Benchmark single-step ethylene purification from ternary mixtures by a customized fluorinated anion-embedded MOF

Ethylene (C2H4) purification from multi-component mixtures by physical adsorption is a great challenge in the chemical industry. Herein, we report a GeF62- anion embedded MOF (ZNU-6) with customized pore structure and pore chemistry for benchmark one-step C2H4 recovery from C2H2 and CO2. ZNU-6 exhibits significantly high C2H2 (1.53 mmol/g) and CO2 (1.46 mmol/g) capacity at 0.01 bar. Record high C2H4 productivity is achieved from C2H2/CO2/C2H4 mixtures in a single adsorption process under various conditions. The separation performance is retained over multiple cycles and under humid conditions. The potential gas binding sites are investigated by density functional theory (DFT) calculations, which suggest that C2H2 and CO2 are preferably adsorbed in the interlaced narrow channel with high aff0inity. In-situ single crystal structures with the dose of C2H2, CO2 or C2H4 further reveal the realistic host-guest interactions. Notably, rare C2H2 clusters are formed in the narrow channel while two distinct CO2 adsorption locations are observed in the narrow channel and the large cavity with a ratio of 1:2, which accurately account for the distinct adsorption heat curves.


REVIEWER COMMENTS
Reviewer #1 (Remarks to the Author) In this work authors reported GeF6 2anion incorporated MOF named ZNU-6. The MOF has optimum pore structure and environment for one step ethylene purification and also high selective for carbon dioxide separation as well. This research area is very promising especially today when we need to find better alternative for energy storage and separation. The authors investigated the MOF in different conditions and proved its viability and effectiveness. The chemical approach of this work is strong, and the results are clearly visible from the experiments However here are some of my questions and concerns 1. The topic of ethylene purification/ separation using MOF is not very new, even incorporation of anions in MOFs has been reported by the authors group recently (Wang et al, Nano Research 2022). It would be great if authors can describe the novelty of this research as suitable of Nature common. 2. Its seems that the customized pore environment helps to achieve higher selectivity however DFT studies, and in-situ X ray have been performed to corroborate the statement. Have the authors performed any in-situ IR to verify the conclusions? This evidence will help strengthen the case 3. Heat of adsorption values and isotherm dependency with temp indicates the physisorption nature of guest /host chemistry. However, the additional interaction with anions and even TPA can provide some chemisorption in CO2. Have the authors performed any Temp programed desorption studies in terms of gases of choice to see the binding behavior? 4. What is the relative humidity used in this study? 5. We can see ZNU-6 is strong water adsorbent however in humid conditions shows negligible deterioration of separation performance. However, the reasoning and data is not very clear, has the authors tried to quantify the effect of humidity in this case. 6. In similar topic, what happen in desorption cases in breakthrough, I mean when the fixed bed is undergoing regeneration, have the authors quantified the effect of humidity in gas desorption cases and in cyclic performance?
Reviewer #2 (Remarks to the Author) The authors reported a GeF6 2anion embedded MOF, ZNU-6, with optimized pore structure and environment for highly efficient C2H4 recovery from various C2H2/CO2/C2H4 ternary mixtures. The material exhibited good recyclability and resistance toward moisture during breakthrough experiments. Ethylene (C2H4) purification from multi-component mixtures by physical adsorption is an important industrial challenging, and this work will be very interesting for the readers. The experiments were well performed and the manuscript has been written well. I agree that this manuscript to be accepted for Nature Communications after some minor revision. The following are the suggestions for the authors to improve the manuscript. 1. Figure 1 Caption, (f) is inconsistent with other numbering. 2. Supplementary Figure 10-11, the authors should provide the pressure of these IAST selectivities. 3. For the in-situ gas-loaded MOF structure, the authors should add a figure to describe the C2H2 gas molecule cluster and the interactions between the molecules, either in main text or SI. A figure illustrating the interaction between adsorbed water and gas molecules should also be added.
Reviewer #3 (Remarks to the Author) Thanks to its potential to introduce energy-efficient, single-step ethylene purification approaches, this manuscript by Jiang et al. is of high topical relevance to gas purifications, and MOFs for separations. Considering high importance of the new findings, and their general relevance in controlling the pore electrostatics (F···C=O interactions) driven gas separation/purification properties, I support publication subject to necessary revisions as follows: 1. Molecular formulae are missing including that of the new as-synthesized ZNU-6 (although I find this formula in Table S7, should be clearly noted in the manuscript too). All molecular formulae for the gas-loaded phases should be clearly written in the main article, if needed, using a table with analysis of sorbate-sorbent interactions (distances) 2. The authors should not present the adsorption data with respect to molecules per GeF6 2anion, this makes the data skewed in favour of this adsorbent, ZNU-6. Also, the units for adsorption uptakes and pressure need to be consistent. Right now, with mixed use of cm 3 /g, mmol/g, mol/mol and mmol/g, these are all mixed up; same mix-up is observed in the units used for pressure, bar and kPa. Collectively, all these mistakes come together in the manuscript Figure 2. I strongly recommend replotting the isotherm-based uptakes in one consistent pair of units: mmol/g and bar. This not only makes this manuscript coherent but also helps the whole community working in this area with regard to comparing performance parameters across different adsorbents. 3. Page 2, lines 33-34, the authors need to discuss "Presently, multi-step purification process is adopted for purification of C2H4 from C2H4/C2H2/CO2 mixtures." This needs to take cognisance of the recent reports on single-step C2H4 purification from ternary (1:1:1 for C2H2/C2H4/C2H6) and quaternary (1:1:1:1 for C2H2/C2H4/C2H6/CO2) gas mixtures. Some examples include: Cao, JW. et al., Nat Commun. 2021, 12, 6507. Xu, Z. et al., Nat Commun. 2020. I don't quite agree with the authors use of the word "static" in the heading of Figure  2. Should just be written as "The sorption performance." In another instance, the authors use this phrase "static adsorption" (page 11, line 202), which I object to. 5. It is stated that the C2H4 productivity is 309 mL/g, again this unit is far from the standard unit typically used in other literature reports in this area, i.e., mol kg -1 h -1 6. Importantly, the authors have calculated productivity by integrating the effluent flow rate of C2H4 (cm 3 /min). However, all the breakthrough curves (in Fig. 5) are showed in with the effluent concentration (C/C0). It is obvious that integrating the concentration curve cannot give the amount, although a few references adopt this wrong method. Therefore, it is necessary to show the details for direct measurement and/or indirect calculation of the effluent flow rate and concentration in this study.
See Shen, J. et al., Nat Commun. 2020, 11, 6259 as a reference. 7. How common / rare is the ith-d topology among the anion-pillared MOF library? It would greatly help the article/future readers, if the authors can address this by a thorough literature-based contextualisation of the structural topology of ZNU-6. 8. Section 7 in the supplementary information has a header "Breakthrough simulations and experiments", the following figure captions for Supplementary Figures 28-43 are all including "Experimental" in their figure captions, nonetheless. I wonder where the simulation-derived breakthrough data is? Overall, an interesting idea executed by the authors that should advance this area in the near future. I will be glad to look at a suitably revised article, when ready.
Reviewer #4 (Remarks to the Author) This manuscript by Jiang et al, report a metal organic framework (ZNU-6) for the application of simultaneous removal of C2H2 and CO2 from C2H4 stream. This area is of important practical applications and has been extensively investigated in the past few years. The overall quality of this manuscript is high with detailed investigation of the structure characterisation of the MOF, study of its adsorption properties using isotherm and breakthrough experiments, and thorough investigation of the host-guest interactions. Below are some comments and questions from me, and I would recommend the publication of this manuscript after the authors fully address them: 1. I think it is necessary that the authors reference some highly relevant publications in the introduction, e.g. Hexafluorogermanate (GeFSIX) Anion-Functionalized Hybrid Ultramicroporous Materials for Efficiently Trapping Acetylene from Ethylene, Ind. Eng. Chem. Res. 2018, 57, 21, 7266-7274. 2. The cif. File for CO2 loaded ZNU-6 shows an O-C-O bond angle of CO2 being 157° instead of the theoretical 180°, could the authors please explain this discrepancy. 3. Figure 2f, the Qst for CO2, why it showed a sudden drop at 2 mmol/g coverage? 4. From the isotherms for CO2 and C2H4 (Figure 2b, 2d), the saturation uptake for CO2 and C2H4 are very close; and the kinetic data (supplementary figure 25) of CO2 and C2H4 are also similar with C2H4 being slightly faster in adsorption. For Qst, CO2 started higher than C2H4 then fall to be lower than C2H4. These three parameters (uptake, kinetic, Qst), all seem indicating the interaction between the gas molecules and the framework is very similar. In this case, how do the authors rationalise the observed separation in breakthrough experiments? what do the authors think is the really reason that ZNU-6 can retain CO2 from mixtures containing mainly C2H4 and small percentage of CO2.

Comments from Reviewer 1:
Overall comment. In this work authors reported GeF6 2anion incorporated MOF named ZNU-6. The MOF has optimum pore structure and environment for one step ethylene purification and also high selective for carbon dioxide separation as well. This Compared with those, the investigation on the single-step C2H4 purification from C2H2/CO2/C2H4 has been investigated much less (no more than 10 materials). The reference mentioned in comment 1 (Wang et al, Nano Research 2022) is for C2H2 purification from other mixtures, different from this work. So far, TIFSIX-17-Ni, NTU-65 and NTU-67 are the optimal materials. However, each material has significant drawback. The capacity of C2H2 (3.30 mmol/g) and CO2 (2.20 mmol/g) is relatively low in TIFSIX-17-Ni due to the over-contracted channel. NTU-65 capture C2H2 and CO2 by tuning the gate opening, but the applied temperature must be at 263 K because lower temperatures lead to the adsorption of all the gases while higher temperatures cause the exclusion of CO2. NTU-67 displays modest C2H2 (3.29 mmol/g) and CO2 (2.04 mmol/g) capacity as well as reduced C2H2/C2H4 and CO2/C2H4 selectivity compared to TIFSIX-17-Ni. Besides, the separation performance of NTU-67 is deteriorated under humid conditions. Therefore, the trade-off between adsorption capacity and selectivity in the C2H4 purification from C2H2/CO2/C2H4 is still challenging to overcome by existing porous materials. ZNU-6 reported in this work is a novel MOF with both narrow interlaced channels, large pores and abundant functional sites for record C2H2/CO2/C2H4 separation. MOFs with such kind of optimal pore chemistry and pore size/shape are rare.
Besides, such high stability in humid air and water is difficult to realize in anion pillared MOFs due to the weak Cu-N coordination bonds. Notably, our study include in-situ single crystal structure analysis, in-situ IR spectra analysis and DFT calculation. All the results are consistent and lead to the same conclusion. There has never been a report that provides such detailed mechanism study in C2H2/CO2/C2H4 separation. Therefore, there are several novel points in this research and all of these can be found in the main text.

Comment 2.
Its seems that the customized pore environment helps to achieve higher selectivity however DFT studies, and in-situ X ray have been performed to corroborate the statement. Have the authors performed any in-situ IR to verify the conclusions? This evidence will help strengthen the case Author response: Thank you for your suggestion, the in-situ IR spectroscopy was conducted on gas-loaded and activated samples, and the results have added into supplementary materials. New and obvious stretching bands that belong to C2H2 and CO2 are observed in the C2H2 and CO2 dosed single crystals. The νas(C2H2) and ν(C≡C) stretching band of adsorbed C2H2 down-shifted to 3160 and 1720 cm -1 respectively with reference to the gas-phase value at 3287 and 2500-1900 cm -1 , indicating the existence of guest-host interactions. Similarity, ν(CO2) band also undergoes a downward shift from gas-phase value 2349 cm -1 to 2335 cm -1 , showing the interactions between CO2 and framework. In contrast, the stretching band of C2H4 is not obvious in C2H4@ZNU-6.
All the IR spectroscopic data are recorded in a Nicolet iS5 ATR-FTIR spectrometer.
The samples of gas-loaded crystals were prepared by the method described in Preparation of gas loaded ZNU-6 in manuscript.
As shown in the Supplementary Figure 5, new and obvious stretching bands that belong to C2H2 and CO2 are observed in the C2H2 and CO2 dosed single crystals. The νas(C2H2) and ν(C≡C) stretching band of adsorbed C2H2 down-shifted to 3160 and 1720 cm -1 respectively with reference to the gas-phase value at 3287 and 2500-1900 cm -1 , indicating the existence of guest-host interactions. Similarity, ν(CO2) band also undergoes a downward shift from gas-phase value 2349 cm -1 to 2335 cm -1 , showing the interactions between CO2 and framework. In contrast, the stretching band of C2H4 is not obvious in C2H4@ZNU-6. Author response: Thank you for your suggestion. Firstly, there is no hysteresis in the desorption curves, indicating that the interactions between gas molecules and framework are not too strong. Secondly, we have performed the desorption at room temperature. The reactivation condition between the 5 th cycle and 6 th cycle adsorption experiment is under dynamic vacuum at room temperature for three hours, as shown in Supplementary Fig. 26, the uptake of 6 th cycle is similar with that of 5 th cycle. Such mild regeneration condition shows that the interactions between CO2 and ZNU-6 are relatively week. Thus, it belongs to physisorption rather than chemisorption. Finally, as shown in in-situ crystals, the binding sites of CO2 are close to GeF6 2anion in either large cage or small interlaced channel, this shows that the interactions between CO2 and GeF6 2anion are stronger than those between CO2 and TPA. Besides, due to the steric hinderance of pyridine rings, the CO2 molecules are difficult to approach TPA ligands. In summary, we believe that the whole CO2 adsorption behavior belongs to physisorption and the strongest interaction is between CO2 and GeF6 2anion.

Comment 4. What is the relative humidity used in this study?
Author response: The relative humidity used in the breakthrough experiments is 60%.

Comment 5. We can see ZNU-6 is strong water adsorbent however in humid conditions
shows negligible deterioration of separation performance. However, the reasoning and data is not very clear, has the authors tried to quantify the effect of humidity in this case.
Author response: We have quantified the effect of humidity on the breakthrough by calculating the C2H4 productivity. As shown in Table S9 in supplementary, the C2H4 productivity is 13.81 and 13.79 mol/kg respectively under dry and humid conditions. The productivity decrease is only 0.14%, in the range of measurement error . The reasons for the negligible deterioration are mainly attributed to the slow diffusion of water molecules in the framework. As shown in supplementary Fig 28 and 48, the H2O adsorption is very slow. In real breakthrough conditions, less than 30% of saturated H2O amount is adsorbed within 10.6 h while the gas mixture breakthrough experiments at 298 K are all finished within 200 min. The quite slow diffusion may be resulted from the small hydrophobic widows between large cage and interlaced channel. On the other hands, our in-situ single crystal structure have showed the water can be co-adsorbed in ZNU-6 without the reduction of gas loading. In the original main text (line 226-230), we have showed the reasons: "Although many water molecules can be adsorbed in ZNU-6, as described in in-situ crystals and water adsorption isotherms ( Supplementary Fig. 27), the presence of humid has negligible influence on the separation performance (Fig. 5f). This is probably due to the co-adsorption of water and target gases as well as the fast C2H2/CO2/C2H4 diffusion kinetics ( Supplementary Fig. 29-31)."   conditions, the results are shown in the graph below. It is obviously that the influence of moisture is negligible on the process of the regeneration. This is also reflected by the overlapping (adsorption) breakthrough curves (Fig 5f).

Comments from Reviewer 2:
Overall comment. The following are the suggestions for the authors to improve the manuscript. Comment 1. Figure 1 Caption, (f) is inconsistent with other numbering.
Author response: Thanks for your reminder. We have corrected the Figure 1 Caption

Modification:
Manuscript: Page 5 Fig. 1   Fig. 1: a-c Exquisite control of pore size/shape and pore chemistry in ZNU-6 from pillared (3,4)-connected pto network to GeF6 2embedded ith-d topology framework; d Overview of ZNU-6 structure with cage-like pores and interlaced channels. e Structure and size of the cage-like pore. f Structure and size of the interlaced channel connecting four cages. Figure 10-11, the authors should provide the pressure of these IAST selectivities. There are two kinds of interactions between C2H2 molecules in the site I. One is the C···H interactions, whose distances are between 2.2 and 2.6 Å, and the other is π···π interactions between C≡C bonds, which are all in the distance of 2.4 Å.
Manuscript: Page 11 Fig. 4 This not only makes this manuscript coherent but also helps the whole community working in this area with regard to comparing performance parameters across different adsorbents.
Author response: Thanks for your suggestion, to make the manuscript coherent and also helps the whole community working in this area with regard to comparing performance parameters across different adsorbents. we have unified the pressure units to bar, and the uptake units to mmol/g in the manuscript, especially in the Fig. 2.
The Fig. 2c that was the adsorption data with respect to molecules per anion before has been replaced by the adsorption data in the units of mmol/g as suggested and the origin Fig.2c has been moved to supplementary. However, we want to make an interpretation why we chose different units to present the adsorption data at first.
Different units have different significance, mmol/g (or STP cm 3 /g) is the basic uptake unit, the data in this units can be straightly obtained from the adsorption equilibrium measurements. While the density of single crystal is identified, the uptake data in cm 3 /g can be converted to that in cm 3 /cm 3 to evaluate the uptake of the adsorbent at a certain volume which is more useful for industrial application. To evaluate the influence of anions on the adsorption accurately, and to eliminate the effect brought by density and molecular mass simultaneously, mol/mol was chosen to be the uptake unit. The uptake data in mol/mol represents the number of the gas molecules those are adsorbed by per anion.

Manuscript: Page 3 Line 47
However, the capacity of C2H2 (3.30 mmol/g) and CO2 (2.20 mmol/g) is relatively low due to the over-contracted channel.

Manuscript: Page 3 Line 58-61
Static gas adsorption isotherms showed that ZNU-6 takes up 1.53/8.06 mmol/g of 2.21 mmol/g (Fig. 2b), even higher than the uptakes of many porous materials at 1 bar and 298 K, for example, TIFSIX-17-Ni (3.30/2.20 mmol/g). 36 In the meantime, the C2H4 uptakes on ZNU-6 at 0.01 and 0.1 bar are only 0.15 and 1.07 mmol/g, much lower than those of C2H2 and CO2 under the same conditions. The C2H2, CO2 and C2H4 adsorption isotherms were further collected at 278 and 308 K (Fig. 2d) Cao, JW. et al., Nat Commun. 2021, 12, 6507. Xu, Z. et al., Nat Commun. 2020 Author response: Thanks for your suggestion. However, we have to make a clarity: lines 33-34 in page 2 shows the industrial situation instead of the current status of C2H4 purification by physical adsorption. To avoid misunderstanding, we have modified the manuscript according to your advice, and we have added the references to the corresponding places in page 3 line 41-42.

Manuscript: Page 3 Line 41-42
Besides, single-step purification of C2H4 from ternary C2H2/C2H4/C2H6 33, 34 or quaternary C2H2/C2H4/C2H6/CO2 35 mixtures has also been realized by several porous materials. Author response: Thank you for your suggestion. We have modified the Figure 2 caption and the sentence in the manuscript.
Author response: Thank you for your question. However, we need to clarify one point that we choose mL/g as the unit of x-axis instead of cm 3 /min, the normalized flow volume is calculated by the formula "Flow volume (mL/g)= flowrate (mL/min) × time (min) /sample weight (g)". The real flow rates and original figures of experiments with time (min) as x-axis had been presented in the supplementary. According to formula and the figure below, the flow rates have been considered in the calculation of C2H4 productivity. Therefore, the productivity can be straightly calculated out by the y-axis which represents effluent concentration (C/C0). Comment 3 Glad about the authors edits on including a few updated citations from 2020, 2021, and 2022.
Author response: Thank you for your positive comment.
Comment 4 I appreciate the authors omitting the word "static" before "adsorption".
Author response: Thank you for your positive comment.
Comment 5 I do not quite agree with the authors arguing that use of the more commonly used unit for productivity, mol kg -1 h -1 will neglect flow rate. The calculation of productivity with respect to this unit takes into account the flow rate used, and therefore, the authors are again advised to use the unit mol kg -1 h -1 for quantifying the C2H4 productivity.
Author response: Thank you for your suggestion. We have added the productivity in the unit of mol kg -1 h -1 to the corresponding places. We have chose two time period, one is from 0 to Time 1, and the other is from 0 to Time 2 as shown in Supplementary Table 10..

Modification:
Manuscript: Page 12 Line 220-222 C2H4 productivity with the unit of mol/kg/h is also calculated for comparison (Supplementary Table S10). ZNU-6 with the productivity of 15.93 mol/kg/h is still the best material.
Supplementary: Author response: Thank you for your positive comment.

Comment 7
The authors' response (in the revised introduction) on quantifying the rarity of ith-d topology among anion-pillared family of MOFs is satisfactory.