Microporous polymer adsorptive membranes with high processing capacity for molecular separation

Trade-off between permeability and nanometer-level selectivity is an inherent shortcoming of membrane-based separation of molecules, while most highly porous materials with high adsorption capacity lack solution processability and stability for achieving adsorption-based molecule separation. We hereby report a hydrophilic amidoxime modified polymer of intrinsic microporosity (AOPIM-1) as a membrane adsorption material to selectively adsorb and separate small organic molecules from water with ultrahigh processing capacity. The membrane adsorption capacity for Rhodamine B reaches 26.114 g m−2, 10–1000 times higher than previously reported adsorptive membranes. Meanwhile, the membrane achieves >99.9% removal of various nano-sized organic molecules with water flux 2 orders of magnitude higher than typical pressure-driven membranes of similar rejections. This work confirms the feasibility of microporous polymers for membrane adsorption with high capacity, and provides the possibility of adsorptive membranes for molecular separation.

The manuscript is suitable for publication in Nature Communications, but some technical concerns should be addressed. There are some concerns about the technical aspects of the results and analyses.
1. It is an innovative idea to tune the charge of the polymer by pronation and deprotonation of the AO groups and use the membranes to separate anionic and cationic dyes. One technical problem is that the protonation of AO groups in acidic solution could lead to significant decay of mechanical properties. Did the authors measure the mechanical properties of membranes treated at low pH? 2. The adsorption mechanisms need in-depth study and analysis. It is known that for these microporous materials with charges, the adsorption of dye molecules may be dominated by electrostatic adsorption on the surface of membranes instead of purely adsorption in the micropores. The schematic diagram shown in Fig. 1c is likely incorrect. Do you have direct evidence to claim that all the dye molecules are adsorbed in the micropores? 3. The adsorption capacity presented in Figure 3e needs to be explained carefully. Because the molecules are likely adsorbed on the surface of membranes and form a cake layer, the adsorption of these molecules are apparent values instead of intrinsic values. Also the capacity should be linked with the pore volume of the membranes available for adsorption instead of membrane area, because the thickness of these membranes are different. So the adsorptive capacity unit should be changed to g/m3 instead of g/m2. The authors tried to compare the rejection and flux trade-off with conventional separation membranes based on size and charge. In my opinion, this is not a fair comparison. When the porous polymers are used as adsorbents, the separation performance are dynamically changing, both rejection and water flux are dynamic values. Therefore, the adsorption performance in terms of flux and rejection are not intrinsic properties of the membranes. If the authors intend to compare the performance with other membranes, they should clarify that these membranes are adsorbents, and compare with previously reported membrane adsorbers. For example, MOF membrane adsorbers show water permeance of 20,000 L m-2 bar-1 h-1, which is much higher than the water permeance achieved in this study. The MOF membrane adsorber also showed high rejection at the beginning of filtration tests. References: High-permeance metal-organic framework-based membrane adsorber for the removal of dye molecules in aqueous phase. Environ. Sci.: Nano, 2017,4, 2205-2214. The multicycle dynamic adsorption and desorption process shown in Figure 4 needs to be carefully explained. Normally, for these adsorbents, as the high-surface area membranes adsorb more molecules and reach equilibrium, the membrane will lose adsorption capacity and the membrane fouling will become dominant. The authors explained that the molecules adsorbed on the surface and inside of the membrane, and likely the water flux decreases. In this case, the phenomenon is membrane fouling. To avoid confusion, the authors should report the original data of the filtration, particularly, the dye concentration of the feed, retentate, and the permeate, so you can estimate the mass balance. 5. It is recommended to compare the adsorption separation performance with control experiments using AO-PIM powders. For example, you can prepare AO-PIM particles with hierarchical porosity and load the particles into a column and study the adsorption and separation performance. 6. The authors need to change some of the expressions, such as extremely high adsorption capacity, unprecedented capacity. 7. The authors are recommended to perform a deeper literature search, and include relevant reports in the references, for example, recent report on preparation of hierarchical AO-PIM membranes by NIPS method. Nature Sustainability, 5, 71-80 (2022). 8. To some extent, the AO-PIM membrane reported by the authors are still ultrafiltration membranes. It will be good to compare with recent development of ultrafiltration membranes in the literature. Some recent studies on ultrafiltration membranes also show very high water flux and high rejection towards dye molecules. For example, recent work published in Separation and Purification Technology, 2022, 283, 120163, showed, a pure water flux of 110.4 LMH with 99.2% rejection to dye molecules (Congo red). 9. Figure S11, it would be better to provide more analysis f the dye molecules, for example, molecular volume and size.

Response:
Thanks for reviewer's keen questions. As shown in Figure R1, AOPIM-1 favors the capture of negatively charged MO molecules under acidic conditions while capturing the oppositely charged MB under alkaline conditions, exhibiting a pHtunable adsorption feature. And much lower equilibrium adsorption capacity is observed under neutral pH conditions where the amidoxime possess minimal chargeability, which reveals that the pH-tunable affinity sites make major contribution to the adsorption capacity. Therefore, from the interpretation of experimental data, the dye molecules sieving by the AOPIM-1 adsorptive membrane should be mainly attributed to electrostatic interactions. Figure R1. Equilibrium adsorption capacity (q e ) of AOPIM-1 for MO and MB dye molecules at different pH conditions. The discussion regarding Figure 2b has been revised as follows (page 7, line 22; page 8, line 1-4): "…And much lower equilibrium adsorption capacity is observed under neutral pH conditions where the amidoxime possess minimal chargeability, which reveals that the pH-tunable affinity sites make major contribution to the adsorption capacity, and that the dye molecules sieving should be mainly attributed to electrostatic interactions.".

Reviewer's comment: 3)
For the mechanism part, the molecular dynamics simulations on polymers with water or dye molecules separation might be supported as you have highlighted in your author contribution part. Response: According to the reviewer's suggestion, the molecular dynamics simulations has been conducted on polymers with water and dye molecules (MB and MO) separation in alkaline condition. Energy optimized molecular model and adsorption energies of AOPIM-1 to MB, MO and H 2 O molecule in alkaline condition are shown in Figure R2. The absorption energy (E ads ) of AOPIM-1 to MB, MO and H 2 O molecule are -1587.7, -745.4, and -742.3 kcal/mol, respectively. Obviously, the absorption energy (E ads ) of AOPIM-1 to MB is the highest, which is very consistent with the experimental result that MB molecule was selectively adsorbed and separated by the AOPIM-1 membrane.
The simulation method and results summarized as follows and the discussion "Energy optimized molecular model and adsorption energies of AOPIM-1 to MB, MO and H 2 O molecule in alkaline condition are shown in Figure S8. The absorption energy (E ads ) of AOPIM-1 to MB, MO and H 2 O molecule are -1587.7, -745.4, and -742.3 kcal/mol, respectively. Obviously, the absorption energy (E ads ) of AOPIM-1 to MB is the highest, which is very consistent with the experimental result that MB molecule was selectively adsorbed and separated by the AOPIM-1 membrane." have been added in the revised manuscript (page 8, line 14-20) and the revised supporting information as Figure S8.

Simulation method
All simulations were performed by LAMMPS 1 with the condensed-phase optimized molecular potentials for atomistic simulation studies 2 (COMPASS) force field. The initial charge of molecules was assigned by COMPASS. To find its optimized structure, AOPIM-1 chain and all the small molecular (MB, MO, and H 2 O) has been annealed from 2000 K to 300 K. AOPIM-1 chains are randomly initiated throughout the simulation box and grown by randomly choosing one of the two chiral monomers and one of the two possible orientations until the target chain length of 12 monomers is achieved or an overlap is the result of adding either monomer type in either orientation. After the initial geometry optimization, the adsorption energy was calculated by: E ads =E AOPIM-1+MO/MB/H2O -E AOPIM-1 -E MB/MO/H2O , where E AOPIM-1+MO/MB/H2O is the total energy of small molecule (MB/MO/H 2 O) adsorbed on AOPIM-1, the E AOPIM-1 and E MB/MO/H2O are the energy of AOPIM-1 and small molecules, respectively. Reference 61. Plimpton, S. Fast parallel algorithms for short-range molecular dynamics. J. Comput. Phys. 117, 1-19 (1995). 62. Sun, H. Compass: an ab initio force-field optimized for condensed-phase applications overview with details on alkane and benzene compounds. J. Phys. Chem. B. 102, 7338-7364 (1998).

Reviewer's comment: 4)
The core size distribution of the prepared membrane should be given as it affects the permeance of the membrane. Response: According to the reviewer's suggestion, we measured the pore size distribution of membranes prepared in different coagulation baths by the bubble point method. As shown in Figure R3, it can be seen that the mean pore size of M1, M2 and M3 is 271 nm, 247 nm and 211 nm, respectively. The pore size decreases with increasing ethanol content in the coagulation bath. This is also consistent with the variation trend of membrane flux. The pore size distribution of the membranes has been added in the supporting information as Figure S9. To achieve API separation, a tightened membrane pores (water flux of ~121.3 L m -2 h -1 bar -1 , MWCO of ~20 kDa) are prepared, and the mean pore size of M6 is 3 nm as shown in Figure R4 ( Figure  S18 in supporting information).  Reviewer's comment: 5) How does the concentration of the solutes influence the separation performance of the membrane? Response: To address the reviewer's query regarding the effect of solution concentration on the membrane, the separation performance of M3 membrane were measured by using 10/20/30/50 mg L -1 RHB feed solution. As the results shown in the figure below, the permeation volume (rejection > 99%) of 10/20/30/50 mg L -1 RHB solution is 1067.6, 628, 251.2, 125.6 L m -2 , respectively. And the flux of the 10/20/30/50 mg L -1 RHB solution varies from 297.79 L m -2 h -1 bar -1 to 292.12 L m -2 h -1 bar -1 , from 288.79 L m -2 h -1 bar -1 to 283.57 L m -2 h -1 bar -1 , from 276.93 L m -2 h -1 bar -1 to 270.24 L m -2 h -1 bar -1 and from 272.97 L m -2 h -1 bar -1 to 269.4 L m -2 h -1 bar -1 . The mean flux slightly decreases with the increase of RHB solution concentration, while the total adsorption capacity of the adsorptive membrane is almost unchanged (11.2-12.7 g m -2 ). It is obvious that he concentration of solutes have very little influence on the separation performance of our adsorptive membrane.
The relevant data and discussion have been added in the revised supporting information ( Figure S13): "The solutes concentration influence on the membrane separation performance is further investigated by separating 10/20/30/50 mg L -1 RHB solution. In the Figure S13, it can be seen that the flux of the 10/20/30/50 mg L -1 RHB solution varies from 197.21 L m -2 h -1 bar -1 to 193.46 L m -2 h -1 bar -1 , from 191.25 L m -2 h -1 bar -1 to 187.79 L m -2 h -1 bar -1 , from 183.39 L m -2 h -1 bar -1 to 178.97 L m -2 h -1 bar -1 and from 180.78 L m -2 h -1 bar -1 to 178.44 L m -2 h -1 bar -1 .And the permeation volume (rejection > 99%) of 10/20/30/50 mg L -1 RHB solution is 1067.6, 628.1, 251.2, 125.6 L m -2 , respectively. The mean flux of membrane slightly decreases with increase of RHB solution concentration, while the adsorption capacity is almost unchanged (11.2-12.7 g m -2 ). Obviously, the concentration of solutes has very little influence on the separation performance of our adsorptive membrane.". Reviewer's comment: 6) How to understand the efficient membrane separation? Some related work should be useful to understand this. (e.g. Nature communications 2018, 9 (1), 1-8; Journal of Materials Chemistry A 2018, 6 (42), 21104-21109). Response: According to the reviewer's reference and our understanding, an efficient separation membrane for industrial filtration and separation processes should be endowed with high flux and high selectivity capability and be stable in aqueous and various organic solvents, but should also be stable in harsh environments. Our AOPIM-1 membrane achieves >99.9% removal of various nano-sized organic molecules with water flux 2 orders of magnitude higher than typical pressure-driven membranes of similar rejections. And they are very stable in various pH condition and behave ultrahigh adsorptive capacity, which could be regarded as an efficient separation membrane. The references provided by the reviewer are very helpful for the understanding of the topic, and they have been referred (9-10) in the revised manuscript (page 3, line 8-10): "However, trade-off between membrane permeability and selectivity is an inherent shortcoming of membrane-based molecular separation 7-10 .".

Reviewer's comment: 7)
Although the adsorption capacities for ionic dyes were outstanding in acidic or alkaline conditions, the adsorption capacity in neutrally and practical condition is more meaningful, because it is impossible or too difficult to adjust the pH values of the wastewater during the application. The major adsorption studies and analysis should be performed in neutrally condition. Meanwhile, for the comparation with other works, the adsorption capacities in this work should be the values in neutrally condition. Response: We could not agree with the reviewer that the adsorption capacity in neutral condition is more meaningful. Wastewater could have very different characters in different industries, it is common to encounter acidic or alkaline wastewater. Taking textile wastewaters as an example, their pH values are easily under 2 or above 12, as the textile is frequently treated in acidic or alkaline conditions to enhance color fixing capability of dyes. Besides wastewater treatment, other industrial separation processes like active pharmaceutical ingredient extraction, food industry concentration and purification, etc., a lot of target feed solutions possess acidic or alkaline pH conditions. Our amidoxime modified polymer favors the capture of negatively charged molecules under acidic conditions while capturing the oppositely charged molecules under alkaline conditions. It demonstrates a pH-tunable adsorption feature, which could be applied in various pH conditions for different dyes treatment. This unique character of AOPIM-1 membrane is also the highlight of this work.
Regarding the separation behavior of AOPIM-1 membranes in neutral pH environment, the static adsorption capacities at various pH values has been demonstrated in our manuscript (Figure 2b). The adsorption capacity of the polymer is found to be as high as 445.02 mg g -1 (MO, pH = 3.3) and 735.75 mg g -1 (MB, pH = 10.9), respectively. And lower equilibrium adsorption capacity (70-120 mg g -1 for MB and MO) is observed under neutral pH conditions. Furthermore, the dynamic adsorption capacity of our membrane in neutral pH condition is measured. As shown in the Figure below, the membrane adsorption capacity in pH 7 is 0.3 and 7.1 g/m 2 for MO and RHB, respectively, much lower than that in acid or alkaline condition. This is ascribed to the minimal chargeability in neutral condition. It also illustrates that the pH-tunable affinity sites make dominant contribution to the adsorption capacity of our membrane. The relative discussion and result have been added in the revised manuscript (page 11, line 16-21) and supporting information ( Figure S15).  Table S1 for comparation.

Reviewer's comment: 9)
The mechanical properties like tensile strength of the membrane should be tested. Response: According to the reviewer's suggestion, we have tested the tensile strength of AOPIM-1 adsorption membrane under different pH values and calculated their tensile modulus respectively. The tensile stress at break is 2.7, 3.1 and 3.3 MPa, and the modulus is 20.5, 47.9, 54.0 MPa in pH=3, 7 and 10, respectively. The mechanical strength at pH=3 is the lowest, likely due to the decreased H-bond interaction in acid pH environment. The relevant discussion has been added in the revised manuscript (page 6, line 21-22; page 7, line 1-2): "All the membranes in various pH condition demonstrate sufficient mechanical strength for pressurized permeation tests ( Figure S6), while the mechanical strength appears lower in acid condition, likely due to decreased interchain H-bond interaction.".

Reviewer #3:
The manuscript reports the development of microporous polymer as adsorptive membranes for separation applications. PIM polymers are an important class of polymers for separation applications. Conventional PIMs are difficult to process due to their solubility in a limited range of solvents. Recent studies on amidoxime-functionalized PIMs have shown great promise owing to their solubility in polar solvents such as DMF, DMSO, NMP. Therefore, solution processing of these polymers by non-solvent-induced phase separation (NIPS) into hierarchically porous membranes is quite novel. This manuscript reports the processing of AO-PIM polymer into sponge-like porous membranes and demonstrated their application as membrane adsorbers. Compared to previous work on membrane adsorbers such as MOF-based membranes, these PIM membranes show a great advantage in terms of processability and functionality. The authors demonstrated the selective removal of dye molecules in acidic and alkaline solutions. These adsorptive membranes provide adsorption functions that cannot be achieved by conventional nanofiltration and ultrafiltration membranes. The demonstration of separation of API shows promising potential of these membranes. While these membranes show impressive apparent adsorption capacity and high water flux, the data interpretation and comparison need to be considered carefully because the membranes work as membrane adsorbers instead of separation membranes. The manuscript is suitable for publication in Nature Communications, but some technical concerns should be addressed. There are some concerns about the technical aspects of the results and analyses. Reviewer's comment: 1) It is an innovative idea to tune the charge of the polymer by pronation and deprotonation of the AO groups and use the membranes to separate anionic and cationic dyes. One technical problem is that the protonation of AO groups in acidic solution could lead to significant decay of mechanical properties. Did the authors measure the mechanical properties of membranes treated at low pH? Response: According to the reviewer's suggestion, we have measured the mechanical properties of membrane treated at low pH, and compared with those of membrane treated at higher pH as shown in Figure R8. The tensile stress at break of membrane treated at low pH is 2.7 Mpa. The mechanical strength of membrane treated at pH=3 is lower than those of membrane treated at higher pH (3.1 Mpa for pH=7 and 3.3 Mpa for pH=10). This is mainly attributed to protonation of AO groups in acidic solution, resulting the decreased interchain H-bond interaction. In spite of this, the membranes treated at low pH still have strong-enough mechanical strength for all the pressurized permeation tests conducted in this study. The relevant discussion has been added in the revised manuscript (page 6, line 21-22; page 7, line 1-2): "All the membranes in various pH condition demonstrate sufficient mechanical strength for pressurized permeation tests ( Figure S6), while the mechanical strength appears lower in acid condition, likely due to decreased interchain H-bond interaction.".

Reviewer's comment: 2)
The adsorption mechanisms need in-depth study and analysis. It is known that for these microporous materials with charges, the adsorption of dye molecules may be dominated by electrostatic adsorption on the surface of membranes instead of purely adsorption in the micropores. The schematic diagram shown in Fig. 1c is likely incorrect. Do you have direct evidence to claim that all the dye molecules are adsorbed in the micropores? Response: It should be clarified in the first place that we did not claim that all the dye molecules are adsorbed in the micropores of the AOPIM-1 membrane. And we agree that the chargeability of the membrane could unavoidably causes adsorption of dye molecules on the membrane surface. But we also have justifications and multiple evidences to support the adsorption of dye molecules in the micropores of the polymer. Firstly, as an amidoxime-modified polymer with intrinsic porosity, AOPIM-1 has high specific surface area and interconnected micropores that provides volume for dye molecule adsorption, and the amidoxime groups further promote adsorption process via electrostatic interactions. The static adsorption experiment performed in this work and previously reported data (B. Satilmis, et al J Polym Environ. 28, 995-1009(2020; Zhang, C. et al Chem. Eng. Res. Des. 109, 76-85 (2016);Xu, et al, J. Colloid Interface Sci. 607, 890-899 (2022)) confirm the high bulky-adsorption capacity (490~760 mg g -1 ) of AOPIM-1 through this mechanism. Moreover, we have measured the effective separation pore size of the AOPIM-1 membranes fabricated and included the result in the revised manuscript ( Figure S9), which are around 200 nm. This result reveals that the membranes exhibit a hierarchical pore feature, containing macropores generated from the membrane formation process, and intrinsic micropores existing within the polymer matrix. As dye molecules possess sizes in the nanometer range, they are more likely to penetrate into the membrane and get adsorbed into the polymer matrix instead of only remaining on the membrane surface. In addition, we find that the Brunauer-Emmet-Teller (BET) surface area of the AOPIM-1 adsorptive membrane is reduced from 552 m 2 g -1 to 415m 2 g -1 after the adsorption test, and it can be easily regenerated to the original level after desorption (Figure 4b). And it can be seen in Figure 4c that the pore size distribution (PSD) of AOPIM-1 is also restored to the original level after cleaning. Specifically, the amount of 0.5-0.7 nm size micropores reduces the most substantially, indicating that smaller micropores are mostly filled with dye molecules. The variation of membrane's BET and PSD are direct evidences of dye molecules adsorbed in the micropores.  Figure 3e needs to be explained carefully. Because the molecules are likely adsorbed on the surface of membranes and form a cake layer, the adsorption of these molecules are apparent values instead of intrinsic values. Also the capacity should be linked with the pore volume of the membranes available for adsorption instead of membrane area, because the thickness of these membranes are different. So the adsorptive capacity unit should be changed to g/m 3 instead of g/m 2 . The authors tried to compare the rejection and flux trade-off with conventional separation membranes based on size and charge. In my opinion, this is not a fair comparison. When the porous polymers are used as adsorbents, the separation performance are dynamically changing, both rejection and water flux are dynamic values. Therefore, the adsorption performance in terms of flux and rejection are not intrinsic properties of the membranes. If the authors intend to compare the performance with other membranes, they should clarify that these membranes are adsorbents, and compare with previously reported membrane adsorbers. For example, MOF membrane adsorbers show water permeance of 20,000 L m-2 bar-1 h-1, which is much higher than the water permeance achieved in this study. The MOF membrane adsorber also showed high rejection at the beginning of filtration tests. References: High-permeance metal-organic framework-based membrane adsorber for the removal of dye molecules in aqueous phase. Environ. Sci.: Nano, 2017,4, 2205-2214. Response: We are very grateful that the reviewer has provide in-depth discussion and constructive comments to improve our manuscript. Regarding the adsorption of dye molecules on the membrane surface, we have provided detailed discussion in the response to Comment #2 from the reviewer. And we agree that the adsorption capacity of the membrane closely relates to the pore volume. In fact, we have demonstrated in the manuscript that the processing capacity of M3 with sponge-like structure is much higher than that of M1 with finger-like structure, revealing that the separation performance of adsorptive membrane is closely related to the whole membrane structure. As recommended by the reviewer, we calculated processing capacity of the AOPIM-1 membranes in the unit of g/m 3 , which is 28,45,178,191 and 219 kg/m 3 for M1, M2, M3, M4, and M5, respectively. Interestingly, the volumetric processing capacity does not only increase sharply when the membrane changes from finger-like structure to sponge-like structure (M1 to M3), but also found to increase with the membrane thickness (M3 to M5). It appears that the membrane processing capacity is closely related to the length of dye transportation route on top of pore volume. The longer and tortuous route would result in higher processing capacity, which is consistent with the property of adsorption materials. The above results and discussions have been included in the revised manuscript (page 10, line 10-19). However, when benchmarking this work with other reported adsorptive membranes, it is difficult to compare the processing capacity with the g/m 3 unit, as the thickness information of related works are often missing. More importantly, separation behavior based on membrane area is the more widely accepted and comparable criteria for the performance evaluation of membranes.

Reviewer's comment: 3) The adsorption capacity presented in
Regarding the review's concern on the performance comparison with other membranes, we agree that the comparison should primarily be conducted among reported adsorptive membranes and we put such comparison in Figure 3e and 3f. Indeed, it is difficult to make comparation with other size/charge selective membranes including nanofiltration and ultrafiltration membranes, but we believe it is of great importance to benchmark our work with membranes that are targeting similar application scenarios for the interest of a wider spectrum of readers. We therefore adopted similar experimental conditions in this study including feed variety, concentration, trans-membrane pressure, temperature, etc. comparable to that of other size/charge selective membranes. The trickiest part is how to collect the effective flux and rejection of adsorptive membranes, as "both rejection and water flux are dynamic values" as mentioned by the reviewer. We hereby clarify that the effective flux and processing capacity of the membrane in this study is calibrated under the criterion of 99% rejection ratio of dye molecule. This criterion is consistent with size/charge selective membranes, and widely accepted by previously published work regarding adsorptive membranes.
Lastly, for the research article referred by the reviewer, although very high permeance has been reported for the MOF membrane, the separation behavior is quite different comparing with our work and other reported adsorptive membranes. According to Figure 8 of the article (show Figure R10 for reference), the dye concentration increases sharply with less than 2 mg rose bengal adsorbed, and the highest membrane rejection recorded in the beginning of filtration tests is around 95% based on reading the first data point in each figure. Although good rose bengal adsorption capacity of 181.2 mg g −1 is reported for the membrane, the dynamic separation behavior of the membrane appears different from our study and other reported adsorptive membranes. Therefore, we feel it inappropriate to include this MOF membrane in the separation performance benchmarking figure (Figure 3f).

Reviewer's comment: 4)
The multicycle dynamic adsorption and desorption process shown in Figure 4 needs to be carefully explained. Normally, for these adsorbents, as the high-surface area membranes adsorb more molecules and reach equilibrium, the membrane will lose adsorption capacity and the membrane fouling will become dominant. The authors explained that the molecules adsorbed on the surface and inside of the membrane, and likely the water flux decreases. In this case, the phenomenon is membrane fouling. To avoid confusion, the authors should report the original data of the filtration, particularly, the dye concentration of the feed, retentate, and the permeate, so you can estimate the mass balance. Response: Thanks for reviewer's comment. The multi-cycle adsorptive separation performance stability of the AOPIM-1 adsorptive membrane was performed in a dead-end filtration cell with an effective membrane area of 3.14 cm 2 for 8 cycles. For a typical cycle, the 500 L m -2 RHB solution (20 mg/L, pH = 10) was filtrated through the membrane at 0.2 MPa. The original data of every cycle is summarized in the below table and added in the revised supporting information as Table S5. The eluent concentration is provided instead of retentate because the rejected dye is adsorbed on the membrane. It can be seen that permeate concentration slightly increases with cycle number increasing. Thus, the rejection rate slightly decreases with cycle number increasing due to possible membrane fouling in the cycle filtration process, which is in accordance with the reviewer's comment. However, the membrane fouling could be relieved by longer time desorption treatment as confirmed by the decreased permeate concentration of cycle 8.  Reviewer's comment: 5) It is recommended to compare the adsorption separation performance with control experiments using AO-PIM powders. For example, you can prepare AO-PIM particles with hierarchical porosity and load the particles into a column and study the adsorption and separation performance. Response: Thanks for reviewer's suggestion. When AOPIM-1 is made into particles or fibers, the specific surface area might be even larger than that of AOPIM-1 membranes and the adsorption capacity would be optimized. This has been verified by the reported works. Xu et al (Journal of Colloid and Interface Science 607 (2022) 890-899) prepared AOPIM-1/Alg composite beads for cationic dyes adsorption from aqueous solution, achieving high adsorption ability and outstanding regeneration ability. Bekir et al (Applied Surface Science 467-468 (2019) 648-657) has prepared AOPIM-1 fiber ultrafine fibers for rapid removal of uranyl ions from water. However, our work emphasizes the fabrication of an adsorptive membrane with microporous polymer to make the full use of the advantages of AOPIM-1 including high surface area and good solution processability, while simultaneously utilize the size sieving effect of the membrane to achieve highly efficient separation of complex systems such as API feed streams. Despite of the high adsorption capacity, adsorbents in the form of packing columns lack such sieving feature, and are not suitable to be Reviewer's comment: 9) Figure S11, it would be better to provide more analysis of the dye molecules, for example, molecular volume and size. Response: Thanks for reviewer's kind suggestion. We have added the molecule model and molecule size including 3D size, arithmetic mean radius, and geometric mean radius of dye molecules in the Figure S14 of the revised supporting information.