Dynamic donor–acceptor [2]catenanes

Abstract

Donor–acceptor [2]catenanes based on cyclobis(paraquat-p-phenylene) as the π-acceptor ring have been used prominently in the construction of functional molecular devices. We report here their thermodynamically controlled synthesis from isolated π-donor and π-acceptor rings under the catalytic influence of tetra butylammonium iodide. The initial nucleophilic attack of iodide ion, which opens up the π-acceptor ring, is followed by complexation to the π-donor ring and the subsequent catenation of the π-donor ring by the π-acceptor ring [2]catenane. The reaction is general in scope and proceeds in high yields, without giving rise to side-products.

Topologically nontrivial molecules dubbed catenanes (1) contain two or more mutually interlocked, inseparable rings, arranged mechanically like the links in a chain. [2]Catenanes with several complementary recognition sites expressed as matching pairs between their two rings can be rendered bistable (2) or multistable (3) and hence endowed with switching properties (4), paving the way for molecular device applications, straddling diversity from unidirectional motors (5, 6) through reconfigurable switches (7, 8) to components of electronic displays (9, 10).

Although the stabilizing noncovalent interactions associated with the matching recognition sites are used routinely in a templating fashion to preorganize precursors in the kinetically controlled preparation of [2]catenanes, they have been used very rarely (11, 12) as the thermodynamic driving force in their template-directed synthesis under equilibrium control. The vast majority of preparative strategies to access catenanes, and organic compounds in general, still relies on irreversible, kinetically controlled reactions. The distribution of competing products in kinetic reactions is controlled by the relative stabilities of the corresponding transition states. Kinetically controlled reactions remain central in synthesis on account of their relatively fast reaction times and high selectivities for the most part. Recently, however, dynamic covalent chemistry (DCC) (13, 14) has begun to emerge as a powerful synthetic complement, if not an alternative, to the kinetic approach. Reactions involving DCC operate under thermodynamic control where covalent bonds are reversibly formed and broken until equilibrium is reached. At equilibrium, the distribution of products is dictated only by their relative thermodynamic stabilities. More stable products, in either a molecular or a supramolecular sense, profit from reversibility in this scenario, because a cascade of error-checking and proof-reading processes ultimately provides the energetically most favored compounds in superior yields. The equilibration products can be “fixed” to give isolable compounds, typically by an irreversible chemical transformation, e.g., reduction in dynamic imine chemistry, acidification in dynamic disulfide chemistry, etc.

Molecular Borromean rings (15) and Solomon knots (16) are among the eminent examples of complex mechanically interlocked molecules constructed expediently as a result of dynamic imine formation chemistry. Sanders and coworkers have used dynamically forming hydrazone imines (17) and disulfides (18) to select and amplify hosts suitable for natural and unnatural guests at the expense of ill-fitting hosts.

Given that catenanes are typically more stable than the sum of their isolated constituent rings, because of the pairwise stabilizing interactions that promote the formation of a mechanically interlocked structure in the first place, they represent an ideal test bed for the development of thermodynamically controlled methods. Indeed, they have been prepared using DCC (13, 14) by us (19) and others (20, 21). Alkene metathesis has been used to open up a crown ether ring reversibly and subsequently close it again around a suitable binding site to afford a [2]catenane (19–21). Although not strictly belonging to the DCC domain, reversible metal coordination has also been used by Sanders and colleagues (20, 22) and Fujita and colleagues (20, 23) to construct [2]catenanes. Despite a number of device-driven applications (7, 8), bistable donor–acceptor [2]catenanes based on the cyclobis(paraquat-p-phenylene) (CBPQT⁴⁺) (24) as the π-acceptor ring (Fig. 1A) have so far only been constructed under kinetic control (25, 26). Herein, we present a general, practical catalytic, and high yielding approach to the template-directed synthesis (27) of donor–acceptor [2]catenanes incorporating the CBPQT⁴⁺ tetracationic cyclophane that utilizes dynamic nucleophilic substitution as the thermodynamically controlled means of its production. We demonstrate that, starting from isolated π-donor and π-acceptor rings, catalytic amounts of tetrabutylammonium iodide (TBAI) can be used to open up the π-acceptor ring selectively and close it around the π-donor ring subsequently, affording [2]catenanes. Reactions which proceed under thermodynamic control to establish an equilibrium that is shifted toward the formation of [2]catenanes can be conducted preparatively to produce them in high yields without the generation of side-products.

Fig. 1.

Structural formula (A) and solid-state structure (B) of CBPQT⁴⁺. Strained nature of the CBPQT⁴⁺ ring is evident from its solid state structure. The six aromatic rings are bent out of plane to relieve this strain, causing the bowing of the cyclophane ring sides. These deformations are the most obvious in the bending of the exocyclic C–CH₂ and N⁺–CH₂ bonds emanating from the p-phenylene and bipyridinium residues: these bonds are at angles of 14° and 23° relative to each other (see ref. 24). The strain in the ring system is coupled with the inherently higher reactivity of benzylic position and with the nucleofugality of the bipyridinium to render the CBPQT⁴⁺ ring reactive toward nucleophilic substitution.

Results and Discussion

A cursory glance at the structure (24) of CBPQT⁴⁺ suggests a facile nucleophilic attack on the benzylic methylene groups that should displace (Fig. 2A) the positively charged bipyridinium group as a good nucleofuge or leaving group, aided and abetted by the considerable relief of ring strain (Fig. 1B) within the tetracationic cyclophane (28). Indeed, the sensitivity of CBPQT⁴⁺ toward nucleophiles has loomed large over previous synthetic endeavors (29) orchestrated under kinetic control. If the attacking nucleophile is also of significant nucleofugality and, provided it is in substoichiometric amounts to reduce the likelihood of manifold attack on CBPQT⁴⁺, the reaction illustrated in Fig. 2A is expected to be reversible. This equilibrium will be shifted toward 1³⁺ if its effective concentration is continuously decreased by complexation. Because electron-rich crown ethers, exemplified in Fig. 2 by bis-p-phenylene[34]crown-10 (2a), form 1:1 complexes, or [2]pseudorotaxanes, with analogs (ref. 30 and references therein) of 1³⁺ but not with CBPQT⁴⁺, they should remove 1³⁺ from the equilibrium, arresting it (Fig. 2B) in the form of [1 ⊂ 2a]³⁺. The reverse nucleophilic attack by the pyridyl nitrogen lone pair can now occur in [1 ⊂ 2a]³⁺, expelling the original nucleophile as a leaving group and generating (Fig. 2C) the [2]catenane. Because the benzylic functionality is restored in the [2]catenane, it can undergo the reverse transformation. The catenation-decatenation process is, therefore, taking place under equilibrium control, albeit shifted toward the [2]catenane because of the inherently higher stability of the mechanically interlocked structure relative to the individual structures of the collective noncatenated rings.

Fig. 2.

Proposed mechanism of [2]catenane formation involving dynamic nucleophilic substitution. (A) Nucleophilic opening of the CBPQT⁴⁺ ring. (B) Threading of π-acceptor 1³⁺ through the crown ether 2a. (C) Nucleophilic closure of the CBPQT⁴⁺ ring around the crown ether to form the [2]catenane 3a⁴⁺. Note that, in keeping with its role as a catalyst, the iodide (I⁻) ion is consumed in the first step (A) and regenerated in the third one (C). Further, note that, to carry out this dynamic nucleophilic substitution on the CBPQT⁴⁺ ring, it has to be rendered soluble in MeCN by the presence of soft counterions, e.g., 4PF₆⁻.

With this hypothesis established, we set out to find a suitable nucleophile to test it. In a model reaction, CBPQT·4PF₆^† was exposed to a number of nucleophilic reagents in acetonitrile (MeCN). The appearance of ring-opened, nonradical derivatives of CBPQT⁴⁺ could be expected if the nucleophilicity of the reagents was sufficient, although the absence of complexing species means that materials other than 1³⁺ could be formed as well. Common-or-garden nucleophilic catalysts, such as N,N-dimethylaminopyridine (DMAP) and Bu₃P proved to be too strong as reducing agents, resulting in one-electron transfer to CBPQT⁴⁺, triggering its subsequent decomposition. Noninvasive 4,4′-bipyridine was inert, even after 7 d at 82°C, and the same was true for KI, presumably because of its low solubility in MeCN. Finally, we found the desired nucleophile in TBAI: exposure of CBPQT·4PF₆ to it (30 mol%, 82°C, MeCN, 6 d) resulted in the opening of the tetracationic cyclophane with the concomitant formation of several nonradical species, as indicated by ¹H NMR spectroscopy. This preliminary result established that CBPQT⁴⁺ does undergo nucleophilic attack and encouraged further experimentation aimed at channeling this reactivity toward catenane formation.

Synthetic Studies.

The crucial experiment (Fig. 3) was performed by combining equimolar amounts of CBPQT·4PF₆ and 2a with 30 mol% TBAI in acetonitrile-d₃ (CD₃CN) to allow monitoring (Fig. 4) by ¹H NMR spectroscopy. In the initial reaction mixture, only CBPQT·4PF₆ and 2a can be identified spectroscopically: see the ¹H NMR spectra shown in Fig. 4 A–C. On heating (82°C), another set of signals becomes apparent (Fig. 4D) and starts growing in size relative to those of CBPQT ·4PF₆ and 2a. The new peaks can be assigned (Fig. 4F) unambiguously (25, 26) to the [2]catenane 3a·4PF₆. The reaction reaches equilibrium after 6 d at ≈80°C when the final molar ratio of the [2]catenane to the individual rings is ≈70:30. Because the reaction only proceeds at elevated temperatures (≈80°C), equilibration is essentially shut down upon cooling, providing a rather convenient way of terminating a thermodynamically controlled reaction, because no chemical transformation is required. After workup, pure 3a·4PF₆ was isolated in 46% yield from the crude reaction mixture. The reaction is catalytic in TBAI, proceeding at 30 mol% loading; albeit much slower, catenation also occurs with TBAI loadings as low as 0.5 mol%.

Fig. 3.

Preparation of different [2]catenanes by dynamic nucleophilic substitution. Dynamic nucleophilic substitution can be used to prepare compounds 3a–c·4PF₆ in high yields and without side-products. All of the prepared catenanes are two-station systems, because their π-donor rings have two sites that can bind to the CBPQT⁴⁺ (or 1³⁺). While 3a·4PF₆ and 3b·4PF₆ are degenerate in this respect, [2]catenane 3c·4PF₆ is not. Its tetrathiafulvalene-based binding site has an inherently higher affinity for CBPQT⁴⁺ than the naphthalene-based one. However, oxidation of the tetrathiafulvalene converts it into its dication, which strongly electrostatically repels the CBPQT⁴⁺ ring and causes it to move to the secondary, naphthalene-based site. This behavior constitutes the base of bistability and an entry into a number of molecular device applications (see refs. 7, 9, and 10).

Fig. 4.

Synthesis of 3a·4PF₆ monitored by ¹H NMR spectroscopy. Partial ¹H NMR (500 MHz) spectra (recorded in CD₃CN at 298 K) of the pure 2a (A), the pure CBPQT·4PF₆ (B), a reaction mixture at t = 0 d (C), after heating for t = 1 d (D) and t = 7 d (E) at 82°C in the presence of TBAI (30 mol%), and of pure [2]catenane 3a·4PF₆ (F) are shown. Light red, 2a; light blue, CBPQT·4PF₆; gray, 3a·4PF₆.

Performing the reaction with the excess of the crown ether pushed the conversion to completion (≈100%) by ¹H NMR spectroscopy and led to the isolation of 3a·4PF₆ in 66% yield. Although either the donor or the acceptor ring can be used in excess, we found it more practical to use a surfeit of the neutral crown ether on account of its easier separation from the tetracationic [2]catenane 3a·4PF₆. The isolated yield is low in light of the complete conversion observed by ¹H NMR spectroscopy. Tentatively, this shortfall can be rationalized in terms of the rather weak binding between 2a and 1³⁺, which gives rise to an appreciable concentration of free 1³⁺ in solution, enabling it to undergo side reactions, most likely polymerization to give insoluble materials (judged by the white deposit on the walls of the NMR tube).

The exclusive formation of trication 1³⁺ by the nucleophilic opening of CBPQT⁴⁺ is crucial for the success of reaction. Treatment of 1,4-bis(bromomethyl) benzene and the dicationic 1,1′-[1,4-phenylene bis(methylene)]-bis-4,4′-bipyridinium hexafluorophosphate, the two precursors used in the kinetic “clipping” approach to 3a·4PF₆ (25, 26), with TBAI at 80°C resulted in formation of intractable mixtures of products. Presumably, 1³⁺ forms at elevated temperatures but then reacts with the remaining starting materials in a random, rather than templated fashion. When it is present alone in the reaction medium, having been generated from CBPQT⁴⁺ in the presence of TBAI, trication 1³⁺ forms a [2]pseudorotaxane with 2a and then undergoes intramolecular ring closure to form the [2]catenane.

In the context of reaction efficiency and conversion, the use of 1,5-dinaphtho[38]crown-10 (2b) as the donor ring proved advantageous. On account of the higher stabilities of dioxynaphthalene–CBPQT⁴⁺ counterparts, the equilibrium now lies virtually completely on the [2]catenane side (>95:<5 by ¹H NMR, 30 mol% TBAI, 82°C, MeCN, 6 d). These favorable characteristics render the reaction, workup, and isolation (93% isolated yield) of 3b·4PF₆ remarkably facile. The two reactants were combined with TBAI in MeCN, yielding a brownish solution. After several days of heating, the solution became deep purple, indicative of dioxynaphthalene–CBPQT⁴⁺ charge transfer absorption band. Filtration of the crude mixture through a plug of silica and counterion exchange provided (31) pure 3b·4PF₆ as the precipitate.

Finally, and perhaps most significantly from the viewpoint of molecular device fabrication (7–10), the bistable switchable [2]catenane 3c·4PF₆ can be obtained more efficiently by using this “magic ring” protocol. Under conditions identical to those used in the synthesis of 3b·4PF₆, 3c·4PF₆ can be isolated in 60% yield; that is, almost three times greater than that observed (23%) by using conventional kinetically controlled reaction conditions (2). We believe that the yield obtained in this case was lower (relative to 3b·4PF₆) because of some decomposition of the tetrathiafulvalene-containing crown ether on prolonged heating.

The identity of all three [2]catenanes was confirmed by comparing their ¹H NMR and mass spectra with those reported (2, 24, 31). The structure of 3a·4PF₆ was confirmed by x-ray crystallography as well.

Preliminary Mechanistic Investigations.

The reaction mechanism proposed in Fig. 2 could not be probed by solution-state kinetic measurements. The reasons are manifold. Addition of TBAI to CBPQT·4PF₆ results in counterion exchange, producing several species associated with different counterions. In MeCN, it is not unlikely that some (or all) of these species exist as tight ion pairs, obscuring the real identities of CBPQT⁴⁺-containing species. The situation is further complicated by the insolubilities of crown ethers and iodide salts of CBPQT⁴⁺ in MeCN at the high concentrations which would be required for ¹H NMR spectroscopic investigations. Finally, some of the added TBAI most likely is oxidized to elementary I₂, judged by the immediate development of a brown coloration upon mixing with the other two reactants. These factors make estimates of the solution concentrations of reactive species and, in turn, the quantitative evaluation of kinetic parameters, a dubious prospect.

Qualitatively, our observations point quite persuasively to the ring-opening step (Fig. 2A) being the rate-limiting one. The reactions proceed apace with 30 mol% of TBAI and significantly slower with 0.5 mol% suggesting that TBAI is involved in the rate-limiting step. The original experiment performed on 2a (with 1:1 ratio of CBPQT⁴⁺ and crown ether) reaches equilibrium after 6 d. With excess of CBPQT⁴⁺, the reaction is complete in only 2 d; on the other hand, with excess of the crown ether 2a, it still takes at least 6 d to reach full conversion. These observations indicate the involvement of CBPQT⁴⁺, but not the crown ether in the rate-limiting step, a conclusion which is further confirmed by the insensitivity of the time needed to complete the catenation on the nature of crown ether: 2b and 2c still require 6 d, despite their better association with CBPQT⁴⁺. Overall, these findings suggest that a relatively slow opening of CBPQT⁴⁺ is followed by a rapid association with the crown ether component to give the pseudorotaxane [1 ⊂ 2]³⁺ and its subsequent closure to form the [2]catenane. Counterintuitively, the strained tetracationic cyclophane is unusually robust toward nucleophiles.

The assumption of reversibility involving all steps shown in Fig. 2 is reasonable, but nonetheless speculative. To confirm the reversibility of nucleophilic closure of [1 ⊂ 2a]³⁺ into [2]catenane 3a⁴⁺, the preformed catenane 3a·4PF₆ was exposed (Fig. 5) to a twofold excess of 2b and 30 mol% of TBAI. This synthetically superfluous experiment (because 3b·4PF₆ can be prepared directly, without the intermediacy of 3a·4PF₆) is no more than a demonstration that the two [2]catenanes can be equilibrated; that is, provided the nucleophilic attack is indeed reversible, then it is to be expected that the higher affinity of dioxynaphthalene-based system for CBPQT⁴⁺ will drive the equilibrium toward 3b·4PF₆. ¹H NMR spectra recorded (Fig. 6) during the reaction confirmed this hypothesis. At the outset of the reaction (Fig. 6 A–C) peaks (δ 6.75–6.00 ppm) for two species can be observed in the partial ¹H NMR spectrum (CD₃CN) of the mixture of the catenane 3a·4PF₆ and excess of the crown ether 2b. As the heating (≈82°C) commences, the signals assigned to these two species start decreasing in intensity at the expense of signals for a newly formed species, namely, the free crown ether 2a, identified by the diagnostic singlet at δ 6.72 ppm, and the dioxynaphthalene-based catenane 3b·4PF₆, identified (Fig. 6 D and E) by broad doublets at δ 6.34 and 6.06 ppm. After 8 d, the reaction reaches equilibrium, with <5% of 3a·4PF₆ remaining. The ¹H NMR spectrum shown in Fig. 6F was recorded after reaction for 11 d.

Fig. 5.

Replacement of one donor ring by a different one in a [2]catenane. Nucleophilic attack of TBAI onto 3a·4PF₆ opens up the π-acceptor ring of the catenane and allows the decomplexation of the π-acceptor (1³⁺) and π-donor (2a) components. Free 1³⁺ then complexes with 2b as the better π-donor into [2]pseudorotaxane [1 ⊂ 2b]³⁺. The reverse of the original nucleophilic attack occurs in this [2]pseudorotaxane: I⁻ is expulsed, and the CBPQT⁴⁺ π-acceptor ring is regenerated to constitute the more stable catenane 3b·4PF₆. To highlight the difference between the starting and ending materials, the color scheme is modified slightly relative to Fig. 3, with 2a shown in red and 2b in burgundy.

Fig. 6.

Replacement of the donor ring in 3a·4PF₆. Partial ¹H NMR (500 MHz) spectra (recorded in CD₃CN at 298 K) of pure 2b (A), pure 3a·4PF₆ (B), a reaction mixture at t = 0 d (C), after heating for t = 1 d (D), t = 5 d (E), and t = 11 d (F) at 82°C in the presence of TBAI (30 mol%) are shown. Light red, 2a; gray, 3a·4PF₆; burgundy, 2b; orange, 3b·4PF₆.

Conclusions

In summary, we have developed a general, and aesthetically and practically appealing protocol to prepare [2]catenanes from their constituent rings. Commercially available TBAI acts as a nucleophilic catalyst which “magically” opens the acceptor ring. Complexation of the now-opened π-acceptor with the crown ether ensues. Roles of reaction partners then invert and iodide becomes a nucleofuge and gets expelled by the bipyridine nitrogen as the nucleophile to form the [2]catenane. The reaction, which is easy to perform and work up, proceeds cleanly in high yields. This discovery is likely to find immediate applications synthetically, as an approach to donor–acceptor polycatenanes and in the arena of molecular device fabrication.

The dynamic character of the covalent bond formed between aliphatic carbon and iodine atoms in the presence of iodide ions has been known for more than 70 years. In the 1930s, the critical catalytic role of iodide ion in bimolecular nucleophilic substitutions of alkyl iodides by a mechanism involving an inversion of configuration at the saturated carbon atom bonded to the iodine atom was demonstrated (32) in a now classical series of experiments involving optically active and radioactively labeled starting materials. Given its long history nucleophilic substitution has been unjustly overlooked and underutilized in its dynamic context of late. It is our hope and belief that its potential in this regard will now be rediscovered and used in other situations in chemistry.

Materials and Methods

All reactions were carried out, without a protecting atmosphere, in anhydrous MeCN, by using standard laboratory techniques. Yields refer to materials purified by chromatography. Reagents of the highest commercial purity were purchased and used as such, or prepared according to literature procedures. The [2]catenanes described herein were characterized by the comparison of their ¹H NMR and mass spectra with those reported in the literature. In a typical preparative experiment, CBPQT·4PF₆ (30.0 mg, 0.028 mmol) and the crown ether 2b (17.5 mg, 0.028 mmol) were mixed in anhydrous MeCN (25 ml). TBAI (3.0 mg, 0.008 mmol) was added to give a light brown mixture. The reaction mixture was heated under reflux for 6 d, during which time the solution became intensely purple. The solution was filtered through a plug of silica gel to remove uncharged organic material. Subsequent elution of the purple band with a solution of NH₄PF₆ in Me₂CO (1% by weight) provided the crude product. The eluent containing the [2]catenane was concentrated to a small volume, before being treated with cold H₂O, causing precipitation of a purple powder that was filtered and identified as pure 3b·4PF₆ (44.0 mg, 93%). For further experimental details, see supporting information (SI).

Abbreviations

CBPQT

cyclobis(paraquat-p-phenylene)

DCC

dynamic covalent chemistry

TBAI

tetrabutylammonium iodide.