Efficient ethylene purification by a robust ethane-trapping porous organic cage

The removal of ethane (C2H6) from its analogous ethylene (C2H4) is of paramount importance in the petrochemical industry, but highly challenging due to their similar physicochemical properties. The use of emerging porous organic cage (POC) materials for C2H6/C2H4 separation is still in its infancy. Here, we report the benchmark example of a truncated octahedral calix[4]resorcinarene-based POC adsorbent (CPOC-301), preferring to adsorb C2H6 than C2H4, and thus can be used as a robust absorbent to directly separate high-purity C2H4 from the C2H6/C2H4 mixture. Molecular modelling studies suggest the exceptional C2H6 selectivity is due to the suitable resorcin[4]arene cavities in CPOC-301, which form more multiple C–H···π hydrogen bonds with C2H6 than with C2H4 guests. This work provides a fresh avenue to utilize POC materials for highly selective separation of industrially important hydrocarbons.

al. previously reported a boronate ester cage with potential for C2 hydrocarbon separations, albeit based on Henry and IAST calculated selectivities pairs of ethane, ethene and acetylene (DOI 10.1002/chem.201802123).
Although the experimental measurements seem to be robust and well described, the report of the molecular modelling used to explore the adsorption of the guest, and hence thee origin of the selectivity is rather unclear. The main text states that DFT-d methods were used to determine the primary guest adsorption site. However, this cannot be the method used to initially locate these sites. The methods section suggests the initial sites were determined using a simulated annealing approach, but the detail is too sparse here and explaining the use of DMol3. There is no further information in the ESI. It is impossible to judge whether the methods use to determine the sites are appropriate and if they have been appropriately applied, and therefore whether the results shown are likely to be robust. The authors should provide a fuller description of the process for modelling the adsorption sites.
Notwithstanding uncertainty about the simulation methods, the authors propose that interactions between the guest hydrogen atoms and the arene are stronger for the C2H6 than C2H4 because it is more polarizable, but they then go onto say that it is because there are more C-H...pi H-bonds. This section is quite confusing, partly because collectively it is true that the binding energy suggests stronger interaction between ethane and the host, but arguably if these H-bonds dominate, as they are more numerous, individually each C-H...pi is weaker. The evidence here for the polarization theory is not particularly strong. The authors could consider looking at the electrostatic potential of the host and guest to explore this, for example, mapping on the Hirshfeld surface. They should also should include any relevant references to studies of selectivity in other organic materials to strengthen their proposed mechanism. Currently, although it is intuitive, it is not fully supported by the results presented and the authors do not provide any literature to show it has been established previously. Although the system may not be at equilibrium or saturation, the authors should comment on whether additional adsorption sites are expected other than the six equivalent primary sites from the measured gas uptake, or whether all six sites are occupied.
In summary, the manuscript reports the application of an example of a relatively recent class of materials (POCs) to an industrially relevant separation. Although the experimental measurements of gas adsorption and breakthrough separations are satisfactory, the interpretation of the results in the context of the wider field are limited. It is unclear how the molecular simulations included were performed and this undermines the results. The interpretation is also somewhat superficial and confusing and this aspect also lacks reference to previous studies of these types of guests in porous organic hosts. Without major revision, particularly of the simulation sections, I would not recommend publication in Nature Communication. As an experimental study alone, it is slightly limited in scope, but could perhaps be augmented, for example, by measurements of other cages in the family.

Point by point response to the reviewers' comments
Dear Reviewers: We wish to express our appreciation to the referees for their great efforts and suggestions for our manuscript. These comments are valuable and very helpful for our revisions and improvements to our paper. We have tried our best to improve the manuscript and made some changes in the manuscript and the supplementary information. We hope that the revised manuscript will be accepted.

Response:
We thank you for your positive comments on our work. We have tried our best to get the crystal structure with C2 hydrocarbon guest molecules by single-crystal X-ray diffraction. Although this method for structure analysis has been achieved in porous metal-organic framework (MOF) system, it was unsuccessful in our case due to the following two reasons: (1) the peak intensities of x-ray diffraction from CPOC-301 are relatively weak, because there is no heavy atom in the molecular structure of CPOC-301. In fact, it is not able to locate the guest solvent molecules in the inner cavity of CPOC-301, even we have tried many times to collect the single crystals by increasing the exposure time; (2) the crystals of CPOC-301 easily crack into pieces when removed from the mother liquor for a short time, and quickly turn into pieces or powder under vacuum, because the packings of these organic cages are via weak supramolecular interactions. Notably, the pieces or powder have been characterized by PXRD, H-NMR, mass spectrometry and gas sorption, which showed that CPOC-301 is robust.
According to your suggestions: Q1. The desorption curve of N2 gas sorption isotherm is strange. It should be coincided with the adsorption curve since there is no strong interaction. The results should be checked.

Response:
Thanks for the useful suggestion and comments. We have repeated the desorption curve of N2 gas sorption isotherm of CPOC-301 several times, which all showed the same trend.
In fact, all the desorption curve of nitrogen gas sorption isotherms of the reported  Ed. 2020, 59, 19675-19679).

Response:
Thanks for this useful suggestion. We have changed the unit of gas uptake in Qst curve of the C2 hydrocarbons to be consistent with that in adsorption isotherms.
Q3. The most intuitive evidence to verify stability of materials should be the N2 gas sorption isotherm. The N2 adsorption results as well as the PXRD of the sample after exposed to air should be supplied.

Response:
Thanks for the useful suggestion and comments. We have added the N2 adsorption results ( Figure C1) as well as the PXRD of the sample after exposed to air ( Figure C1) in the revised supporting information. Figure C1. N2 gas sorption isotherm at 77 K for CPOC-301 after being exposed to air for a week. Figure C2. PXRD of CPOC-301 after being exposed to air for a week.

Response:
Thanks for the useful suggestion and comments. We have added the breakthrough performance of C2H2/C2H4/C2H6 (1:1:1) ternary mixture of CPOC-301 ( Figure C3), which further confirmed our experimental Qst results (the host-guest interactions between CPOC-301 and C2 hydrocarbon are in the order of C2H2<C2H4<C2H6).  27529-27541). The molecular input of CPOC-301 was generated starting from its crystal structure, in which one of two disordered linkers was selected and the isobutyl group was reduced to a methyl group for simplifying the simulation. And then the initial structure was optimized by the semiempirical extended tight-binding (xtb) program package developed by the Grimme group. The C2 molecules were placed manually at the center of mass of the CPOC-301. To screen for different binding sites, the noncovalent interaction (NCI)/iMTD algorithm in CREST is employed. CPOC-301 was free of any constraints, allowing structural relaxation and adaption to the C2 guest. From the CREST calculation, a structure ensemble of NCI complexes within a 6 kcal mol −1 energy window is obtained. The energetically lowest conformation was selected as the first binding site.
The free CPOC-301, C2 guests and binding site C2@CPOC-301 were further optimized by first-principles dispersion-corrected density functional theory (DFT-D) method by the Dmol3 module as implemented in the Accelrys Materials Studio package. The widely used generalized gradient approximation (GGA) with the Perdew-Burke-Ernzerhof (PBE) functional and the double numerical plus d-functions (DNP) basis set, Grimme method for DFT-D correction were used. The energy, force and displacement convergence criterions were set as 1 × 10 −5 Ha, 2 × 10 −3 Ha and 5 × 10 −3 Å, respectively. Binding energies (ΔE bind in kJ mol −1 ) are calculated as the differences in total energies E between the fully optimized C2@CPOC-301 and the free CPOC-301 and C2 guests (ΔE = EPOC+gas − EPOC − Egas). In fact, the calculated binding sites are consistent with the calculated results last time, and the variation trend of the bonding energy is also consistent. We have added fuller description of the process for modelling the binding sites in the revised manuscript in binding energy calculations section as well as the calculated result in the revised supporting information (Table C1).

Response:
Thanks for the useful suggestion and comments. We have revised "the interactions between the guest hydrogen atoms and the arene are stronger for the C2H6 than C2H4 because it is more polarizable". Moreover, we have also added the Hirshfeld surface analysis of C2 hydrocarbon guests with the calix[4]resorcinarene cavity host to show their intermolecular interactions ( Figure C4). Notably, the strong intermolecular interactions between the C2 hydrocarbons and calix[4]resorcinarene cavity are indicated as bright red spots on the Hirshfeld surface ( Supplementary Fig. 19), which are close to the abovementioned interaction regions of C−H···π bonds between the C2 guests and calix[4]resorcinarene host. We have cited the related references of other organic materials by using the DFT simulation method to calculate the interactions between the C2 gas molecules and framework host as well as their selectivity (see example as J. Am. Chem. Soc. 2020, 142, 633-640;Angew. Chem. Int. Ed. 2021, 60, DOI: 10.1002  Without major revision, particularly of the simulation sections, I would not recommend publication in Nature Communication. As an experimental study alone, it is slightly limited in scope, but could perhaps be augmented, for example, by measurements of other cages in the family.

Response:
Thanks for the useful suggestion and comments. We have added the C2H6 and C2H4 gas sorption curves and experimental column breakthrough results (Figures C5-C8) of other cages in the family including one hexameric octahedra with functional methyl group (CPOC-301-Me) and a trimeric triangular prism (CPOC-201). All the aforementioned three organic cages show the preferential adsorption with C2H6 over C2H4, and can also directly separate C2H4 from the C2H4/C2H6 mixture by column breakthrough experiments. These data further suggest that the C−H···π interactions existing between the C2H6 and calix[4]resorcinarene cavity are stronger than those of the C2H4 molecule. Moreover, the host-guest interactions between CPOC-301 and C2 hydrocarbon are in the order of C2H2<C2H4<C2H6, which have also been validated by the breakthrough performance of C2H2/C2H4/C2H6 mixtures of CPOC-301 ( Figure C3).   Reviewer #1:

Comments:
The detailed modeling studies and experiments have been conducted in the revised manuscript and its quality has been significantly improved. I suggest acceptance for publication after addressing following issues.

Response:
We are thankful to you for accepting the revised manuscript and recommending for publication.
Q1. The C 2 H 2 adsorption curve and C 2 H 6 /C 2 H 2 selectivity should be provided.

Response:
Thanks for this useful suggestion and comment. We have added the C 2 H 2 adsorption (Fig.   S1) curve and C 2 H 6 /C 2 H 2 selectivity (Fig. S2) in the revised supporting information.  Q2. The water stability of POCs is needed.

Response:
Thanks for this useful suggestion and comment. We have added the water stability of CPOC-301 (Fig. S3) in the revised supporting information. Q3. The C 2 H 6 /C 2 H 4 separation performance comparisons of this work and the reported materials is needed to be included in the form of figure or table.

Response:
Thanks for this useful suggestion and comment. We have added the C 2 H 6 /C 2 H 4 separation performance comparisons of CPOC-301 and the reported materials (Fig. S4) in the revised supporting information.

Fig. S4
The reported C 2 H 6 -selective porous organic molecular materials, and several selected C 2 H 6 -selective porous framework materials S1-S9 . Note: the C 2 H 6 Uptake of boronic ester cage was measured at 273 K, and its actual C 2 H 6 /C 2 H 4 separation performance by breakthrough experiment were not investigated.
Q4. The author added some other POCs. The PXRD pattern comparisons of the synthesized samples and the simulated structures is needed. The C 2 H 6 /C 2 H 4 selectivity should be provided.

Response:
Thanks for this useful suggestion and comment. We have added the PXRD patterns (Figs.