A Kinetic Self‐Sorting Approach to Heterocircuit [3]Rotaxanes

Abstract In this proof‐of‐concept study, an active‐template coupling is used to demonstrate a novel kinetic self‐sorting process. This process iteratively increases the yield of the target heterocircuit [3]rotaxane product at the expense of other threaded species.


Small Azide Screen for Kinetic Self-Sorting
In order to be suitable for the Kinetic Self-Sorting protocol, the azide stopper has to be both passable by large macrocycle 1 and impassable to small macrocycle 2c. Three candidates were chosen as shown in Figure S1. Figure S1. Azide stoppers screened for the Kinetic Self-Sorting protocol.
To test their applicability for the active template-copper-mediated azide-alkyne cycloaddition (AT-CuAAC) reaction, all three azides were reacted with large alkyne 4 and [Cu(MeCN) 4 ]PF 6 to form triazole threads (Scheme S1). After 30 minutes, thread formation was detected by 1 H NMR and the triazoles isolated and characterised. Repeating these experiments in the presence of small macrocycle 2c duly afforded quantitative conversion to [2]rotaxanes S3-S5 by 1 H NMR.
Scheme S1. Test of small azide candidate stoppers in CuAAC and AT-CuAAC reactions.
Satisfied that the azides were suitable for the AT-CuAAC reaction and capable of retaining the small macrocycle 2c, test reactions were undertaken with each azide in the presence of large macrocycle 1 to attempt to form the heterocircuit [3]rotaxane (Scheme S2). When azides 11a or 11b were used, 1 H NMR analysis of the crude reaction mixture after workup with NH 3 -EDTA indicated consumption of 1 and formation of metastable [2]pseudorotaxanes 12a and 12b respectively. [3]rotaxane 15a was not observed in the case of 11a whereas only a trace amounts (< 2.5%) of 15b was observed in the case of 11b. The test reaction with azide 11c, by contrast, afforded 19% conversion to heterocircuit [3]rotaxane alongside recovered macrocycle 1, thread S8 and [2]rotaxane S5 (Scheme S3). No metastable [2]pseudorotaxane 12c was observed. Heterocircuit [3]rotaxane 15c was isolated by column chromatography.
As a result of these tests, ester azide 11c was taken forward: small macrocycle 2c is retained and large macrocycle 1 may pass, with no unexpected metastable products exist as with the other systems tested.

Analogous Reaction with Small Acetylene S10
To further validate the stereoselectivity of the reaction we repeated the self-sorting reaction using small acetylene S10 and azide 3 (Scheme S4). To test the suitability of alkyne S10 for the CuAAC reaction and AT-CuAAC reaction, it was reacted with large azide 3 with [Cu(MeCN) 4 ]PF 6 as a catalyst, either in the presence of absence of small macrocycle 2c. Both [2]rotaxane S11 and thread S13 were made and isolated. The reaction of S10 and 3 in the presence of 1 and 2c only afforded thread S13, [2]rotaxane S13 and recovered macrocycle 1 in keeping with our previous observations: the larger macrocycle, always held over the alkyne as per the proposed mechanism, is on the same side as the smaller steric barrier rendering any [3]rotaxane formed unstable to dethreading.

Optimisation of alkyne and azide equivalents in the synthesis of 15c
We examined the optimal number of equivalents of 11c and 4 to afford complete consumption of 2c and maximise the yield of 15c. The results suggest that the formation of heterocircuit [3]rotaxane 15 continues until smaller macrocycle 2c is entirely consumed and this is achieved with 1 equivalent each (vs total macrocycle present) of the azide and alkyne component. The only practical benefit of this optimisation is to reduce the amount of thread S8 and [2]rotaxane 12 by-products in the final reaction mixture. This ratio of components was used in the iterative selfsorting reaction.

S-19
Notes Aliquots for analysis were diluted with CH 2 Cl 2 (1mL) then washed with NH 3 -EDTA (1 mL). The organic layer was passed through MgSO 4 , reduced in vacuo and analysed by 1 H NMR to determine the ratio between products using the characteristic signals of macrocycle 1 and [3]rotaxane 15c at 7.94 ppm and 6.30 ppm respectively (see Fig 2 of  manuscript).

"All-in-One" Experiment for Comparison
A solution of macrocycle 1 (5.4 mg, 0.010 mmol), macrocycle 2c (24.1 mg, 0.050 mmol), azide 11c (35.2 mg, 0.120 mmol), alkyne 4 (65.1 mg, 0.120 mmol) and [Cu(MeCN) 4 ]PF 6 (21.5 mg, 0.0576 mmol) in CH 2 Cl 2 (1 mL) was stirred at 100 °C in a 150W microwave reactor for 2 hours. The solution was allowed to return to room temperature before dilution with further CH 2 Cl 2 (50 mL) and washing with NH 3 -EDTA (50 mL). The organic layer was retained and the aqueous layer extracted with CH 2 Cl 2 (2 × 50 mL). The organic extracts were combined, dried over MgSO 4 and dried in vacuo. 1 H NMR analysis of the crude reaction mixture revealed 28% conversion of macrocycle 1 of the target [3]rotaxane 15c, based on their characteristic signals at 7.94 ppm and 6.30 ppm respectively (vide supra). Column chromatography (0-30% MeCN/ 1:1 Hexane:CH 2 Cl 2 ; +0.25% EtOH throughout) afforded the product as a white foam (5.3 mg, 28%), with characteristic data consistent with that stated above.  Analysis of the reaction mixture prior to work-up (Fig S2a) indicated complete consumption of macrocycle 1 and the presence of a new interlocked species (signals shown in red). Analysis by MS revealed a mass ion corresponding to [12a+Cu+H] 2+ suggesting that Cu remained coordinated. After aqueous work-up of a portion of the reaction mixture using NH 3 -EDTA a new threaded species was observed (signals shown in orange), which was assigned as metastable [2]pseudorotaxane 12a by analogy with previously isolated [2]rotaxanes, alongside some recovered 1 (blue). When this mixture was heated at 80 °C for 1 h the majority of the threaded species was destroyed, producing non-interlocked thread and macrocycle 1. This demonstrated the instability of 12a with respect to dethreading. In contrast, when the reaction mixture heated for a further hour at 80 °C no significant changes were observed; the species assigned as [Cu12a] persists and macrocycle 1 (or its copper complex [Cu1]) remains absent. Furthermore, even after heating the reaction mixture for 16 h at 80 °C, aqueous work-up with NH 3 -EDTA confirmed the presence of 12a. Taken together, these results not only confirm the presence of a metastable threaded product both before an after work-up, they strongly suggest that Cu stabilises the threaded species.

Study of the stability of metastable [2]pseudorotaxane 12a
These results shed some light on the diminished efficiency of the self-sorting process in later rounds of AT-CuAAC; although in the first step 19% of macrocycle 1 is consumed, in later iterations the reaction appears to consume 12-8% of the remaining macrocycle. It seems likely that, although [2]pseudorotaxane 12c and [3]pseudorotaxane 13c dethread rapidly in the absence, Cu coordination stabilises the threaded species prior to work up and thus reduces the amount of free 1 available in the reaction mixture.