Efficient and selective capture of thorium ions by a covalent organic framework

The selective separation of thorium from rare earth elements and uranium is a critical part of the development and application of thorium nuclear energy in the future. To better understand the role of different N sites on the selective capture of Th(IV), we design an ionic COF named Py-TFImI-25 COF and its deionization analog named Py-TFIm-25 COF, both of which exhibit record-high separation factors ranging from 102 to 105. Py-TFIm-25 COF exhibits a significantly higher Th(IV) uptake capacity and adsorption rate than Py-TFImI-25 COF, which also outperforms the majority of previously reported adsorbents. The selective capture of Py-TFImI-25 COF and Py-TFIm-25 COF on thorium is via Th-N coordination interaction. The prioritization of Th(IV) binding at different N sites and the mechanism of selective coordination are then investigated. This work provides an in-depth insight into the relationship between structure and performance, which can provide positive feedback on the design of novel adsorbents for this field.

Thank you for your useful comments on improving the quality of this manuscript.
We have revised the manuscript according to your comments. The revised parts are highlighted in red in the revised manuscript.
The point-to-point responses are summarized below: Comment 1: Authors did not discuss the challenges behind the selective separation of thorium from rare earth elements or uranium.
Response: Thank you for your careful review and for pointing this out. The separation of thorium from rare earth elements and uranium is quite challenging, because of their similar chemical properties (Nat. Commun. 2023, 14, 261;Nanoscale. 2020Nanoscale. , 12, 1339Nanoscale. -1348.

Correction:
However, the separation of thorium from rare earth elements and uranium is quite challenging, because of their similar chemical properties 11,12 .
(Line 41-43, Page 3) Comment 2: CIF files and check if reports should be provided if applicable. Also, details on methodology for structure simulation should be provided. Space group P1 sounds suspicion for such symmetric structure.
Response: Thank you for your valuable and thoughtful comment. We have added the CIF files for studying the selective adsorption mechanism in the supplementary information of the revised manuscript. The Material Studio software was used to construct the Py-TFImI-25 and Py-TFIm-25 COF and was not analyzed by the X-ray single crystal data, hence there is no checkcif report. The details on methodology for structural DFT-simulation are as follows and also added in the supplementary information of the revised manuscript.
In this work, the crystal models of Py-TFImI-25 and Py-TFIm-25 COF were constructed with the Crystal Building module of the Material Studio software. Then, a series of geometry optimizations were performed with UFF force field at the ultra-fine quality in the Forcite module, allowing optimization of all lattice parameters and atomic coordinate. All the optimized COF materials were used for subsequent simulation of the powder diffraction patterns with the Reflex module in the Material Studio software.
Both the eclipsed AA and staggered AB stacking modes were tried for the Py-TFImI-25 and Py-TFIm-25 synthesized here. Compared the experimental XRD results with the simulated XRD, finally, the eclipsed AA stacking modes were settled due to the good match of the experimental XRD results with the simulated XRD on the eclipsed AA stacking mode. Based on the obtained results, periodic density functional theory The structure building expressed that both materials are P1 space group. We suppose that it is because the linking units are composed of 25% BFIIm (or BFIm) and 75% TFTDA, thus leading to such incomplete symmetry.
Comment 3: NMR -supporting information: Assign the peaks of the NMR with the corresponding protons having equivalent chemical environment.
Response: Thank you for the thoughtful comment. The peaks of 1 H NMR spectra of all compounds we synthesized were assigned and the 1 H NMR spectra were supplemented in the revised supplementary information.
Comment 4: How stable is the mesoporous COFs in water? PXRD and BET data should be provided after soaking the COF crystals in water.
Response: Thank you for your professional comments. We have complemented relevant experiments and demonstrated that both materials exhibited good stability after soaking in water and maintained the integrity of the skeleton and the crystal structure (see Fig. R1 and R2).  To provide more details, we added some sentences in the revised manuscript, as follows:

Correction:
The two materials exhibited good stability after soaking in water and maintained the integrity of the skeleton and the crystal structure (Supplementary Fig. 10 and 11). Response: Thank you for the kind comments. It is the as-synthesized samples that were used for the adsorption experiments. During the synthesis of the materials, the solvent molecules were removed as much as possible during repeated washing with THF and acetone, and the detergent was removed by vacuum drying at 120 °C. As shown in Fig. R3, the DTG curves showed that the materials have only one obvious weight loss peak at 535 °C, which was attributed to the sustained thermal decomposition of the skeleton. The BET surface areas of the synthesized materials are about 1400 m 2 g -1 , which is considerably high with the ionic COF reported (Angew. Chem. Int. Ed. 2016, 55, 1737-1741J. Am. Chem. Soc. 2016, 138, 5897-5903;Angew. Chem. Int. Ed. 2017, 56, 4982-4986). Therefore, we believe that the solvent molecules have been removed as much as possible during the synthesis process, and the weight loss of less than 10% of the material in the 25-500°C range is due to the adsorption of water from the air because of the large specific surface area. Therefore, no further activation was performed, which is also consistent with other COF treatments used for the adsorption of metal ions in the literature (Angew. Chem. Int. Ed. 2020, 59, 4168-4175;CCS Chemistry. 2019, 1, 286-295). Comment 6: Is there is any role of particle size (exfoliation of 2D COF sheets) for the adsorption capacity of Th(IV)?
Response: Thank you for your professional question. We supposed that particle size (exfoliation of 2D COF sheets) can play a significant role in the adsorption capacity of Th(IV) by influencing the surface area. The exfoliation of 2D COF sheets can lead to a larger surface area, which can increase the number of effective adsorption sites, thus improving the Th(IV) adsorption capacity. Since Py-TFImI-25 COF and Py-TFIm-25 COF were synthesized with only the linking units BFIIm and BFIm different, all other synthesis conditions were parallel. So, the specific surface areas of the materials are similar (1324.05 m 2 g -1 for Py-TFImI-25 COF, and 1430.15 m 2 g -1 for Py-TFIm-25 COF), which allows us to explore the effect of the different N sites caused by BFIIm and BFIm on the adsorption.

Comment 7:
The experimental condition for the competitive adsorption of Th(IV) from the competing cations are not included in the manuscript?
Response: Thank you for your valuable comment. The experimental condition for the competitive adsorption of Th(IV) from the competing cations is in Section 1.3.4 in the supplementary information. Also, we have added some details to the description of the experimental conditions, as follows.

Adsorption selectivity
A multi-ion solution of Th(IV), U(VI), Sr(II), Cs(I), La(III), Pr(III), Nd(III), Sm(III), Eu(III), and Gd(III) was prepared with each metal ion maintaining at a concentration of around 25 mg L -1 and the pH was adjusted to 4. The solid-liquid ratio was 1:3000 g mL -1 by adding 1.5 mg COF materials into 4.5 mL multi-ion solution. The batch adsorption experiments were carried out at a temperature of 25 °C and a speed of 200 rpm, and then the multi-ion solution after adsorption was separated with a 0.22 μm aqueous nylon filter to measure the concentration of Th(IV) and the competing ions.  We have added the relevant PXRD results in the revised manuscript, as follows:

Correction:
Various characterization approaches were adopted to investigate the interaction mechanism of Py-TFImI-25 COF and Py-TFIm-25 COF towards Th(IV). The obtained PXRD patterns of Py-TFImI-25 COF and Py-TFIm-25 COF after Th(IV) adsorption exhibited consistent diffraction peaks with the pristine ones ( Supplementary Fig. 16), indicating that the adsorption of Th(IV) would not disrupt the crystalline structures. Response: Thank you for your careful review. It is neglected because the distribution coefficients of the two materials for competing ions are too small. We have added the following discussion in the manuscript as follows.
(Line 191-197, Page 11) Comment 12: The experimental details for the activation (degassing) of the COF samples for the BET experiment are missing.
Response: Thank you for your careful review. We have supplemented the activation of the COF samples for the BET experiment as follows.

Correction:
The fresh sample was activated in the degassing station of the instrument at 30 °C for 2 h, and then at 120 °C for 13 h to make the pores guest-free.

(Page 4 in Revised supplementary information)
Special thanks to you for your careful review and constructive suggestions.

Response to reviewer #2:
Thank you very much for your valuable comments. We have added control experiments in response to the comments, revised the manuscript and marked all changes in red on the revised manuscript.
The point-to-point responses are summarized below: Comment 1: The authors state that based on their calculation the Th(IV) affinity follows the order NIm > N-C=N-> NIm-CH 3 . That means imidazole nitrogens bind to Th(IV) more strongly than imine nitrogens. Interestingly, their own previously reported 3D COF, namely COF-DL229, (https://doi.org/10.1016/j.seppur.2022.121413) possessed exclusively imine linkages and exhibited much higher Th(IV) uptake capacity (~500 mg/g vs ~ 140 mg/g of Py-TFIm-25 of this study). We have to mention that the COF-DL229 has a lower surface area (569 m 2 /g vs 1430 m 2 /g of Py-TFIm-25) and a narrower pore size (1.68 nm vs 2.3 nm). That indicates that Th(IV) can penetrate through Py-TFIm-25 more easily (because of larger pores) and be more exposed to nitrogen sites (because of higher surface area). Yet, much lower capacity was obtained with Py-TFIm-25. Does it disprove the claimed higher affinity of imidazole nitrogen mentioned above?
Response: Thank you for your careful review. Since the adsorption capacity is related to the initial concentration of the solution. When the initial concentration of the Th(IV) solution was 300.2 mg L -1 , the Th(IV) uptake capacity of COF-DL229 was 508.1 mg g -1 . When the Th(IV) initial concentration was 274.5 mg L -1 , the adsorption capacity of Py-TFIm-25 COF could reach 822.6 mg g -1 ( Supplementary Fig. 13). With similar Th(IV) initial concentrations, Py-TFIm-25 COF containing imidazole nitrogens exhibited a higher adsorption capacity than COF-DL229 containing imine nitrogens.
Combined with the adsorption behavior of its analog Py-TFImI-25 COF, we suggested that imidazole nitrogen possesses a higher affinity, which is also consistent with the theoretical calculations. The role of the fluorinated fragment from TFTDA was added in the revised manuscript, as follows:

Correction:
The introduction of TFTDA is to enhance the molecular interlayer interaction between the fluorinated build block and non-fluorinated build block 48 , which can improve the crystallinity and chemical stability of the COFs. can see that at pH 5 where the nitrogens are most neutral, Py-TFImI-25 and Py-TFIm-25 perform nearly equivalent. That appears to us that the imine nitrogens play a significant role in the capturing power. At too low pHs (pH 1 and 2), the imine bond is more susceptible to hydrolysis and therefore the same phenomenon is observed. At pH3 and 4, bigger capacity differences are recorded due to the fact that the imines become protonated and hence exist as cations that exert electrostatic repulsion to Th(IV) and this phenomenon is more dramatic in the Py-TFImin-25.
Response: Thank you for your careful review and for pointing this out. With regard to this concern, we would like to interpret from three points. (1) we do agree that imine nitrogen is the actual binding site in Py-TFImI-25 COF, since the nitrogens in the imidazolium have no free electrons to coordinate. However, as shown in Fig. 1b and   Fig. 5c, the nitrogens in benzimidazolium linkers unbound to the methyl group in Py-TFIm-25 COF possess a higher localized charge density, which could be the potential electron donor. (2) Since nitrogens are most neutral at pH 5, as shown in Table R1, the Th(IV) removal rates of Py-TFImI-25 COF and Py-TFIm-25 COF are as high as 99.8% and 99.9%, respectively, and the concentration of Th(IV) after adsorption is close to the detection limit, so the initial concentration of ~25 mg L -1 at pH 5 is not sufficient to reflect the difference in the capture of the two materials, which is also the reason why pH 4 was chosen for subsequent experiments. However, as the initial concentration of Th(IV) continued to increase, the difference in the adsorption capacity caused by different N sites of the two materials became apparent ( Supplementary Fig. 13). (3) As interpreted comment 4, both two materials exhibited good stability after soaking in water. At low pH (pH 1 and 2), the protonation of the nitrogen sites is the reason for the decreased adsorption capacity. At pH 3 and 4, more nitrogen sites become available, resulting in an increased adsorption capacity. The capacity difference is due to the different nitrogen adsorption sites of the two materials (N-C=Nof Py-TFImI-25 COF vs NIm of Py-TFIm-25 COF).  Fig. R1 and R2, both two materials exhibited good stability after soaking in water and maintained the integrity of the skeleton and the crystalline structure. In addition, in our subsequent study of the fluorescence property of Py-TFImI-25 COF, as shown in Fig. R3, Py-TFImI-25 COF have not been hydrolyzed in acid solutions, otherwise, the fluorescence intensity would not be enhanced with increasing acidity.   We have supplemented the water stability of the two COFs in the revised manuscript, as follows:

Correction:
The two materials exhibited good stability after soaking in water and maintained the integrity of the skeleton and the crystal structure ( Supplementary Fig. 10 and 11). Once again, thank you very much for your comments and suggestions.

Response to reviewer #3:
Thank you very much for your important and helpful comments. We have revised the manuscript following the comments and marked all the amends in red on our revised manuscript.
The revision details are listed as follows: Comment 1. There are structural formula errors in Fig.1. Please note the correct drawing of the pyrene unit.
Response: Thank you for your careful review. We have corrected them in the revised manuscript, as follows.  (Table   R1). So, we have removed the relevant description and made the following revisions in the revised manuscript.

Correction:
The results showed that although Py-TFImI-25 COF and Py-TFIm-25 COF have analogous structures, the new N sites on the benzimidazolium linkers of Py-TFIm-25 COF could be responsible for the improved sorption performance.

Response
ions are preferable to interact with the Py-TFIm-25 COF by occupying the adsorption sites, resulting in the significantly high adsorption selectivity towards Th(IV) for the adsorption experiments in this work. When multiple metal ions in solution compete for adsorption sites, the metal ions with a strong affinity for the adsorption site will preferentially occupy the adsorption site, preventing those with a relatively weak affinity from binding to the sites. This phenomenon is consistent with the previous report (Chem. Eng. J. 2018, 344, 594-603).

Comment 6:
The source of the main materials should be marked.

Response:
Thank you for your careful review and valuable advice. We have supplemented the sources of the main materials in the supplementary information of the revised manuscript.

(Page 3 in Revised supplementary information)
Special thanks for the constructive comments, which greatly helped us to improve the manuscript.

Reviewers' Comments:
Reviewer #1: Remarks to the Author: All my comments have been satisfactorily addressed. Publication is recommended.
Reviewer #2: Remarks to the Author: Liu et al. made commendable efforts to respond to our comments; however, we still find the manuscript insufficient for acceptance in Nature Communications. The main purpose of the paper is to map Thorium adsorption activity according to the electronic nature of different nitrogen functionalities. The authors' calculations show that the charge density of imidazole nitrogen is greater than that of imine and imidazolium, and they correlate this to the observed adsorption activity. Specifically, they attribute the greater capacity of Py-TFIm-25 to the presence of the neutral imidazole nitrogen. However, we believe that the experimental data do not support this claim. We think that the major adsorption sites are still the imine nitrogens, expressed by a significant capacity of the charged Py-TFImI-25 at high concentrations (>400 mg/g, Fig. S13). The reason why Py-TFIm-25 has a much greater maximum capacity (822 mg/g, Fig. S13) is because of its less severe positive chargepositive charge repulsion with thorium cations. Such repulsion becomes more significant when the pores are more packed with higher quantities of cations, provided by higher concentrations. This interaction also affects the observed adsorption kinetics. At low concentrations, the data show that the two materials perform quite similarly (Fig. 3a).
We have four main further comments on the manuscript: Comment #1: The authors pointed out that at a similar concentration, the uptake capacity of Py-TFIm-25 COF is higher than that of COF-DL229. Although this is the observed data, it does not necessarily correlate to the higher affinity of imidazole nitrogen since it can still be due to the higher porosity of TFIm-25 COF. Furthermore, with the particular structures, the imine's electron density is lower than that of imidazole since it is highly deactivated by the electron-withdrawing effect of the fluorinated moiety. Therefore, it would be beneficial to have a non-fluorinated version of the series to have a fairer assessment of the relative affinity.
Comment #2: The authors explained their choice of the fluorinated TFTDA building block to enhance the interlayer interaction between the fluorinated moieties and non-fluorinated ones, which can lead to an improvement in crystallinity. However, the XRD spectra revealed an eclipsed AA stacking structure, which means the fluorinated benzene rings are right on top of each other. Does this contradict the stated electronic interaction? In our opinion, it would be more informative if the authors obtained nonfluorinated structures so that none of the nitrogen functionalities' electronics is compromised to have a more conclusive judgement.
Comment #3: The authors defended their claim of the role of imidazole nitrogen as the major binding force by presenting the increasing capacity gap between Py-TFImI-25 and Py-TFIm-25 when the initial thorium concentration is increased. We do not find this a convincing proof since when binding strength is under focus, sorbate concentration should be examined at low levels at which theoretical capacity (capacity at which all binding sites in a given structure reach their stoichiometric capture of the target) has not been reached to obtain an accurate insight into chemical interaction. Any excess capture beyond theoretical capacity must be the result of other secondary effects such as the reduction of the cations or interaction between adsorbed cations. For that, we would like to know the theoretical capacity of this series. Furthermore, the authors say that the data at pH 4 (bigger capacity gap) should be based on rather than those at pH 5 (little gap) since 1) at pH 5 and 25 ppm, the uptake in both materials are not at saturation points, and 2) at pH 4, more protonation would reduce the uptake capability, leading to capacities obtained from the available nitrogen sites of the two materials (N-C=N-of Py-TFImI-25 COF vs NIm of Py-TFIm-25 COF). While the former statement may be true, the latter is not quite accurate since, given the higher charge density calculated for the Py-TFIm-25 imidazole nitrogen, this nitrogen site would be protonated before the imine nitrogens. Moreover, the majority functional group in both materials is still the imine. As a result, based on the collected data, we believe that the imine nitrogen is the main binding site.
Comment #4: The authors demonstrate the water stability of the imine COFs by showing the XRD patterns before and after treatment with water. However, we would like to know the mass recovery of the materials after the treatment. Did the authors recover the same quantity of materials after treatment as they started with? Also, the stability should be tested by dispersing the COF powder in aqueous solutions for prolonged time. Not one hour.
Reviewer #3: Remarks to the Author: The authors have noticed the inadequacies in the manuscript and have revised them carefully according to the questions and suggestions raised by the reviewers. I think the revised manuscript can now be accepted for publication in Nature Communications.

Response to reviewer #2:
Thank you for your professional comments on improving the quality of this manuscript, which also help us to understand more deeply. We have interpreted and revised the manuscript according to your comments. The revised parts are highlighted in red in the revised manuscript.
The point-to-point responses are summarized below: Liu et al. made commendable efforts to respond to our comments; however, we still find the manuscript insufficient for acceptance in Nature Communications. The main purpose of the paper is to map Thorium adsorption activity according to the electronic nature of different nitrogen functionalities. The authors' calculations show that the charge density of imidazole nitrogen is greater than that of imine and imidazolium, and they correlate this to the observed adsorption activity. Specifically, they attribute the greater capacity of Py-TFIm-25 to the presence of the neutral imidazole nitrogen. However, we believe that the experimental data do not support this claim. We think that the major adsorption sites are still the imine nitrogens, expressed by a significant capacity of the charged Py-TFImI-25 at high concentrations (>400 mg/g, Fig. S13). The reason why Py-TFIm-25 has a much greater maximum capacity (822 mg/g, Fig. S13) is because of its less severe positive charge -positive charge repulsion with thorium cations. Such repulsion becomes more significant when the pores are more packed with higher quantities of cations, provided by higher concentrations. This interaction also affects the observed adsorption kinetics. At low concentrations, the data show that the two materials perform quite similarly (Fig. 3a).

Comment 1:
The authors pointed out that at a similar concentration, the uptake capacity of Py-TFIm-25 COF is higher than that of COF-DL229. Although this is the observed data, it does not necessarily correlate to the higher affinity of imidazole nitrogen since it can still be due to the higher porosity of TFIm-25 COF. Furthermore, with the particular structures, the imine's electron density is lower than that of imidazole since it is highly deactivated by the electron-withdrawing effect of the fluorinated moiety. Therefore, it would be beneficial to have a non-fluorinated version of the series to have a fairer assessment of the relative affinity.
Response: Thank you for your careful review and for pointing this out. We agree that it is not available to conclude that imidazole nitrogen has higher affinity by   electronics is compromised to have a more conclusive judgement.
Response: Thank you for your professional comment, which helped us to reconsider the structures of the two COFs. The two COFs are eclipsed AA stacking modes, but with the anti-isomer stacking mode, as shown in Fig. R2. The anti-isomer mode is more stable than the current syn-isomer mode due to the interlayer interaction (J. Am. Chem. Soc. 2013, 135, 546-549). Since the adsorption sites are mainly located in the N atoms and not in the interlayer, the stacking of COFs in the z direction has little effect on the adsorption property. However, to ensure the rigorousness of the data, we redid relevant quantitative calculations based on the anti-isomer mode, and the results showed that the calculation results based on the anti-isomer mode did not change much compared with the previous ones (Table R2). Corresponding amendments have been highlighted in red in the revised manuscript.
As interpreted in Comment 1, since the effect of F atoms on N atoms in both COFs is the parallel, we suppose that it would not affect the convincingness of the conclusion.
As shown in Table R1, the affinity order of the N sites in the non-fluorinated structure is consistent with that of the fluorinated structures.  we would like to know the theoretical capacity of this series. Furthermore, the authors say that the data at pH 4 (bigger capacity gap) should be based on rather than those at pH 5 (little gap) since 1) at pH 5 and 25 ppm, the uptake in both materials are not at saturation points, and 2) at pH 4, more protonation would reduce the uptake capability, leading to capacities obtained from the available nitrogen sites of the two materials (N-C=N-of Py-TFImI-25 COF vs NIm of Py-TFIm-25 COF). While the former statement may be true, the latter is not quite accurate since, given the higher charge density calculated for the Py-TFIm-25 imidazole nitrogen, this nitrogen site would be protonated before the imine nitrogens. Moreover, the majority functional group in both materials is still the imine. As a result, based on the collected data, we believe that the imine nitrogen is the main binding site.
Response: Thank you for the thoughtful comment. As shown in Table R3, the theoretical capacities of Py-TFImI-25 COF and Py-TFIm-25 COF are 890.9 mg g -1 and 1206.4 mg g -1 , respectively, based on the Th and N coordination ratio of 1:1 (Fig. 6a).
The equilibrium adsorption capacities obtained so far are lower than the theoretical capacities.
Since the imidazole nitrogen site would be protonated before the imine nitrogen, we analyzed the occupation of the nitrogen sites at different thorium concentrations.
When the initial concentration of Th(IV) is about 22 mg L -1 and the pH is 4, the initial concentration is 0.1 mmol L -1 for Th(IV) and 0.1 mmol L -1 for H + , while the content of imidazole nitrogen (NIm) is 0.35 mmol L -1 (Table R4). Moreover, the hydrogen ions will not be completely bound to imidazole nitrogen sites due to the presence of protonation constant. Therefore, at this concentration, the main adsorption site of Py-TFIm-25 COF is the imidazole nitrogen, while the adsorption site of Py-TFImI-25 COF is the imine nitrogen.
When the initial Th(IV) concentration increased 12 times, the uptake capacity of Py-TFImI-25 COF for Th(IV) increased by 10.9 times and the uptake capacity of Py-TFIm-25 COF for Th(IV) increased by 13.4 times (Table R4 and Table R5). In this adsorption system, there are hydrolysis equilibriums of thorium ions, proton equilibriums, and reaction equilibriums. When the pH is fixed, the difference between Py-TFImI-25 COF and Py-TFIm-25 COF adsorption systems should focus on the difference of the reaction equilibrium constants, k1 for the imine site and k2 for the imidazole site. The 10.9-fold increase in the uptake capacity of Py-TFImI-25 COF is due to the contribution of k1, while the 13.4-fold increase in the uptake capacity of Py-TFIm-25 COF is due to the combined contribution of k1 and k2. Therefore, at higher concentrations, the contribution of imidazole nitrogen cannot be neglected. COFs N-C=N-content (mmol g -1 ) NIm content (mmol g -1 ) NIm-CH 3 content (mmol g -1 ) Total N content (mmol g -1 ) Theoretical capacity (mg g -1 ) Py-TFImI-25 3.84 0 1.92 5.76 890.9 Py-TFIm-25 4.16 1.04 1.04 6.24 1206.4 showing the XRD patterns before and after treatment with water. However, we would like to know the mass recovery of the materials after the treatment. Did the authors recover the same quantity of materials after treatment as they started with? Also, the stability should be tested by dispersing the COF powder in aqueous solutions for prolonged time. Not one hour.
Response: Thank you for your professional comments. Unfortunately, we were unable to obtain the mass recovery due to the unavoidable loss of samples during the transfer process. However, we hope that the N content in water before and after water treatment could indicate the stability of the two COFs. The N content before and after water treatment was carried out on HQ40D-Multi (Hach). As shown in Table R6, the loss of both COFs is less than 1.5%, indicating good water stability. Since all adsorption experiments were completed within 1 h, we choose 1 h to test the water stability.
Meanwhile, 6 h was set to test the water stability. As shown in Fig. R3, both materials exhibited good stability after soaking in water and maintained the integrity of the skeleton and the crystal structure. We appreciate your professional work earnestly. Thank you very much for your comments and suggestions.