Chemical Consequences of the Mechanical Bond: A Tandem Active Template‐Rearrangement Reaction

Abstract We report the unexpected discovery of a tandem active template CuAAC‐rearrangement process, in which N2 is extruded on the way to the 1,2,3‐triazole product to give instead acrylamide rotaxanes. Mechanistic investigations suggest this process is dictated by the mechanical bond, which stabilizes the CuI‐triazolide intermediate of the CuAAC reaction and diverts it down the rearrangement pathway; when no mechanical bond is formed, the CuAAC product is isolated.


General Experimental
Synthesis: Unless otherwise stated, all reagents, including anhydrous solvents, were purchased from commercial sources and used without further purification. All reactions were carried out under an atmosphere of N2 using anhydrous solvents unless otherwise stated. Petrol refers to the fraction of petroleum ether boiling in the range 40-60 °C. EDTA-NH3 solution refers to an aqueous solution of NH3 (17% w/w) with 0.1 M sodiumethylenediaminetetraacetate. Flash column chromatography was performed using Biotage Isolera-4 or Biotage Isolera-1 automated chromatography system, employing Biotage SNAP or ZIP cartridges. Analytical TLC was performed on precoated silica gel plates (0.25 mm thick, 60F254, Merck, Darmstadt, Germany) and observed under UV light or with potassium permanganate solution. Microwave heating of reactions was achieved using a Biotage Initiator+ microwave system. Reactions were run at a maximum power level of 400 W in crimp-cap sealed vials (CEM Ltd.). The temperature was monitored automatically and maintained at the set level throughout the reaction after an initial ramp period, typically ~ 1 minute.
Analysis: NMR spectra were recorded on Bruker AV400, AV3-400 or AV500 instrument, at a constant temperature of 298 K. Chemical shifts are reported in parts per million from low to high field and referenced to residual solvent. Coupling constants (J) are reported in Hertz (Hz). Standard abbreviations indicating multiplicity were used as follows: m = multiplet, quint = quintet, q = quartet, t = triplet, d = doublet, s = singlet, app. = apparent, br = broad. Signal assignment was carried out using 2D NMR methods (HSQC, HMBC, COSY, NOESY) where necessary. All melting points were determined using a Griffin apparatus. Low resolution mass spectrometry was carried out by the mass spectrometry services at the University of Southampton (Waters TQD mass spectrometer equipped with a triple quadrupole analyser with UHPLC injection [BEH C18 column; MeCN-hexane gradient {0.2% formic acid}]). High resolution mass spectrometry was carried out by the mass spectrometry services at the University of Southampton (MaXis, Bruker Daltonics, with a Time of Flight (TOF) analyser; samples were introduced to the mass spectrometer via a Dionex Ultimate 3000 autosampler and uHPLC pump in a gradient of 20% acetonitrile in hexane to 100% acetonitrile (0.2% formic acid) over 5 min at 0.6 mL min; column: Acquity UPLC BEH C18 (Waters) 1.7 micron 50 × 2.1mm).

Rotaxane 4
Prepared according to general procedure B with 1a (              in MeOH was added and the reaction mixture and stirred for two days at rt until the solution turned black. N2 was bubbled through the reaction mixture to remove the solvent. The residue was diluted with CH2Cl2 (20 mL) and washed with H2O (10 mL). The aqueous layer was extracted with CH2Cl2 (2 × 10 mL). The combined organic extracts were washed with brine (10 mL), dried (MgSO4), filtered and the solvent removed in vacuo. Chromatography (petrol with a gradient of 0 to 50% EtOAc) gave S10 as a white foam (4.7 mg, 19%). Decomposition to unidentified species was observed during purification, accounting for the low isolated yield. 1   . Chromatography (petrol with a gradient of 0 to 40% Et2O) gave S11 as a white foam (17.0 mg, 72%). 1

Single crystal X-ray analysis of rotaxanes 4, 5, 6 and 8
Crystals of 4 were grown by vapour diffusion of Et2O into a CH2Cl2 solution. Crystals of 5, 6 and 8 were grown by vapour diffusion of pentane into a Et2O solution.
Data were collected at 100 K using a FRE+ HF diffractometer equipped with a Saturn 724+ enhanced sensitivity detector. Cell determination, data collection, data reduction, cell refinement and absorption correction were performed with CrysalisPro. The structures 4 and 8 were solved using SUPERFLIP, [8,9] 5 was solved using ShelXT, [10] and 6 was solved using ShelXS. [ 11 ] All structures were refined against F2 using anisotropic thermal displacement parameters for all non-hydrogen atoms using ShelXL [10] and software packages within. Hydrogen atoms were placed in calculated positions, except structure 5 H(1) which was located in the difference map, and all were refined using a riding model.  0.31/-0.24 Figure S115 Ellipsoid plot of the asymmetric unit of 6 (ellipsoids shown at 50% probability). Hydrogen atoms omitted for clarity.
A "B-level" alert was detected using the IUCR checkcif algorithm. This was determined to be due to the geometry of the acrylamide combined with the macrocycle orientation which forces the amide proton and one of the alkene protons into close proximity. Thus, this alert is not a crystallographic error but an unusual feature of the interlocked structure which sterically constrains the covalent subcomponents in an otherwise energetically disfavoured arrangement.   11] the lower selectivity of TBAF may be due to fluoride-mediated decomposition of the tetrabutyl ammonium cation) [ 12 ] Interestingly, when [Cu(MeCN)4]PF6 is replaced by CuSO4/Na-ascorbate, KNO3 outperforms KF, suggesting that the role of the inorganic salt is quite complex.   8 mL) in a microwave vial. The orange mixture was stirred at 70 °C (µW) for 1 h. The reaction mixture was diluted with CH2Cl2 (10 mL) and washed with EDTA-NH3 solution (10 mL). The aqueous layer extracted with CH2Cl2 (2 × 50 mL). The combined organic extracts were washed with brine (50 mL), dried (MgSO4), filtered and the solvent removed in vacuo. Analysis by 1 H NMR revealed S12 to be the sole product.

Figure S119
Stacked partial 1 H NMR (400 MHz, CDCl3) spectra of (from top to bottom) 1c, the crude reaction mixture, triazole axle S12 and acrylamide S13.  8 mL) in a microwave vial. The yellow mixture was stirred at 70 °C (µW) for 1 h. The crude reaction mixture was diluted with CH2Cl2 (10 mL) and washed with EDTA-NH3 solution (10 mL). The aqueous layer was extracted with CH2Cl2 (2 × 50 mL). The combined organic extracts were washed with brine (50 mL), dried (MgSO4), filtered and the solvent removed in vacuo. Analysis by 1 H NMR revealed S12 to be the sole product.
Figure S120 Stacked partial 1 H NMR (400 MHz, CDCl3) of (from top to bottom) the crude reaction mixture, triazole axle S12 and acrylamide S13. Step 2A: The residue from step 1 was dissolved in a mixture of KF(aq) (0.1 M, 0.1 mL) and THF (0.9 mL). The orange mixture was stirred at 70 °C (µW) for 1 h. The reaction mixture was diluted with CH2Cl2 (20 mL) and washed with EDTA-NH3 solution (10 mL). The aqueous layer was extracted with CH2Cl2 (2 × 10 mL). The combined organic extracts were washed with brine (10 mL), dried (MgSO4) and the residue analysed by 1 H NMR.
Step 2D: The residue from step 1 was dissolved in a mixture of HPF6(aq) (0.13 M, 0.1 mL, 1 eq,) and THF (0.9 mL). The orange mixture was stirred for 1 h at rt. The reaction mixture was diluted with CH2Cl2 (20 mL) and washed with EDTA-NH3 solution (10 mL). The aqueous layer was extracted with CH2Cl2 (2 × 10 mL). The combined organic extracts were washed with brine (10 mL), dried (MgSO4) and the solvent removed in vacuo. The residue was analysed by 1 H NMR. Production of 11 was confirmed by 1 H NMR. In all cases, triazole rotaxane S9 was not observed. Due to broadening of the signals corresponding to triazolide 12, conversion could not be quantified in the case of step 2A. However, the presence of the doublet at 6.21 ppm which is assigned to 12, indicates that consumption of 12 is incomplete.

Control experiments
To rule out the conversion of 12 to rotaxane S9 followed by reaction to produce 11, control experiments with rotaxane S9 were performed under the same conditions: Step 1 Step 2A Step 2B Step 2C Step 2D No Reaction S88 μL, 0.08 M, 0.0017 mmol), in a sealed microwave vial. The mixture was stirred at 70 °C (µW) for 1 h. The reaction mixture was diluted with CH2Cl2 (10 mL) and washed with EDTA-NH3(aq) (5 mL). The aqueous layer was extracted with CH2Cl2 (2 × 10 mL). The combined organic extracts were washed with brine (10 mL), dried (MgSO4), filtered and the solvent removed in vacuo. Analysis of the residue by 1 H NMR showed a small amount of decomposition but no formation of 11.

Figure S122
Stacked partial 1 H NMR (400 MHz, CDCl3) of the product of conditions A and B, compared with triazole rotaxane S9.
Step A Step B Step 2: A portion (2.0 mL) was removed and treated with Tf2O (0.1 M in CDCl3, 0.05 mL, 0.05 mmol) and the reaction mixture was stirred at rt for 20 mins. 1 H NMR analysis of the reaction mixture reveals the species assigned as 12 has largely been consumed to produce a new major species whose 1 H NMR resonances are consistent with cumulene 13; macrocycle protons HA and HB (8.3 ppm and 8.1 ppm respectively) appear as single environments suggesting the axle stereogenic unit has been lost. This assignment is supported by LCMS analysis of a portion of the reaction mixture; the major species observed has m/z =984.8 (retention time = 3.02 min) which is consistent with 13 (calc. for C64H79CuN3O2 = 984.6). Fractions were also observed corresponding to macrocycle 1a (2.14 min), rearranged product 11 and also 12-OH, suggesting that S14 persists in the reaction mixture ([M-OTf] + ), or that the corresponding cation is a stable intermediate in the case of 12.
Step 3a: A portion of the solution produced in Step 2 (0.5 mL, 0.0125 mmol) was treated with KCN (7.2 mg, 0.110 mmol) in H2O (1 mL) and the mixture stirred at rt for 16 h. The reaction mixture was diluted with CH2Cl2 (20 mL), washed with H2O (3 × 3 mL), dried (MgSO4) and the solvent removed in vacuo. The crude residue was analysed by 1 H NMR to reveal 11 as the major product (11 : S12 ~ 95 : 5).
Step 3b: A portion of the solution produced in Step 2 (0.5 mL, 0.0125 mmol) was treated with H2O (1 mL) and the mixture stirred at rt for 16 h. The reaction mixture was diluted with CH2Cl2 (20 mL), washed with H2O (3 × 3 mL), dried (MgSO4) and the solvent removed in vacuo. The crude residue was analysed by 1 1 H NMR analysis (control ii) revealed the major product to be triazolide 12 (11 and S12 observed in trace amounts).
Control ii.

S91
These experiments suggest that Tf2O triggers the extrusion of N2 from triazolide 12 and provide evidence that a cumulene of the form 13 is an intermediate in the reaction. The control experiment with TfOH rules out the in situ hydrolysis of Tf2O to produce acid that then triggers the rearrangement. The slow reaction of 13 with H2O in contrast with KCN(aq) suggests that the Cu I ion stabilises the cumulene structure, although the lower pH of the KCN solution may also play a role in accelerating the nucleophilic attack. These results also suggest that the hydrolysis of the cumulene intermediate under the optimised reaction conditions is accelerated by H + , given that 12 rearranges to 11 rapidly at rt in the presence of HPF6 without added KF.

S92
ii. In situ 1 H NMR analysis of the reaction of 1a, 2a and 3a followed by Tf2O To provide evidence that the same mechanism observed for 12 is in operation with other substrates we investigated the reaction of 1a, 2a and 3a under anhydrous conditions with Tf2O and analysed the results by 1 H NMR: Following the same procedure as above with 1a, 2a and 3a gave a similar outcome as above. After step 1 a new species was observed that was consistent with triazolide S15. Treatment with Tf2O (step 2) led to ~70% conversion of S15 to produce a species assigned as cumulene S17 (ratio of signals at ~5.0 ppm [2H of S15] to doublet at ~8.2 ppm [2H of S16] = 1 : 2.6). Treatment of this product with KCN(aq) (step 3a) gave acrylamide 5 as the major product (5 : 4 = 90 : 10). Conversely, treatment of a solution of cumulene S17 with H2O (step 3b) resulted in incomplete consumption of S17 to produce 5 (5 : 4 = 90 : 10).

Preliminary computational analysis of the mechanism of the rearrangement process
In order to provide further information on the pathway of the rearrangement process and in an attempt to identify the role of the Cu I ion in the process, we carried out a preliminary computational investigation of the reaction pathway.
i. Preparation of a truncated model Ia of triazolide S15 Preliminary molecular modelling was carried out to assess the feasibility of the mechanism proposed based on the studies in sections S6-S10. Due to the large size of the interlocked intermediates of the reaction a truncated model was used.
A model of triazolide S15 was prepared using Spartan '10 (Wavefunction Ltd.) and a conformation search performed using mechanics (MMFF, vacuum). The lowest energy structure identified was optimised using the PM6 semi-empirical method (vacuum). The model was then truncated to provide a starting point (Ia) for DFT calculations below ( Figure S126). Gaussian '09 (DFT-rB3LYP-631G) was used for subsequent calculations with the default H2O solvation model. [13] The reaction pathway obtained, computed structures obtained and their relative energies are shown in Figure S127 and Figure S128. VIIa +139.8

S95
H + was added to either N 3 of the triazole or the OH of structure Ia and the geometry of these species was optimised. No stable structure could be identified for the O-protonated species in which the C-O bond was maintained. The energy difference between Nprotonated (IIa) and O-protonated (IIIa) models was found to be 17.9 kJmol -1 .
The departed H2O leaving group was removed from the model of the carbocation and the geometry was optimised again (IVa). This structure was used as a starting point for further calculations and its computed energy was used as the new baseline.
A scan (unrestricted, 10 steps of 0.1 Å) was performed using the C-N 3 bond length as the redundant coordinate. The outcome of this scan indicated that an energy maximum was reached with C-N 3 = 1.82276 Å (Figure S129a). A transition state calculation was performed using the maximum energy structure found in this scan as a starting point. A species (TS-Va) with a single imaginary frequency was identified. An IRC calculation (20 steps forward and 20 steps reverse, final structures optimised, Figure S129b) confirmed that this transition state connected the starting cation (IV) and a cumulene species (VI) in which N2 had been extruded. The reaction is exergonic by 44.9 kJmol -1 and TS-Va lies 77.7 kJmol -1 above cation Va. Examining the calculated structure of cumulene (VI), it is worth noting that a) the Cu I -ion remains associated with the p-system and b) the C-C-C-N unit is not linear. This may indicate that the true structure lies somewhere between the limiting resonance structures in which the Cu I -ion engages the cumulene through a p-metal interaction and a s-metal interaction (Figure S127Figure S129).
We were unable to locate a transition state for the stepwise opening of the triazole to give vinyl diazonium VIIa, the expected intermediate if N2 loss were to proceed in a stepwise manner. This is unsurprising as this species was found to lie 139.8 kJmol -1 above that of cation IVa (i.e. 62.1 kJmol -1 above TS-Va) when prepared and optimised directly. The IRC plot shows an inflection before TS-Va and examining the computed structures, it is clear that whereas early in the process the reaction coordinate is primarily associated with the stretching of the N 1 -N 2 bond, as would be expected en route to VIIa, the inflection point is associated with the stretching of the C-N 3 bond starting to contribute to the pathway. iii. DFT evaluation of the pathway of N2 loss from truncated triazole model Ib Finally, for comparison, we repeated the above calculations for the corresponding reaction of triazole starting material Ib (pathway shown in Figure S130, calculated structures and energies shown in Figure S131). The reaction of Ib was found to proceed via a stepwise pathway in which vinyl diazonium VIIb is an intermediate lying in a shallow minimum ~1 kJmol -1 below the transition state which is itself found 49.1 kJmol -1 above IVb. The transition state for loss of N2 from VIIb was found to lie ~ 90.9 kJmol -1 above this intermediate and thus, including the pre-equilibrium between IVb and VIIb, the transition state of the rate limiting loss of N2 lies 139 kJmol -1 above cation IVb. Furthermore, an IRC calculation suggests that the product of this pathway is best represented by limiting the limiting resonance structures shown rather than a cumulene structure analogous to VIa as migration of the C-H bond was not found to take place spontaneously during loss of N2.

Figure S130
Intermediates in the calculated path from truncated axle Ia to cumulene VI.

iv. Conclusions
Comparing the computed reaction pathways for loss of N2 from Ia and Ib, the electron rich C-Cu I bond of the triazolide appears favour the opening of the triazole compared with the C-H bond of the simple triazole in three ways: i) the stability of the intermediate formed by protonation of the OH group (III) compared to the intermediate protonated on N (II) is enhanced by the Cu-C bond, biasing this pre-equilibration step towards key reactive intermediate III (IV); ii) the pathway of the ring opening process is altered as the electron rich Cu-C bond is eliminated during the loss of N2, leading directly to a stable cumulene structure instead of a stepwise process via vinyl diazonium VIIa; iii) the C-H bond does not participate in N2 loss. As a consequence, loss of N2 leads to unstable vinyl cation product IX with a consequently higher reaction barrier.