, my previous comment (comment 7th in the previous letter) related to the relative energies between both cages (exo and endo) has not properly addressed. The authors should provide the energy difference between both structures, which can help to see which structure is more stable.
Typos. Page 4 in the revised version. "Optimizations" instead of "Optimizatoins". In the same line, it would be better "…DFT calculations…"

Reviewer 1:
I am pleased to have read the revised version of the manuscript by Professor Zhang and co-workers which has gone through a substantial edit and as a result has been significantly improved.
Answer: First of all, we would like to thank the Reviewer again for carefully reviewing our manuscript and appreciating our research work.

Q1:
The PXRD part of the new version of the manuscript has been thoroughly revised and I am glad to see that a correlation between the simulated and recorded powder data now exists. The peak widths of the experimental patterns are rather large considering that single crystals were used in data collection. This could be explained, however, by partial desolvation of the crystals if the capillaries used in data collection were loaded with dried crystals and not together with mother liquor. This small detail, i.e. were the capillaries loaded with or without crystallization solvent, is an important information and should be added in the experimental detail in the supporting information to avoid confusion.

A1:
Thanks for the reviewer's comments. The Reviewer is indeed correct. We therefore have revised the experimental details as "Slow evaporation of the settled solution yielded high-quality single crystals, which were then loaded into a quartz glass capillary with a diameter of 0.7 mm without mother liquor", which was highlighted in yellow in Powder X-ray Diffraction section (section 2.6, Page S5) of the Revised Supplementary Information.

Q2:
The given densities in single crystal X-ray diffraction data tables (Supplementary Tables 2-6) should be in g/cm3 (not in M g/cm3).

A1:
Thanks for the reviewer's careful remark. The density units of the single crystal X-ray diffraction (Supplementary Tables 2-6, Page S24-S28) should be either Mg/m 3 or g/cm 3 , and we therefore used "g/cm 3 " in the revised version, which are highlighted in yellow in the Revised Supplementary Information. 2

Reviewer 2:
Despite numerous arguments adduced by the authors, I am still not convinced that the idea of conformation capturing" represents a novel concept (see below for detailed discussion) and constitutes the main value of the paper. However, because I don't neglect experimental facts, but only their interpretation, and I really appreciate the intriguing structures of the cages and the huge amount of experimental work on their properties, I can recommend the paper for publication.
First of all, we would like to thank the Reviewer again for carefully reviewing our manuscript and appreciating our research work.

Q1:
However, I think that the title and the abstract should be modified according to the current content.
The authors mention the energy difference for 1 as one of the arguments in the discussion (However, as we calculated the conformation energy of the two conformers of molecule 1 (Supplementary Fig. 9 and Table 1), we found that their energy difference is negligible (ca. 0.1 kJ mol-1)." For a dynamic system, the relative stability of the products matters not of the substrates. Not to mention that for identification of conformers it is the barrier of interconversion, not the relative stability that matters.

A1:
Thanks for the reviewer's comments. The reviewer's opinion is right and we have made the following corrections in the Revised Manuscript.

1).
The title has been amended as "Diphanes: A Class of Twin-Cavity Cages".
2). The abstract has been revised to emphasize the structure of the twin-cavity cages (diphane) and their properties as follows: Covalent organic cages recently have attracted wide interest in the fields of recognition/separation, sensor, catalysis, etc. A variety of organic cages with different fascinating geometries have been developed during the last decade, but most of them exhibit a single cavity. We envisioned that a cage with a pair of cavities could open the way for the search of novel porous materials with emmergent properties and functions. Here, as a proof of concept, we rationally designed a threedimensional hexaformyl precursor 1, which exhibits two types of conformers, i.e. Conformer-1 and -2 with different cleft positions and sizes. Aided by molecular dynamics simulations, we selected two triamino conformation capturers (denoted CC). Small-sized CC-1 selectively captured Conformer-1 by matching its cleft size, while large-sized CC-2 was able to match and capture both conformers. This strategy allowed the formation of two sandglass-shaped and one dumbbell-like compounds with twin cavities, which we coined diphane. The self-assembly of the three diphanes in turn led to the discovery of supramolecular materials with tunable proton conductivity, which reached up to 1.37×10 -5 S cm -1 , approximately 10 3 times higher than bulk water. Depending on the configuration of diphanes, their conductivity can be tuned by an order of magnitude of 10 4 .

Q2:
Size match -the authors argue that the results are not intuitive because the average size of the cleft does not agree with the observed selectivity. However, this is not the average size, but the minimum size is the factor that matters and, in this context, the results are intuitive. The authors themselves admit this "as the maximally stretched amines of CC-1 still cannot reach the three formyl groups within each cleft" A2: Thanks for the reviewer's comments. We agree with the reviewer's opinion that the minimum size of the clefts play a more important role in the formation of diphanes with different geometry. It proves we still cannot convince the reviewer with the size-matching strategy as the novelty. We had already changed the focal point and interpretation of the experimental results in the first revision, and therefore have revised the title and abstract in the second revision to comply with the Reviewer's comment.

Q3:
The sentence "conversion rates of endo-and exo- [1,2,5]diphane' were different, as the former was generated faster than the latter, leading to a product distribution of 75% endo-and 25% exo- [1,2,5]diphane', respectively." is not precise, because it is not the kinetics but thermodynamics that determines the final distribution. It should be corrected.

A3:
Thanks for the reviewer's comments. The reviewer's opinion is correct and the term "faster" is not rigorous enough. We therefore revised the phrase as "The evolution profile illustrates that the consumption of molecule 1 occurred simultaneously with the formation of two diphane isomers until the equilibrium, which led to a product distribution of 75% endo-and 25% exo-[1,2,5]diphane', respectively." in Page 10 of the Revised Manuscript. 4

Reviewer 3:
Although the authors have satisfactorily addressed most of my previous comments, there are some theoretical points that need a better explanation for a total recommendation of the manuscript in Nature Communications.
Answer: First of all, we would like to thank the Reviewer again for carefully reviewing our manuscript and appreciating our research work. Nevertheless, I do not see why the authors have decided to preserve the results at the B3LYP-D3 level since the energy order using total energies or free energies is similar at the M062X level. A justification should be provided. In Table 1, the units are missing (I guess kJ/mol).

A1:
Thanks for the reviewer's comments.

1).
As per request of the Reviewer in the first revision, we employed M062X-D3, and compared with the results of B3LYP-D3 in the original manuscript. We found that the energy difference of the Conformer-1 and -2 calculated with B3LYP-D3 (-0.12 kJ/mol) and M062X-D3 (0.72 kJ/mol) functionals is very similar within the range of experimental error, which confirms their fast interchange in CHCl3. According to the Grimme's work (Phys. Chem. Chem. Phys., 2011, 13, 6670-6688), B3LYP-D3 is indeed not the overall applicable functional (which might be the concern of the Reviewer). The authors showed that among all tested 23 hybrids, M062X-D3 is statistically the best of all hybrids. On the other hand, they also observed SCF-convergence problems for all Minnesota functionals, even for simple atomic systems. In our calculations, we also found that some of the optimized relaxed structures through M062X-D3 functional were indeed difficult to converge. Under this consideration, the results reported in this paper are thus all calculated by B3LYP-D3 method in chloroform unless with specific instructions. During the revision, we have provided a justification which is highlighted in yellow in DFT calculations section (starting from Page S18) of the Revised Supplementary Information.

2).
We have added the unit of kJ/mol in Supplementary Table 1 of the Revised Supplementary Information.

Q2:
The computed torsion potential for the rotation of the carbon-carbon bond of the central benzene ring and the subsequent sp3 carbon atom does not seem to suggest an easy rotation (according to the energetics) and the simple cartoon used in Figure 1, middle-left cannot be appropriate. A clarification of this should be provided. The type of torsion potential, either rigid or fully relaxed should be indicated in the text or supplementary information. Finally and for consistency with other energies, the units in the torsion graph should be in kJ/mol.

A2:
Thanks for the reviewer's comments.

1).
Following the reviewer's suggestion, we have differentiated the rotational C-C bonds with letter a and b in Fig. 1 of the Revised Manuscript, and revised the second sentence of the caption as "The threedimensional molecule 1 experiences spontaneous interconversion between Conformer-1 with a pair of upper and lower clefts and Conformer-2 with a pair of left and right clefts (left column), of which C-C bonds a are much more liable to rotate than C-C bonds b, as the latter has a higher rotational energy barrier determined by relaxed potential energy scan." Meanwhile, we have also revised the sentences of the Revised Manuscript (line 2, in Page 4) as "molecule 1 exists infinite forms of conformers via all possible C-C single bond rotations, of which C-C bonds a are much more liable to rotate than C-C bonds b, as the latter has a higher rotational energy barrier ( Supplementary Fig. 9). It therefore leads to the formation of two types of conformers, Conformer-1 and -2 ( Fig. 1 and Fig. 2a)." 2). The conformation energies were determined by relaxed rather than rigid potential energy scan. We

Q3:
The reply to my previous comment (comment 6th in the previous letter) related to MD calculations is elusive unsatisfactory. First, the sentences added explain the type of DFT calculations that have been performed over the two Conformers-1 and -2. A call for the MD calculations is appropriate when Figure   1 is explained because, based on those calculations, the cartoon in Figure 1 (middle-left) is created.
Second, I keep thinking that the histograms in Supplementary Figure 10 have been able to be interchanged.
If we compare the cartoon for Conformer-2 with its histogram there are possible events with distances lower than 0.4 nm, which does not match with the cartoon (distances between 0.6 -1.3 nm). This should be clarified if there are no mistakes.