Symmetry breaking has received considerable attention for its potential to unravel the origin of homochirality as found in nature.1–3 The construction of artificial symmetry broken systems is also of practical significance in pharmaceutical,4 biology,5 chiral electronics,6 asymmetric catalysis7 and other fields.8 It avoids the prerequisite of enantiopure precursors in the fabrication of chiral materials, and has unique advantages in reducing cost and expanding applications.9 Symmetry breaking normally occurs during self-assembly wherein achiral molecules adopt spontaneously acentric stacking or racemates are deracemized.10–12 For instance, the tiny chiral bias generated by achiral precursors during assembly can be amplified to produce considerably unequal amounts of right- and left-handed nanostructures.13 Symmetry breaking has also been occasionally observed in liquid crystal,14,15 Langmuir films,16 and supramolecular gels.17 Although symmetry breaking occurs in a variety of forms, the previous reports have prevalently focused on the spontaneous emergence of homogeneous chirality in the aggregation of discretely dispersed molecules in solution. Symmetry breaking in intrinsically confined environments (e.g. in a solid state) remains unexplored, although it has crucial potential implications for revealing the origin of homochirality in nature, such as that in seashells and insect elytron.
The precise manipulation of intermolecular interactions between building blocks plays a pivotal role in symmetry breaking.18–20 As an important weak interaction, halogen bond has been widely used in organic crystal engineering, profiting from its exclusive directionality, tunability and hydrophobicity.21–23 However, co-crystallization of halogen bond donors (e.g. halides24) with curved polycyclic aromatic molecule (e.g. helicenes25,26)-derived halogen bond acceptors to regulate the accumulation patterns and prepare advanced chiral materials is solely unexplored.
Herein, we report the preparation of azahelicene-derived halogen-bonded cocrystals and a unique solid-state symmetry breaking process emerged during the dissociation of halogen bonds by removing halides from the cocrystals. Due to the relatively weak halogen bonding with azahelicenes, the halides in the cocrystals can be completely removed through a vacuum–heating treatment. Then the residuary azahelicenes spontaneously rearranged into ordered conglomerates, revealing a distinctive solid-state reordering process. Especially, the removal of halides from the racemic cocrystals of 2-aza[4]helicene (2N4H), where equal amounts of P- and M-conformers are immobilized, produced emergent chiroptical active solid with well-preserved integral morphology. Such solid-state symmetry breaking was instantly amplified with the extension of vacuum–heating treatment and eventually generated chiral conglomerates. In the presence of additional enantiopure 4-aza[6]helicene (4N6H) with fixed chirality, the chirality of symmetry breaking of 2N4H can be precisely controlled by the “sergeants and soldiers” effect.
Pentafluoroiodobenzene (1IFB), 1,4-diiodotetrafluorobenzene (1,4-2IFB), 1,2-diiodotetrafluorobenzene (1,2-2IFB) and 1,3,5-trifluoro-2,4,6-triiodobenzene (3IFB) were selected as halogen bond donors to prepare halogen bond-woven cocrystals with azahelicenes. 4N6H was first employed to prepare the cocrystals, typically by diffusion of n-hexane vapor into an acetone solution of the mixed molecules. Although 4N6H failed to strongly associate with 1IFB but generated single crystal alone, it co-crystallized other three halides and produced diversified halogen bond-woven cocrystals (Figure 1).
For P-4N6H/1,4-2IFB cocrystal, P-4N6H and 1,4-2IFB was crystalized in a 2:1 ratio and an interspaced layered packing along a axis was clearly observed (Figures 1, S10). As seen from the bc plane, each 1,4-2IFB combined with two P-4N6H molecules through C–I···N halogen bonds, forming dumbbell-like subunits, which were staggered and extended into a single layer. The two iodine atoms on 1,4-2IFB were both involved in the formation of C–I···N bonds, with lengths of 2.9 Å and 2.8 Å, and angles of 179.0° and 172.1°, respectively. In P-4N6H/1,2-2IFB cocrystal, P-4N6H and 1,2-2IFB formed herringbone trajectories along c axis, which were further arranged in a parallel fashion along b axis (Figures 1, S11), mainly through intermolecular halogen bonding. Probably due to the steric hindrance, only one iodine atom on 1,2-2IFB was connected to P-4N6H through a C–I···N bond, with a length of 3.0 Å and an angle of 160.8°. Obviously, the strength of such halogen bonding was weaker than that between P-4N6H and 1,4-2IFB. This was presumably caused by the weakened positivity of the end region of C–I bond for 1,2-2IFB. The co-crystal of P-4N6H and 3IFB revealed an interwoven weak-bonding network (Figures 1, S12). Interestingly, in addition to the C–I···N halogen bonding between P-4N6H and 3IFB (length of 2.9 Å and angle of 166.9°), a second C–I···π halogen bond with a length of 3.6 Å and an angle of 177.8° was observed between the adjacent 3IFB molecules. The presence of intermolecular halogen bonds was also strongly evidenced by the displacement of characteristic peaks in X-ray photoelectron spectroscopy (XPS), Fourier transform infrared (FT-IR) spectra and Raman spectra (Figures S13–S17). On the other hand, compared with the circular dichroism (CD) signal of P-4N6H single crystal, which reached a maximum value at 362 nm, the CD signals of halogen bond-woven cocrystals obviously blue shifted (Figure S18), again reflecting the existence and effect of halogen bonds.
Subsequently, thermodynamic properties of the above co-crystals were studied by thermogravimetric analysis (TGA). As shown in Figure 2a–c, the above cocrystals all showed two periods of weight loss during the heating process. P-4N6H/1,4-2IFB, P-4N6H/1,2-2IFB, P-4N6H/3IFB cocrystals lost 38%, 55% and 60% of weight in the first period, respectively, perfectly matching with the weights of corresponding halides in the cocrystals. Notably, the starting departure temperatures of 1,4-2IFB, 1,2-2IFB and 3IFB from the cocrystals were 100, 120 and 135 °C, respectively, which were correlated with their boiling points. All the second weight loss occurred around 260 °C, which can be attributed to the volatilization of P-4N6H (Figure S20). In addition, the melting point of P-4N6H was determined to be 222 °C (Figure S20). It appeared that the halides could leave from the cocrystals by heating below the melting temperature of P-4N6H.
The cocrystals were subsequently heated at 150 °C to remove the halides. After 1 h, 19F NMR spectra of the residual solids showed that the halides were completely eliminated (Figures S21–S23). Powder X-ray diffraction (PXRD) patterns (Figure 2d) showed that the residual solids almost lost the crystallinity. To reduce the thermal interference, vacuum treatment was employed to remove the halides at a lower temperature (Figure 2e). Typically, a glass tube containing the cocrystals was placed in a glass desiccator, which was connected to a vacuum pump and heated on a heating block. After 2 h of vacuum treatment at 60, 95 and 105 °C, respectively, 19F NMR spectra showed that 1,4-2IFB, 1,2-2IFB and 3IFB were completely removed from the cocrystals (Figures S24–S26) and 1H NMR spectra showed that the remaining solids were pure P-4N6H (Figures S27–S29). As shown by XRD patterns (Figure 2f–h), with the extension of vacuum–heating treatment, a series of new peaks appeared and the original peaks of cocrystals disappeared. Finally, the residual aggregate exhibited characteristic peaks that were coincided with P-4N6H single crystal. Hence, the removal of halides allowed P-4N6H to directly transform from cocrystals to single crystals completely in a solid-state. It was worth mentioning that along with the crystal-to-crystal transition, the corresponding Cotton effect of solids also returned to the characteristic signal of P-4N6H single crystal (Figure S30).
The change of crystal morphology during the removal of halides was revealed by scanning electron microscope (SEM). The original flat crystal surface of the cocrystals became irregularly rough after heating at 150 °C for 1 h (Figure S31 middle panels and S32), reflecting the decrease of crystallinity as shown by XRD patterns (Figure 2d). In contrast, after the vacuum–heating treatment, the crystal surface turned crisscrossed with clear edges (Figure 3a, Figure S31 right panels). For P-4N6H/3IFB cocrystal, SEM images (Figure 3a) showed that after vacuum treatment at 105 °C for 1 h, the edge of the original micron sheet decomposed into regular blocks, while the central region remained intact. X-ray diffraction structural analysis showed that the intact structure remained to be P-4N6H/3IFB cocrystal, while the dissociated areas was corresponded to P-4N6H single crystal (Figure S33), revealing a unique in-situ crystal-to-crystal transformation process. As 3IFB was completely removed, the morphology of the solid changed into tightly packed slim blocks. It was postulated that after the removal of 3IFB, the remaining P-4N6H molecules start to flip and slip to some extent due the emergence of free space (Figure 3b). Further, re-nucleation of agglomerates begins due to the association between adjacent P-4N6H molecules, possibly through intermolecular C–H···N bonds. Then, more molecules were gradually involved in the re-crystallization and eventually generated pure P-4N6H crystals.
Subsequently, 2-aza[4]helicene (2N4H), which possesses chiral conformations but rapidly enantiomerizes in solution at ambient condition, was employed to grow cocrystals with 1IFB, 1,4-2IFB and 3IFB by diffusion of n-hexane vapor into an acetone solution of the mixed molecules. The resulting 2N4H/1IFB, 2N4H/1,4-2IFB and 2N4H/3IFB cocrystals revealed inter-locked, layered and interleaved molecular packings, respectively, where the P- and M-conformers of 2N4H were equally immobilized (Figure 4a). Specifically, 1IFB and 2N4H were connected via a single C–I···N bond (Figure S34), while 1,4-2IFB and 3IFB both generated C–I···N bonds with two 2N4H molecules, representing stronger halogen bond synergies.
Then the cocrystals were heated under vacuum to remove the halides. 19F NMR spectra showed that 1IFB, 1,4-2IFB and 3IFB were completely removed from the cocrystals after 1 h of vacuum treatment at 40, 95 and 105 °C (Figures S37–S40), respectively. 1H NMR spectra showed that the remaining solids were pure 2N4H (Figures S41–S43). The stacking of the residuary 2N4H molecules was revealed by XRD patterns (Figure 4b–d). For 2N4H/1IFB cocrystal, the original peaks disappeared after the removal of 1IFB, while a series of new peaks appeared (Figure 4b). Surprisingly, it was found that the emerging peaks were in perfect agreement with the characteristic pattern of P-2N4H single crystal, indicating that the residual 2N4H molecules spontaneously reordered into homochiral conglomerate in the solid state. In contrast, for 2N4H/1,4-2IFB cocrystal, the residual 2N4H molecules recrystallized into rac-2N4H single crystal (Figure 4c). In addition, there is no obvious crystallinity of the solid obtained after the removal of 3IFB (Figure 4d), indicating the loss of ordered stacking. It seemed that the selectivity of reordering was highly dependent on the vacuum–treating temperature. The reduced crystallinity was probably caused by the remarkably low melting point of 2N4H, which was determined as 85 °C (Figure S35), obviously below the removal temperatures of 1,4-2IFB and 3IFB.
For 2N4H/1IFB cocrystal, the samples retained the platelet morphology during the vacuum–heating process, although the surface gradually became rough (Figures 5a, S44). On the contrary, the regular prisms of 2N4H/1,4-2IFB and 2N4H/3IFB cocrystals melted into chunks after the vacuum–heating treatment (Figure S45). The optical activity of the aggregates obtained during the removal of halides was recorded by circular dichroism (CD) spectra. The samples were dispersed into KBr and pressed to eliminate interference from linear dichroism and other artifacts. With the departure of 1IFB, CD signals emerged in the region of 280~400 nm, corresponding to the absorption region of 2N4H (Figure S46), and gradually increased with the extension of the vacuum–heating treatment (Figure 5b). Time-dependent XRD patterns also indicated the generation of homochiral aggregates (Figure S47). Analysis of 25 parallel samples showed batch-dependent CD signals and intensities, revealing a representative spontaneous symmetry breaking process (Figure 5c). Notably, mirror-image CD signals were found in a series of pairs of chiral aggregates (Figure 5d), further confirming the symmetry breaking behavior. Moreover, such symmetry breaking process has also been observed during the removal of halide from the complex formed by 2N4H and 2,4-difluoro-1-iodobenzene (1I24FB) (Figure S48).
Control experiments showed that after being treated in the same vacuum–heating condition, the single crystal of 2N4H or the physical mixture of 2N4H and 1IFB failed to generate chiroptical activity (Figure S49). It appeared that the participation and departure of halide in the cocrystal provided essential space for the molecular rearrangement of 2N4H, which played an important role in symmetry breaking. As shown in Figure 5e, after the removal of 1IFB from the cocrystal, molecular re-nucleation probably produces an occasional enantiopure region, which subsequently becomes the homochiral nuclei. The surrounding P and M isomers gradually associate with the homochiral nuclei and are instantly converted into homochiral conformation to maximize the intermolecular C–H···π forces. Hence, the chirality of the reorganized aggregate amplifies along the growth of homochiral domain, until forming a homochiral aggregate.
The above results suggested that the solid-state symmetry breaking was strictly dependent on the association of the cocrystal. It was concluded that the generation and intensity of chiroptical activity are mainly modulated by two factors (Figure 6). First, a single halogen bond with weak bond energy allows the temperatures of dissociation and rearrangement below the melting point of 2N4H, which is conducive to the symmetry breaking to form homochiral aggregates (Figures S50, S51). On the contrary, multiple halogen bonds cause a large dissociation temperature that higher than the melting point of 2N4H, resulting in forming racemic aggregates (Figures S52, S53). Second, the homogenous structure greatly contributes to the chirality transfer and amplification process. One cocrystal should produce only one handedness (P or M) of homochiral aggregate upon the vacuum–heating treatment, leading to the strong chiroptical activity. For instance, one long-range-ordered 2N4H/1IFB cocrystal produced strong Cotton effects after being heated at 40 °C under vacuum for 1 h (Figure S54a). However, after the same vacuum–heating treatment, the mixture of scattered 2N4H/1IFB cocrystals produced apparently weaker Cotton effects (Figure S54b). Since each cocrystal has an equal probability of producing a P- or M-chiral aggregate, the system of multiple cocrystals generally exhibited substantially reduced homochirality.
To accurately manipulate the absolute configuration generated by the solid-state symmetry breaking, enantiopure 4N6H with fixed chirality was employed as chiral inducer. The 2N4H/1IFB cocrystal was soaked with a n-hexane solution containing 0.01 eq. of P-4N6H and then dried at room temperature to allow a uniform coating of P-4N6H. Subsequently, the complex was heated at 40 °C under vacuum for 1 h to remove the halide. The sample generally retained the crystal morphology after the vacuum–heating treatment (Figure S55). XRD patterns showed that the characteristic peaks of the treated sample were in perfect agreement with enantiopure (P- or M-) 2N4H (Figure 7a). The presence of negative CD signals in the region of 280~400 nm (Figures 7b, S56) further confirmed the generation of P-2N4H aggregates (Figure S59). On the other hand, positive CD signals (corresponding to M-2N4H aggregates) were observed for the sample derived from the cocrystal doped with 0.01 eq. of M-4N6H. These results were further supported by 10 parallel experiments (Figure 7c). As postulated in Figure 7d, during the re-nucleation of 2N4H, P-4N6H selectively bind with P-2N4H to form P-chiral nuclei. Then, with the stacking of 2N4H molecules around the nuclei and their conversion into P-chiral conformations, the overall chirality constantly amplifies until the complete formation of P-2N4H aggregate. Correspondingly, M-4N6H selectively associate with M-2N4H to form M-chiral nuclei and to induce the formation of M-2N4H aggregate.
In conclusion, we have presented a unique solid-state symmetry breaking process of dynamically chiral azahelicene. Under the vacuum treatment at a slight elevated temperature, the halide can be removed from the cocrystal, allowing the residual azahelicene to rearrange into homochiral conglomerates. The introduction of a tiny amounts of chiral inducers in the re-nucleation further allows the precise manipulation of the chirality. This strategy should also be applicable to a diverse array of molecules with dynamically chiral conformations, such as π-extended nanographenes27,28 and polygonal ring-containing curved nanographenes.29 Consequently, such solid-state symmetry breaking from racemic precursor would remarkably expands the efficiency for the fabrication of chiral organic materials.