Guest‐Dependent Isomer Convergence of a Permanently Fluxional Coordination Cage

Abstract

A fluxional bis‐monodentate ligand, based on the archetypal shape‐shifting molecule bullvalene, self‐assembles with M²⁺ (M=Pd²⁺ or Pt²⁺) to produce a highly complex ensemble of permanently fluxional coordination cages. Metal‐mediated self‐assembly selects for an M₂L₄ architecture while maintaining shape‐shifting ligand complexity. A second level of simplification is achieved with guest‐exchange; the binding of halides within the M₂L₄ cage mixture results in a convergence to a cage species with all four ligands present as the “B isomer”. Within this confine, the reaction graph of the bullvalene is greatly restricted, but gives rise to a mixture of 38 possible diastereoisomers in rapid exchange. X‐ray crystallography reveals a preference for an achiral form consisting of both ligand enantiomers. Through a combination of NMR spectroscopy and DFT calculations, we elucidate the restricted isomerisation pathway of the permanently fluxional M₂L₄ assembly.

The first metallo‐supramolecular M₂L₄ cages bearing shape‐shifting bullvalene ligands are prepared. The enormous potential complexity is limited by geometric constraints in solution, which converge to a single isomer in the solid state. Through detailed experimental and computational analysis, we map the fluxional nature of these systems.

Through unending sequences of sigmatropic rearrangements, bullvalene exists as an ensemble of >1.2 million degenerate isomers and has no permanent carbon–carbon bonds. ^[1] The introduction of substituents gives rise to ensembles of non‐degenerate isomers, whereby substituents may adopt all possible structural relationships. While the chemistry of this remarkable hydrocarbon dates back to the 1960s, ^[2] its development was slowed by arduous synthetic access. A recent revival of interest has been led by Bode, ^[3] Echavarren, ^[4] and one of our laboratories. ^[5] In a recent report, we developed easy synthetic access to a range of diaryl substituted bullvalenes. ^[5c]

The coordination chemistry of bullvalene as a shapeshifting ligand is completely unexplored, save for a few early reports of metal–diene complexation to the parent hydrocarbon. ^[6] Bidentate substituted bullvalenes as ligands offer the potential to impart dynamic host‐guest binding. Pioneering studies by Bode have demonstrated both fullerene and polyol host‐guest binding to multi‐functionalised bullvalenes in highly complex, though structurally ill‐defined, settings.[ ^3b , ^3c , ^3g ]

In conceptualising bis‐pyridyl bullvalene as a bis‐monodentate ligand, a range of N−N bite angles are possible through interconversion of its major isomers (Figure 1). Given this geometric variability, metal coordination by fluxional ligands has the potential to produce highly complex and dynamic mixtures. Considering this, we prepared a 3‐substituted bis‐pyridyl bullvalene to examine its coordination chemistry in the structurally well‐defined setting of M₂L₄ coordination cages (Figure 1).

Figure 1.

Metal‐mediated self‐assembly of bis‐3‐pyridyl bullvalene forms a complex mixture of fluxional M₂L₄ cage isomers. Guest exchange restricts the fluxional behaviour of the assembly resulting in a significant convergence to an “all‐B” isomer form.

M₂L₄ coordination cages are accessible metallo‐supramolecular hosts, typically assembled from concave‐shaped ditopic ligands and square planar metal centres. ^[7] Their simple and symmetrical structure has served as a platform to introduce various stimuli responsive properties ^[8] and control self‐sorting phenomena through ligand design. ^[9] Most of these advances have been achieved with symmetrical and static ligands that produce only single cage isomers upon complexation with metal ions. ^[9b] On the other hand, utilising unsymmetrical ligands ^[10] or ligands with rotational isomerism ^[11] results in isomeric cage mixtures that require steric or geometric constraints to achieve defined structures. In this context, coordination cages composed of permanently fluxional, shape‐shifting ligands are hitherto unexplored.

Herein we present the self‐assembly and guest‐induced isomer amplification of an M₂L₄ coordination cage assembled from bis‐3‐pyridyl bullvalene. Whilst a highly complex isomeric mixture is obtained in the presence of BF₄ ⁻ counterions, the presence of a suitable guest results in a significant convergence to a defined M₂L₄ cage isomer that remains permanently fluxional through a restricted pathway. Through a combination of NMR spectroscopy, X‐ray crystallography, and DFT calculations, we map out the fluxional pathway of the bullvalene derived ligand within the confines of the M₂L₄ assembly.

Our study commenced with the synthesis of bis‐3‐pyridyl bullvalene via a Suzuki coupling between bis‐(Bpin) bullvalene and 3‐bromopyridine in excellent yield (Figure 2a). Variable temperature (VT) ¹H NMR spectroscopy in CD₃CN revealed that the ligand exists in a dynamic equilibrium between three major isomers: A, B, and C in a 48 : 38 : 13 ratio respectively, in rapid exchange at room temperature (Figure S1).

Figure 2.

a) Synthesis of bis‐3‐pyridyl bullvalene, Pd₂L₄ cage assembly, guest exchange, and NMR analysis. i) Pd(PPh₃)₄ 5 mol %, NaOH, THF/H₂O, 65 °C, 92 %; ii) [Pd(CH₃CN)₄](BF₄)₂, 0.5 equiv, CD₃CN, 60 °C, 10 min. ¹H NMR spectra (−35 °C, CD₃CN) of b) 1⋅BF₄ (ESI‐HRMS shown in inset); c) 1⋅Cl; d) 1⋅I.

Treatment of the ligand with 0.5 equivalents of [Pd(CH₃CN)₄](BF₄)₂ initially led to an intractable complex mixture denoted 1⋅BF₄ (Figure 2b). The broad signals observed in the room temperature ¹H NMR spectrum resolved somewhat at −35 °C, and the NCH pyridyl resonances were observed downfield relative to the free ligand indicating coordination to Pd²⁺ (Figure 2b, Figure S1). The anticipated Pd₂L₄ structure was confirmed by ESI‐HRMS, with evident peaks of the cage at 479, 762, and 1611 m/z corresponding to [Pd₂L₄+n BF4]⁴⁻ⁿ (n=1–3) (Figure 2b, Figure S8). Consistent with the complex mixture, the ¹⁹F NMR spectrum of 1⋅BF₄ revealed a multitude of resonances corresponding to various isomers encapsulating BF₄ ⁻ (Figure S7).

The addition of one equivalent of chloride to 1⋅BF₄ resulted in complete exchange of the encapsulated anion, producing a significantly resolved ¹H NMR spectrum containing one dominant cage species, 1⋅Cl (Figure 2c). Anion exchange of 1⋅BF₄ or 1⋅Cl with an equivalent of iodide produced the corresponding Pd₂L₄ cage 1⋅I (Figure 2d). ^{[12] 1}H NMR analysis of 1⋅Cl and 1⋅I at −35 °C revealed the dominance of isomer B of the ligand in solution, indicating a complete change in the isomer distribution of the bullvalene relative to the parent complex 1⋅BF₄ . However, the dynamic nature of the bullvalene ligand core persists as shown by characteristically broad signals in the room temperature ¹H NMR spectra (Figure S19). ESI‐HRMS was indicative of a Pd₂L₄ structure in each case, with prominent [Pd₂L₄+X+n BF4]³⁻ⁿ (X=Cl⁻ or I⁻, n=0–1) peaks supporting the encapsulation of the respective anion within the central cavity of the cage.

Several other Pd₂L₄ cages were prepared via anion exchange with 1⋅BF₄ , in order to compare binding preferences of the system. Larger counterions such as PF₆ ⁻ do not undergo anion exchange with the M₂L₄ assembly, leading to complex spectra analogous to that in Figure 2b (see Supporting Information for full details). Anion exchange with NO₃ ⁻ on the other hand, produced a ¹H NMR spectrum with the isomer B ligand predominating amongst other unresolvable species (Figure S9). Across the anionic guests that were investigated, only halides resolved the fluxional mixture to a dominant isomer B cage species, with preferential binding observed for larger halides (e.g. I⁻ over Cl⁻). Analysis of the HRMS spectra of these samples confirmed the Pd₂L₄ structure and the absence of higher nuclearity M _n L_2n species (see Supporting Information for full details) that could be afforded by other isomers of bis‐3‐pyridyl bullvalene.

We also prepared the Pt₂L₄ analogues of the BF₄ ⁻, Cl⁻, and I⁻ encapsulated cages (denoted as 1′⋅BF₄ , 1′⋅Cl, and 1′⋅I); heating at 85 °C was required to induce anion exchange for the Pt₂L₄ cages indicating a dissociative exchange mechanism. Each of these cages displays near identical characteristics to their corresponding palladium cage by VT NMR spectroscopy. Together, these results indicate that the fluxional behaviour of the ligand within the M₂L₄ structure occurs without metal‐ligand dissociation as a requirement. ¹H EXSY NMR measurements on 1⋅I revealed the exchange of particular bullvalene proton signals, indicative of isomer B to B′ isomerism with a rate constant of ≈0.7 s⁻¹ at −20 °C (see Supporting Information for full details). This establishes the mode of isomerism within the cage, and this rate constant is broadly consistent with the calculated barrier of ligand isomerisation (see below).

The conceivable complexity of this M₂L₄ system is built upon the reaction graph of the bullvalene ligand. For disubstituted bullvalene networks of this type, there are 15 possible isomers including three enantiomer pairs, (Figure 3b) where nodes represent isomers and edges Cope rearrangement pathways. ^[5c] For bis‐3‐pyridyl bullvalene there are three populated isomers, A, B, and C, that rapidly interconvert through a lower energy circuit including intermediate isomer D. Isomer C also possesses a degenerate rearrangement pathway with itself (Figure 3b). DFT calculations provide a predicted energy dimension to this reaction graph (colour coded in Figure 3b, see Supporting Information for full details).

Figure 3.

a) Isomer distribution of the ligand. b) Ligand reaction graph. c) Stereoisomerism of isomer B. d) Network diagram of the all‐B M₂L₄ cage. e) X‐ray structures of “1243” bullvalene cage isomer i) A perspective view of 1 highlighting the symmetry elements and directionality of the cyclopropane ring in each bullvalene ligand enantiomer ; ii) 1⋅BF₄ ; iii) 1′⋅Cl; iv) 1⋅I. f) DFT analysis of Pd₂L₄ cages.

In combining four bullvalene ligands into an M₂L₄ cage, the potential level of complexity is enormous. If all possible bullvalene isomers could combine in all possible permutations, there would be approximately two hundred thousand possible isomeric cages. ^[13] However, all experimental evidence indicates isomer B as the predominant ligand isomer within the cage. This isomer is axially chiral, and its Plus/Minus (P/M) descriptors are defined by considering an axis that runs from the centre point of the cyclopropane through the apical bridgehead (or apex) of bullvalene. ^[5d] Within the cage, isomer B also possesses a vertical directionality relative to an arbitrary top‐view of the cage, which we define as the up/down tilt of the apex. Thus, each coordinated ligand possesses two independent stereochemical degrees of freedom that may dynamically interconvert. This gives four possible ligands that make up the “all‐B” cage which we label 1 (P↑), 2 (P↓), 3 (M↑), and 4 (M↓) (Figure 3c). A reaction graph of all possible all‐B cages is shown in Figure 3d. From this we have an isomer list with 38 possible all‐B cages, including 15 enantiomer pairs (red) and 8 achiral cages (blue). Each all‐B isomer cage may interconvert with every other, as represented by the graph edges.

The isomerism, directionality, and chirality of the bullvalene ligand within the M₂L₄ cage was determined by single‐crystal X‐ray analysis. Single crystals of 1⋅BF₄ , 1⋅NO₃ , 1⋅I, and 1′⋅Cl were obtained by slow vapour diffusion of diisopropyl ether into CD₃CN solutions of the respective samples. All four samples crystallise in the monoclinic space group C2/m, and the contents of the asymmetric unit for each complex comprises of two crystallographically unique B ligand isomers (M and P enantiomers) coordinating to a metal centre of a separate M₂L₄ cage. ^[14] The remaining ligands from each identical but crystallographically independent cage molecule are generated by a horizontal mirror plane and C₂ axis of symmetry. The crystallised M₂L₄ isomer for each sample is composed of an all‐B cis‐configuration of P and M bullvalene enantiomers, giving a “1243” or P↑‐P↓‐M↓‐M↑ configuration (Figure 3e). Across the three Pd₂L₄ structures, the average Pd⋅⋅⋅Pd distance is 8.49±0.07 Å, whilst the Pt₂L₄ assembly has a slightly shorter metal separation of 8.40 Å. In all structures the respective anion was located inside the central cavity of the cage.

While the exclusive preference for the 1243 B isomer (across a multitude of counterions) in the crystal lattice cannot be explained energetically, it is rational. This isomer represents the maximum degree of entropy, with respect to its stereochemical degrees of freedom, as well as maximum symmetry, bearing both an internal mirror plane and a C₂ axis. ^[15]

To understand the cage isomer convergence in the presence of halide guests, we undertook VOIDOO calculations on the X‐ray structures of 1. Based on a 1.4 Å probe, a cavity volume of 69.7 ų was obtained for 1. For BF₄ ⁻, NO₃ ⁻, I⁻, and Cl⁻ as the counterions, the packing coefficients are 76 %, 59 %, 50 %, and 34 % respectively (Supporting Information, Table S5). As the decreasing packing coefficient correlates well to the experimental convergence of M₂L₄ isomers in solution, we postulate that optimal host–guest interactions, particularly for smaller spherical guests, drive the cage isomer convergence.[ ^11a , ¹⁶ ]

Together, these results show that the guest‐driven convergence of the M₂L₄ system significantly simplifies the reaction graph of bullvalene, constraining it to a dominant B‐isomer form. To further rationalise the structural preferences and dynamic behaviour of the Pd₂L₄ cages, we turned to computational modelling. Geometry optimisations were performed for a series of cage isomers at the B3LYP/Def2‐SVP level on the respective tetracations in the gas phase. While this computational methodology doesn't carry the expectation of reliable chemical accuracy, the qualitative relative energies and geometry constraints are indicative.

Six structures were considered as exemplars of the AAAA, BBBA, BBBB, BBBC, BBBD, and CCCC cage isomers (Figure 3f). The BBBB “1243” isomer was found to be the lowest energy of the structures considered with a Pd−Pd distance of 8.46 Å (in excellent agreement with the crystal structure's average Pd−Pd distance of 8.49 Å). Isomerisation of one of the B ligands to its corresponding C ligand gives the BBBC cage which is only 3 kJ mol⁻¹ elevated in energy and nearly identical in its geometric parameters (Pd−Pd distance 8.3 Å). Further isomerisation to the CCCC cage gives a structure with similar overall geometry and somewhat elevated energy. Conversion of the ligand from the B isomer to the A isomer is more problematic. Firstly, the B isomer must transit isomer D, whereby the BBBD cage incurs a small contraction of the Pd−Pd distance to 8.0 Å and a moderate energy increase. From there, the D isomer must transit to isomer A giving the intermediate BBBA cage.

While the Pd−Pd distance of this cage is the same as that of BBBD, the energy is significantly elevated and there is a pronounced distortion of the regular M₂L₄ geometry whereby the planes of the square planar coordination around palladium are tilted by 12° with respect to one another (see Supporting Information for full details). Further isomerisation to the AAAA cage gives a structure that is elevated in energy by 46 kJ mol⁻¹ and has a short Pd−Pd distance of 5.7 Å (Figure 3f). Such a cage would have an internal cavity too small to encapsulate common anions (27.4 ų as indicated by VOIDOO calculations), and would have a significantly shorter metal–metal distance than any M₂L₄ cage reported in the literature (the shortest appears to be 6.5 Å reported by Puddephatt). ^[17]

From this we may rationalise the preponderance of ligand isomer B in M₂L₄ cages, their rapid interconversion through a B‐C‐B′ pathway, and the exclusion of isomer A from the reaction graph of the cage. Within the cage, each ligand may be considered as geometrically “pinned”, with its pyridyl nitrogen–nitrogen distances fixed in place. This excludes transition to the more contracted A isomer, and that region of the reaction graph is shut down. However, rapid B‐B′ isomerism is still possible via isomer C, leading to dynamic diastereoisomer ensembles of “all‐B” M₂L₄ cages. While Bode ^[3] has previously demonstrated abstract shifts in bullvalene isomer distributions through host–guest chemistry, this is the first time whereby imposed changes on the reaction graph of a bullvalene have been structurally mapped.

In summary, we have reported and structurally characterised the first permanently fluxional coordination complex of bullvalene. Self‐assembly of bis‐3‐pyridyl bullvalene with Pd²⁺ or Pt²⁺ metal ions in the presence of BF₄ ⁻ counterions produced a complex M₂L₄ coordination cage mixture. Anion exchange with smaller anionic guests (e.g. Cl⁻ and I⁻) led to a dramatic convergence to a well‐defined all‐B isomer cage complex; in solution this isomeric form remains fluxional, even within the kinetically inert Pt₂L₄ cage scaffold. X‐ray crystallography revealed that the all‐B cage is selective for the “1243” isomer. EXSY data and computational methods revealed that the all‐B isomer host‐guest complex undergoes a restricted B‐C‐B′ isomerisation, thus removing a huge fraction of complexity by shutting down other potential isomerisation circuits. We hope this work will inform the design of dynamic metallo‐supramolecular systems that can undergo stimuli‐responsive shape and size adaptability.

Conflict of interest

The authors declare no conflict of interest.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

A. P. Birvé, H. D. Patel, J. R. Price, W. M. Bloch, T. Fallon, Angew. Chem. Int. Ed. 2022, 61, e202115468; Angew. Chem. 2022, 134, e202115468.

Data Availability Statement

The data that support the findings of this study are available in the Supporting Information of this article.