Polyyne Rotaxanes: Stabilization by Encapsulation

Varying the Length of the Polyyne

Polyyne rotaxanes 2b–f·M1, with 8–24 contiguous sp-hybridized carbon atoms, were prepared using an active Cu(I)-template homocoupling strategy (Scheme 1, Table 1).⁴ In a typical procedure, the 1:1 complex of CuI and the phenanthroline-based macrocycle M1 (CuI·M1) reacted with a slight excess of the terminal polyyne 1b–f in THF at 60 °C in the presence of K₂CO₃ and iodine. Attempts to prepare the butadiyne-linked rotaxane 2a·M1 from the alkyne 1a were not successful; it seems that the Tr*-capped polyyne products need at least 8 sp-carbon atoms to provide space for the macrocycle. Preparation of a rotaxane with 32 sp-carbon atoms was also unsuccessful, and we observed complete decomposition of octayne 1g in the reaction mixture. This result indicates that the limit for the current coupling conditions has been reached at the stage of 2f·M1 with a C₂₄ dodecayne dumbbell.

The product yield for the series of rotaxanes 2b–f·M1 decreased with increasing length of the polyyne (Table 1), probably as a consequence of the lower stability of the longer terminal polyynes. Use of lower reaction temperatures suppresses decomposition of the starting materials, but does not increase the yield of rotaxane. In the synthesis of rotaxane 2f·M1, the reaction was stopped after 16 h, when TLC still showed traces of unreacted 1f, to minimize decomposition of the product. In all cases, the concentration of macrocycle M1 was 5–10 mM; increasing the concentration to 30 mM (for conversion of 1c to 2c·M1) did not noticeably alter the yield.

The copper-catalyzed cross-coupling of acetylenes, first reported by Cadiot and Chodkiewicz,²⁸ has been successfully utilized for the synthesis of polyynes,^9b,31 rotaxane-based shuttles¹⁹ and catenanes,²² but the reaction had not been explored for the synthesis of polyyne rotaxanes. We chose to test the synthesis of rotaxanes via Cadiot−Chodkiewicz coupling using triyne 1c and the corresponding bromotriyne 3 (prepared by AgNO₃-catalyzed bromination of 1c with NBS).³² Equimolar amounts of triyne 1c and bromotriyne 3 were added to a solution of the CuI·M1 complex in THF, and the mixture was stirred under nitrogen at 60 °C. The reaction was complete after 4 h (monitored by TLC), and the corresponding rotaxane 2c·M1 was isolated in 38% yield (Table 2, entry 1). The yield of rotaxane from this cross-coupling reaction is slightly higher than that from homocoupling (32%). In the latter case, a slight excess of 1c (2.5 equiv) was used, thus in the next experiment we tested an excess of both 1c (1.2 equiv) and 3 (1.2 equiv), relative to the macrocycle. The reaction was complete after 4 h, and the rotaxane was isolated in 43% yield (Table 2, entry 2). We repeated this reaction keeping the amount and concentration of 1c constant (1.2 equiv) while varying the amount of 3 (1.3–1.5 equiv) and always obtained the rotaxane in good yields (47–53%, Table 2). Finally, we carried out the reaction at 20 °C using stoichiometric amounts of 1c and 3. The starting materials were consumed after 76 h, giving rotaxane 2c·M1 in 26% yield (Table 2, entry 5), while at 40 °C the reaction was complete after 24 h, and the product yield was 36% (Table 2, entry 4). The synthesis of hexayne rotaxanes via cross-coupling is more effective than homocoupling, and proceeds at lower temperatures (e.g., homocoupling failed at 40 °C). However, cross-coupling is not applicable to the synthesis of long polyyne rotaxanes due to the low stability of halogenated polyyne precursors.

Mechanistically, it is believed that Cadiot–Chodkiewicz coupling proceeds via oxidative addition of the bromoacetylene to a Cu-acetylide producing a Cu(III) intermediate, which undergoes reductive elimination affording the cross-coupled product.^17,22,33 However, competing homocoupling of the alkynyl halide is often observed, via halogen-metal exchange,^17,28,34 and it has been difficult to achieve high selectivity in cross-coupling reactions of polyynes. In some cases, selectivity can be obtained by careful choice of the amine base, solvent and reagent concentrations,³⁵ or by applying a polymer-supporting technique.³⁶ To test the selectivity of the cross-coupling reaction, the supertrityl diyne 1b (1.0 equiv) and bromotriyne 3 (1.1 equiv) were reacted with CuI·M1 complex (1.0 equiv) under the conditions described above (THF, 60 °C, 12 h). The reaction gave a mixture of two rotaxanes, as confirmed by the ¹H NMR and MALDI spectra (Figure S15). The ratio of hexayne and pentayne rotaxanes (4:1) was estimated from ¹H NMR spectrum. Thus, homocoupling of the bromotriyne 3 is indeed competitive with the desired heterocoupling reaction.

We envisioned that changing the coupling acetylene partners could provide higher selectivity in cross-coupling,^28,34 so the synthesis of porphyrin–polyyne mixed rotaxanes was investigated. Porphyrin rotaxanes represent an important class of molecular machines and photoresponsive assemblies,³⁷ and the macrocycle M1 has previously been used in the active-metal template homocoupling of meso-ethynyl-porphyrins affording porphyrin-capped rotaxanes.²⁰ Thus, porphyrin 4 (1.0 equiv), supertrityl bromotriyne 3 (1.5 equiv) and CuI·M1 complex (1.0 equiv) were reacted in toluene/THF (2:1) under nitrogen at 60 °C to give rotaxane 5a·M1 in 19% yield (Scheme 2). To our surprise, the formation of unthreaded hexayne or bis-porphyrin rotaxanes was not observed by TLC analysis, ¹H NMR, or MALDI spectra of the crude reaction mixture.

In rotaxane 5a·M1, the second meso position of the porphyrin can serve as a coupling partner after desilylation with TBAF. Thus, rotaxane 5b·M1 was subjected to conditions similar to that for 5a·M1 to give [3]rotaxane 5c·(M1)₂ in 23% yield (Scheme 2). Again, the reaction was highly selective, affording the product without formation of 2c·M1 or bis(porphyrin)-capped rotaxanes.

Synthesis of “Masked” Polyyne Rotaxanes

One limitation of preparing rotaxanes by oxidative coupling of terminal and bromo-polyynes is the instability of the starting materials, and this encouraged us to explore alternative synthetic strategies. One approach is based on the use of masked oligoynes, in which the polyyne framework is assembled in a protected form.³⁸ In the final step, the linear carbon chain is constructed via elimination of masking functional groups. Recently, the Fritsch–Buttenberg–Wiechell (FBW) rearrangement of carbene/carbenoid intermediates has evolved into a valuable synthetic methodology for the preparation of polyynes from geminal dihaloolefin-masked acetylene precursors.²⁹ The basis of the FBW method is the treatment of 1,1-dibromo-2,2-dialkynylethenes with n-BuLi, which leads to the in situ formation of a carbenoid species, followed by 1,2-migration to yield the corresponding linear polyyne (Scheme 3).²⁹ We are interested in utilizing gem-dibromoolefins as masked polyynes in the synthesis of rotaxanes. Compound 6a did not undergo homocoupling in the presence of the copper(I) complex of macrocycle M1. However, the reaction of 6a with the bromo-derivative 6b under cross-coupling conditions furnished the rotaxane 7a·M1 in 9% yield (Scheme 4). Increasing the length of the acetylenic axle dramatically improved the yield of this synthesis. Homocoupling of 6c, in the presence of CuI·M1 gave rotaxane 7b·M1 in 78% yield (Scheme 4). However, all attempts at FBW rearrangement of 7b·M1 were unsuccessful, despite testing a range of reaction conditions (Scheme 5 and Table S1); rotaxane 7b·M1 reacts with butyl lithium to give a complex mixture of products. It is not clear why this reaction fails, but it appears that the proximity of the phenanthroline macrocycle adversely affects the reactivity of the carbenoid intermediate, promoting alternative pathways.

Varying the Size of the Macrocycle

The supertrityl end-group is large enough to prevent slippage of macrocycle M1, and we were interested to test whether it is an effective stopper for even larger macrocycles. However, use of smaller macrocycles is appealing because it should allow the synthesis of polyyne rotaxanes with simpler terminal groups. In many rotaxanes, the macrocycle protects the axle from the external environment,^1,2,39 preventing enzymatic digestion,⁴⁰ or chemical degradation,⁴¹ and in these cases a “tight” fit between macrocycle and thread is desirable. We tested a variety of macrocycles for the synthesis of hexayne rotaxanes (Scheme 6 and Table 3). The triyne 1c and bromotriyne 3 were chosen for rotaxane synthesis since a hexayne axle is long enough to eliminate steric interactions with the supertrityl end-group. The size of macrocycle M1 can easily be adjusted by changing the length of the alkyl bridge. Decreasing the size from C₆ (M1) to C₃ (M2) or C₄ (M3) worked well, yielding hexayne rotaxanes 2c·M2 and 2c·M3 under cross-coupling conditions in 21% and 43% yields, respectively. Under homocoupling conditions, the same macrocycle gave lower yields and the rotaxane 2c·M2 was isolated in only 5% yield. A macrocycle with a C₈ linker (M4)⁷ gave rotaxane 2c·M4 in 41% yield under cross-coupling (9% from homocoupling) but no rotaxane was isolated with the larger C₁₀ macrocycle (M5),⁷ probably because this macrocycle slips off the dumbbell, as indicated by inspection of CPK models.

The phenanthroline-based macrocycle M6 with a C₁₀ alkyl strap gave the corresponding rotaxane 2c·M6 in 17% and 26% yields from homo- and cross-coupling, respectively. The macrocycle M7(15,42) with a more rigid p-tolyl ether framework gave 2c·M7 in 23% and 54% yield, for homo- and cross-coupling respectively (Table 3). Finally, a bipyridine-based macrocycle⁴³ gave rotaxane 2c·M8 in 23% and 26% yield, from homo- or cross-coupling, respectively. Generally, small macrocycles reduce the yield of rotaxanes in both homo- and cross-coupling, except in the case of M7. This contrasts with the results obtained for the synthesis of rotaxanes via active template copper-catalyzed azide–alkyne cycloaddition reaction, where smaller macrocycles afforded better yields.⁴³

Synthesis of [3]Rotaxanes with Two Threaded Polyyne Chains

Rotaxanes with multiple axles passing through a single ring are very rare.⁴⁴ The challenge in preparing this type of molecular architecture is to satisfy the structural demands of both components: The macrocycle may require more than one template site to assemble multiple-threads, and its cavity must be large enough to accommodate two axles, yet small enough to prevent the dethreading. Additionally, the macrocycle-thread interactions that direct the assembly process, must overcome steric hindrance between crowded dumbbell units. So far, single-macrocycle threaded [3]rotaxanes have been synthesized utilizing hydrophobic interactions,^44a octahedral metal centers as templates,^44b,44c hydrogen-bond formation between a thread and axles,^44d and an active-metal templated acetylene homocoupling.^44f The axles can be identical^44b−44e or different^44a and can be assembled using the same or different type of reactions for each step of the threading.

To test the threading of two identical polyynes, the CuI·2c·M1 stoichiometric complex (1.0 equiv) was prepared and mixed with 1c (1.2 equiv) and 3 (1.6 equiv) in THF (Scheme 7). The oxygen-free reaction mixture was stirred for 36 h at 60 °C in the dark. After workup followed by silica and size-exclusion chromatography, the [3]rotaxane (2c)₂·M1 was obtained in 6% yield, and 70% of the 2c·M1 rotaxane starting material was recovered. The product was characterized by mass spectrometry, NMR and UV–vis absorption spectroscopy, and the structure of the molecule was determined by X-ray crystallography. The [3]rotaxane (2c)₂·M1 is stable as a crystalline solid at 4 °C, while at ambient conditions the yellow solid darkens slowly over weeks. To improve the yield of the double-threaded product, a larger macrocycle was tested, to reduce steric crowding. We were pleased to find that under similar reaction conditions, rotaxane 2c·M4, with a larger macrocycle, yields the corresponding [3]rotaxane (2c)₂·M4 in 18% (Scheme 7). The stability of (2c)₂·M4 is comparable to that of (2c)₂·M1, and the solid discolors slowly under ambient conditions, but is stable indefinitely at −20 °C.

Spectroscopic Characterization of Rotaxanes

The polyyne rotaxanes were characterized by MALDI mass spectrometry, UV–vis absorption, and NMR spectroscopy. Threading does not significantly perturb the electronic structure of polyynes and the absorption spectra of the rotaxanes resemble the sum of the macrocycle and polyyne absorptions.^4,26 For example, the absorption spectra of the rotaxanes 2b–f·M1 (Figure 1a) are essentially the sum of the spectra of their components, 2b–f and M1. The polyyne absorption bands in the rotaxanes are shifted to lower energy by about 4 nm, compared to the unthreaded analogs²⁶ (Figure 1b).

In double-threaded [3]rotaxanes, the absorption in the polyyne region is double that of the parent rotaxanes, as expected. Normalization of the spectra of rotaxanes 2c·M1 and (2c)₂·M1 at the absorption maximum (317 nm) shows that the second vibronic band at ∼297 nm is red-shifted by 1 nm and is slightly more intense in the [3]rotaxane (Figure 2).

The ¹H NMR spectra of rotaxanes 2b–f·M1 reveal that the interactions between the supertrityl end groups and the macrocycle become weaker as the polyyne becomes longer (Figure 3). Upon threading, the chemical shift of proton H_f of the macrocycle resorcinol moiety increases, while those of protons H_h and H_g decrease. As the length of the polyyne chain is increased, these changes become insignificant. Thus, in rotaxane 2f·M1, resonances H_f, H_g, and H_h become almost identical to those from free macrocycle M1. The aromatic protons of the supertrityl end-group also move to higher chemical shift in response to polyyne elongation (H₁ and H₂, Figure 3).

In the ¹H NMR spectrum of the [3]rotaxane (2c)₂·M1, protons labeled H_d, H_b, H_e, H_h, H_g, H_j, and H_i move to lower chemical shift, while the chemical shift of proton H_c increases, compared with the corresponding [2]rotaxane 2c·M1 (Figure 4). Resonances of the protons H₁ and H₂ from supertrityl group are unaffected. The two hexayne chains in (2c)₂·M1 are undistinguishable by ¹H and ¹³C NMR spectroscopy in CD₂Cl₂, both at 298 K and at 193 K (Figures S16 and S17).

Thermal Stability

It was hypothesized decades ago that mechanical encapsulation via rotaxination should stabilize extended polyynes.^8,12,13 While remarkable stabilization for cumulene rotaxanes was demonstrated recently,¹⁵ the thermal stability of polyyne rotaxanes was a key aspect that we sought to address in this study. The thermal stabilities of rotaxanes 2c–f·M1 and the corresponding polyynes 2c–f were compared by differential scanning calorimetry (DSC), by heating the samples to 400 °C at 10 °C/min under an atmosphere of nitrogen (Table 4), and observing the exothermic thermal decomposition. Naked polyyne dumbbells show a sharp decrease in thermal stability with increasing chain length,²⁶ whereas there is less change in stability for the rotaxanes.

Rotaxane 2c·M1 decomposes at a lower temperature than the bare hexayne 2c,²⁶ which is probably because the rotaxane melts (melting point: 216 °C; dec 287 °C), whereas the dumbbell decomposes without melting (dec 323 °C). Here the threaded macrocycle serves to reduce the symmetry of the polyyne and to disrupt crystal-packing interactions, reducing the melting point and consequently reducing the thermal stability. However, with further increase of the polyyne length, all of the compounds undergo thermal decomposition without melting and the rotaxanes become more stable than the corresponding naked dumbbells. The difference in stability increases with increasing polyyne chain length. The greatest enhancement in thermal stability is observed for 2f·M1, as DSC shows a decomposition peak at 228 °C, while for the free dumbbell 2f, the decomposition peak occurs at 168 °C (Figure 5). As the polyyne chain gets longer, steric shielding of the carbon chain by the supertrityl groups decreases, providing an opportunity for the macrocycle to act as an additional shield and suppress the polyyne degradation.

X-ray Crystallography

Here we report crystal structures⁴⁵ of polyyne rotaxanes 2c·M2, 2c·M6, 2c·M7 and 2d·M1, dibromolefin rotaxane 7a·M1, porphyrin rotaxane 5a·M1, and [3]rotaxanes 5c·(M1)₂ and (2c)₂·M1. Full crystallographic details regarding crystal growth, data collection, analysis, and crystal packing are given in the SI. Crystallographic data (excluding structure factors) have been deposited with the Cambridge Crystallographic Data Centre (CCDC 1437276–1437283) and can be obtained via www.ccdc.cam.ac.uk/data_request/cif. The crystal structures of rotaxane 2c·M1 and the free hexayne 2c have been published previously^4,26 and are included in the discussion for comparison. Selected parameters are summarized in Table 5.

X-ray crystallography studies of polyynes provide insights into the structures of these molecules and into their noncovalent interactions.⁴⁶ For example, an analysis of the structures of a series of t-butyl-capped polyynes⁴⁷ indicated that infinitely long polyynes reach a saturation in the bond-length-alternation (BLA), which implies that carbyne has alternating single and triple bonds.⁴⁸ This result is in agreement with data from the electronic absorption spectra of Tr*-capped polyynes which predict a finite optical bandgap (E_g = 2.56 eV) for carbyne.²⁶

It is expected that threading will affect the conformation and packing of a polyyne. For example, the unthreaded hexayne 2c is centrosymmetric,²⁶ whereas in all hexayne rotaxanes the macrocycle breaks the symmetry. Thus, the hexayne chain in 2c·M2 is slightly bent in a helical shape and the average C≡C—C(sp) angle is 177.6(14)°. Similarly, in rotaxanes 2c·M6 and 2c·M7 the hexayne axles are semihelical and S-shaped, with average C≡C—C(sp) angles of 175(2)° and 176(2)°, respectively. In 2d·M1, the octayne chain is slightly curved in a helical fashion and the average C≡C—C(sp) angle is 177.0(19)° (Figure 6). Only three X-ray structures of octaynes have been reported previously.^9a,47,49

The angle between two terminal sp-carbon atoms and the centroid of the central C—C bond of the polyyne chain (φ) is a useful parameter for quantifying the bending of a polyyne (Table 5).¹⁶ In this family of polyyne rotaxanes, the axle in 2c·M7 exhibits the most bending with φ = 165(1)°, however, 2c·M6 has the smallest average C≡C–C(sp) angle (175(2)°). All the polyyne [2]rotaxanes have mean BLAs of approximately 0.14–0.15 Å (Table 5), with the exception of 2c·M6 (BLA = 0.160(8) Å); it is not clear whether the unusually high BLA in 2c·M6 is a result of conformational adjustment due to the threaded macrocycle or other effects (such as packing interactions).

The cylindrical π-systems of polyynes can become involved in CH/π_sp interactions. Traditionally, CH/π_arene interactions have received more attention due to their abundance in biology,⁵⁰ whereas CH/π_sp interactions are rarely discussed,⁵¹ and the few existing reports focus on terminal alkynes rather than polyynes. We have analyzed the intermolecular CH/π_sp and π_arene/π_sp interactions in polyynes having four or more triple bonds from the Cambridge Structural Database (CSD),⁵² and herein we compare these data with results for polyyne rotaxanes. The following discussion of CH/π_sp interactions offers, to the best of our knowledge, the first rigorous analysis of this secondary bonding motif.

The polyyne π-systems and macrocycles interact in the solid state via CH/C_sp and C/C_sp short contacts (Figure 6). The phenoxyl groups of the macrocycles interact with the polyyne π-system and the CH/C_sp distances are within the range of values found in CSD (2.65–2.75 Å; see SI for details). Additionally, in rotaxane 2c·M6, the alkyl strap of the macrocycle makes weak contacts with the polyyne π-system (c and d contacts in Figure 6b). In 2c·M7, the shortest CH/C_sp distance is 2.702 Å, which is significantly shorter than the mean value found for tetraynes and longer homologues (2.82(7) Å, SI). In addition to the many CH/π_sp interactions, the phenoxyl groups of the compact macrocycle M7 interact with the hexayne π-system through π_arene/π_sp interactions (Figure 6c). A search of the CSD revealed that π_arene/π_sp interactions are rare in comparison to CH/π_sp interactions (14 vs 120 observed short contacts, respectively). Interestingly, the closest C_arene/C_sp contact in 2c·M7 (Figure 6c: 3.193 Å) is the shortest distance compared to other molecules found in the CSD (mean value for C_arene/C_sp contacts: 3.36(4) Å). Once the macrocycle cavity becomes smaller (M7, M2) the C_arene/C_sp interactions become significant, in addition to CH/C_sp contacts (Figure 6).

Another consequence of mechanical encapsulation is the larger distances between neighboring polyynes. For instance, the inter-sp-chain distance for 2c·M7 is about 9.7 Å, for 2c·M6 it is 11.6 Å and for 2c·M2 it is 12.9 Å, much longer than the shortest distance between neighboring molecules of free hexayne 2c (8.1 Å).²⁶ These distances are far from that required for topochemical polymerization of the polyynes (ca. 4 Å).³⁰ While this result is not surprising for rotaxinated polyynes, the crystallographic data clearly illustrate the protective role of the threaded macrocycles.

In the [3]rotaxane (2c)₂·M1, two hexayne chains are arranged in a crossed geometry (Figure 6e), and the macrocycle M1 sits around the middle of the two hexayne chains (designated as A and B). The angle between chains A and B is 74° (measured as a torsional angle between the quaternary sp³ carbons of the Tr* end groups and the centroids between pairs of these quaternary carbon atoms). In the solid state, the molecule is chiral, due to the helical arrangement of the two hexayne chains inside the cavity of the macrocycle, but the crystal is racemic (P-1). The closest contact between the two hexayne chains is 3.290 Å (C6_A–C6_B), which is less than the sum of van der Waals radii of two sp-carbon atoms (3.4 Å).⁵³ Few crystal structures with π_sp/π_sp interaction are known⁵⁴ and in all cases the alkyne chains adopt a crossed alignment, which may prevent significant orbital overlap between the two close-lying alkynes, conferring stability to the compounds.

The two hexayne chains in (2c)₂·M1 have the same curvature within error: the average ∠C—C≡C angles are 176(2)° and 176(3)° for chains A and B, respectively. In chain A, the average BLA is 0.150(11) Å and in chain B it is 0.175(44) Å, which is surprisingly high (Figure 6f). As a general rule, in extended polyynes, BLA gradually decreases toward the middle of the chain.⁴⁷ For example, in rotaxane 2c·M1 the C≡C triple bonds get longer and the single C—C bonds get shorter toward the middle of the chain, as depicted in Figure 6f (red line). In (2c)₂·M1 both chains deviate from this trend, especially C—C single bonds at C6, the position where polyyne chains form a van der Waals contact. This aberration is clearly reflected in BLA values of A and B chains of (2c)₂·M1 (Figure 6f). To our knowledge, this is the first time that such a long single bond has been found in the middle of a hexayne chain.

In rotaxane 7a·M1, the macrocycle sits on top of one dibromoethene moiety, while the second dibromoethene moiety points in the opposite direction with a slight twist (torsion angle 141.85(8)°, Figure 7a). Both central triple bonds have a similar length (1.202 Å), and the length of the single bond between them is 1.39 Å. The macrocycle interacts with the axle through several CH/π short contacts formed between the π-system of the axle and alkyl chain of the macrocycle. In addition, a bromine atom of the second dibromoethene moiety forms van der Waals contacts with the aromatic π system of the resorcinol part of the macrocycle (Figure 7a).

In the porphyrin rotaxanes 5a·M1 and 5c·(M1)₂, one molecule of methanol is coordinated to the Zn center (Figure 7b,c). In both cases, the tetrayne chains are slightly arced, with an average ∠C—C≡C angle of 176.3(13)° in 5a·M1.