Polymer‐Grafting from MOF Nanosheets for the Fabrication of Versatile 2D Materials

Two‐dimensional (2D) materials based on metal‐organic framework (MOF) nanosheets are highly useful for electronic devices, catalysis, and energy storage systems because of their high structural designability and anisotropy. However, their high surface energy often causes aggregation and low dispersibility, which reduces their processability and performance. To address these fundamental issues, it is reported a new approach involving polymer grafting from the surfaces of MOF nanosheets. A newly discovered one‐step synthetic pathway provides MOF nanosheets functionalized with a polymerization initiator agent on the surface with over 90% yield. Successive nitroxide‐mediated living polymerization of vinyl monomers from the MOF nanosheet surfaces provides polymer‐grafted nanosheets with superior dispersibility in many organic solvents. The polymer‐decorated MOF nanosheets exhibit versatile property modulation, which allows for their protection and efficient use in applications such as photoredox catalysts and oil/water separation membranes. This polymer‐grafting approach can facilitate the fabrication of surface‐functionalized MOF nanosheets, which can lead to the wider application of 2D materials in various fields.


Introduction
Two-dimensional (2D) materials have attracted considerable interest owing to their unique properties originating from their anisotropic sheet configurations. [1]4] Recently, metal-organic framework (MOF) nanosheets have emerged as a new family of 2D materials that offer a highly DOI: 10.1002/adfm.202312265designable 2D platform, [5][6][7] further expanding their application scope and requiring more elaborate structural designs and modifications. [8,9]The highly anisotropic and nanoporous nature of MOF nanosheets makes them suitable for highperformance membrane applications. [10]urthermore, MOF nanosheets have recently been used as heterogeneous photocatalysts for controlled radical polymerization, demonstrating their versatility in a wide range of applications. [11,12]urface modification and functionalization of MOF nanosheets are indispensable for meeting the many requirements of practical applications.Owing to their high surface energies, most 2D MOFs tend to form aggregates, eventually decreasing the synthetic yield and undermining their performance in various applications. [13]Although several approaches for synthesizing highly dispersive MOF nanosheets have been reported, these examples are limited to specific frameworks, which preclude the universal chemical modification of nanosheets. [14,15]Therefore, the development of a facile and rational approach to resolving these fundamental issues in MOF nanosheet synthesis and functionalization is still awaiting further research.
Herein, we report a new approach for fabricating stable and highly dispersive MOF nanosheets by grafting polymers from their surfaces (Figure 1).A monodentate ligand with a polymerization-initiating moiety (L) was designed and incorporated into 2D MOF crystals of Zn-TCPP (hereafter referred to as 1, TCPP: tetrakis(4-carboxyphenyl)-porphyrin).The ligand, which was axially coordinated between the 2D MOF layers, facilitated the fabrication of MOF nanosheets via solvent-assisted exfoliation.Using a post-synthetic surface-initiated polymerization technique, the MOF nanosheets were decorated with polymer chains, allowing the protection and versatile property modulation of the MOF nanosheets.
Current state-of-the-art methods for 2D MOF nanosheet synthesis can be divided into top-down [16] and bottom-up [17] approaches.The top-down approach involves the exfoliation of bulk MOF crystals.Although this approach is facile with high scalability, it risks structural degradation and ill-controlled sheet thickness due to technical reasons.The MOF exfoliation process generally involves harsh conditions, including sonication, [18] grinding, [19] shearing, [20] and chemical reduction, [21,22] which leads to low yields (typically < 20%). [23,24]In contrast, the bottom-up approach uses interfacial reactions [25,26] or solutionphase batch reactions with external modulators such as surfactants [24] and coordinative organic additives [27,28] to facilitate the anisotropic growth of MOF crystals.Modulator methods have broad applicability to many MOF systems and permit control of the thickness and dispersibility of the product [13,[29][30][31] as well as facile functionalization of 2D MOF surfaces for further applications. [32]Instead of conventional low-molecular-weight modulators, polymer decoration has been effectively used to fabricate MOF-based composite materials.For example, poly(methyl methacrylate) (PMMA) has been used for surface grafting of MOF nanoparticles. [33]The aggregation behavior [33,34] and surface wettability [35] can be easily controlled by tuning the side chains of the methyl methacrylate monomers.Polymers other than vinyl polymers can also be used for the surface modification of MOFs, such as polyethylenedioxythiophene (PEDOT) [36] and polyimides. [37]However, in the reported methodologies, the synthesis of MOF particles and the installation of surfaceanchored initiator ligands are separated.Therefore, this conventional approach is not applicable to the fabrication of surfacefunctionalized MOF nanosheets, because most MOF nanosheets are likely to form aggregates that hamper the second ligand installation step.
In this study, we synthesized a polymerization initiator ligand, (4-(1-((2,2,6,6-tetramethylpiperidin-1-yl)oxy)ethyl)pyridine, L) (Figure 1a), and fabricated Zn-TCPP (1) MOF nanosheets possessing L as an axially coordinating ligand on the surface (hereafter referred to as 1 NS -L) (Figure 1b).Owing to the newly discovered one-step synthetic pathway, 1 NS -L MOF nanosheets were obtained with a high yield of ∼91%.The successive nitroxidemediated polymerization (NMP) of styrene (St) from 1 NS -L surfaces provided polystyrene-grafted MOF nanosheets (1 NS -PSt) (Figure 1e) with superior dispersibility in many organic solvents.Because of the TCPP center on the MOF nanosheet surface, 1 NS -PSt showed photoredox catalytic activity with a turnover number (TON) of 822, which is higher than that of conven-tional Zr-based TCPP-MOFs in the bulk crystalline form. [38]Generally, MOFs comprising paddle-wheel Zn clusters are unstable in moisture.Poly(pentafluorostyrene) (PPFSt)-grafted MOF nanosheets (1 NS -PPFSt) exhibited superhydrophobicity and considerably improved water stability.Furthermore, because of its high dispersibility in organic solvents, we fabricated a deposited membrane of 1 NS -PPFSt.Owing to its superhydrophobic nature, the membrane showed a high-efficiency separation of oil over water from the emulsion mixture.Because the PPFSt chains are firmly anchored on the nanosheets via coordination bonding, membrane degradation hardly occurred, as evidenced by the negligible selectivity and flux loss in 20 repetitive separation cycles.

Two-Step Synthesis of Initiator Functionalized MOF Nanosheet (1 NS -L)
We used a well-known TCPP-based MOF ([Zn 2 (ZnTCPP)] n , 1) [39] as the parent material.Generally, 1 is synthesized via a typical solvothermal synthesis with Zn(NO 3 ) 2 , resulting in a bulk crystal with a layered structure of 2D coordination sheets (1 NS ), in which ZnTCPP is interconnected with Zn 2 (COO) 4 paddle-wheel clusters (Figure S1-S3, Supporting Information).To functionalize the 1 NS sheet surfaces, we synthesized a functional monodentate pyridine ligand (L) with an alkoxyamine moiety that served as an NMP initiator (Figure S4-S6, Supporting Information). [40]n this study, we attempted a modulator-assisted method as our first approach because it is expected to yield MOF nanosheets in a one-step reaction. [30]This method involves the use of a surfactant as a coordination modulator to control crystal growth, facilitating the formation of MOF nanosheets instead of bulk crystallization.To follow this strategy, we added an excess amount of L as a functional modulator to the mixture of Zn(NO 3 ) 2 and TCPP, using a mixed solvent of N,N-diethylformamide (DEF)/EtOH (3/1, v/v), and carried out the solvothermal reaction at 85 °C for 24 h.However, contrary to our expectations, the reaction gave a bulk crystalline phase (hereafter referred to as 1 bulk -L) instead of dispersed nanosheets.Optical microscopy and scanning electron microscopy (SEM) revealed micrometer-scale brick-shaped crystals (Figure S7, Supporting Information).Based on these results, we believe that ligand L is not capable of stabilizing the nanosheet because of its relatively short solubilizing group compared to other reported surfactant modulators. [26]he obtained bulk crystals of 1 bulk -L were analyzed using powder X-ray diffraction (PXRD), which revealed clear diffraction peaks (Figure 2a).By using full profile fitting analysis, these peaks were identified as belonging to an orthorhombic lattice with cell parameters of a = 16.0934Å, b = 16.5676Å, c = 29.8651Å (Figure S8 and Table S1, Supporting Information).The a and b axis lengths were found to be consistent with the lattice of the previously reported parent 1 crystal structure (a = b = 16.6814Å, c = 17.4944Å). [39] Compared to the structure of reported 1, the layer distance was expanded to ∼1.5 nm, which corresponded to the approximate thickness of the [Zn 2 (ZnTCPP)] n layer with two axial ligands, L, on both sides (∼1.5 nm), indicating that L was intercalated into the parent crystal structure of 1.This intercalation of L was also supported by 1 H NMR analysis of the crystal digested in DMSO-d 6 /CF 3 COOD (9/1, v/v) solution (Figure S9, Supporting Information).NMR analysis showed that L was incorporated into the crystal at a ratio of 1.2 per one ZnTCPP unit, suggesting that the axial Zn sites on the ZnTCPP units were not fully occupied in 1 bulk -L.
Based on the reported crystal structure of 1, we constructed a crystal structure model of 1 bulk -L in which the L ligand was intercalated at the interstices of the [Zn 2 (ZnTCPP)] n layers via coordination between the pyridyl group of L and the Zn center of the ZnTCPP units (Figures S10 and S11, Supporting Information).We considered that L coordinates only to the Zn center of the ZnTCPP unit and not to the axial sites on the Zn paddle wheel clusters because the latter coordination geometry results in a considerably larger layer distance (∼1.9 nm (Figure S10, Supporting Information).The simulated PXRD pattern of the modeled crystal structure agreed with the observed PXRD pattern of 1 bulk -L (Figure 2a), supporting the formation of the anticipated MOF nanosheet structure in the crystal (Figure S11, Supporting Information).The specific coordination behavior in the present system can be attributed to the combined effects between the significant steric hindrance caused by L and optimal packing structure of the MOF nanosheets.
We then investigated the solvent-assisted exfoliation of 1 bulk -L into individual nanosheets and found that the exfoliation reaction depended on the solvent used.Soaking 1 bulk -L in common polar solvents such as DEF and acetone did not result in exfoliation.Thus, 1 bulk -L retained its original PXRD pattern (Figure S12, Supporting Information).However, the use of alkylbenzene solvents, such as toluene, ethylbenzene, and cumene, significantly changed the PXRD pattern and led to exfoliation (Figure S12, Supporting Information).In the resulting PXRD profiles, all peaks broadened, and the intensities of the 110 (2 = 7.52°) and 002 (2 = 5.82°) diffractions decreased, suggesting the formation of a turbostratic phase.To further investigate this structural change, we monitored the transition process using in-situ time-resolved PXRD (Figure S13, Supporting Information).The results revealed that cumene, which had the largest alkyl substituent among those tested, showed the fastest transformation into the turbostratic phase, with prominent decay in the 110 (2 = 7.52°) and 002 (2 = 5.82°) peak intensities within 6 h.Benzene did not induce the formation of a turbostratic phase or exfoliation (Figure S13, Supporting Information).This suggests that the structure of the solvent, particularly the length of the alkyl chains substituted on the phenyl ring, is a crucial factor in disrupting the layered structure of 1 bulk -L.We speculate that the neighboring layers are held together by weak van der Waals forces and interdigitated shape matching.However, when encapsulating alkylbenzene guests, the interaction between layers is disrupted, leading to the 1 bulk -L crystals to swell between the layers.The SEM images showed that the resulting turbostratic phase exhibited the typical morphology of a layered structure (Figure S13, Supporting Information).After the turbostratic phase was obtained, the material was further exfoliated in cumene by stirring at 25 °C, resulting in 1 NS -L with a yield of 82% yield.It is worth noting that from the turbostratic phase, 1 bulk -L underwent exfoliation even in polar solvents such as DEF and tetrahydrofuran (THF), which were not effective for the assynthesized bulk phase.After exfoliation, the resulting colloidal suspension of 1 NS -L remained stable for more than two weeks at 25 °C (Figure S14, Supporting Information).Therefore, 1 NS -L was successfully synthesized through a two-step route involving solvothermal synthesis and exfoliation.
Dynamic light scattering (DLS) measurements indicated that the average particle size d a of 1 NS -L produced via the two-step route was 282 nm (Figure 2b).The nanosheets were imaged using atomic force microscopy (AFM) and were found to be ∼100-200 nm (Figure 3a and Figure S15, Supporting Information).The height of the sheets, which is approximately 1.5 nm, corresponds to the thickness of the ZnTCPP monolayer decorated with two L ligands coordinated to the Zn center of the TCPP unit (Figure 3b-d).SEM images of the nanosheets showed the typical flower-like morphology of the folded sheets (Figure 3e).Transmission electron microscopy (TEM) revealed wrinkled and ruptured sheet morphologies (Figure 3f), indicating the thin and flexible nature of the nanosheets.
Furthermore, high-resolution TEM (HRTEM) allowed latticeresolution imaging of individual 1 NS -L sheets, which revealed a lattice structure in the form of a fringe pattern (Figure 3g).The spacing of the lattice fringes observed using HRTEM was approximately 1.1 nm, corresponding to the 110 spacing of 1 crystal structure (Figure 3d and Figure S16, Supporting Information).The selected area electron diffraction (SAED) pattern of the sheet displayed clear spots at 110 and 100, indicating that crystallinity was maintained even after the exfoliation process (Figure 3g).

One-Step Synthesis of 1 NS -L
From the results of the two-step synthesis, we anticipated that the addition of cumene to the solvothermal reaction of 1 bulk -L would produce a nanosheet (1 NS -L) in a single step.Thus, we conducted the solvothermal reaction of Zn(NO 3 ) 2 and TCPP using the mixed solvent of DEF/EtOH/cumene (3/1/1, v/v/v) at 75 °C for 12 h and obtained a suspension of the product (see Experimental Section).Although a small number of large MOF particles formed simultaneously, we could easily separate the particles from the suspension by centrifugation.The remaining 1 NS -L in the supernatant constituted 91% of the total product, indicating that this method yielded stable nanosheets more effectively than the two-step route.
1 NS -L was collected by ultracentrifugation (25,000 g, 10 min) and subjected to PXRD measurements.The PXRD data of 1 NS -L obtained by one-step synthesis revealed a characteristic peak pattern similar to that of 1 bulk -L, suggesting the formation of a 2D framework of 1 (Figure 2a).The sharp diffraction peaks compared to those of the two-step 1 NS -L suggest higher structural integrity and crystallinity of the 2D framework obtained by the one-step route.Interestingly, the out-of-plane diffraction (e.g., 002) corresponding to the layer order was completely absent in the one-step 1 NS -L, whereas other in-plane diffractions remained clear.This PXRD feature indicated disorder in the stacking direction, representing the effect of cumene, which could lubricate the layers and facilitate in situ exfoliation.The AFM images showed that the one-step 1 NS -L nanosheets were micrometer-sized and had a thickness of ∼1.5 nm, which was the same as that of the two-step 1 NS -L (Figure 4a,b and Figures S17 and S18, Supporting Information).Despite their large sheet size, the one-step 1 NS -L nanosheets exhibited good dispersibility in organic media (Figure 4c).DLS measurements also showed d a of 1,256 nm (Figure 2b), consistent with the AFM results, and significantly larger than that of the two-step 1 NS -L.The high crystallinity observed in the PXRD patterns reflects the high structural integrity of each crystalline [Zn 2 (ZnTCPP)] n layer, which can result in the formation of large monolayer sheets.
Through 1 H NMR analysis on an acid-digested sample of the one-step 1 NS -L, we determined the actual incorporation ratio of L to be 1.96 per one ZnTCPP unit (Figure S19, Supporting Information), satisfying the ideal ratio and supporting the surface anchoring of L on both sides of the nanosheets.The surface anchoring of L was also confirmed by X-ray photoelectron spectroscopy (XPS) (Figure 4d and Figures S20 and S21, Supporting Information), where the Zn photoemission spectra identified a specific Zn-N coordination on the surface of 1 NS -L.The signal of Zn 2p 3/2 can be divided into two peaks with a 1:2 ratio of area, corresponding to the ratio between Zn at the porphyrin center and the paddle-wheel cluster.The Zn-porphyrin signal of the pristine 1 was observed at 1,022.0 eV, while that of 1 NS -L was observed at a lower binding energy of 1,023.5 eV due to the increased electron density of the Zn center resulting from the axial coordination of L. In contrast, Zn in the paddle-wheel cluster did not show any binding energy shift in 1 NS -L compared to the reference signal of pristine 1, validating our model crystal structure of 1 bulk -L where L coordinates only to Zn of ZnTCPP (Figure 3d).Finally, electron microscopy was performed on one-step 1 NS -L.SEM and TEM showed sheet-like structures (Figure 4e,f), and HRTEM revealed characteristic fringe patterns with a spacing of 1.1 nm corresponding to the 110 spacings of 1 crystal (Figure 4g).The SAED pattern shows clear diffraction spots at 100 and 110, providing evidence of the high stability and structural integrity of one-step 1 NS -L (Figure 4g and Figure S22, Supporting Information).Therefore, we successfully demonstrated a one-step synthetic route using an alkylated aromatic compound (cumene) as a secondary solvent, which plausibly assists in the in situ nanosheet exfoliation by stabilizing the intermediate turbostratic phase during the solvothermal reaction.

Surface-Initiated Polymerization from 1 NS -L
Numerous studies have discussed effective methods for tailoring the interfacial structure of nanomaterials. [41,42]One such method involves the use of controlled radical polymerization techniques, such as reversible addition-fragmentation chaintransfer (RAFT) polymerization, [33] atom transfer radical polymerization (ATRP), [34] and NMP, [43] to modify the surface of various materials.As described in the previous section, we successfully synthesized 1 NS -L nanosheets in a one-step method in which the ligand L capable of mediating NMP, was anchored on the MOF nanosheet surfaces.We confirmed that free L alone can mediate NMP of styrene at 130 °C, producing monodisperse polystyrene (PSt) with an M w of 9,900 (Ð = M w /M n = 1.03) in 72 h (Figures S23 and S24, Supporting Information).Using the same polymerization conditions, surface-initiated NMP was performed on 1 NS -L.We dispersed 1 NS -L, obtained through the one-step synthesis, in distilled styrene and heated it to 130 °C (see Experimental Section).The reaction time was varied from 6 to 24 h to obtain grafted PSt chains of different lengths (Table 1).During polymerization, the reaction mixture became a viscous gel-like suspension.After the reaction, the solid fraction in the resulting suspension was washed with tetrahydrofuran (THF) to remove any PSt that was not attached to the 1 NS surface.The solid was then collected by ultracentrifugation (25,000 g, 10 min) to produce PSt-grafted 1 NS .The PSt-grafted 1 NS synthesized using different reaction times (6, 8, 12, 24, 36, and 48 h) are referred to as 1 NS -PSt X (X = 8k, 13k, 20k, 62k, 95k, and 138k, respectively), indicating the M w of the grafted PSt as measured by size-exclusion chromatography (SEC) (vide infra).
The PXRD data of all 1 NS -PSt samples showed a characteristic 110 peak (2 = 7.52°), indicating that the internal 2D [Zn 2 (ZnTCPP)] n framework was preserved even after polymerization (Figure 5a).The broad peak (2 = 19.5°)was attributed to the typical amorphous halo of the PSt chain.The PSt grafted from the 1 NS surfaces was isolated by acid digestion of the respective 1 NS -PSt samples, followed by precipitation from MeOH (Figure S25-S30, Supporting Information) (see Experimental Section).This allowed for SEC measurements of the grafted chains to analyze their molecular weights (MW) and MW distributions (Ð = M w /M n ) (Table 1 and Figure 5b).Depending on the reaction time, the PSt chains grafted from the surface had MWs ranging from M n = 5,700 to 93,400 g mol −1 (Table 1).The linear increase in MW with reaction time indicates the livingness of the surface-initiated NMP mediated by L, which also reflects the relatively narrow MW distribution of the PSt chains (Figure S31, Supporting Information and Table 1).For the short PSt detached from the nanosheets (8k, 13k, and 20k), the presence of a pyridine terminal group was identified by 1 H NMR, verifying the successful initiation reaction from L anchored to the surfaces of the nanosheets (Figure S25-S27, Supporting Information).Based on 1 H NMR data of the digested PSt-grafted nanosheets (1 NS -PSt 13k and 1 NS -PSt 20k ) under acidic conditions, we calculated the graft-ing density of PSt on the surface of the nanosheets to be 0.296-0.334chains/nm 2 (Figures S32 and S33 and Table S2, Supporting Information).
Thermogravimetric analysis (TGA) of 1 NS -PSt 20k showed the disappearance of the weight loss corresponding to the thermal decomposition of L, which initially appeared at ∼180 °C for 1 NS -L, (Figure S34, Supporting Information), supporting the involvement of L in the polymerization reaction.HRTEM images of the PSt-grafted nanosheets also show a 110-fringe pattern of 1 NS with a spacing of 1.1 nm, ensuring the preservation of the high structural integrity of 1 NS even after grafting PSt from the surfaces (Figure 5d).In contrast, TEM and AFM revealed a rougher surface for 1 NS -PSt 20k compared to bare 1 NS (Figure 5c,e,f), which can be attributed to varying length caused by molecular weight distribution of surface-grafted PSt chains and the presence of unreacted L on the surfaces.The AFM data for 1 NS -PSt 20k indicates a nanosheet thickness of approximately 50-90 nm (Figure 5e,f, and Figure S35, Supporting Information).If PSt chains are grafted on both sides of the nanosheet, the theoretical thickness should equal the sum of the nanosheet thickness (1.5 nm) and twice the theoretical contour length of the 20k PSt chain (PSt 20k, calculated as 42.6 nm).This alignment is logical, given a graft density of 0.334 chains/nm 2 , which appears sufficient to fully extend and align the grafted PSt chains perpendicular to the surface, causing them to stand upright on the nanosheet surface. [44]In contrast, surfaces reported earlier with lower grafting densities exhibited PSt chains in a downward position, resulting in a thickness considerably smaller than the contour length of PSt with an equivalent MW, and closer to the corresponding average square radius of gyration (R g ). [45]Therefore, the part of MOF nanosheets could have the densely grafted PSt chains aligning perpendicular to the nanosheet surfaces.1 NS -PSt 95k and 1 NS -PSt 138k , characterized by longer PSt chains, displayed substantially thicker PSt layers (>50 nm) in the AFM images (Figures S36 and S37, Supporting Information).The proportional increase in PSt layer thickness with MW suggests that chain growth is initiated from the 1 NS surfaces.To study the thermal properties of the PSt grafted from the nanosheet surface, we analyzed the glass transition temperatures (T g ) of 1 NS -PSt and detached PSt using differential scanning calorimetry (DSC).The DSC profiles of 1 NS -PSt 8k , 1 NS -PSt 13k , and 1 NS -PSt 20k were significantly higher T g than that of free PSt detached from the nanosheets, namely PSt 8k , PSt 13k , and PSt 20k , respectively (Figure 6).When polymer chains are covalently bonded to substrates, it is common to observe an increase in T g owing to restricted chain mobility. [46]The increase in T g depended on the MW of the grafted PSt.The largest T g increment of 20 °C was observed in 1 NS -PSt 8k , which had relatively shorter PSt chains, whereas 1 NS -PSt 20k with longer MW PSt showed an increase of 3 °C (Figure 6).We attribute this trend to the grafting effect, whereby shorter PSt chains grafted from the surface have more strongly confined segments than longer PSt chains. [45]As the MW increased, the T g difference between the grafted and free PSt chains decreased (Figures S38 and S39, Supporting Information), and both T g values gradually approached the same constant value of ∼106 °C, corresponding to T g of the pristine PSt.metal sites on the surface and ultrathin thickness, which is beneficial for large surface areas. [47,48]TCPP has been extensively used as a photocatalytic motif and incorporated into MOFs for photocatalytic reactions such as photodegradation, photooxidation, and hydrogen generation. [47]However, in general, MOFs containing Zn 2 (COO) 4 paddlewheel clusters have poor stability under moisture and acidic/basic conditions [49] ; thus, the 2D [Zn 2 (ZnTCPP)] n framework (1) has been less frequently employed as a catalyst.It is widely known that the combination of MOF and polymers helps improve the stability of coordination frameworks in recent years. [50]Therefore, we anticipated that, owing to the polymer protective layer, the synthesized 1 NS -PSt nanosheets would have improved stability compared to bare 1, providing a robust photoredox catalyst even in harsh environments.In this context, we selected the photooxidation of phenylboronic acid [38] as a model reaction to investigate the catalytic performance of 1 NS -PSt.It is worth noting that such acidic substrates and products quickly cause poisoning of TCPP catalysts in general. [51]o a DMF solution of phenylboronic acid and N,Ndiisopropylethylamine (DIPEA), 1 NS -PSt 20k (0.01 eq. in TCPP unit with respect to phenylboronic acid) was added, and the mixture was exposed to a blue LED light ( = 455 nm) with vigorous stirring in an air atmosphere for 24 h at 25 °C (see Experimental Section).Control reactions were also performed using H 4 TCPP(H 2 ), bulk crystals of [Zn 2 (ZnTCPP)] n (1), and 1 NS -L as catalysts.The conversion-based yield was determined by gas chromatography (GC), and the results are summarized in Table 2.The mixture without any catalyst showed no conversion regardless of photoirradiation (Entry 1 and 2).TCPP alone (Entry 4) and 1 (Entry 5) showed moderate activity under blue light, provide 18% and 57%, respectively.Compared to Entry 5, in which bulk 1 crystals were dispersed in the mixture, 1 NS -L showed relatively higher photocatalytic activity owing to better solution dispersibility, which would facilitate access of the substrate to the catalytic center on the TCPP units.

2D MOF/COF nanosheets have recently emerged as promising heterogeneous catalysts owing to their highly exposed active
Meanwhile, 1 NS -PSt 20k showed the highest activity (79% GC yield), which was ascribed to the enhanced solution dispersibility and substrate accessibility to the catalytic center owing to the PSt-grafted structure.It should be noted that simply mixing bulk   1 and PSt separately, synthesized using L as the initiator, did not improve the catalytic performance (Table S3, Supporting Information).Therefore, the molecularly hybridized structure, where the PSt chains are anchored on the nanosheet surface, effectively stabilizes the ZnTCPP nanosheets and enables them to exert their full catalytic performance, even for otherwise unfavorable acidic substrates.We also noted that the addition of TEMPO or 1,4-quinone, which are a stable radical and superoxide radical scavenger, respectively, inhibited photooxidation.This suggests that the reaction is a free-radical process and superoxide radicals are the main reactive oxygen species. [52,53]he cycling stability of 1 NS -PSt 20k was investigated through recycling experiments (Figure 7a).The results indicate that no significant activity loss of 1 NS -PSt 20k occurred even after ten cycles, whereas bulk 1 crystals and bare nanosheets 1 NS -L lost their activity after three cycles.These results emphasize the function of the grafted polymer layer in protecting the nanosheets from poisoning by the acidic substrates.Interestingly, the turn-over-number (TON) of 1 NS -PSt 20k is 822, which exceeds the performance of other MOF catalysts, such as the representative Zr-based MOF (MOF-525), [38] although some porphyrinbased COF has been reported to show higher performance (Table S4, Supporting Information). [54]The PXRD measurements confirmed that 1 NS -PSt 20k maintained its crystallinity even after ten cycles (Figure 7b).DLS analyses indicated that the particle size of 1 NS -PSt 20k used in the photocatalytic reaction decreased to 352 nm from its original 1,232 nm after ten cycles (Figures S40 and S41, Supporting Information).A possible explanation for this size reduction is the loss of the grafted PSt chains.To verify this hypothesis, we stirred a DMF solution of 1 NS -PSt 20k for ten days and measured the SEC of the supernatant.The SEC results showed the existence of PSt with the same MW as the PSt isolated by acid digestion of 1 NS -PSt 20k (Figure S42, Supporting Information).Therefore, the decomposition of 1 NS -PSt 20k during the catalytic reaction can be divided into two steps) gradual detachment of the surface-grafted PSt chains and 2) immediate decomposition of the nanosheet in a harsh reaction environment owing to the lack of a surface protection layer.

Superhydrophobic PPFSt-grafted MOF Nanosheet for Fabrication of Oil/Water Emulsion Separation Membrane
The polymer grafting modification approach allows for the precise manipulation of MOF nanosheet properties and surface functionalities, making it valuable for the fabrication of highperformance membranes.MOF nanosheets have gained significant attention in the development of separation membranes because of their designability, which enables precise control over the molecular transport pathways, size, and chemical affinity for specific species. [55,56]The high aspect ratio of the 2D morphology also contributes to the formation of membrane structures with a preferential orientation of pores and penetration paths, resulting in high permeation and selectivity.However, the fabrication of 2D MOF nanosheet membranes is often challenging owing to their low dispersibility.This limits the use of various membrane fabrication methods, including vacuum-assisted filtration assembly, to achieve a controlled membrane thickness.
59][60][61] This is primarily because of the instability of 2D MOF nanosheets in water.In this context, post-synthetic polymer grafting modification offers a powerful way to alter the hydrophobicity and stability of MOF nanosheets, thereby enabling the development of high-performance membranes for water treatment.In this study, we synthesized a superhydrophobic MOF nanosheet via surface-initiated NMP from 1 NS -L using PPFSt as the monomer instead of PSt (see Experimental Section).The AFM images confirmed that the resulting 1 NS -PPFSt nanosheet retained its original sheet morphology after polymerization (Figures S43 and S44, Supporting Information).The thickness of the sheet was approximately 25 nm, indicating the successful grafting of PPFSt chains from the surfaces. 1H NMR and SEC analyses of the digested nanosheets confirmed the presence of PPFSt chains with M n = 21,500 g/mol (Table 1 and Figures S45 and S46, Support- ing Information).The PXRD data for 1 NS -PPFSt exhibit a characteristic pattern comparable to that of 1 NS -PSt 20k , confirming the preserved in-plane structural integrity of the nanosheets (Figure S47, Supporting Information).Furthermore, DSC analyses on free PPFSt and 1 NS -PPFSt revealed that the observed T g in 1 NS -PPFSt is approximately 129 °C, ∼21 °C higher than that of free PPFSt (T g = 108 °C) (Figure S48, Supporting Information).This trend is consistent with that observed for 1 NS -PSt, further supporting the PPFSt-grafted structure on the 1 NS surface.
To investigate the effect of polymer grafting on the hydrophobicity of the material, water contact angle (CA) measurements were performed (Figure 8 and Figure S49-S52, Supporting Information).First, we evaluated the CA of bulk 1 and 1 NS -L as reference materials.Bulk 1 exhibited a CA of 26.9 ± 2.09°, indicating its hydrophilic nature resulting from the exposed Zn metal sites that may attract water molecules for coordination.In contrast, 1 NS -L displayed a significantly high CA of 95.7 ± 3.60°, indicating its hydrophobic nature.This hydrophobicity can be attributed to the aliphatic tetramethylpiperidine moiety of L which caps the Zn sites, thus exposing the nanosheet surfaces.We discovered that 1 NS -PPFSt exhibited an extremely high CA of 150.8 ± 4.31°, confirming the expected superhydrophobicity.Since the CA for 1 NS -PPFSt even exceeded that observed for 1 NS -PSt 20k (106.5 ± 5.84°), we conclude that the excellent hydrophobicity is caused by the layers of grafted PPFSt chains on the nanosheets.
Owing to its inherently superhydrophobic nature, the 1 NS -PPFSt nanosheet demonstrated significantly enhanced stability in water.PXRD analyses confirmed that 1 NS -PPFSt maintains its crystal structure even after soaking in water for two days at 25 °C (Figure S47, Supporting Information).Therefore, grafting PPFSt chains provides excellent superhydrophobicity and water stability to the moisture-sensitive [Zn 2 (ZnTCPP)] n MOF nanosheets.
To demonstrate the potential of surface-initiated polymer grafting to create versatile 2D nanomaterials, we fabricated a membrane composed of superhydrophobic 1 NS -PPFSt nanosheets for oil/water emulsion separation applications.[64] Because the high stability of emulsions makes it challenging to apply conventional phase separation methods, membrane separation is considered an efficient approach.To prepare the membrane, we employed a well-established filtration method (vacuum-assisted filtration assembly) [61] to deposit 1 NS -PPFSt nanosheets onto the surface of a commercially available PTFE filter with a pore size of 0.45 μm.This process resulted in the formation of a thin laminar membrane consisting of superhydrophobic 1 NS -PPFSt nanosheets on a porous PTFE support.Membranes fabricated from MOF nanosheets can be divided into two categories: porous and laminar. [61]The former comprises nanosheets with intrinsic nanopores that enable the selective permeation of ions and molecules, whereas the latter consists of nanosheets (porous or nonporous) arranged in a laminar structure where the channels between the layers act as pathways for molecular transport.The 1 NS -PPFSt nanosheet membrane fabricated in this study is categorized as the latter case, in which the hydrophilic interlayer channels serve as oil penetration paths.
Single-component permeation tests revealed that the 1 NS -PPFSt membrane effectively prevented water penetration (Movie S1, Supporting Information), whereas typical organic solvents, such as n-hexane (Movie S2, Supporting Information) and dichloromethane (DCM) permeated the membrane rapidly.This indicates that the 1 NS -PPFSt membrane can be effectively employed in oil/water separation applications.To assess the separation capabilities of the 1 NS -PPFSt membrane for practical mixtures, a toluene/water (98/2, v/v) emulsion was prepared using ultrasonic emulsification.A photograph of the emulsion obtained using an optical microscope is shown in Figure 9a.When this emulsion was passed through the 1 NS -PPFSt membrane, only toluene permeated, whereas the dispersed water droplets were unable to pass through the interlayer superhydrophobic channels of the membrane (Movie S3, Supporting Information).Consequently, a clear filtrate was obtained after passing the solution through the membrane, and no dispersed water droplets were observed under an optical microscope (Figure 9b).This membrane separation technique was also applicable to other combinations of organic solvents and water, including nhexane, DCM, and aqueous ethyl acetate (EA) aqueous emulsions.To accurately assess the effectiveness of the membranes in the separation processes, the efficiency of separation and the rate of permeation, namely, the selectivity (R%) and flux (F), respectively, were measured, as well as the separation performance of the 1 NS -PPFSt membrane for aqueous emulsions containing n-hexane, DCM, and EA (Figure 9c, Table S5, Supporting Information, see Experimental Section).Compared with previously reported MOF composite membranes for oil/water emulsion separation (Table S6, Supporting Information), the 1 NS -PPFSt membrane demonstrated high selectivity and separation flux.In the representative toluene/water emulsion separation, the selectivity of the separation exceeded 99.95%.Among them, the toluene/water emulsion exhibits the highest separation flux at 2,955 ± 30.46 L m -2 h -1 , followed by n-hexane/water emulsion at 2,770 ± 17.46 L m -2 h -1 .The separation throughput for EA/water emulsion reaches 2,317 ± 89.56 L m -2 h -1 .However, DCM shows the best solubility with PPFSt and consequently the DCM/water emulsion displays the lowest flux at 1,383 ± 509.9 L m -2 h -1 .The low flux can be attributed to partial swelling of the 1 NS -PPFSt membrane, which reduced the size of the interlayer channels.Although some recent studies have achieved higher separation fluxes of up to 15,600 L m -2 h, -1 [65] the separation selectivity for emulsions (90%) was lower than that of the 1 NS -PPFSt membrane.
In addition, we investigated the stability of the 1 NS -PPFSt membrane during the separation process of toluene/water emulsion using multiple-cycle separation experiments (Figure 9d).Interestingly, the membrane demonstrated excellent durability with no significant separation selectivity or flux changes even after 20 separation cycles (Figure 9d, Figures S53 and S54, Table S7, Supporting Information).This suggests that surface grafting of the polymer is a promising method for surface modification and for improving the stability of MOF nanosheets.The slight decrease in the separation flux after the 10th cycle may be attributed to the damage to the hydrophobic polymer layers.After the 20th cycle of toluene/water emulsion separation, the membrane was examined, and its surface F and Zn elemental compositions were determined by SEM-energy dispersive X-ray spectroscopy (EDX) elemental analysis.The original membrane had surface F and Zn contents of 17.3% and 6.5%, respectively (Table S8, Supporting Information).However, analysis of the membrane showed a slight decrease in the surface fluorine content to 14.8% and an increase in the zinc content to 8.8% (Table S9, Supporting Information).This could be attributed to the removal of a portion of the PPFSt chains grafted from the nanosheet surfaces by the toluene phase during separation, causing a slight reduction in the flux rate.Notably, a mixed-matrix membrane made from a simple mixture of 1 NS -L and free PPFSt did not exhibit such a high selectivity (Figure S55, Supporting Information).These results confirm the importance of anchoring the polymer on the surface of the nanosheets to enhance the performance of MOF nanosheet membranes.

Conclusion
We developed a new approach to directly graft polymers from the surfaces of 2D MOF nanosheets, enabling the fabrication of stable and highly dispersible 2D nanosheets.This was achieved by incorporating a specially designed monodentate ligand with a polymerization-initiating group into 2D MOF crystals of ZnTCPP.The ligand facilitated the formation of a turbostratic phase in the MOF crystals and enabled the one-step synthesis of single-layer nanosheets through an in-situ solvent-assisted exfoliation process.Subsequently, using a post-synthesis surfaceinitiated polymerization technique, PSt was grafted from the surfaces of the 2D MOF nanosheets, providing protection against moisture and allowing for versatile property modulation.The resulting PSt-grafted nanosheets exhibit excellent dispersibility in organic solvents, and the grafted polymers enhance the catalytic activity and stability of the MOF nanosheets in water.Furthermore, a superhydrophobic 2D MOF nanosheet was fabricated by grafting PPFSt from the nanosheet surfaces.A membrane composed of PPFSt-grafted nanosheets demonstrated highly efficient separation of oil/water emulsions, with minimal degradation even after 20 separation cycles.This study presents a promising method for fabricating surface-functionalized MOF nanosheets with improved properties and demonstrates their potential for various applications, including catalysis and separation processes.Furthermore, this could be extrapolated to the synthesis of polymer-functionalized 2D conductive MOF nanosheets, a progression that would hold substantial promise for the fabrication of electronic, sensing, and energy storage devices. [66,67]pecifically, the adaptation of the current approach to conduc-tive MOFs could facilitate the development of 2D MOF-based thin films for electronic applications or advanced solid-state energy storage solutions. [68,69]Thus, the versatility of the presented method will lead to significant contributions to the wide range of materials science fields in which further technological breakthroughs are required.
General Instrumentation: 1 H NMR spectra were recorded using a JEOL ECS-400 spectrometer operating at 400 MHz.Powder X-ray diffraction (PXRD) data were recorded using Cu K radiation on a Rigaku SmartLab X-ray diffractometer.MDI Jade 6 was used for data processing, full-profile fitting, and peak indexing of the PXRD profile.Materials Studio software package (ver.19.1, Dassault Systèmes BIOVIA) was used to create the model structure for the PXRD refinement.Size-exclusion chromatography (SEC) measurements were performed at 40 °C on a Shodex GPC-101 system with two polystyrene gel columns in series (Shodex KF-806 M) and equipped with a refractive index detector and a UV detector.The mobile phase was tetrahydrofuran (THF) at a 1.0 mL mi −1 n flow rate.The column pressure was 6.0 MPa.Differential scanning calorimetry (DSC) was performed using a Hitachi High-Tech DSC7020 instrument at a heating rate of 10 K/min under continuous nitrogen flow.Scanning electron microscopy (SEM) was performed using a Hitachi High-Tech S-3000N instrument at an accelerating voltage of 5 kV.The samples were deposited on a conducting carbon tape attached to an SEM sample holder and coated with platinum.Atomic force microscopy (AFM) was performed using an Asylum Research MFP-3D Origin.An Al-coated silicon cantilever (OMCL-AC160TS, Olympus) was used.The polymer sample was dissolved in chloroform and cast on a mica or HOPG substrate by spin-coating at 2500 rpm for 10 s.Dynamic light scattering (DLS) measurements were performed using a Malvern model Zetasizer equipped with a He-Ne laser ( = 633 nm) at an angle of 6°-7°at 25 °C.Gas chromatography (GC) was performed using a Shimadzu GC-2014 system equipped with a flame ionization detector (GC-FID).A GC column (Rtx-1; Agilent Technologies) was used.The injector and detector temperatures were kept constant at 200 °C, and the column temperature was controlled from 100 °C to 300 °C at a ramp rate of 5 °C/min.Two-step Synthesis of 1 NS -L: TCPP (40 mg, 0.050 mmol, 1 eq), Zn(NO 3 ) 3 •6H2O (45 mg, 0.15 mmol, 3 eq), L (130 mg, 0.50 mmol, 10 eq) were dissolved in a mixed solvent (DEF/EtOH = 3/1, v/v, 10 ml).The resulting solution was then sonicated for 2 min.The solution was divided into ten glass vials (1 mL each).The solution was heated in a convection oven to 75 °C for 12 h, then cooled to room temperature.The product was collected by centrifugation (25,000 g, 10 min) and washed by DMF (three times) and acetone (three times) successively, affording 1 bulk -L. 1 H NMR of the material was measured to determine its chemical formula after digestion in d 6 -DMSO/CF 3 COOD (9/1, v/v).
To exchange the guest solvents, the as-synthesized 1 bulk -L was soaked in DMF for two days.The supernatant was replaced with fresh DMF daily.This procedure was repeated using acetone as the solvent.The resulting 1 bulk -L including acetone, was soaked in different alkylbenzene solvents (toluene, ethylbenzene, and cumene) for another three days.In the alkylbenzene solution, 1 bulk -L changes to a turbostratic phase.This phase shift was monitored by in-situ PXRD measurements using a sealed glass capillary (Figure S13, Supporting Information).The resulting material was suspended in an organic solvent (THF, EtOH, or iPrOH) and stirred overnight (2000 rpm).After standing for 12 h, the exfoliated nanosheets, 1 NS -L, in the supernatant were collected by centrifugation (25,000 g, 10 min).
General Procedure of Surface-initiated Polymerization from 1 NS -L1 NS -L: (10 mg) was suspended in 1,4-dioxane (1 ml).Distilled styrene (St) (1 ml) was then added to the mixture using a syringe.The resulting mixture was degassed by three freeze-pump-thaw cycles and heated to 130 °C for 6 h, 12 h, and 24 h under vigorous stirring.After a specified polymerization time, the reaction mixture was cooled to room temperature.The resulting purple gel-like material was washed with THF (ten times) by dispersion and sedimentation cycles to remove the polymer not attached to the nanosheets.The solid residue was collected by centrifugation (25,000 g, 10 min) to obtain 1 NS -PSt. 1 NS -PPFSt was synthesized using the same procedure, except that 2,3,4,5,6-pentafluorostyrene (PFSt) was used as the monomer with a polymerization time of 24 h.The polymers grafted from the nanosheets were detached by digesting the nanosheets in trifluoroacetic acid (TFA) and isolated by reprecipitation from MeOH.
General Procedure of Catalytic Photoredox Reactions: Phenylboronic acid (30 mg, 0.25 mmol, 1 eq) and the photocatalyst (0.01 eq) were placed in a glass tube.DIPEA (87 μL, 0.5 mmol, 2 eq) and DMF (1 mL) were added to the glass tube.The mixture was exposed to blue light ( = 455 nm, 12 W at the distance, 2 cm from the LED light source) with vigorous stirring in the air.After 24 h, the photocatalyst was collected by centrifugation (25,000 g, 10 min) and recycled if required.Then, 1 M HCl aq.(1 mL) was added to the supernatant solution at 0 °C and left for 1 h.The resulting mixture was diluted ten times by adding DMF, and the diluted solution was analyzed using GC to calculate the conversion yield.
Preparation of Laminar Nanosheet Membranes: Laminar membranes of 1, 1 NS -PSt 20k, and 1 NS -PPFSt were fabricated using a vacuum-assisted filtering assembly.Respective THF suspension of 1, 1 NS -PSt 20k, and 1 NS -PPFSt (10 mg in 20 ml each) were filtered using commercial PTFE membranes (Omnipore, 0.45 μm, Merck), affording the laminar membranes deposited on the porous PTFE support.The membranes were used for the oil/water emulsion separation experiments without further treatment.
Oil/water Emulsion Separation Experiments: The toluene/water emulsion separation was described in detail as a representative procedure.A mixture of water (2 ml) and toluene (98 ml) was sonicated for 1 h to form an emulsion.For the subsequent flux and selectivity measurements, the water phase was dyed with methylene blue (0.5 mg mL −1 ) prior to use.To evaluate the separation performance of the nanosheet membrane (1 NS -PPFSt), vacuum-assisted filtration was performed using a diaphragm pump (ULVAC, DAP-15, working pressure: 39.9 kPa).The separation flux and selectivity were calculated using a previously reported method. [70]The water content of the organic filtrate was determined by measuring the absorbance at 644 nm using a predetermined calibration line for methylene blue concentration (Figure S53, Supporting Information).Selectivity (R%) and flux (F) were calculated using the following equations: where c 1 and c 0 represent the concentrations of water in the filtrate and original mixture, respectively.A, V, and t denote the effective membrane area, filtration volume, and filtration time, respectively.The separation of the other oil/water mixtures (DCM, EA, and hexane) was performed in an identical manner.
Statistical Analysis: For the convenience of presentation, the profiles measured by size exclusion chromatography (SEC) are normalized based on RI intensity.In in-situ PXRD performed in a capillary, each measured profile is subtracted from the background signal measured in an empty capillary tube under the same step length, exposure time, and optical system parameters.Contact angles for all material samples were presented as the mean ± SEM of four independent measurements.The separation fluxes for the different water-in-oil emulsions measured for the same membrane sample are presented as the mean ± SEM of three independent measurements.

Figure 1 .
Figure 1.Overview of the synthesis of polymer-grafted MOF nanosheets.a) Synthesis of dispersed MOF nanosheets functionalized with polymerization initiating agent (L) on the surfaces (1 NS -L), which is achieved by both two-step and one-step routes.H 4 TCPP(H 2 ) denotes protonated version of TCPP in a free base form.b) Subsequent surface-initiated polymerization from 1 NS -L provides c) polymer-grafted MOF nanosheet (1 NS -Polymer).Chemical structure of the ZnTCPP unit in d) 1 NS -L and e) polystyrene (PSt)-grafted MOF nanosheet (1 NS -PSt).

Figure 2 .
Figure 2. a) PXRD patterns of (red) 1 bluk -L, (blue) 1 NS -L obtained in the one-step synthesis, and (green) 1 NS -L obtained in the two-step synthesis.A black line shows a simulated pattern of 1 bulk -L.b) Particle size distribution of (green) 1 NS -L obtained in the one-step synthesis and (orange) 1 NS -L obtained in the two-step synthesis, measured by DLS in THF at 25 °C.

Figure 3 .
Figure 3. Characterization of 1 NS -L obtained in the two-step synthesis.a) AFM image of the 1 NS -L.b) Height profiles along (black) Line 1 and (red) Line 2 shown in the AFM image (panel a).c) The unit chemical structure of 1 NS -L, showing the calculated thickness of the sheet.d) Modeled crystal structure of 1 NS -L.(top) Side view along a axis, showing the calculated thickness of the sheet.(bottom) Top view along c axis, showing the d-spacing of 1.1 nm corresponding to the (110) plane.e) SEM and f) TEM image of the 1 NS -L.(g) HRTEM image of the 1 NS -L, showing a fringe pattern corresponding to the (110) plane with a regular spacing of d = 1.1 nm (inset: SAED pattern of the corresponding area).

Figure 4 .
Figure 4. Characterization of 1 NS -L obtained in the one-step synthesis.a) AFM image of the 1 NS -L.b) Height profile along Line 1 shown in the AFM image (panel a).c) Digital photograph of the Tyndal effect of 1 NS -L dispersed in isopropanol.d) XPS spectra of Zn 2p 3/2 for (top) the 1 NS -L and (bottom) 1 in bulk form.The purple circles and orange line represent the original data and the result of peak fitting, respectively.The Zn 2p 3/2 peak is split into (yellow) higher-and (blue) lower-energy components by the peak deconvolution analysis.e) SEM and f) TEM image of the 1 NS -L.g) HRTEM image of the 1 NS -L, showing the specific fringe pattern of the (110) plane, as also seen in that obtained in the two-step synthesis (inset: SAED pattern of the corresponding area).

Figure 7 .
Figure 7. a) Results of the recycle experiments for the photocatalytic reaction using (orange) 1 NS -L, (green) bulk 1, and (purple) 1 NS -PSt 20k as the catalyst.b) PXRD patterns of the 1 NS -PSt 20k catalyst after a given cycle of the recycle experiments.

Figure 9 .
Figure 9. Digital photograph and microscope image of a) water-in-toluene emulsion (toluene/water) before separation and b) the filtrate obtained after the separation using the 1 NS -PPFSt membrane.c) Separation flux of different water-in-oil emulsions.d) Separation performance of 1 NS -PPFSt membrane in 20 separation cycles of the toluene/water emulsion.

Table 1 .
Results of the surface-initiated polymerizations.
a) General condition of surface-initiated polymerization:1 NS (10 mg) was dispersed in 1 ml St or PFSt and then heated to 130 °C (110 °C for PFSt) after degassing by three freeze-pump-thaw cycles.

Table 2 .
Results of catalytic photoredox oxidation reaction of phenylboronic acid.