Symmetry breaking in self-assembled M₄L₆ cage complexes

Abstract

Here we describe the phenomenon of symmetry breaking within a series of M₄L₆ container molecules. These containers were synthesized using planar rigid bis-bidentate ligands based on 2,6-substituted naphthalene, anthracene, or anthraquinone spacers and Fe^II ions. The planarity of the ligand spacer favors a stereochemical configuration in which each cage contains two metal centers of opposite handedness to the other two, which would ordinarily result in an S₄-symmetric, achiral configuration. Reduction of symmetry from S₄ to C₁ is achieved by the spatial offset between each ligand’s pair of binding sites, which breaks the S₄ symmetry axis. Using larger Cd^II or Co^II ions instead of Fe^II resulted, in some cases, in the observation of dynamic motion of the symmetry-breaking ligands in solution. NMR spectra of these dynamic complexes thus reflected apparent S₄ symmetry owing to rapid interconversion between energetically degenerate, enantiomeric C₁-symmetric conformations.

Symmetry breaking must occur before complexity can develop (1, 2). In cosmology, the perfect symmetry of the singularity at the inception of time (3) evolved under the direction of physical laws to produce the present-day universe filled anisotropically with asymmetrical pieces of matter. In biology, a zygote must break its symmetry before the functional specialization involved with cell differentiation and tissue architecture development can occur (1, 4). Studies on model systems for biological symmetry breaking reveal that the complexity on larger scales is often underpinned by asymmetry on smaller scales (4, 5), which is a consequence of dynamic interactions at the molecular level (4). Investigations of symmetry breaking during complex molecular self-assembly phenomena thus present an opportunity to shed light upon the foundations of the evolution of matter toward complexity.

To contribute to the understanding of symmetry breaking, in the present study we demonstrate a rational method of systematic symmetry breaking within a series of M₄L₆ metal–organic tetrahedra through the control of linker geometry.

Metal–organic polyhedra (6–13) have attracted significant attention due to their host–guest behavior that can be applied in molecular storage (14–17), separation (18), and catalysis (19–22). Platonic or Archimedean metal–organic polyhedra can be constructed by a careful choice of ligands and metal ions (8, 23–27), where the geometries of the individual building blocks define the symmetry axes of the polyhedron. Although high-symmetry architectures represent a great achievement in terms of logical molecular design, lower-symmetry ones are possibly of still greater interest. Besides the fundamental interest in symmetry breaking mentioned above (1), an asymmetric capsule could achieve recognition of asymmetric substrates (28) and possibly catalyze asymmetric transformations (29). So far all reported metal–organic capsules have high symmetry (27, 30, 31), although in some cases they are chiral (29, 32–34), and specific guest molecules are observed to stabilize lower-symmetry structures (32, 35).

M₄L₆ capsules are constructed from octahedrally coordinated metal ions and linear bis-bidentate ligands. Because each vertex is defined by a metal stereocenter (Δ or Λ), an M₄L₆ structure can have T (ΔΔΔΔ/ΛΛΛΛ, homochiral), C₃ (ΔΔΔΛ/ΛΛΛΔ, heterochiral), or S₄ (ΔΔΛΛ, achiral) symmetry (36), with the homochiral cage being most commonly observed (33, 37–43). When the T-symmetric configuration is adopted, all metal-to-metal distances are the same, avoiding strain potentially incurred by differences in these distances present in the C₃- or S₄-symmetric configurations (44). The geometries and steric properties of the ligands can nonetheless overrule the preference for T symmetry. For example, in cages constructed from pyridyl-imine ligands via subcomponent self-assembly (45), S₄ symmetry, rather than T, was found to be the lowest-energy stereochemical configuration when the two terminal phenylene rings within the ligand spacer are forced to be coplanar (44), a configuration that is optimal for the four syn ligands that an S₄-symmetric cage requires to connect metal centers with opposite stereochemistry.

Building upon the S₄-symmetric framework, we hypothesized that the S₄ symmetry axis could be broken if an offset was introduced between the two coordination sites within the same ligand. We reasoned that bis(pyridylimine) ligands derived from the planar, rigid diamines A, B, or C (Fig. 1) could be used to introduce such an offset.

Fig. 1.

Synthesis of M₄L₆ (where L₆ = 2L_anti + 4L_syn) structures that contain two Δ (shown in purple) and two Λ (green) metal vertices along with two anti and four syn ligands. The anti ligands are colored blue and the syn ligands red.

Results and Discussion

Synthesis and Characterization of Asymmetric Fe^II₄L₆ Cages.

As illustrated in Fig. 1, the reactions between diamines A, B, or C (3 eq), 2-formylpyridine (6 eq), and either Fe(ClO₄)₂ or Fe(SO₃CF₃)₂ (⁻OTf, 2 eq) in acetonitrile provided a new set of M₄L₆ cages as the uniquely observed products in solution, as verified by ¹H NMR and electrospray ionization (ESI)-MS and detailed below.

NMR spectra (Fig. 2 and SI Appendix, Figs. S4, S5, S7, S8, S13, S14, S17, and S18) for the corresponding products 1a, 1b, and 1c (Fig. 2) contained more peaks than could be attributed to any single diastereomer of T, C₃, or S₄ symmetry, or a mixture of all three of them. Diffusion-ordered spectroscopy (DOSY) ¹H NMR spectra were consistent in each case with the presence of only a single species. Although there was significant peak overlap in both the ¹H and ¹³C NMR spectra, it was possible to identify 12 imine carbon signals in the ¹³C NMR and heteronuclear multiple quantum correlation spectra (HMQC), indicating the presence of 12 ligand environments with equal populations. These observations suggest that all of the ligand hydrogen atoms are in inequivalent environments, and the cage is therefore asymmetric. A change in counter anion or in solvent (e.g., using nitromethane instead of acetonitrile) did not affect the symmetry and only resulted in slight shifts for some signals in the NMR spectra (e.g., SI Appendix, Figs. S4 and S7).

Fig. 2.

The ¹H NMR spectra for Fe^II₄L₆ cages 1a·OTf, 1b·OTf, 1c·OTf, with imine protons in red.

Vapor diffusion of diisopropyl ether into a nitromethane solution of cage 1a·ClO₄ allowed the isolation of single crystals suitable for X-ray analysis. The crystal structure (Fig. 3) confirmed the presence of the M₄L₆ cage framework. The unit cell contains two cages that are enantiomers of each other. The cage is crystallographically asymmetric yet has many features in common with an S₄-symmetric framework: Two Fe^II centers are of the same handedness whereas the other two are of the opposite handedness; each pair of Fe^II centers of the same stereochemistry is connected by ligands adopting anti conformations, whereas the other four ligands are syn.

Fig. 3.

(Upper) Crystal structure of Fe^II cage 1a·ClO₄ (counter anions, solvent molecules and hydrogen atoms are omitted for clarity; only one conformation of the disordered naphthyl group is shown). (Lower) A schematic illustration of how the S₄ axis of symmetry is broken. The Δ and Λ metal centers are colored purple and green, respectively. The anti ligands are colored blue and the syn ligands red. The parts of the naphthyl spacer that point out of and into the plane of the page are colored orange and maroon, respectively.

The offset geometry of the naphthyl spacers causes them to adopt an arrangement whereby within each ligand one ring is oriented inward, roughly toward the center of the cage and one ring points outward. As shown in Fig. 3, an S₄ operation (consisting of a 90° rotation about the axis that bisects both anti ligands of 1a, followed by a reflection through a mirror plane perpendicular to this axis) results in a configuration in which all syn ligands are indistinguishable from their arrangement before the operation, but where the anti ligands have adopted a different arrangement. These two configurations are enantiomeric mirror images of each other. If the naphthyl groups of the anti ligands were to adopt a conformation twisted by 90° in either direction along their N-N axes (i.e., where the ligand becomes planar), the S₄ operation would result in a configuration indistinguishable from the initial one, and the molecule would possess S₄ symmetry. Models suggest (SI Appendix, Fig. S37), however, that such an S₄-symmetric configuration would result in steric clashes between naphthyl rings.

The asymmetric configurations assumed by the ligands thus allow the naphthyl rings to avoid steric clashes with each other near the metal centers, reducing strain. The ligands pack tightly together (SI Appendix, Fig. S33), which eliminates any cavity space and provides additional stabilizing CH–π interactions between neighboring naphthalene rings.

Fe–Fe distances in 1a are in the range of 10.368(4)–10.724(4) Å; the two Fe^II centers linked by an anti ligand are 0.3 Å farther away from each other than those connected by a syn ligand. The Fe^II–N bond lengths are in the range of 1.96–2.01 Å, consistent with the low-spin configuration observed by NMR.

Either of the enantiotopic conformations of the anti ligands of 1a (Fig. 3) could be adopted with minimal energetic perturbation to the overall cage framework, as highlighted by disorder observed in the X-ray structures of both the perchlorate and the trifluoromethanesulfonate salts of 1a. Whereas the syn ligands in this structure (and all others reported herein) were observed to be well-ordered, the naphthyl groups of one (in 1a·ClO₄) or both (in 1a·OTf; SI Appendix, Fig. S34) of the anti ligands were observed to be disordered across two positions.

X-ray–quality crystals of anthracene-edged cage 1b·OTf were obtained by vapor diffusion of diethyl ether into an acetonitrile solution of the cage. Cage 1b·OTf is also asymmetric in the solid state with no disorder observed (Fig. 4). The Fe–Fe distances are 12.131(3)–12.894(4) Å, a range of variation 0.3 Å wider than the naphthalene-edged cage 1a·ClO₄. The longer edges also engender a discrete void space within cage 1b, whereas 1a showed no internal void. The VOIDOO program revealed the shape of this 16.8 ų cavity (Fig. 4), showing how the C₁ symmetry of host 1b is imposed on the shape of its cavity. The small cavity size is attributed to the packing of the anthracene spacers to minimize steric clashes, as with 1a.

Fig. 4.

The X-ray crystal structure of cage 1b, superimposed upon a plot of the asymmetric cavity (16.8 ų) found with VOIDOO (46).

The ¹H NMR spectra of dissolved crystals of 1a (i.e., both 1a·ClO₄ and 1a·OTf) and 1b were indistinguishable from that of the corresponding cage freshly prepared from subcomponents in acetonitrile, indicating the rotation of the anti ligands about their N-N axes is slow on the NMR time scale, and therefore C₁ symmetry is the lowest energy configuration for the system in solution. For all of the three Fe^II cages 1a, 1b, and 1c, when the temperature was increased to 343 K (SI Appendix, Figs. S12, S16, and S20), neither decomposition nor coalescence was observed in the ¹H NMR spectrum, indicating relatively high thermal stability of the complexes and limited rotational freedom of the ligands.

Co^II₄L₆ and Cd^II₄L₆ Cages.

Co(BF₄)₂ (4 eq) was also observed to form M₄L₆ structures with both diamines A and B (6 eq) and 2-formylpyridine (12 eq); when C was used an insoluble product was obtained. These products, 2a and 2b, were characterized in solution by paramagnetic ¹H NMR spectroscopy and ESI-MS. In similar fashion to the Fe^II cage 1a, Co^II cage 2a also exhibited 12 imine signals in the ¹H NMR spectrum (SI Appendix, Fig. S21), indicating that cage 2a is also asymmetrical.

The crystal structure obtained for 2a·BF₄ (SI Appendix, Fig. S35) shows the same overall configuration as the iron analog 1a. The coordination bond lengths are in the range of 2.117(6)–2.197(6) Å, on average 0.15 Å longer than that for Fe–N in 1a·ClO₄. The average bond length for Co–N_imine is around 0.03 Å longer than that for Co–N_py, and no significant Jahn–Teller distortion was observed, consistent with the presence of Co^II ions in the high-spin state. There is a wider range of Co–Co distances in 2a·BF₄ [10.339(3)–10.947(5) Å] compared with the Fe–Fe distances in 1a and correspondingly greater distortions from idealized octahedral geometry in the Co^II cage.

Intriguingly, the analogous tetrahedron 2b·BF₄ has a much simpler ¹H NMR spectral profile (SI Appendix, Fig. S25). Only three imine signals were observed, consistent with S₄ symmetry on the NMR time scale. We infer that the anthracenyl groups of the anti ligands of 2b were undergoing rapid rotation, as opposed to adopting an S₄-symmetric lowest-energy state, based upon our observations of their analogs and examination of models, which suggested that for Co^II, as for Cd^II and Fe^II, the S₄-symmetric configuration would experience steric clashes (SI Appendix, Fig. S37).

The ¹H NMR peaks for 2b·BF₄ were observed to sharpen as the temperature increased, indicating fast thermal motion within the complex; these peaks broadened into the baseline below 15 °C, consistent with a slowing of the anti ligands’ rotation about their N-N axes below this temperature (SI Appendix, Fig. S26). No coalescence of the ¹H NMR signals of 2a·BF₄ was observed upon increasing the temperature to 70 °C (SI Appendix, Fig. S23), indicating that the anti ligands do not rotate about their N-N axes at a rate comparable to the NMR time scale at this temperature. We attribute the faster ligand rotation in 2b than in 2a to the looser geometrical requirements imposed by the longer Co–N bonds, which allow the individual ligands to rotate past each other with a lower energetic penalty and to the greater lability of these bonds, which may loosen along one N–Co–N axis by undergoing Jahn–Teller distortion during ligand rotation.

Cd^II was not observed to template the formation of a single product with B or C under the conditions that gave clean cage formation with Fe^II; instead, mixtures of soluble and insoluble products were obtained. Cage 3a was generated cleanly, however, when diamine A was used, as confirmed by its ¹H NMR spectrum (SI Appendix, Fig. S27). NMR spectra of 3a were consistent with rapid rotation of the anti ligands even at −35 °C in solution (SI Appendix, Fig. S32), in contrast to its Fe^II or Co^II analogs. As with 2b, we infer 3a to undergo interconversion between enantiomeric C₁-symmetric configurations rapidly on the NMR time scale, such as to exhibit apparent S₄ symmetry, as opposed to adopting an S₄-symmetric lowest-energy state. We hypothesize that steric clashes between ligands during these interconversions would incur a lower energetic penalty owing to the larger ionic radius of Cd^II, resulting in the more rapid rate of rotation observed in the case of this metal ion. We infer that the mechanism of ligand rotation may involve distortion of the coordination sphere of Cd^II, but not breaking of the Cd–N_imine linkage on the NMR time scale, because satellite peaks were associated with the imine ¹H NMR resonances of 3a, attributable to J-coupling with the two spin-1/2 isotopes of cadmium (SI Appendix, Fig. S27).

Because Cd^II has a d¹⁰ electron configuration, cage 3a does not benefit from crystal field stabilization, and its metal–ligand bonds are thus weaker than in complex 1a. When Fe(ClO₄)₂ (4 eq per cage) was added to an acetonitrile solution of cage 3a·ClO₄, the orange solution turned dark purple immediately, suggesting rapid formation of a Fe^II complex. Cage 1a·ClO₄ was the only observed complex in both the ¹H NMR spectrum (SI Appendix, Fig. S36) and ESI-MS of the reaction mixture, with no evidence for the presence of 3a or any mixed-metal species, consistent with a clean metal displacement where Fe^II substituted all Cd^II to transform cage 3a·ClO₄ into cage 1a·ClO₄ (Fig. 5), halting the anti ligands’ rapid rotation. Although transmetallation has been demonstrated in macrocyclic complexes (47) and helicates (48), to our knowledge this is a unique example of the conversion of one metal–organic polyhedron into another via metal exchange.

Fig. 5.

Transformation of 1c·ClO₄ into 1a·ClO₄ via metal exchange.

Conclusions

We have demonstrated that by introducing an offset between the two metal binding sites within a linear bis-bidentate ligand an asymmetric M₄L₆ cage structure could be generated. This symmetry breaking was achieved as a consequence of three features of the system. First, this class of cages (44) is known, when both ends of the diamine spacer are forced to be coplanar, to adopt a configuration where two metal centers have the opposite handedness to the other two, resulting ordinarily in an S₄-symmetric structure. Second, the diamine spacer’s offset geometry breaks the erstwhile S₄ symmetry axis, due to steric interactions between the naphthyl groups. Third, racemization of these asymmetric cages, via rotation of the spacer groups of the anti ligands, can be slowed through the use of Fe^II as the metal ion. The shorter Fe^II–N bonds led to higher energetic barriers to ligand torsion, locking the cage’s stereochemistry.

The phenomenon of asymmetric cage formation was observed to be general across three metal ions and three offset spacer groups; other metal ions and spacers also thus show promise because the steric effects giving rise to symmetry breaking operate next to the metal centers (SI Appendix, Fig. S37), independently of ligand length. Although the cavity of the largest cage structurally characterized herein, 1b (Fig. 4), is too small for even the smallest chiral guest, the methods outlined here demonstrate how larger asymmetrical cages, imprinted with stereochemical information (29, 32–34), might selectively encapsulate and asymmetrically transform low-symmetry, information-rich guest species.

Materials and Methods

Syntheses and characterization (i.e., ¹H NMR, ¹³C NMR, mass spectra, elemental analysis, and crystallography) of all new compounds are described in the text of SI Appendix. Details of experimental instrumentation are summarized in the text of SI Appendix. Further discussion on the mechanism of symmetry breaking is available in SI Appendix.