Introduction

Chirality pervades nature, playing an essential role in biological systems, chemical synthesis, and materials science1,2. Among its myriad forms, helical chirality is distinguished by its highly ordered structural motifs and diverse functionalities3. Naturally occurring helices, such as DNA double helices and α-helical proteins, exemplify the critical importance of helical chirality in biological processes4. Beyond biology, synthetic methodologies have facilitated the creation of helical architectures in covalent organic polymers (COPs)5, inorganic metal-based materials6, and supramolecular soft gels7, unlocking their potential for cutting-edge applications.

An especially elegant approach to helicity control is supramolecular transcription, which amplifies molecular-level chirality into mesoscale helical structures8,9,10,11. This strategy employs chiral soft templates to orchestrate the self-assembly and crosslinking of precursor molecules via noncovalent interactions, including electrostatic forces, hydrogen bonding, and π–π stacking. Remarkably, even products derived entirely from achiral monomers can exhibit strong chiroptical activity12,13. Chiral inversion can be achieved by modifying the building blocks10,14, adjusting external conditions (e.g., solvent, light, or pH)15, or reconfiguring the assembly pathway16,17, with these changes typically reflected in altered chiroptical responses or static structural shifts. In most reported systems, supramolecular chirality is directly transferred and amplified from the chiral template to helical organic polymers, establishing a one-to-one correction between helical handedness (P- or M-helix) and chiroptical response9,18,19,20,21. For example, in a single system, P-helical COPs exhibit positive circular dichroism (CD), whereas their M-helical counterparts show negative CD, and vice versa. This pattern was also corroborated by our recent studies8,15,22. In enantiomerically pure template systems, the chiroptical activity of mesoscopic COPs was precisely modulated using achiral metal ions or comonomers, accompanied by a coupled switch in mesoscopic helicity. These systems thus enabled precise control over both helicity and optical properties. Nevertheless, whether an inherent mechanistic link exists between helicity and chiroptical activity remains an open question. While mesoscale helicity is readily observable, the molecular-level chiral organization and stacking are often difficult to characterize, underscoring the need for further investigation.

Covalent organic frameworks (COFs) are porous crystalline polymers formed by covalently linked organic monomers, renowned for their precisely designed topologies and highly tunable porosity23,24. Chiral COFs (CCOFs), a key subset, have emerged as a focal point in chiral materials chemistry due to their unique chirality and excellent crystallinity, offering significant potential in chiral recognition, enantiomer separation, asymmetric catalysis, and optical applications owing to their adjustable pore structures and robust stability25,26,27. Traditionally, CCOF synthesis relies on three approaches28,29: (1) post-synthetic modification of achiral COFs to introduce chirality, (2) direct synthesis using optically pure monomers, and (3) chirality-induced synthesis from achiral precursors via chiral catalysts. All these approaches are rooted in molecular chirality. However, the inherent conflict between the low symmetry of chiral monomers and the high symmetry required for COF crystallization hampers their development30. Moreover, conventional CCOFs rely strictly on the molecular chirality of the building blocks or inducers to establish chiroptical activity, necessitating enantiomerically pure starting materials to obtain mirror-image CCOFs, a process both cumbersome and inefficient. To address these limitations, our recent work introduced a strategy that bypasses molecular chirality, focusing instead on mesoscale helical chirality in COFs9. Through self-crystallization or monomer exchange, helical COPs were converted into CCOFs composed entirely of achiral monomers, retaining chiroptical activity closely correlated with the helicity of the parent COPs. This approach not only streamlines chirality incorporation but also opens unexplored avenues for CCOF design. Notably, recent reports have revealed isomerism in COFs31,32,33, where identical monomers crystallize into different topologies under varied catalysts, solvents, or temperatures, yielding distinct properties. Inspired by these findings, several intriguing questions arise. Within a single helical COF framework, can parameters such as interlayer stacking be further modulated to generate internal structural diversity? Moreover, would such variations in interlayer arrangements, in turn, influence the material’s chiroptical activity? Addressing these questions is essential for understanding the origin of chiroptical activity in mesoscale helical CCOFs constructed entirely from achiral monomers.

Here, we report a groundbreaking phenomenon of chiroptical isomerization in amorphous COPs and crystalline COFs exhibiting mesoscopic helical chirality, where identical mesoscopic helicity yields opposing chiroptical responses. Using a standard supramolecular transcription approach, we synthesized several amorphous helical polyimines from amine and aldehyde monomers, with their mesoscopic helicity directed by the molecular chirality of the template agent (Fig. 1a). Intriguingly, subtle adjustments to a single reaction parameter, such as the selection of structurally similar amine or aldehyde monomers, the type of good solvent, or the ratio of good to poor solvents, triggered diverse, including reversed chiroptical responses in these polyimines, despite their uniform mesoscopic helicity following the template’s chirality (Fig. 1b). Combined experimental and theoretical analyses revealed that this chiroptical inversion likely stems from opposite chiral arrangements of sub-nanometer units, while the mesoscopic twist direction remains fixed under the template’s chiral control. Moreover, by precisely varying the binary composition of achiral good solvents (dioxane and CH3CN), we achieve continuous tuning of the Cotton effect from negative to positive in helical polymers of identical chemical composition. Leveraging the symmetry of diamine and trialdehyde building blocks, these polyimines were converted into thermodynamically stable crystalline COFs through self-crystallization or monomer exchange, preserving both helicity and chiroptical isomerization. Remarkably, even when both diamine and trialdehyde units in the parent COP were simultaneously replaced to form a β-ketoenamine COF with entirely new building blocks, the original mesoscopic helicity and its chiroptical isomerization traits were seamlessly retained (Fig. 1c). Fundamentally, this discovery challenges the long-standing assumption in supramolecular and polymer systems that mesoscale helicity directly dictates the sign of chiroptical responses. Instead, it establishes that chiroptical activity is not solely determined by the overall helical geometry but is also highly sensitive to the sub-nanometer-scale chiral arranging—an insight that provides a refined conceptual framework for chirality transfer and amplification across hierarchical length scales, and offers a versatile strategy for designing chiral materials with tunable optical responses independent of molecular chirality, thus expanding their potential in chiral photonics, enantioselective sensing, and optoelectronic applications.

Fig. 1: Supramolecular templating and linker-exchange strategies enabling helical polyimines and COFs with identical mesoscopic helicity but opposite chiroptical responses.
Fig. 1: Supramolecular templating and linker-exchange strategies enabling helical polyimines and COFs with identical mesoscopic helicity but opposite chiroptical responses.
Full size image

a Schematic illustration of the preparation of helical amorphous polyimines using a 2,3-DAP-mediated supramolecular template method. b By varying a single experimental parameter in the supramolecular transcription system, such as structurally similar but distinct amine or aldehyde monomers, the type of good solvent, or the ratio of good to poor solvents, the resulting polyimines displayed identical mesoscopic helical chirality but opposite chiroptical activities. c Schematic illustration of the preparation of CCOFs with identical chemical composition and mesoscopic helical chirality but opposite chiroptical responses through a stepwise linker exchange approach.

Results

In this study, we focused on the synthesis of polyimines exhibiting mesoscopic helical chirality, achieved through our recently developed supramolecular templating strategy mediated by 2,3-diaminopyridine (2,3-DAP)15. The template molecules employed were (R)- and (S)-PhgC16, long-chain amino acid derivatives synthesized via the acylation of phenylglycine with hexadecyl chloride (Supplementary Figs. 1 and 2). The preparation of these polyimines proceeded as follows: Initially, (R)/(S)-PhgC16, 2,3-DAP, and diamine monomers were dissolved in good solvent, followed by the addition of water as a precipitating agent under vigorous stirring to drive the self-assembly of chiral supramolecular architectures. At ambient temperature, the introduction of 1,3,5-triformylbenzene (Tf) or terephthalaldehyde (TPA) initiated a Schiff base condensation at the assembly interfaces. After polymerization, the resulting solids were collected via filtration and purified by extraction with anhydrous ethanol to eliminate residual templates and oligomeric imine byproducts. The resulting products included amorphous COPs: those derived from the reaction of o-tolidine (BD-Me2) with Tf and TPA were designated as XCOP-1 and XCOP-2, respectively, while the COP from biphenyldiamine (BD) and TPA was named XCOP-3 (Fig. 2a). Here, X denotes the molecule chirality of (R)- and (S)-PhgC16, being either R or S.

Fig. 2: Structural control of helical polyimines revealing identical mesoscopic helicity but reversed molecular-scale chiral stacking.
Fig. 2: Structural control of helical polyimines revealing identical mesoscopic helicity but reversed molecular-scale chiral stacking.
Full size image

a Structurally similar amine-formaldehyde monomers form RCOP-1/2/3. b–d SEM images of (b) RCOP-1, (c) RCOP-2 and (d) RCOP-3. e–g TEM images of (e) RCOP-1, (f) RCOP-2 and (g) RCOP-3. h DRCD and DRUV-Vis spectra of RCOP-1/2/3 and SCOP-1/2/3. i Reverse chiral stacking of aromatic segments within the same mesoscopic helical skeleton.

Typically, the chiroptical responses of helical polymers derived from the same supramolecular transcription system are closely correlated with their helicity. However, in this study, we reported an unusual phenomenon where helicity and ground-state chirality were decoupled. By systematically varying a single parameter within the transcription system, such as the selection of structurally similar amines or aldehydes, the type of the good solvent, or the ratio of good to poor solvents, the resulting polyimines exhibited diverse, including opposite, chiroptical responses. Within the same templating system, utilizing the template molecule (R)-PhgC16, scanning electron microscopy (SEM) images revealed that RCOP-1, RCOP-2, and RCOP-3 all exhibited distinct left-handed helical morphologies (Figs. 2b–d). Their hollow structures were clearly discernible in transmission electron microscopy (TEM) images (Figs. 2e–g). Specifically, RCOP-1 displayed lengths ranging from a few micrometers to tens of micrometers, with diameters of 90–150 nm and helical pitches of 200–300 nm. In comparison, RCOP-2 exhibited slightly reduced diameters and pitches, with a less uniform sense of twist than RCOP-1, though its overall helical character remained distinctly recognizable. RCOP-3 had diameters comparable to RCOP-1 but featured a reduced pitch of 90–200 nm. Similarly, SCOP-1, SCOP-2, and SCOP-3 exhibited corresponding right-handed helical morphologies (Supplementary Fig. 3). The helicity of these COPs corresponded directly to the molecular chirality of the (R)/(S)-PhgC16 template employed. Fourier-transform infrared (FTIR) spectra confirmed that RCOP-1, RCOP-2, and RCOP-3 were all linked by imine units, with the polymer backbones containing unreacted amino and aldehyde groups (Supplementary Fig. 4). Powder X-ray diffraction (PXRD) analysis revealed that RCOP-1 and RCOP-2 were amorphous. In contrast, RCOP-3 exhibited several intense reflections in the 20–30° range; however, its low BET surface area and the absence of N2 uptake in the microporous region indicated a semi-crystalline structure (Supplementary Figs. 5 and 6)34. Thermogravimetric analysis (TGA) demonstrated that all three polymers possessed favorable thermal stability, with RCOP-1 and RCOP-2 remaining stable up to 410 °C under a nitrogen atmosphere, and RCOP-3 exhibiting stability up to ~500 °C (Supplementary Fig. 7).

To probe the ground-state chirality of these COPs, we employed solid-state diffuse reflectance circular dichroism (DRCD) and ultraviolet-visible absorption (DRUV-Vis) spectra (Fig. 2h). Unexpectedly, the helicity of these COPs did not dictate the sign of their Cotton effect. For instance, RCOP-1 and RCOP-2, synthesized using the same amine monomer (BD-Me2) but different aldehyde monomers (Tf versus TPA), exhibited opposing chiroptical responses. RCOP-1 displayed two negative Cotton effects at 338 nm and 457 nm, consistent with its UV-Vis absorption profile, whereas RCOP-2 showed two positive Cotton effects at 303 nm and 469 nm. To elucidate the origin of these Cotton effects, we measured the DRCD and DRUV–Vis spectra of the individual monomers (Supplementary Fig. 8) (Tf, TPA, BD-Me2, and BD). All monomers were CD-silent due to their achiral nature. Apart from BD, which is prone to oxidation and exhibits a broad absorption from 200–800 nm, the absorption of Tf, TPA, and BD-Me2 was confined below 400 nm. Therefore, the Cotton effects observed in the 400–500 nm region for RCOP-1 and RCOP-2 can be attributed to the chiral stacking of extended conjugated segments formed by imine condensation between amine and aldehyde units35,36,37,38. In addition, the short-wavelength Cotton effects, including RCOP-1 at ~330 nm and RCOP-2 at ~300 nm, indicate chiral packing of local aromatic rings. Thus, despite sharing left-handed helical morphologies at the mesoscopic scale, RCOP-1 and RCOP-2 exhibited opposing chiral stacking of aromatic units at the molecular level (Fig. 2i). It should be noted that, owing to the amorphous nature of the COPs, such chiral stacking interactions are short-range. A similar divergence was observed between RCOP-2 and RCOP-3, which shared the same aldehyde monomer but differed in the amine monomer by the presence of two ortho-methyl groups. RCOP-3 displayed three negative Cotton effects at 290 nm, 361 nm, and 478 nm, in stark contrast to the positive Cotton effects of RCOP-2. Likewise, SCOP-1 and SCOP-2, as well as SCOP-2 and SCOP-3, showed opposing Cotton effects and molecular-scale chiral stacking patterns. The characteristic Cotton effects of the (R)/(S)-PhgC16 supramolecular self-assemblies appeared below 300 nm, while the main CD peaks of the (R)/(S)-PhgC16 co-assembled with 2,3-DAP were observed around 280 nm and 370 nm (Supplementary Fig. 9). In contrast, the CD maxima of R/SCOP-1/2/3 were located near 450 nm, indicating that the ground-state chirality arising from residual template molecules could be excluded. Diffuse reflectance linear dichroism (DRLD) measurements revealed negligible linear dichroism (LD) in all COPs, confirming that LD contributions to the CD signals were minimal (Supplementary Fig. 10). To elucidate the origin of the discrepancy between chiroptical activity and mesoscopic helicity, we optimized the ground-state geometries of pentamer segments of COP-1, COP-2, and COP-3 using simplified time-dependent density functional theory (sTD-DFT) calculations, assuming an identical anticlockwise chiral stacking arrangement (Figs. 3a–c, and Supplementary Figs. 11-13). The simulated ECD spectra, calculated over the same wavelength range as in the experiments (220–800 nm), exhibited comparable exciton-coupled Cotton effects, characterized by positive signals in the long-wavelength region (Fig. 3d). Correspondingly, pentamers constructed in a clockwise chiral stacking mode yielded ECD spectra that were nearly mirror images of their anticlockwise counterparts, displaying negative Cotton effects in the same spectral region (Supplementary Fig. 14). These results indicate that when COP-1, COP-2, and COP-3 share the same chiral stacking mode at the molecular level, their ground-state chirality presented by CD profiles should be comparable, exhibiting similar positive or negative signals in the CD spectra within the absorption region of the conjugated imine segments. Thus, the experimentally observed inversion of chiroptical activity is more plausibly attributed to an opposite chiral stacking mode rather than intrinsic molecular structural differences. Such contrasting stacking modes might arise from the distinct physical properties of the building blocks, for example, the higher hydrophilicity and water solubility of TPA compared with Tf, or the steric hindrance introduced by the ortho-methyl substituents in BD-Me2 relative to BD. These differences likely lead to variations in the microscopic chiral packing of COP oligomers during the early stages of polymerization, even under the same supramolecular templating conditions. These findings highlight that the interplay between helicity and ground-state chirality is highly intricate, governed by both molecular architecture and the subtle microscopic stacking dynamics during assembly.

Fig. 3: Anticlockwise chiral stacking of COP pentamers and their computed ECD signatures.
Fig. 3: Anticlockwise chiral stacking of COP pentamers and their computed ECD signatures.
Full size image

a–c DFT-optimized geometries of pentamer segments for (a) COP-1, (b) COP-2, and (c) COP-3 in the anticlockwise chiral stacking mode (top and side views). d Simulated ECD spectra of the pentamer segments for anticlockwise-stacked COP-1, COP-2, and COP-3 obtained using the Multiwfn program.

Beyond the influence of structurally similar amines or aldehydes, which could lead to a mismatch between helicity and chiroptical activity, the type of achiral good solvent and the ratio of good solvent to poor solvent (H2O) in the supramolecular transcription system also induced chiroptical isomerization in COPs with identical chemical compositions (Fig. 4a). For example, in the synthesis of RCOP-1, using different achiral good solvents (e.g., dioxane, acetone, DMF, DMAC, DMSO, MeOH, EtOH, NMP, CH3CN, and i-PrOH) consistently produced COPs with well-defined left-handed helical morphologies (Figs. 4b–k). Their molecular chemical structures remained essentially identical, as confirmed by FTIR spectra (Supplementary Fig. 15), and all displayed long-range disordered amorphous structures, as evidenced by WAXRD patterns (Supplementary Fig. 16). However, the chiroptical responses of these COPs varied significantly depending on the achiral solvent system (Fig. 4m). Specifically, RCOP-1 synthesized in dioxane, acetone, DMF, DMAC, DMSO, MeOH, and EtOH possessed two negative Cotton effects, whereas those prepared in NMP, CH3CN, and i-PrOH displayed two positive Cotton effects. In the DMF-based system, keeping all other parameters constant but increasing the proportion of the poor solvent (H2O)—adjusting the good-to-poor solvent ratio from 1:3 to 1:5 and further to 1:7—preserved the left-handed helical morphology of RCOP-1 (Fig. 4l, and Supplementary Fig. 17), yet induced pronounced changes in the Cotton effects, culminating in a complete sign inversion at a ratio of 1:7 (Fig. 4n). Here, the solvent ratios refer exclusively to the DMF/H2O environment governing supramolecular assembly, without considering the additional DMF/H2O mixture used solely for dissolving Tf. Given that the dominant chiroptical band arose from the chiral stacking of large aromatic conjugated segments formed by the imine linkage between Tf and BD-Me2 units, the opposing Cotton effects indicated reverse stacking modes. This chiral inversion phenomenon was also perfectly reproduced in the SCOP-1 system (Supplementary Figs. 18 and 19).

Fig. 4: Solvent-mediated chiral isomerization in helical polyimines leading to opposite molecular-scale chiral stacking.
Fig. 4: Solvent-mediated chiral isomerization in helical polyimines leading to opposite molecular-scale chiral stacking.
Full size image

a Chiral isomerization of R/SCOP-1 mediated by achiral good solvents or their ratio with poor solvents. b–k SEM images of RCOP-1 prepared in (b) Dioxane, (c) Acetone, (d) DMF, (e) DMAC, (f) DMSO, (g) MeOH, (h) EtOH, (i) NMP, (j) CH3CN, (k) i-PrOH. l SEM image of RCOP-1 obtained with a DMF-to-water ratio of 1:7. m DRCD and DRUV-Vis spectra of RCOP-1 prepared in different achiral good solvents. n DRCD and DRUV-Vis spectra of RCOP-1 obtained with a DMF-to-water ratio of 1:7. o Schematic diagram of identical mesoscale helicity with opposing molecular-scale chiral stacking.

In addition to chirality inversion, some systems exhibited spectral shifts. For example, the CH3CN-based RCOP-1 showed a blue-shifted Cotton effect compared with its dioxane-based counterpart. However, their DRUV–Vis spectra were broadly similar, each featuring a broad absorption band in the 400–500 nm region, which can be attributed to the superposition of absorptions from conjugated segments of varying chain lengths. These differences likely stem from the distinct polarity, dielectric constant, hydrophilicity, and viscosity of the solvent systems, which can modulate both the chain extension of the dominant chiral conjugated segments during supramolecular transcription and the local chiral stacking arrangement. The combined effect of these factors results in Cotton effect shifts, even for RCOP-1 samples sharing identical mesoscopic helical handedness. Similar phenomena have been reported in solvent-polarity-controlled supramolecular assemblies16,39,40, where supramolecular chirality in soft gels can be tuned by solvent polarity, leading to both Cotton effect inversion and partial spectral shifts, in agreement with our observations. For the structurally similar RCOP-2 synthesis system, no comparable chiroptical isomerization was observed. In all the aforementioned solvent systems, RCOP-2 consistently exhibited well-defined left-handed helical senses (Supplementary Figs. 20a–k) and positive Cotton effects (Supplementary Figs. 20m and 20n). In contrast, RCOP-3 formed random aggregates in MeOH and EtOH systems (Supplementary Figs. 21f and g), displaying CD silence (Supplementary Fig. 21m). In other good solvent systems, RCOP-3 showed well-defined left-handed helical morphologies (Supplementary Figs. 21a–j). Specifically, it exhibited negative Cotton effects in solvent systems such as dioxane, acetone, and DMF, but displayed an opposite positive Cotton effect in the CH3CN system (Supplementary Fig. 21m), demonstrating a solvent-dependent chiroptical isomerization phenomenon similar to that observed in RCOP-1. However, adjusting the DMF-to-water ratio to 1:7 did not induce a comparable chiral inversion phenomenon (Supplementary Fig. 21n).

To elucidate the mechanism of this solvent-dependent chiral response isomerization, we selected dioxane and NMP systems as representative examples. Prior to polymerization (before the addition of Tf solution), the supramolecular co-assemblies in both the dioxane and NMP systems were isolated via vacuum filtration (Supplementary Fig. 22). Both resulting supramolecular co-assemblies exhibited left-handed helicity (Supplementary Fig. 23), and their CD signals were similar, featuring a positive Cotton effect near 350 nm and two negative Cotton effects around 240 nm and 268 nm (Supplementary Fig. 24). These results indicated that the chiral stacking modes of the supramolecular co-assemblies in the dioxane and NMP systems were identical prior to polymerization. Thus, the solvent effect did not directly alter the helicity or chiroptical response of the supramolecular soft templates. We then monitored the polymerization process over time (Supplementary Fig. 25). In the dioxane system, the helical structure began to be transcribed after about 5 minutes and approaches equilibrium after 3 h. In the NMP system, the reaction proceeded more slowly, with transcription of the helical structure beginning after 1 h and approaching equilibrium after 10 h. This time-dependent evolution was further corroborated by DRCD measurements (Supplementary Figs. 26 and 27). As the reaction proceeded, the chiroptical activity of RCOP-1 in both systems gradually intensified until equilibrium was established, with the dioxane system reaching CD saturation significantly faster than the NMP system, consistent with the equilibrium times observed in the time-dependent SEM studies. Remarkably, even at the earliest measurable time points, RCOP-1 already displayed opposite Cotton effects in the two solvent systems, despite the absence of any discernible mesoscale helical morphology. These findings provide direct evidence that the kinetics of helical transcription and amplification are strongly governed by the solvent environment, and that the conjugated imine oligomers formed in the initial stage adopt opposite chiral stacking modes in the two solvent systems. Logically, helical polymers derived from supramolecular assemblies with identical helicity and chiroptical activity via chiral transfer and amplification should have exhibited the same chirality. However, in this system, we observed an achiral solvent-dependent chiroptical activity phenomenon. We hypothesized that during the initial nucleation stage of RCOP-1, differences in the polarity, hydrogen-bonding capacity, viscosity, and other solvent-related parameters between dioxane and NMP synergistically led to distinct chiral stacking directions of the initial oligomeric units. However, as polymerization progresses, the mesoscopic twisting direction was governed by the helicity of the supramolecular assemblies formed by the (R)-PhgC16-based supramolecular template, resulting in identical helicity. For instance, the oligomers formed by the polymerization of Tf and BD-Me2 initially assembled in a right-handed helical manner (Fig. 4o). Yet, as polymerization continues, the sub-nanometer units grew in a left-handed twisted form (Fig. 4o). This evolutionary process resulted in identical mesoscale helicity but opposing chiroptical responses. This phenomenon highlighted a significant distinction between molecular chiral stacking at the sub-nanometer scale and helical growth at the mesoscale.

Given that achiral good solvents were mutually miscible and their composition ratios could be precisely controlled, we explored their competitive effects in the chiral supramolecular transcription system (Fig. 5a). In the RCOP-1 synthesis system, we selected dioxane (exhibiting negative Cotton effects) and CH3CN (exhibiting positive Cotton effects) and systematically varied their volume ratios. The resulting RCOP-1 consistently exhibited left-handed helicity (Figs. 5b–k), but their CD signals varied continuously with the volume ratio (Fig. 5l). When the proportion of dioxane decreased from 100% to 80%, the intensity of the two negative Cotton effects decreased significantly. At 70%, a weak positive Cotton effect emerged near 410 nm, corresponding to a negative Cotton effect at 465 nm. This positive-negative excitonic coupling Cotton effect suggests stronger chiral coupling between the large conjugated segments of RCOP-1. Further reducing the dioxane proportion led to a gradual increase in the intensity of the positive Cotton effect, accompanied by a redshift. At 10%, the negative Cotton effect disappeared entirely, with only two positive Cotton effects observed near 299 nm and 417 nm. This work provides a previously unexplored demonstration that the ground-state chirality of a polymer with identical helicity could be controllably tuned from positive to negative circular dichroism in a stepwise manner. Through Gaussian fitting (Fig. 5m)41,42,43, we found that when the volume fraction of dioxane in the binary solvent mixture ranged from 20% to 80%, the relative CD intensity of the resulting RCOP-1 followed a well-defined trend described by:

$$y={y}_{0}+\frac{A}{w\sqrt{\pi /2}}{{{{\rm{e}}}}}^{-2\frac{{\left(x-{x}_{c}\right)}^{2}}{{w}^{2}}}$$
$$y={y}_{0}+\frac{A}{w\sqrt{\pi /2}}{{{{\rm{e}}}}}^{-2\frac{{\left(x-{x}_{c}\right)}^{2}}{{w}^{2}}}$$
Fig. 5: Cosolvent-regulated chiroptical modulation and nonlinear CD response in helical RCOP-1.
Fig. 5: Cosolvent-regulated chiroptical modulation and nonlinear CD response in helical RCOP-1.
Full size image

a Competitive effect of achiral cosolvents on the circular dichroism of RCOP-1. b–j SEM images of RCOP-1 obtained with varying volume ratios of dioxane to CH3CN (the ratios indicated in the figure represent the proportion of dioxane). k Schematic structure of the left-handed helical RCOP-1. l DRCD and DRUV-Vis spectra of RCOP-1 prepared in different volume ratios of dioxane to CH3CN. m Nonlinear relationship between the CD intensity of RCOP-1 and the volume ratio of dioxane ranging from 20% to 100%.

With parameters \({y}_{0}\) = 0.03, \({x}_{c}\) = 1.16, w = 0.67, and A = 0.89. Here, x represents the volume fraction of dioxane in the dioxane/acetonitrile binary solvent system (0.2–1.0), and y represents the CD intensity of RCOP-1 in the mixed solvent relative to that obtained in pure dioxane. This nonlinear trend underscored the complex chiral stacking modes during the supramolecular transcription process of RCOP-1 in binary competitive achiral solvents.

The constituent monomers of RCOP-1, Tf and BD-Me2, were structurally symmetric and linked via dynamic, reversible imine bonds, enabling their transformation into thermodynamically more stable COFs through monomer exchange (Figs. 6a, b). In the COP synthesis, 2,3-DAP was introduced as a structural modulator. Although 2,3-DAP contains amino groups capable of reacting with aldehydes, its pyridyl substituent exerts a strong electron-withdrawing effect15, thereby reducing the nucleophilicity of the amino groups. Combined with the relatively mild reaction conditions (room temperature) and its much lower molar ratio compared with the main amine monomers, the asymmetric 2,3-DAP was not significantly incorporated into the polymer backbone. Consequently, it did not interfere with the subsequent transformation of helical COPs into helical COFs. To further confirm the molecular backbone structures of these imine-linked COPs, we performed 1H NMR analysis after decomposing the frameworks in a mixture of deuterated trifluoroacetic acid and D2O (v/v = 10:1). As shown in Supplementary Fig. 28, the 1H NMR spectra of RCOP-1, RCOP-2, and RCOP-3 closely match those of their corresponding achiral control COPs synthesized directly from the respective monomers, with no detectable signals corresponding to 2,3-DAP. These results demonstrate that 2,3-DAP, owing to its low nucleophilicity under room-temperature conditions, functions as a removable structural modulator that promotes the formation of helical COPs while exerting negligible influence on the subsequent COP-to-COF conversion. In a single-step transformation (Fig. 6a), RCOP-1 prepared in dioxane and NMP solvent systems served as the precursor. Tf was replaced with triformylphloroglucinol (Tp) to yield R-Tp-BD-Me2, while BD-Me2 was substituted with 1,3,5-tris(4-aminophenyl)benzene (TAPB) or tris(4-aminophenyl)amine (TAPA) to produce R-Tf-TAPB and R-Tf-TAPA, respectively. Despite the covalent bond cleavage and reformation inherent to monomer exchange, the resulting COFs retained the mesoscopic helicity of RCOP-1. Notably, after transformation, all helical COFs—R-Tp-BD-Me2, R-Tf-TAPB, and R-Tf-TAPA—largely preserved the hollow nanotubular architecture of the parent RCOP-1 while exhibiting newly emerged mesopores (Supplementary Figs. 29-31). These mesopores likely originated from the partial decomposition of the less cross-linked amorphous COP domains during the dynamic exchange process, where the released monomers migrated and recondensed into highly crystalline COF regions. Such reconstruction could induce structural defects in the original amorphous zones, resulting in the formation of mesopores within the nanotubes. Derived from RCOP-1 in both solvent systems, R-Tp-BD-Me2 exhibited robust left-handed helicity in both solvent systems (Figs. 6c, f), with diameter and pitch nearly identical to those of the parent polymer. Similarly, R-Tf-TAPB (Figs. 6d, g) and R-Tf-TAPA (Figs. 6e, h) also displayed pronounced left-handed helicity, albeit with slightly increased diameter and pitch, possibly due to the larger molecular volumes of TAPB and TAPA relative to BD-Me2, which might induce spatial expansion of the framework upon imine bond reconnection.

Fig. 6: One-step linker-exchange transformation of helical RCOP-1 into corresponding helical COFs.
Fig. 6: One-step linker-exchange transformation of helical RCOP-1 into corresponding helical COFs.
Full size image

a Schematic diagram illustrating the one-step transformation of RCOP-1 obtained from dioxane or NMP systems into corresponding helical COFs through amine or aldehyde exchange. b Molecular structures of the COF monomers used. c–e SEM images of (c) R-Tp-BD-Me2, (d) R-Tf-TAPB, and (e) R-Tf-TAPA prepared from RCOP-1 derived from the dioxane system. f–h SEM images of (f) R-Tp-BD-Me2, (g) R-Tf-TAPB, and (h) R-Tf-TAPA prepared from RCOP-1 derived from the NMP system. i DRCD and DRUV-Vis spectra of helical COFs prepared from the transformation of RCOP-1 derived from dioxane and NMP systems. j-l PXRD patterns of (j) R-Tp-BD-Me2, (k) R-Tf-TAPB, and (l) R-Tf-TAPA in both dioxane and NMP systems. m Nitrogen adsorption−desorption isotherms of helical COFs prepared from the transformation of RCOP-1 derived from dioxane and NMP systems. n Solid-state 13C CP-MAS NMR and o FTIR spectra of RCOP-1 from the dioxane system and its derived helical COFs.

PXRD analysis revealed that both RCOP and derived helical COF obtained from dioxane and NMP systems exhibited essentially similar diffraction peaks, and RCOP-1 was amorphous, while helical COF showed high crystallinity. R-Tp-BD-Me2 showed a prominent (001) peak at 3.5°, with (110) and (001) peaks at 6.1° and 26.1°, consistent with an eclipsed-AA stacking mode (Fig. 6j). R-Tf-TAPB displayed (100), (110), (200), and (001) peaks at 5.7°, 9.9°, 11.5°, and 25.7°, also matching the eclipsed-AA mode (Fig. 6k). R-Tf-TAPA exhibited slightly lower crystallinity than R-Tp-BD-Me2 and R-Tf-TAPB. Their PXRD patterns displayed four main diffraction peaks at 6.7°, 11.1°, 13.1°, and 22.3°, corresponding to the (100), (110), (200), and (001) planes, respectively (Fig. 6l). Further profile refinement of the PXRD data yielded satisfactory refinement factors, confirming that R-Tf-TAPA also adopted an eclipsed-AA stacking arrangement and could be indexed to the P3 space group (Supplementary Figs. 32 and 33). Nitrogen adsorption-desorption isotherms indicated that RCOP-1 possessed low specific surface areas of 26.7 m2/g in dioxane and 53.0 m2/g in NMP, with pore volumes of 0.11 cm3/g and 0.22 cm3/g, respectively (Fig. 6m). The type IV isotherms with H3 hysteresis indicated the presence of mesopores44. Notably, this mesoporosity was in good agreement with the hollow tubular morphology observed in TEM images (Supplementary Fig. 29), suggesting that the mesopores mainly originated from the internal cavities of the nanotubular structure. This observation was further supported by the nonlocal density functional theory (NLDFT)-derived pore size distributions (PSD) (Supplementary Fig. 34). In contrast, R-Tp-BD-Me2, R-Tf-TAPB, and R-Tf-TAPA presented significantly higher surface areas—1258.8, 1420.4, and 825.3 m2/g in dioxane, and 1368.3, 1428.7, and 828.0 m2/g in NMP—with pore volumes of 0.76, 0.83, and 0.67 cm3/g in dioxane, and 0.86, 1.08, and 0.86 cm3/g in NMP, respectively (Fig. 6m). These helical COFs displayed type I isotherms with a hysteresis loop at high relative pressure, indicating microporosity alongside retained mesoporosity. The presence of mesopores was consistent with TEM observations (Supplementary Figs. 29–31), which revealed hollow nanotubular architectures derived from RCOP-1 as well as additional mesoporous features generated during the COP-to-COF transformation. NLDFT analysis confirmed narrow micropore distributions for R-Tp-BD-Me2 and R-Tf-TAPB (Supplementary Figs. 35a–d), consistent with reported achiral AA-stacked COFs45,46. For R-Tf-TAPA, the micropores were mainly concentrated at 0.5–0.6 nm. The slightly smaller pore size compared to the theoretical value could be attributed to its relatively weak crystallinity, which may result in limited interlayer slippage, partial blockage of pores by oligomers, or capillary-induced collapse of pore walls during activation, consistent with the reported pore size distribution of achiral Tf-TAPA COFs (Supplementary Figs. 35e, f)47,48. Thermogravimetric analysis (TGA) further revealed that all helical COFs exhibited excellent thermal stability (Supplementary Fig. 36). Solid-state 13C CP-MAS NMR (Fig. 6n) and FTIR (Fig. 6o) spectra were employed to validate the effective exchange of monomers, using RCOP-1 in dioxane system as the parent polymer and a series of derived COFs as illustrative examples. The 13C NMR spectrum of RCOP-1 exhibited a characteristic C = N peak at 155.7 ppm and a methyl peak from BD-Me2 at 17.8 ppm49. In contrast, R-Tp-BD-Me2 lacked the imine peak, instead showing signals indicative of β-ketoenamine units, with a C = O resonance at 184.4 ppm and ene-amine signal at 146.2 ppm50. Meanwhile, R-Tf-TAPB and R-Tf-TAPA retained imine peaks at 157.8 and 158.6 ppm, respectively, but lacked the methyl signal. FTIR analysis further supported these observations: RCOP-1 displayed prominent aldehyde (1701 cm-1) and imine (1624 cm-1) bands51, whereas R-Tp-BD-Me2 exhibited keto (1620 cm-1), C = C (1578 cm-1), and C-N (1276 cm-1) bands52, confirming the formation of β-ketoenamine. Additionally, R-Tf-TAPB and R-Tf-TAPA showed reduced aldehyde intensity, reflecting effective amine-aldehyde reticulation. Together, these findings provided robust evidence for the successful substitution of Tf and BD-Me2 within the COF structures.

Given that the parent RCOP-1 exhibited solvent-dependent chiroptical isomerization in 1,4-dioxane an NMP systems, a key question arises: did this chirality isomerization persist in helical COFs following covalent bond cleavage and reformation during thermodynamic crystallization? DRCD characterization revealed that the chiroptical response and isomerization properties of RCOP-1 were effectively transferred to the resulting helical COFs (Fig. 6i). Taking R-Tp-BD-Me2 in the dioxane system as an example, the transformation of imine units in the framework into β-ketoenamine units led to a shift in both the absorption band and CD signals to the characteristic region of the β-ketoenamine moiety. Compared to RCOP-1, the primary absorption band of R-Tp-BD-Me2 redshifted from 453 nm to 546 nm, with its negative CD signal shifting from 457 nm to 552 nm. Additionally, R-Tp-BD-Me2 manifested a positive CD signal at 446 nm. This dual-peak, exciton-coupled Cotton effect suggested enhanced chiral π–π stacking of the β-ketoenamine units in R-Tp-BD-Me253. In contrast, R-Tp-BD-Me2 in the NMP system displayed markedly different CD behavior, exhibiting a positive Cotton effect at 535 nm within the absorption region of the β-ketoenamine unit54. Notably, the chiral stacking direction of the extended conjugated interlayer segments in R-Tp-BD-Me2 differed between the dioxane and NMP systems55, indicating that the solvent-mediated chiroptical isomerization of the parent polymer was effectively transmitted to the inherited helical COFs, even after multiple covalent bond breaking and reformation events. Similarly, for R-Tf-TAPB, a negative Cotton effect was observed at 424 nm in the dioxane system, while a positive Cotton effect appeared at 415 nm in the NMP system, both aligning well with their respective absorption regions. For R-Tf-TAPA, a negative Cotton effect was detected at 519 nm in the dioxane system, whereas a positive Cotton effect emerged at 504 nm in the NMP system. Additionally, the chiroptical activity of these helical COFs was found to remain stable under various conditions, including heat (120 °C), acid (1 M HCl), base (1 M NaOH), and solvent (DMSO) exposure, confirming the stability of their mesoscale helicity (Supplementary Figs. 37 and 38). To the best of our knowledge, this solvent-dependent chiroptical response isomerization dominated by mesoscale helical chirality has not been previously reported in the COF field, and its tunable chiroptical activity held significant promise for applications in chirality-related fields28,30.

The results above demonstrated that a single-step transformation effectively transferred the helicity and chiroptical response isomerization of RCOP-1 to helical COFs. Building on this, we further explored the replacement of Tf in R-Tf-TAPB or R-Tf-TAPA with Tp through a stepwise conversion process (Fig. 7a). Remarkably, across both dioxane and NMP systems, the resulting R-Tp-TAPB and R-Tp-TAPA frameworks exhibited refined left-handed mesoscopic helicity (Figs. 7b–e), with diameters and pitches showing minimal deviation from their parent R-Tf-TAPB or R-Tf-TAPA structures. PXRD analysis confirmed their favorable crystallinity. For R-Tp-TAPB, a prominent (100) facet peak appeared at 5.6°, with (110) and (200) peaks at 9.7° and 11.4°, respectively, aligning with the simulated eclipsed-AA stacking mode (Fig. 7f). Similarly, R-Tp-TAPA displayed clear (100), (110), and (200) peaks at 6.5°, 11.4°, and 13.1°, respectively, also consistent with the eclipsed-AA stacking configuration (Fig. 7g)56. Notably, compared with their parent R-Tf-TAPA, the crystallinity of R-Tp-TAPA was markedly improved, suggesting that the substitution of Tf with Tp effectively restored the partially collapsed pore structure. Nitrogen adsorption-desorption analysis revealed that, irrespective of whether RCOP-1 synthesized in dioxane or NMP systems served as the parent polymer, all resulting R-Tp-TAPB and R-Tp-TAPA frameworks showed characteristic type-IV isotherms, accompanied by H3-type hysteresis loops indicative of mesoporous features (Fig. 7h)44. PSDs calculated via NLDFT showed that R-Tp-TAPB featured a narrow micropore distribution at 1.2 nm (Supplementary Fig. 39a, b), consistent with prior reports57, while R-Tp-TAPA’s micropore size increases to 0.9 nm compared to R-Tf-TAPA (Supplementary Figs. 34c, d), approaching the theoretical pore dimension of ideal AA stacking56. This observation further corroborated the enhanced structural order associated with the transformation from imine-linked COFs to β-ketoenamine-linked COFs. Solid-state 13C CP-MAS NMR (Fig. 7i) and FTIR (Fig. 7j) spectra further validated the effective replacement of Tf with Tp, with imine units transforming into β-ketoenamine linkages. Consequently, we successfully transformed RCOP-1, originally constructed from A (Tf) and B (BD-Me2) monomers in a [C3 + C2] topology, into R-Tp-TAPB and R-Tp-TAPA with a new [C3 + C3] topology comprising C (TAPB or TAPA) and D (Tp) monomers through stepwise exchange, while preserving mesoscopic helical chirality throughout. DRCD and DRUV-Vis spectra further elucidated their chiroptical properties (Fig. 7k). The absorption regions of the Tp-based COFs corresponded closely to β-ketoenamine units, exhibiting significant redshifts compared to their imine-based parent COFs. For R-Tp-TAPB in dioxane system, a negative Cotton effect at 490 nm aligned with the absorption of its β-ketoenamine linkages, confirming chirality transfer from R-Tf-TAPB58. Strikingly, in NMP system, R-Tp-TAPB displayed a positive Cotton effect at 482 nm, mirroring the chiral isomerization behavior of its parent R-Tf-TAPB. Similarly, R-Tp-TAPA exhibited a negative Cotton effect at 532 nm in the dioxane system but a positive one at 529 nm in the NMP system. These findings revealed that the chirality isomerization of the achiral solvent-mediated parent RCOP-1 was first transmitted to helical imine COFs via amine exchange and subsequently to helical β-ketoenamine COFs through stepwise aldehyde exchange.

Fig. 7: Stepwise linker-exchange conversion of helical RCOP-1 into diverse helical COFs.
Fig. 7: Stepwise linker-exchange conversion of helical RCOP-1 into diverse helical COFs.
Full size image

a Schematic diagram illustrating the stepwise transformation of RCOP-1 obtained from dioxane or NMP systems into corresponding helical COFs through amine and aldehyde exchange. b–c SEM images of the stepwise transformation of RCOP-1 obtained from the dioxane system into (b) R-Tp-TAPB and (c) R-Tp-TAPA. d–e SEM images of the stepwise transformation of RCOP-1 obtained from the NMP system into (d) R-Tp-TAPB and (e) R-Tp-TAPA. f–g PXRD patterns of (f) R-Tp-TAPB and (g) R-Tp-TAPA obtained from RCOP-1 derived from dioxane and NMP systems. h Nitrogen adsorption−desorption isotherms and k DRCD and DRUV-Vis spectra of helical COFs prepared from the stepwise transformation of RCOP-1 derived from dioxane and NMP systems. i Solid-state 13C CP-MAS NMR and j FTIR spectra of R-Tp-TAPB and R-Tp-TAPA prepared from the stepwise transformation of RCOP-1 derived from dioxane system.

Based on these observations, we proposed a possible mechanism to rationalize how helical COFs exhibited consistent mesoscale helicity while manifesting opposing chiroptical responses (Fig. 8). In a supramolecular transcription system utilizing (R)-PhgC16 as the templating agent, achiral solvents such as dioxane or NMP did not perturb the supramolecular co-assembly of (R)-PhgC16, 2,3-DAP, and amine monomers. Upon the introduction of aldehyde monomers, however, a rapid Schiff-base condensation occurs, generating conjugated imine oligomers. Differences in solvent properties—including polarity, hydrogen-bonding capacity, and viscosity—drove distinct assembly behaviors of these oligomers around the chiral template16,40,59. For example, direct the packing behavior of these oligomers around the chiral template, leading to opposite molecular-level helical stacking in dioxane and NMP (Fig. 8a). Nevertheless, both assemblies were governed by robust electrostatic interactions with the supramolecular template, resulting in uniform helical orientations at the mesoscale (Fig. 8b)8,15. Consequently, the resultant imine polymers shared identical mesoscale helicity but displayed contrasting chiroptical responses, a phenomenon attributable to their divergent chiral stacking modes at the molecular level60. Furthermore, helical COFs were synthesized by employing a monomer exchange strategy to substitute amines or aldehydes within the COP framework. Given that the aromatic units in the parent COP were pre-organized in the short-range chiral stacking arrangement, their replacement with more reactive or thermodynamically stable aromatic units preserved this chiral stacking mode (Fig. 8c). This ensured that both mesoscopic helical chirality and chirality-dependent optical isomerization were effectively transferred to the derived helical COFs, providing a robust platform for tailoring their chiroptical properties.

Fig. 8: Formation of CCOFs with identical mesoscale helicity but opposite chiroptical responses.
Fig. 8: Formation of CCOFs with identical mesoscale helicity but opposite chiroptical responses.
Full size image

a Assembly of RCOP-1 chiral oligomers into sub-nanometer units with solvent-dependent, short-range opposing stacking orientations. b Growth of these sub-nanometer units with short-range opposing chiral stacking patterns into amorphous polymers exhibiting identical mesoscale helicity under the strong electrostatic influence of external supramolecular templates. c Transfer of chirality isomerization characteristics from amorphous polymers to mesoscopic helical COFs through monomer exchange.

Discussion

Our study unveiled an unexpected phenomenon of chiroptical isomerization in both helical amorphous COPs and crystalline COFs, where identical mesoscopic helicity gave rise to reversed chiroptical responses. Using a conventional supramolecular templating strategy, we achieved precise control over the chiroptical signatures of structurally similar or identical polyimines by systematically tuning polymerization conditions, including structurally similar monomer selection, solvent composition, and good-to-poor solvent ratios. Notably, fine-tuning the binary mixture of achiral good solvents enabled continuous modulation of the Cotton effect from negative to positive in chemically identical helical polymers. This phenomenon likely arose from opposing molecular-scale chiral stacking of sub-nanometer units, despite the retention of mesoscopic helicity. Remarkably, this chiroptical isomerization persisted through monomer exchange processes involving multiple covalent bond cleavage and reformation, demonstrating exceptional stability even under stepwise linker exchange, where both aldehyde and amine components were replaced concurrently. To our knowledge, this work provides one of the earliest clear demonstrations of chiroptical isomerization in the COF domain. These findings not only deepened our understanding of the correlation between helicity and chiroptical activity in helical polymeric materials but also opened advanced avenues for the rational design of next-generation chiral optoelectronic and photonic materials with tunable chiroptical properties.

Methods

Materials and chemicals

Terephthalaldehyde, 1,3,5-triformylphloroglucinol, 1,3,5-triformylbenzene, and benzidine were procured from Shanghai Haohong Scientific Co., Ltd. o-Tolidine was initially purchased from Shanghai Haohong Scientific Co., Ltd., and subsequently from Shanghai Acmec Biochemical Technology Co., Ltd. (Cat. No. O78390-25g) during manuscript revision. N-methyl-2-pyrrolidone, anhydrous ethanol, dioxane, N,N’-dimethylformamide, 2,3-diaminopyridine, mesitylene, dichloromethane, o-dichlorobenzene, tetrahydrofuran, n-butanol, acetone, acetonitrile and N,N’-dimethylacetamide were sourced from Sinopharm Chemical Reagent Co., Ltd. Dimethyl sulfoxide, methanol, isopropanol and acetic acid were obtained from Energy Chemical. All materials were used as received without additional purification. Deionized water (resistivity: 18.2 MΩ·cm) was obtained using a Millipore DI purification system and employed in all experiments.

Computational Details

To rationalize the inconsistent CD responses of R/SCOP-1, R/SCOP-2, and R/SCOP-3, pentamer models with both clockwise and counterclockwise helical stacking were constructed for COP-1/2/3. The complex was optimized using xtb software at the gfn2-xtb level. The ECD spectrum was obtained by grimme’s sTD-DFT method and the energy range of the spectrum is 0 ~ 7.0 eV. The wB97X-D3 functional and def2-SV(P) basis set were adopted for sTD-DFT calculations. The SMD implicit solvation model was used to account for the solvation effect. All sTD-DFT calculations have been carried out by the latest version of ORCA quantum chemistry software (Version 6.1.0). The Multiwfn software was used to plot ECD spectrum.

Instrumentations

Circular dichroism (CD), UV/Vis diffuse reflectance (DRUV/Vis), and linear dichroism (DRLD) spectra were acquired using a JASCO J-1700 CD spectrometer fitted with an integrating sphere for diffuse reflectance measurements. Powder samples were analyzed with a bandwidth of 5 nm and a digital integration time of 1 s. Powder X-ray diffraction (PXRD) patterns were recorded on a Rigaku Ultimate IV diffractometer with Cu Kα radiation (λ = 1.5418 Å) in step-scan mode. Thermogravimetric analysis (TGA) was conducted on a NETZSCH TG209F3 instrument, with samples heated from 30 °C to 800 °C at a rate of 10 °C/min under a nitrogen atmosphere. Scanning electron microscopy (SEM) images were obtained using a JEOL-7800F field-emission microscope at an accelerating voltage of 10 kV, with samples coated with a thin platinum layer to enhance conductivity. Nitrogen adsorption-desorption isotherms were measured at 77 K on an ASAP-2460 instrument. Transmission electron microscopy (TEM) images were captured using an FEI Tecnai G2 microscope operating at 200 kV. Fourier-transform infrared (FTIR) spectra were recorded on a Bruker Tensor27 spectrometer over a wavenumber range of 400–4000 cm-1. Solid-state 13C cross-polarization magic-angle spinning (CP-MAS) NMR spectra were obtained on a Bruker Avance NEO 400WB spectrometer. Solution-state 1H NMR spectra were recorded at 25 °C on a Bruker ARX400 spectrometer (400 MHz). Structural models of COFs and their simulated PXRD patterns were generated using the Reflex Plus module of Materials Studio software, with atomic coordinates or CIF files sourced from relevant literature. All structural models were optimized using the Forcite molecular dynamics module in Materials Studio.

Synthesis of (R)-PhgC16 and (S)-PhgC16

The chiral template molecules (R)-PhgC16 and (S)-PhgC16, long-chain amino acid derivatives, were prepared via acylation of phenylglycine with hexadecyl chloride. Taking the synthesis of (R)-PhgC16 as an example, 0.05 mol of D-phenylglycine was dissolved in 50 mL of deionized water, followed by addition of 35 mL of acetone and 0.05 mol of NaOH to obtain a clear, homogeneous solution. Under an ice bath, 0.03 mol of palmitoyl chloride and 0.05 mol of NaOH (dissolved in 10 mL of water) were added dropwise simultaneously. The reaction mixture was stirred in the ice bath for 1 h, and then allowed to react overnight at room temperature to ensure completion. After the reaction, the mixture was acidified to pH 1 with hydrochloric acid, and the crude product was collected by filtration. The solid was washed repeatedly with water ( ≥ 5 times) and then with petroleum ether to remove by-products, affording white solid (R)-PhgC16 (17.6 g, 86.3%). The synthesis of (S)-PhgC16 was carried out under identical conditions using L-phenylglycine in place of D-phenylglycine. 1H NMR spectra of the products are shown in Supplementary Figs. 1 and 2.

General procedure for XCOP-1 (X = R or S)

The synthesis of XCOP-1 was conducted using a supramolecular template-directed approach mediated by 2,3-DAP. As a representative example, the preparation of RCOP-1 in 1,4-dioxane is described below: (R)-PhgC16 (0.051 mmol), 2,3-DAP (0.07 mmol), and BD-Me2 (0.24 mmol) were dissolved in 5 mL of dioxane under stirring. Subsequently, 15 mL of deionized water was added to the solution with continuous stirring. After 20 minutes, Tf (0.16 mmol) was dissolved in a mixed solvent of 4 mL deionized water and 0.8 mL DMF using a heating mantle set at 150 °C, at which point the actual solution temperature was ~120 °C. The hot Tf solution was then immediately added dropwise to the reaction mixture. The reaction was allowed to proceed at room temperature for 12 h. The resulting solid was collected by vacuum filtration and purified by Soxhlet extraction with anhydrous ethanol for 48 h (51.3 mg, 75.2%). The product was designated as RCOP-1, where R denotes the chirality of the template molecule. When (S)-PhgC16 was used instead, the product was named SCOP-1.

To investigate the effect of solvent on the helical morphology, RCOP-1 was synthesized in ten different good solvents (dioxane, acetone, DMF, DMAC, DMSO, methanol, ethanol, NMP, acetonitrile, and isopropanol under identical conditions, with the solvent (5 mL) being the only variable. For mixed solvent systems, the total solvent volume was maintained at 5 mL, with varying ratios of the constituent solvents, while all other conditions remained unchanged.

Synthesis of XCOP-2 (X = R or S)

The synthesis of XCOP-2 followed a procedure similar to that of XCOP-1, with the following modifications: Tf was replaced with TPA in a 1:1.5 molar ratio, and dioxane (5 mL) was substituted with 5 mL of DMF. All other reagents, quantities, and reaction conditions were identical to those described for XCOP-1 (58.2 mg, 78.1%).

Synthesis of XCOP-3 (X = R or S)

The synthesis of XCOP-3 was analogous to that of XCOP-1, with the following changes: Tf was replaced with TPA in a 1:1.5 molar ratio, BD-Me2 was substituted with an equimolar amount of BD, and dioxane (5 mL) was replaced with 5 mL of DMF. All other reaction parameters remained consistent with those for XCOP-1 (57.4 mg, 84.7%).

One-step and stepwise conversion of helical imine COPs to COFs

The conversion of chiral COPs to crystalline COFs was investigated using R/SCOP-1, prepared in dioxane or NMP, as the precursor. Both one-step and stepwise transformation strategies were employed to convert imine-based R/SCOP-1 into β-ketoenamine or imine-linked COFs through monomer exchange. All reactions were conducted in Schlenk tubes under inert conditions, and the products were thoroughly purified to ensure high purity.

One-step conversion of RCOP-1 to β-ketoenamine COF

To prepare R-Tp-BD-Me2, RCOP-1 (20 mg) and 1,3,5-triformylphloroglucinol (Tp, 30 mg) were dispersed in a mixed solvent of mesitylene/dioxane (1:1 v/v, 2 mL) in a Schlenk tube. Acetic acid (6 M, 0.3 mL) was added, and the mixture was frozen at 77 K in a liquid nitrogen bath. The tube was degassed through three freeze-pump-thaw cycles, sealed under vacuum, and heated at 120 °C for 3 days. After cooling, the solid product was collected by filtration and washed sequentially with DMF, anhydrous ethanol, acetone, THF, and dichloromethane (each at least twice). The product was dried under vacuum at room temperature for 24 h to yield R-Tp-BD-Me2 (19.7 mg, 87.5%).

One-step conversion of RCOP-1 to imine-linked COFs (R-Tf-TAPB and R-Tf-TAPA)

Synthesis of R-Tf-TAPB

RCOP-1 (20 mg) and TAPB (30 mg) were dispersed in a mixed solvent of mesitylene/dioxane (1:1 v/v, 2 mL) in a Schlenk tube. Acetic acid (6 M, 0.25 mL) was added, and the mixture was frozen at 77 K in a liquid nitrogen bath. The tube was degassed through three freeze-pump-thaw cycles, sealed under vacuum, and heated at 120 °C for 3 days. The solid product was collected by filtration, washed sequentially with DMF, anhydrous ethanol, acetone, THF, and dichloromethane (each at least twice), and dried under vacuum for 24 h to afford R-Tf-TAPB (16.4 mg, 76.1%).

Synthesis of R-Tf-TAPA

The synthesis of R-Tf-TAPA followed the same procedure as for R-Tf-TAPB, except that TAPB was replaced with an equimolar amount of TAPA. The resulting product was purified and dried as described above to yield R-Tf-TAPA (17.0 mg, 91%).

Stepwise conversion of RCOP-1 to R-Tp-TAPB

The stepwise conversion involved two stages. First, RCOP-1 was converted to R-Tf-TAPB following the procedure described above for R-Tf-TAPB. Subsequently, R-Tf-TAPB (20 mg) and Tp (30 mg) were dispersed in a mixed solvent of mesitylene/dioxane (1:1 v/v, 2 mL) in a Schlenk tube. Acetic acid (6 M, 0.3 mL) was added, and the mixture was frozen at 77 K, degassed through three freeze-pump-thaw cycles, sealed under vacuum, and heated at 120 °C for 3 days. The solid product was collected by filtration, washed with DMF, anhydrous ethanol, acetone, THF, and dichloromethane (each at least twice), and dried under vacuum for 24 h to yield R-Tp-TAPB (18.8 mg, 85.5%).

Stepwise conversion of RCOP-1 to R-Tp-TAPA

The synthesis of R-Tp-TAPA followed a similar two-stage process. First, RCOP-1 was converted to R-Tf-TAPA as described above. Then, R-Tf-TAPA (20 mg) and Tp (30 mg) were dispersed in a mixed solvent of mesitylene/dioxane (1:1 v/v, 2 mL) in a Schlenk tube. Acetic acid (6 M, 0.3 mL) was added, and the mixture was frozen at 77 K, degassed through three freeze-pump-thaw cycles, sealed under vacuum, and heated at 120 °C for 3 days. The product was collected by filtration, washed with DMF, anhydrous ethanol, acetone, THF, and dichloromethane (each at least twice), and dried under vacuum for 24 h to yield R-Tp-TAPA (19.4 mg, 86.6%).