Porphyrinoid rotaxanes: building a mechanical picket fence

We demonstrate that the threaded macrocycles in interlocked porphyrin–corrole conjugates provide a mechanical “picket fence” without affecting their electronic properties.


Introduction
The copper-mediated alkyne-azide cycloaddition (CuAAC) 1 reaction has been widely applied in the synthesis of functionalised porphyrinoid macrocycles, 2 including multi-porphyrinoid arrays, 3 for applications in biology, 4 energy transfer, 5 catalysis, 6 self-assembly, 7 and sensing. 8 Indeed, the broad functional group tolerance, mild conditions and readily available starting materials of this archetypal "click" 9 reaction make it an ideal tool for the synthesis of complex non-natural products.
The CuAAC reaction has also been widely applied in the synthesis of interlocked molecules, 10 including examples of rotaxanes and catenanes containing porphyrin sub-units. 11,12 To achieve this, triazole formation is oen the nal step that captures the interlocked structure by introducing a stopper unit in the case of rotaxanes or closing a macrocycle in the case of catenanes. A general feature of such "passive template" syntheses, 13 in which non-covalent interactions pre-organise the covalent subcomponents in a threaded architecture prior to the CuAAC reaction, is that additional functionality must be included in the covalent structure of both components to provide the required pre-organisation. These functional groups remain in the interlocked product and, although the intercomponent interactions they oen engender can be exploited in the design of molecular machines, 13 this approach imposes structural limitations on the products available for study.
The active template (AT) approach to interlocked molecules, 14-16 removes the need for such templating units in the sub-components of the interlocked molecule. The potential of this methodology for the synthesis of porphyrin-containing architectures was demonstrated by Anderson and co-workers in the synthesis of diyne-linked porphyrin rotaxanes and catenanes using an AT-Glaser 17 methodology. 18 Furthermore, by reducing the size of the macrocycle employed, we have demonstrated that Leigh's active template modication of the CuAAC reaction (AT-CuAAC), 15 in which a copper centre bound in the cavity of a bipyridine macrocycle mediates the formation of the triazole, is a general approach to functionalised and functional rotaxanes in excellent yield. 19 Thus, although it has yet to be applied in this context, 20 the AT-CuAAC reaction appears particularly appropriate for the synthesis of triazole functionalised porphyrin rotaxanes without altering their otherwise desirable properties.
Here we demonstrate the utility of the AT-CuAAC reaction in the synthesis of mechanically interlocked analogues of previously studied triazole-linked porphyrin-corrole conjugates, 21 and that, because the covalent structure of the chromophores is not altered, photo-induced electron transfer between the tetrapyrrole chromophores is unaffected by mechanical bond formation. 22 Conversely, the threaded macrocycles signicantly modify the steric properties of the system, creating a "mechanical picket fence" motif that suppresses aggregation and ligand driven self-assembly.

Results and discussion
Synthesis and characterisation 433, 533 2 and 633 4 Porphyrin-corrole diad 433 was synthesised in excellent yield (98% aer size exclusion chromatography) by reaction of azide 1, alkyne 2 and macrocycle 3 in the presence of a Cu I salt (Scheme 1). The extremely high efficiency of the AT-CuAAC reaction allowed this approach to be extended to triad [3]rotaxane 533 2 and pentad [5]rotaxane 633 4 in 96% and 70% yield, respectively aer size exclusion chromatography, (98% and 91% yield per mechanical bond forming step).
The mass spectrum (MS) of [2]rotaxane dyad 433 shows a molecular ion at m/z ¼ 1123.5 consistent with [M + H] 2+ . Comparison of the 1 H NMR spectra ( Fig. 1) of [2]rotaxane 433 with non-interlocked thread 4 and macrocycle 3 further conrmed the formation of the mechanical bond; although many of the resonances associated with the axle remain unaffected by mechanical bond formation (H n , H o , H p , H q , and protons associated with Ar 1 and Ar 2 ), which is in keeping with their location away from the threaded region of the axle, triazole proton H k is shied considerably to lower eld (Dd $ 1.8 ppm). This is consistent with previous observations of C-H$N hydrogen bonding between the polarised triazole-H k and the Lewis basic pyridine nitrogen donors 19 and suggests that the macrocycle is largely localised over the triazole unit. Conversely, benzylic protons H j appear at higher eld in the interlocked structure (Dd $ 1.  Their 1 H NMR spectra ( Fig. 1d and 3a) compared with the noninterlocked components display broadly similar changes to that of 433.
Electronic properties of interlocked corrole-porphyrin conjugates 433, 533 2 and 633 4 Pleasingly, the electronic properties of 433, 533 2 and 633 4 revealed no signicant differences compared with the noninterlocked axles. 21 The interlocked and non-interlocked compounds all display electronic absorption bands at $428, 557 and 598 nm associated with the Zn II -porphyrin unit, and a band at 413 nm accompanied by broad features between 500 and 660 nm assigned to the Cu III -corrole units (Fig. S17 †). Furthermore, in all cases the emission associated with the excited singlet state of a 1 Zn II -porphyrin* core was efficiently quenched in the interlocked corrole-porphyrin conjugates.
Femtosecond transient absorption spectroscopy of 433, 533 2 and 633 4 conrmed that, as in the case of the corresponding non-interlocked axles, 21 the quenching of Zn II -porphyrin luminescence is due to efficient and rapid electron transfer from the 1 Zn II -porphyrin* excited state to the Cu III -corrole moieties; transient peaks were observed corresponding to the reduced Cu IIcorrole moiety, and a broad peak appeared corresponding to the Zn II -porphyrinc + radical cation. 23 Using the transient signal of the Cu II -corrole moiety, the rate constant for the charge separation process was evaluated to be k CS $ 10 11 s À1 for 433, 533 2 and 633 4 with subsequent charge recombination rates of k CR ¼ 2.1 Â 10 10 s À1 , 3.4 Â 10 9 s À1 , and 3.9 Â 10 9 s À1 respectively (c.f. k CS ¼ 1.1 Â 10 11 s À1 and k CR ¼ 5.0 Â 10 10 s À1 for 4). 21 These results clearly demonstrate that, as proposed, threading of the macrocycles around the arms of the porphyrin core does not signicantly affect the electronic properties of the system.

Effect of threading on the steric properties of pentad 633 4
Although the electronic properties of the interlocked products are unchanged compared with the axle moiety, rotaxanes 433, 533 2 and 633 4 clearly have very different steric properties; encircling the triazole moieties with macrocycle 3 signicantly increases the steric demand of the linker units. This difference is particularly striking in the case of pentad 633 4 which lacks sterically bulky aryl groups on the central porphyrin unit; the space lling model of 633 4 (Fig. 2a) shows that the Zn IIporphyrin unit is signicantly encumbered by the threaded macrocycles. This steric hindrance was quantied by determining the % buried volume (% V bur ) of spheres centered on the central Zn II ion (Fig. 2b). 24 At low sphere radii (r < 3Å), the values of % V bur for 6 and 633 4 are identical suggesting that the Zn II center is still accessible to small molecules, as required for catalysis or ligand binding. However, as r increases, the values diverge as the threaded macrocycles lead to a higher excluded volume. The comparison between the variation in % V bur of 633 4 and 6 suggests that although the interlocked macrocycles do not affect the accessible volume immediately around the Zn II center, they provide a steric wall at higher radii similar to covalent picket fence porphyrinoids that have been developed for catalytic and photochemical applications. 25,26 An obvious consequence of the difference in steric demand of 633 4 compared with 6 can be found in their 1 H NMR spectra; non-interlocked axle 6 displays resonances that are broadened considerably compared with that of [5]rotaxane 633 4 under the same experimental conditions ( Fig. S31 and S32 †). This difference is exacerbated as the concentration of the sample is increased; the signals of rotaxane 633 4 remain sharp while those of non-interlocked axle 6 broaden and shi. The effect of concentration on the 1 H NMR spectrum of non-interlocked pentad 6 is consistent with aggregation of the Zn II -porphyrin unit through p-stacking interactions, as has been widely reported previously. 27 Conversely, in the case of [5]rotaxane 633 4 , the macrocycles encircling the four arms of the porphyrin unsurprisingly appear to prevent the close approach of the Zn IIporphyrin cores.
To further probe the steric effect of the threaded macrocycles we turned our attention to the well-established ability of the ditopic ligand DABCO (L) to direct the formation of [(Zn IIporphyrin) 2 L] dimers. 29,30 The 1 H NMR spectra (Fig. S36 †) of noninterlocked pentad 6 displayed behaviour consistent with that previously reported as the quantity of L was varied: (i) at L : 6 ratios up to 0.5 : 1 a signal was observed at À4.92 ppm in the 1 H NMR spectrum, along with a new signal corresponding to H n 0 which is consistent with the formation of a [6 2 L] dimer in slow exchange on the NMR timescale; (ii) once the ratio of L : 6 exceeded 0.5 : 1 the signal at À4.92 ppm disappeared and H n 0 moved to progressively lower eld as further ligand was added, stabilising once L : 6 ¼ 1 : 1, consistent with fast ligand exchange once excess L is present and the formation of monomeric complex [6L] in competition with [6 2 L]. This was further conrmed by cooling the equimolar solution of 6 and L; at 273 K a broad signal was observed at À2.96 ppm alongside the reappearance of the signal at À4.92 ppm, consistent with the monomeric species [6L] in equilibrium with [6 2 L] at low temperature. Thus, the speciation of 6 ( Fig. 4a) varies as expected with the ratio L : 6. Non-linear regression analysis (see ESI for details †) allowed the association constants for the stepwise association of 6 to ditopic guest DABCO to be determined as K 1 $ 5 Â 10 6 M À1 and K 2 $ 4 Â 10 7 M À1 , albeit with relatively large associated errors of 30% and 27% respectively. 31,32 The behavior of [5]rotaxane 633 4 is signicantly different (Fig. S34 †). At 298 K progressive addition of L did not lead to the appearance of a signal around À5 ppm corresponding to [(633 4 ) 2 L] but instead the resonances corresponding to triazole proton H k and porphyrin b-protons H n underwent monotonic changes that continued until 1 equivalent of L had been added. Alongside these changes, a new broad signal appeared at À2.94 ppm (Fig. 3b) and increased in intensity until 1 equivalent L had been added at which point it disappeared. These changes are consistent with the formation of [(633 4 )L] that undergoes slow exchange with free porphyrin 633 4 on the 1 H NMR timescale, progressing to fast exchange once excess L is present. Consistent with this, cooling an equimolar mixture of 633 4 and L to 223 K (Fig. 3d) to reduce the rate of ligand exchange resulted in a spectrum consistent with [(633 4 )L]. Thus, the speciation diagram of 633 4 with respect to equivalents of L (Fig. 4b) is signicantly different to that of 6. Nonlinear regression analysis (see ESI for details †) allowed us to determine K 1 to be $ 2 Â 10 5 M À1 at 298 K.
To further examine this unusual observation, we performed a variable temperature 1 H NMR study of 633 4 in the presence of 0.4 equiv. L (Fig. S38 †). As expected, reducing the temperature to 273 K led a sharpening of the signal at À2.94 ppm, consistent with [(633 4 )L]. However, surprisingly, a small signal was also observed at À4.62 ppm. Reducing the temperature further to 248 K led to a number of signicant changes consistent with the formation of [(633 4 ) 2 L] alongside [(633 4 )L] and unbound 633 4 , including the appearance of resonances at À4.62 and 8.6 ppm attributed respectively to L and the porphyrin b protons H n of the dimeric complex. Reducing the temperature further to 223 K (Fig. 3c) (Fig. 4c).
That the [(633 4 ) 2 L] complex forms at all is remarkable given the sterically hindered environment provided by the macrocycles. The effect of temperature suggests that the balance between negative steric interactions and the positive binding interaction between the Zn II center and the N donors of the DABCO ligand are nely balanced. Ultimately, the entropic cost of forming the ternary complex, along with the restrictions to conformational freedom associated with forming such a crowded structure, appear nely balanced against the enthalpic benet of maximising Zn-N interactions, leading to a strongly temperature dependent self-assembly process.
Finally, we examined the speciation of mixtures of L, 6 and 633 4 . In contrast to previous reports in which mixtures of different Zn II -porphyrins in the presence of L led to statistical mixtures of dimeric complexes, at low equivalents of L there is a high selectivity for formation of [6 2 L] (Fig. S38 †). Furthermore, this selectivity is maintained in a 1 : 1 : 1 mixture of L, 6 and 633 4 as the temperature is varied; 6 is selectively consumed in formation of [6 2 L] in keeping with the higher stability constant for dimerisation of the non-interlocked axle and we did not observe any evidence of hetero-complex formation. Thus, the mechanical picket fence provided by the macrocycles leads to self-sorting in a mixture of Zn II -porphyrin hosts.

Conclusions
In conclusion, we have demonstrated that the ease and utility of the CuAAC reaction, which has led to its widespread use in the  design of functional porphyrinoids for various applications, 2 is maintained when the active template modication of this reaction is used to produce interlocked analogues. As the covalent structure of the axle is unaffected by mechanical bond formation, the electronic properties of the porphyrin-corrole dyad, triad and pentad reported here are not affected by threading through bipyridine macrocycles, suggesting that the macrocycle provides an alternative, electronically neutral site for structural diversication. Studies comparing the selfassembly behaviour of pentad [5]rotaxane and the corresponding non-interlocked axle component demonstrate that the mechanical bond provides a sterically hindered environment that can modulate intermolecular interactions including pstacking-driven aggregation and ligand-driven dimerisation. This ability to engineer the steric environment around triazolefunctionalised porphyrinoids, an important variable in determining their utility, 25 without modifying their covalent structure, suggests that such readily available rotaxanes may play a role in the development of novel types of "picket fence" systems. In the longer term, by combining the steric properties demonstrated here with the well-developed chemistry of rotaxane molecular shuttles, 13 it should be possible to extend these results to produce stimuli responsive systems in which the steric environment around the porphyrin core can be modulated to produce "smart" materials and catalysts.

Acknowledgements
THN and JL thank the International Center for Young Scientists (ICYS) for support. This work was partly supported by the World Premier International Research Center Initiative (WPI Initiative), MEXT and JST CREST, Japan. THN thanks the Alexander von Humboldt Foundation and Institut für Chemie und Biochemie, Freie Universität Berlin for the continued support. SMG thanks the University of Southampton, EPSRC (EP/L016621/1) and Royal Society for nancial support. JEML is an EU Marie Skłodowska-Curie Fellow, receiving nancial support from the European Union's Horizon 2020 research and innovation programme under Marie Sklodowska-Curie grant agreement No 660731. SMG is a Royal Society Research Fellow. SMG acknowledges funding from the European Research Council (Consolidator Grant Agreement no. 724987). The authors are also grateful to the US-National Science Foundation (Grant No. 1401188 to FD) for support of this work.