Molecular engineering toward large pore-sized covalent organic frameworks

Molecular engineering toward large


INTRODUCTION
Exploring crystalline porous materials has recently become a focal research topic due to their great potential in solving energy and environmental issues and promoting sustainable social development [1][2][3] .Covalent organic frameworks (COFs) integrate organic building blocks into an ordered lattice through covalent bonds.With the advantages of their ordered and stable pore structure, large specific surface area, and functionalizable skeleton, COFs have elicited growing interest and have been widely applied in numerous fields [4] , such as gas separation and sorption, catalysis, energy storage and conversion, sensors, semiconductors, spintronics, drug delivery, water desalination, etc.Since the first synthesis of boronic esterlinked COF by Côté et al. in 2005 [5] , in 2007, their group reported four examples of three-dimensional (3D) COFs, namely COF-102, COF-103, COF-105, and COF-108 [6] .In 2016, Zhuang et al. reported for the first carbon-carbon double-bond connected stable COF, two-dimensional (2D) poly(phenelyenevinylene) framework (2DPPV), which has an attractive application prospect in the field of optoelectronics due to the high degree of π-π conjugation [7] .In 2017, Jin et al. also reported a similar carbon-carbon double-bond connected COF, sp 2 c-COF, with full sp 2 c conjugation covering the fundamental properties of COFs in terms of magnetism, free radicals, and so on [8] .In the same year, Ma et al. achieved the first large-size, singlecrystal growth and X-ray diffraction (SXRD) structure analysis of 3D COFs based on imine bonding [9] .Over the past few decades, hundreds of COFs with versatile chemical and topological structures have been developed [10][11][12][13][14][15][16][17][18] .The representative advantages of COFs over other porous analogs, such as zeolites, metalorganic frameworks (MOFs), polymers of intrinsic microporosity (PIMs), covalent triazine frameworks (CTFs), etc., lie in the larger pore size, stable and ordered nature, exceptionally high surface area, and controllable functionalization by pore-surface engineering.
To date, most reported COFs are micro-or small mesoporous COFs with pore sizes ranging from 0.5 to 4.0 nm.In sharp contrast, COFs with pore sizes larger than 4.0 nm have rarely been reported [19][20][21][22][23][24] due to difficulty in synthesis and lack of reliable approaches to retain large pore apertures.COFs with larger pores enable the loading, separation, and transfer of larger guest molecules [25][26][27] .Large pore apertures ensure rapid mass transport in hetero-catalysts.For instance, Zhao et al. developed macro-TpBpy-Co with both microand macropores, achieving a high oxygen evolution reaction (OER) activity with an overpotential of 380 mV at 10 mA/cm 2 superior to the microporous COF [28] .For better discussion, we defined pore sizes larger than 4.0 nm in COFs as large pores [29] , in contrast to mostly reported micro or small mesoporous COFs.In this sense, the development of COFs with regular networks that can accommodate macromolecules such as proteins or peptides holds great promise for drug delivery and heterogeneous catalysis by providing sufficient diffusion of substrates and products.To this end, recent efforts have been made to explore synthetic routes for large pore-sized COFs, which are expected to significantly expand the scope of practical applications and improve overall performance.Notably, there are versatile synthesis methods of COFs, such as template-assisted [28,30] , morphology-controlled [31] , mechanochemical, solvothermal synthesis, ionothermal, microwave-assisted, sonochemical [32] , etc.In light of the superiority of bottom-up synthesis in developing atomically precise structures, we mainly focus on the works of controlling the pore sizes of COFs through hydrothermal synthesis.

OVERVIEW OF 2D COFS WITH LARGE PORES
For the skeleton of 2D COFs, organic motifs undergo topological polymerization via condensation to form layered structures, which are further vertically stacked to form periodic columns through interlayer π-π interactions.The pore size and shape of the formed ordered open nano-channels are closely related to the stacking modules of the 2D layers.Conventional strategies to explore large linkers with intense stacking interactions and suitable solubility have enabled the construction of 2D COFs with well-defined porous structures and pore sizes up to 4.0 nm.However, preparing 2D COFs with even larger pore sizes remains challenging.A key fact hindering the development of COFs with larger pores is that the 2D layers tend to stabilize the networks by lateral sliding, a common way of reducing free energy in COFs.In 2010, based on the Density Functional Theory (DFT), Lukose et al. proposed four interlayer stacking modes of 2D COFs, namely, eclipsed (AA), staggered (AB), serrated, and inclined stacking [33,34] .According to this work, 2D COFs tend to stack in a way that exhibits the lowest energy.AA stacking is most likely to construct large pores due to enough space for alleviating steric repulsion, while AB or ABC stacking is difficult to construct large pores due to significant interlayer slipping.Consequently, the pore size and surface area of COFs are diminished because of the structural collapse and pore occlusion, which limits the successful construction of large pore-sized structures.
To date, several promising approaches have been developed to address this crucial issue.They can generally be classified as follows: (A) rational design of 2D COFs with honeycomb (hcb) and kagome (kgm) topologies, which make it easier to form larger pore size compared to other topologies; (B) employing specific linkages such as imides to generate large mesopores in the skeleton of 2D COFs; (C) using large molecule-sized building blocks with charged parts or branched side-chains to prevent the pore shrinkage and interlayer slippage, resulting in the formation of 2D COFs with large pores.
In 2016, Crowe et al. reported three novel COFs containing homogeneous and heterogeneous distributions of π-conjugated dehydrobenzoannulene (DBA), DBA-12, and DBA-18 units with varied triangular molecular sizes [37] .Polycondensation between the C 3 -symmetric monomeric catechol DBA-12 or DBA-18 and C 2 -symmetric pyrene-2,7-dibronic acid (PDBA) yielded three kgm topological COFs, 3.8 nm for Py-DBA-COF 1, 4.3 nm for Py-DBA-COF 2, and 4.1 nm for Py-MV-DBA-COF [Figure 1], respectively.The unique kgm lattice composed of large triangular planar molecules enables the formation of large hexagonal pores in the 2D COFs.It is worth noting that the as-synthesized Py-DBA-COF 2 shows excellent luminescent properties and has the largest pore size among solid-state luminescent 2D COFs, which has great potential in the field of optoelectronic applications.
The polycondensations between two C 2 -symmetric units or a C 3 -symmetric cyclic molecule and a C 2 -symmetric unit can produce kgm topotactic COFs [Scheme 1].Both synthetic routes enable the construction of large pores, demonstrating the superiority of kgm topology in preventing structural collapse.According to the comparison, the C 3 -symmetric cyclic unit-based COFs exhibit larger pores, which could be a promising platform for extending structural diversity.

Linkage-oriented synthetic strategy
COFs can be attached through a variety of methods, such as the boronate esters, triazines, azine bonds, β-ketoenamines, etc.In 2015, Jin et al. reported a series of donor-acceptor COFs named D CuPc -A PyrDI -COF, D NiPc -A PyrDI -COF, D CuPc -A NDI -COF, D NiPc -A NDI -COF, D CuPc -A PDI -COF, and D ZnPc -A PDI -COF with pore sizes of 3.6, 3.6, 3.6, 3.6, 4.0, and 4.0 nm [38] .These COFs with boronate ester linkage were applied in rapid photoinduced electron transfer and charge separation.In the same year, Xu et al. realized the synthesis of D TP -A NDI -COF with a pore size of 5.1 nm and coated it on carbon nanotubes as electrodes for battery devices [20] .In 2020, Zhao et al. reported ester-linked COF-122 with a pore size of 6.2 nm and a Brunauer-Emmett-Teller (BET) specific surface area of 908 m 2 /g, extending the 2D polyester family [24] .In 2014, Fang et al. successfully synthesized imide-linked COFs for the first time via the condensation reaction between extended aromatic dianhydrides and triamine building blocks [21] .Accordingly, a series of highly crystalline polyimide (PI)-COFs were yielded with a large pore size of 5.3 nm and a considerable surface area of 2,346 m 2 /g [Figure 2].These PI-COFs are thermally robust up to 530 °C.Afterward, a large dye molecule, rhodamine B, with dimensions of about 10 × 16 Å 2 , was incorporated into the pore of PI-COF-3 to make temperature-sensing devices, which is not possible for COFs with pore sizes less than 4.0 nm.Similar to PI-COFs, in 2020, Xu et al. reported that Ttba-TPDA-COF, with a pore size of 5.8 nm, exhibited superior dynamic adsorption capability for rhodamine B, highlighting its great potential in environmental remediation applications [23] .The most likely reason for the successful synthesis of large-pore boronate ester and imine-linked COFs is that the identical components of naphthalene diimide or perylene diimide contain fewer structural rotating sites than simple multi-phenyl ligands [Figure 3].

Fragment-oriented synthetic strategies
In 2016, Yu et al. reported an imine-linked polycationic (PC) 2D COF through the condensation of 1,3,5tris(4-aminophenyl) benzene (TAPB) and 1,1-bis(4-formylphenyl)-4,4′-bipyridinium dichloride (BFBP 2+ •2Cl -) [22] .The resultant PC COF exhibits a hexagonal structure with a pore diameter of 5.8 nm [Figure 4].The bipyridine (BIPY)-doped COF exhibits outstanding stability in water due to its PC nature and can almost quantitatively absorb anionic organic dyes in water at low concentrations.This makes the new PC COF a promising adsorbent for the removal of dye residues from water.The unique cationic BIPYderived fragment with a large molecule size is believed to provide the structural basis for stabilizing the large  While the pore sizes of most reported MOFs are smaller than 3.0 nm [39] , in 2015, Cho et al. reported a 4.9 nm-sized MOF with a strategy of incorporating pore-directing "anchors" (i.e., side chains) into building blocks to modulate the interlayer π-π interactions [40] .Along with this scenario, Emmerling et al. successfully obtained dpp-TAB COF and dpp-TAPB COF with pore sizes of 4.8 and 5.8 nm and surface areas of 1,467 m 2 /g and 1,670 m 2 /g, respectively [Figure 5] [29] .The success lies in the use of methoxy groups as pore- orientated "anchors" to prevent pore occlusion and collapse and ultimately produce highly crystalline and porous COFs.
As shown in Figure 6, the incorporation of methoxy substituents close to the imine linkage enables the diminishment of the intrinsic dipole moment and tuning of interlayer interactions.Consequently, the crystallinity of large porous lattices can be well maintained.Reproduced with permission from ref. [29] .Copyright © 2021 The Authors.Published by American Chemical Society.This publication is licensed under CC-BY 4.0.TAPB: 1,3,5-tris(4-aminophenyl) benzene; COF: covalent organic framework.
In 2022, based on the same strategy as above, Mu et al. realized the synthesis of a 2D COF with a record pore size of 10 nm, termed TDCOF-3 [Figure 7] [41] .This milestone is achieved by creatively integrating methoxy side chains and alkyne units into a single monomeric structure.Such a combination not only enlarges the length of building blocks but also overcomes the intrinsic flexibility of growing polymeric chains.The resultant large pore-sized 2D COF is capable of separating 7.0 nm pepsin and protecting tyrosinase from heat-induced inactivation.Furthermore, this work also provides conclusive evidence that large mesoporous COFs can efficiently load enzymes of large molecular size and act as a protective layer to avoid enzyme inactivation under stringent conditions.
For TDCOF-3 with a pore size of 10.0 nm, the attachment of dimethoxy groups to the identical side of the phenyl ring also modulates interlayer interaction induced by local polarity.Besides, the insertion of carboncarbon triple bonds to the main chain enhances the planarity and weakens the steric hindrance, leading to low nonbond energy.As a result, the ordered structure of TDCOF-3 with a record pore size was constructed with low nonbond energy [Figure 8].
For 3D COFs, the synthesis of large pore sizes is determined by various structural factors, and no definite mechanism has been reported in the literature.Therefore, the related explanations have been left for future work.
From the above examples, the approach of enlarging pore sizes of COFs by growing the length of monomer fragments is mainly available for hcb topological structures.For other topologies, the pore deformation might occur more significantly, making it more challenging to retain large pores.
As shown in Tables 1 and 2, the most common strategy for the synthesis of large pores is extending the length of the monomer fragments to increase the pore sizes of COFs, which ultimately leads to more intense pore collapses.To retain the large pore aperture, unique substituents, including methoxy, alkyl, etc., have to be inserted into the building blocks, which increases the difficulty of monomer synthesis and leads to a sharp drop in total yield.Besides, researchers also choose to develop new topologies to create large pores, which in itself is another challenge for COFs.From the above instances, reaction time plays a vital role in the synthesis of 2D large-pore COFs.The dynamic equilibrium of imine bond formation requires a longer time to form the large-pore COF with high crystallinity.Therefore, prolonging the time can be a potential approach to obtain large pores in 2D COFs.  .

OVERVIEW OF 3D COFS WITH LARGE PORES
Compared to 2D COFs with layered structures, the construction of 3D COFs with regular, periodic structures is proven to be a more complicated process.In particular, 3D COFs are generally less stable than 2D layered architectures due to more cavities and the lack of π-π stackings.Furthermore, 3D COFs tend to form cage-like structures when extended in three dimensions, and monomers can continue to react and grow outward in the cavity, which makes it easy for 3D COFs to generate multiple interspersed structures and more prone to collapse and shrinkage.Notably, the interpenetration between pores that occurs in diamondoid (dia) or pts topologies unfortunately leads to a high degree of channel constriction [42] .Taking these facts into account, current strategies for constructing 3D large-pore COFs mainly include (A) synthesizing monomers with side chains to prevent the collapse of large pores and (B) synthesizing novel topologies and fragments to construct large-pore size structures.
In sharp contrast to 2D COFs, the development history of large-pore 3D COFs is relatively new.In 2020, Wang et al. reported the work of expanding the pore size of 3D COFs up to 2.7 nm and achieving a high surface area of 3,023 m 2 /g through steric hindrance engineering [43] .It is worth noting that all JUC-550, 551, a Since the original articles did not mention the locations of the main peaks, we estimated them based on the PXRD patterns in the article.b The pore volumes of COFs were estimated based on the images in the articles.c Pore sizes are decided from the pore size distribution of the nitrogen sorption curves.2D: Two-dimensional; COFs: covalent organic frameworks; DBA: dehydrobenzoannulene; TAPB: 1,3,5-tris(4-aminophenyl) benzene; PXRD: powder X-ray diffraction.
and 552 have non-interpenetrating dia topology by introducing methyl and methoxy groups into the building blocks [Figure 9].For 3D COFs, it appears easier to shrink the pores or collapse the layers.As shown in Figure 10, either the (A) unmodified or (B) partially modified monomers result in shrinkage or interpenetration of the cavities [43] .Only the pores of the 3D COF built with (C) all modified units can be well maintained after the guest removal.Given that 3D cage-cavity organic structures are more prone to collapse, steric hindrance and expanding monomeric fragments are not feasible in producing large-pore 3D COFs, although they work well for creating large-pore 2D COFs.
Since the development of 3D COFs is relatively slow compared to 2D COFs and there are only two cases with pore sizes larger than 4.0 nm, we also include and discuss 3D COFs with pore sizes larger than 3.0 nm in this section.
To date, only three COFs with bor topology have been reported, and all of them possess high specific surface areas and excellent crystallinity.
On this basis, Ding et al. further reported another series of JUC-COFs using HFPTP as one of the monomers [46] .Notably, JUC-640-H achieved a record large pore size of 4.6 nm in 3D COFs with a surface area of 2,204 m 2 /g, which abundantly exposed porphyrin moieties (0.845 mmol/g) [Figure 13].The JUC-640-Co containing the Co metalized porphyrin units was used for the photoreduction of CO 2 to CO, with a high CO generation rate of 15.1 mmol g/h and a selectivity of 94.4%.It is worth noting that the catalytic efficiency of JUC-640-Co exceeds all reported COF-based materials so far.
Since the development of 3D COFs is relatively new, synthetic strategies for 3D large-pore COFs rely on exploring novel topologies and specific fragments, as shown in Tables 3 and 4.However, both methods confront bottlenecks.For instance, it has been reported that only HFPTP can form large pores and react with other building blocks with distinct symmetries.The topologies of large pore apertures are limited to bor and stp, which requires further efforts to develop more topologies to build large pore-sized 3D COFs.Reproduced with permission from ref. [43] .Copyright © 2020, American Chemical Society.3D: Three-dimensional.
Notably, because cage cavity structures are more prone to shrinkage, most works adopted immersion powder for activation, and some of them even utilized supercritical carbon dioxide during the posttreatment of COFs.

CONCLUSION AND OUTLOOK
Similar to other porous materials, the structural characteristics and application performance of COFs are highly dependent on their porosity, especially pore size.In particular, COFs with large mesopores enable the entry of unique large-sized guest molecules, which undoubtedly expands their functionalities.Therefore, it is crucial to develop efficient approaches to design and synthesize large pore-sized COFs rationally.In this work, we review several state-of-the-art strategies for producing large-pore COFs shown in Tables 2 and 4, including (A) designing COFs with specific topologies, such as hcb, kgm, stp, bor, etc.; (B) using imide linkage to prepare target COFs; (C) employing sizeable molecular building blocks with charged part or side chains to prevent shrinkage of large-pore and interlayer slippage; and (D) developing organic unit with unique steric configurations to maintain large pores in 3D COFs.We call upon the fact that despite significant progress in this appealing research area, the synthetic mechanisms remain elusive.Further efforts combining molecular-level simulations and in situ structural characterization may provide a microscopic view of the large-pore formation.Reproduced with permission from ref. [43]   .Copyright © 2020, American Chemical Society.3D: Three-dimensional; COFs: covalent organic frameworks.      .Copyright © 2023, American Chemical Society.HFPTP: 2,3,6,7,14,15-hexa(4′-formylphenyl)triptycene; DCB: dichlorobenzene.
As discussed above, most reported preparations of large pore-sized COFs still rely on simply increasing the length of the building blocks, but this results in low overall reaction yields and a tendency for structural collapse or interlayer interpenetration.Although several fragments with specific steric configurations can yield large pore-sized 3D COFs, the intrinsic relationship between structure and pore size is still unclear.In addition, all COFs with large pores reported to date are prepared using organic units with imine or boronic ester linkages, which also limits stability and subsequent functional exploration.Developing large-pore COFs with other linkages remains a challenge and requires further efforts.
Regardless of 2D COFs or 3D COFs, the pore collapse and blockage of the frameworks due to the extended length and high flexibility cannot be evaded.Therefore, an effective way needs to be proposed to retain the large porous networks.For instance, multi-bond linked COFs can be designed with relatively small pores and robust skeletons.The large-pore COFs can be generated by dissociating one or partial bonds under reverse conditions.Besides, there is an urgent need to explore a facile substituent that can

Figure 8 .
Figure 8. Calculation of the series of TDCOFs with the corresponding nonbond energies.Reproduced with permission from ref.

Figure 10 .
Figure 10.Synthesis for 3D mesoporous COFs for three possible: (A) unmodified; (B) partially modified units; and (C) all modified units.Reproduced with permission from ref.