Furan Release via Force-Promoted Retro-[4+2][3+2] Cycloaddition

Mechanophores (mechanosensitive molecules) have been instrumental in the development of various force-controlled release systems. However, the release of functional organic molecules is often the consequence of a secondary (nonmechanical) process triggered by an initial bond scission. Here we present a new mechanophore, built around an oxanorbornane-triazoline core, that is able to release a furan molecule following a force-promoted double retro-[4+2][3+2] cycloaddition. We explored this unprecedented transformation experimentally (sonication) and computationally (DFT, CoGEF) and found that the observed reactivity is controlled by the geometry of the adduct, as this reaction pathway is only accessible to the endo-exo-cis isomer. These results further demonstrate the unique reactivity of molecules under tension and offer a new mechanism for the force-controlled release of small molecules.


General Experimental Details
Unless otherwise stated, all reagents and solvents were purchased from commercial suppliers. All other chemicals were used without further purification. Compound S1 was prepared according to literature procedure. S1 Gel permeation chromatography (GPC) analyses were performed in THF solution (~1.0 mg mL-1) at 35 °C using a GPC Agilent 1260 Infinity II with 2 × PL gel 10 μm mixed-B and a PL gel 500 Å column, and equipped with a differential refractive index (DRI) detector employing narrow polydispersity polystyrene standards (Agilent Technologies) as a calibration reference. Samples were filtered through a Whatman Puradisc 4 mm syringe filter with 0.45 μm PTFE membrane before injection to equipment, and experiments were carried out with injection volume of 50 μL, flow rate of 1 mL min -1 . Results were analysed using Agilent GPC/SEC Software Version 2.2.
Ultrasound experiments were performed using a Sonics VCX 500 ultrasonic processor set at 25% amplitude and equipped with a 13 mm diameter solid probe or replaceable-tip probe.
The distance between the titanium tip and the bottom of the Suslick cell was 3 cm. The ultrasonic intensity was calibrated using the method outlined by Hickenboth et al. S2

Synthesis of S3
Methyl 4-(bromomethyl)benzoate (5.00 g, 21.83 mmol, 1 equiv.) was added to a round bottom flask (25 mL) and dissolved in acetone (8 mL). Next, sodium azide (1.56 g, 24.01 mmol, 1.1 equiv.) was added to the mixture dissolved in water (2 mL). The resulting mixture was stirred for 12h at room temperature. Next, the reaction was diluted with 30 mL of deionised water and extracted with ethyl acetate (3 x 30 mL). Organic layer was then dried over anhydrous magnesium sulphate, filtered and solvent removed to yield S3 (4.04 g, 21.18 mmol, 97%) as colourless liquid, which was used without further purification.  S3 (2.00 g, 10.46 mmol, 1 equiv.) was dissolved in THF (20 mL) and added into a round bottom flask (100 mL) containing LiOH monohydrate (1.25 g, 52.30 mmol, 5 equiv.) in water and methanol mixture (4 and 20 mL, respectively) at room temperature. Setup was equipped with an air condenser and the mixture stirred vigorously at 50°C for 24 hours. Next, mixture was neutralised with acetic acid and poured onto aqueous solution of HCl (1 M, 30 mL) in a separating funnel, aqueous layer extracted with ethyl acetate (3 x 30 mL) and phases separated. Organic layer was then dried over anhydrous magnesium sulphate, filtered and solvent removed. Crude was subjected to flash column chromatography (PE/EA : 80/20 + 0.5 % AcOH) to yield S4 (1.68 g, 9.52 mmol, 91%) as white solid.

Synthesis of S6 and S7
Freshly distilled furan (0.27 ml, 3.67 mmol, 1 equiv.) is added to a nitrogen flushed microwave vial (20 mL) containing methyl acrylate (1 mL, 11.02 mmol, 3 equiv.), which was filtered through basic alumina. The mixture is shaken well to mix the contents and subsequently cooled to -30°C before BF3 Et2O (0.10 mL, 0.8 mmol, 0.22 equiv.) is added in one portion. The reaction is then left in the freezer (-22°C) for 8 hours. Next, slightly yellow solution is warmed up to room temperature and diluted with deionised water (30 mL). The aqueous phase is then extracted with diethyl ether (30 mL) and phases separated. Organic phase was dried over anhydrous magnesium sulphate, filtered and concentrated. Crude was subjected to flash column chromatography (PE/EA : 80/20) to yield to yield S6 (205 mg, 1.32 mmol, 36%) and S7 (62 mg, 0.4 mmol, 11%).

Synthesis of S12 and S13
S6 or S7 (500 mg, 3.24 mmol, 1 equiv.) was dissolved in THF (5 mL) and added into a round bottom flask (25 mL) containing LiOH monohydrate (680 mg, 16.22 mmol, 5 equiv.) in water and methanol mixture (1 and 5 mL, respectively) at room temperature. Setup was closed and the mixture stirred vigorously at RT for 24 hours. Next, mixture was neutralised with acetic acid (3 mL) and diluted with HCl (1 M, 30 mL), aqueous layer extracted with ethyl acetate (3 x 30 mL) and phases separated. Organic layer was then dried over anhydrous magnesium sulphate, filtered, solvent removed and crude dried on high vacuum for 2-3 days. Crude white solids of S12 and S13 were used without any purification in the next steps.

Representative procedure: synthesis of P-1
Methyl acrylate was passed through basic alumina and degassed with nitrogen for 10 minutes prior to use. Stock catalyst solution was prepared by dissolving copper (II) bromide (5.6 mg) and Me6TREN (16µL) in DMSO (1 mL). 1 (5 mg, 7.1 µmol, 1.0 equiv.) was dissolved in DMSO (0.65 mL), methyl acrylate (0.65 mL, 7.11 mmol, 1000 equiv.) and stock solution of the catalyst (60 µL, 0.48 equiv.) were combined in a microwave vial (5 mL). Extra amount of methyl acrylate (1 mL) was added and the resulting mixture degassed. Next, stirrer bar was wrapped with copper wire (2.5 cm) and submerged in concentrated hydrochloric acid for 10 min, then washed with acetone and dried on high vacuum prior to addition into the reaction mixture. Once the stirrer was added, the mixture was sealed and degassed for further 1 minute with nitrogen. After 35 minutes, reaction was stopped by exposing it to the atmosphere. Next, the viscous solution was added onto stirring methanol (800 mL) in a beaker (1 L) to precipitate the polymer, which was then transferred into a vial and dried on high vacuum for 4 days.

List of polymers
All polymers were analysed using GPC with the obtained results visible in Table S1 and Figures S1 -S4. Polymers P-S14 and P-S15 show a high mass shoulder in their GPC chromatograms. This was attributed to a possibility of the small number of chains interacting with double bond within compounds S14 and S15. Regardless of the origin of the high mass shoulder, majority of P-S14 and P-S15 consists of the correct species. Table S1. Mn and Đ values for polymers P-1 -P4, P-S5, P-S14 and P-S15 as measured by GPC.

Standard Sonication Procedure
The desired polymer (45 mg) was dissolved in the dry acetonitrile (approx. 15 mL in total) and added to a modified Suslick cell. Nitrogen was gently bubbled through the solution as it was sonicated. The mixture was kept in an ice bath for the whole duration of the process and was sampled periodically (varied time intervals) over the course of 240 minutes of total sonication time. After 240 min of sonication time, the mixture was filtered through Whatman Puradisc 10 mm syringe filter with 0.2 μm PTFE membrane to remove metal impurities. After drying under high vacuum for an extended period of time (approximately 24 h), the polymer was washed with methanol (5 mL) before drying again.

Sonication Experiments Overview
The four mechanophore-centred polymers were subjected to mechanical activation following the standard sonication procedure. All samples were found to have fully cleaved after 240 minutes of sonication.

Retention time / minutes
Relative intensity Figure S6. GPC chromatograms of pre-and post-sonication GPC traces (red and black, respectively) of -cis mechanophores. Figure S7. GPC chromatograms of pre-and post-sonication GPC traces (red and black, respectively) of P-S14 reference compound. GPC of the heated sample is shown in black, dotted line. Figure S8. GPC chromatograms of pre-and post-sonication GPC traces (red and black, respectively) of P-S15 reference compound. Figure S9. GPC chromatograms of pre-and post-sonication GPC traces (red and black, respectively) of polymer P-2. a Data referred to the sonication experiment of polymer P-2 for furan observation.

Retention time / minutes
Relative intensity S42 Figure S10. GPC chromatograms of pre-and post-sonication GPC traces (red and black, respectively) of polymer PMA.

Kinetic Investigation Overview
All samples were investigated according to standard sonication procedure. The sonication was sampled at constant time intervals across all samples. The sonications were sampled at sonication times of: 0, 5, 10, 20, 30, 45, 60, 90, 120, 180 and 240 minutes. Upon sonication, the average molecular weight of the sample tends toward 30 kDa (c) and the apparent rate constant (k*) can be derived with Nalepa's method (Table S2, Figure S9-S16). S3
Complete scission of polymer P-1 was confirmed by GPC. However, NMR analysis of the post-sonication solid, it was clear that the footprint of the mechanophore within polymer P-1 was not affected by the sonication (Scheme S5). Application of the force led to random polymer backbone scission within PMA chains attached to the mechanophore, as observed by the GPC of the post-sonication mixture (Scheme S5). The results indicate that endo-exotrans geometry is not capable of force transduction leading to the retro-cycloaddition products ( Figure S19). Figure S19. Partial 1 H NMR stack (500 MHz, acetone-d6, 1024 scans) of 1, P-1 and post-sonication sample of P-1 (top to bottom, respectively).

Sonication Experiment of Polymer P-2
Scheme S6. Sonication of polymer P-2 (endo-exo-cis). Scissile bonds were marked in red. Main products are shown.
Complete scission of polymer P-2 was confirmed by GPC. From the NMR analysis of the post-sonication solid, it was clear that the mechanophore within polymer P-2 had undergone a chemical change. The geometry of the adduct (endo-exo-cis) was found to be mechanically susceptible to undergo a mechanochemically triggered process to yield retro-[3+2] and retro-[4+2][3+2] products; to release free furan, along polymer attached moieties containing acrylate and azido-benzyl derivative (Scheme S6). Indicative of the generation of products of the retro-cycloaddition of the triazoline moiety (retro-[3+2]) was the appearance of the sharp singlet at 4.59 ppm in the 1 H NMR of the post-sonication polymer ( Figure S20), which corresponds to the methylene adjacent to the azido group in the reference polymer P-S5, as well as shift of the aromatic peaks (blue dotted lines). There have been new signals appearing in the region between 5.8 and 6.6 ppm in the post sonication sample, which was attributed to the trace of the P-S14 present in the mixture (black dotted lines) as the product of the retro-[3+2] along the product from retro-[4+2][3+2] whose reference was obtained by heating a P-S14 sample to 70°C for 24h in toluene (red dotted lines). Figure S20. Partial 1 H NMR stack (500 MHz, acetone-d6, 1024 scans) of 2, P-2, post-sonication sample of P-2, heated P-S14, P-S14 and P-S5 (top to bottom, respectively).
Complete scission of polymer P-3 was confirmed by GPC. However, NMR analysis of the post-sonication solid, it was clear that the footprint of the mechanophore within polymer P-3 was not affected by the course of the sonication (Scheme S7). Application of the force led to random polymer backbone scission within PMA chains attached to the mechanophore, as observed by the GPC of the post-sonication mixture. The results indicate that exo-exo-trans geometry is not capable of force transduction leading to the retro-cycloaddition products ( Figure S21).
Complete scission of polymer P-4 was confirmed by GPC. From the NMR analysis of the post-sonication solid, it was clear that the mechanophore within polymer exo-exo-cis had undergone a chemical change. The geometry of the adduct (endo-exo-cis) was found in our study to be mechanically susceptible to undergo a mechanochemical process to yield retro-[3+2] products, to the species as in reference compound P-S15 and azido-benzyl derivative (Scheme S8). Indicative of the generation of products of the retro-cycloaddition of the S51 triazoline moiety (retro-[3+2]) was the appearance of the sharp singlet at 4.59 ppm which corresponds to the methylene adjacent to the azido group in the reference polymer P-S5, as well as shift of the aromatic peaks (blue dotted lines). There have been new signals appearing at 6.4 ppm in the post-sonication sample, which was attributed to the P-S15 present in the mixture (black dotted lines).
Polymer P-4 undergoes partial transformation in the mechanochemical activation where approximately 50% of the mechanophore yields the azido-methylene containing aromatic groups as per reference P-S5 in the mechanical activation. The integration of the aromatic signals matches well with signal at 6.4 ppm ( Figure S22). Figure S22. Partial 1 H NMR stack (500 MHz, acetone-d6, 1024 scans) of 4, P-4, post-sonication sample of P-4, P-S15 and P-S5 (top to bottom, respectively).

Sonication Experiment of Polymer P-S14
Complete scission of polymer P-S14 was confirmed by GPC after 240 min of sonication.

Sonication Experiment of Polymer P-S15
Complete scission of polymer P-S15 was confirmed by GPC after 240 min of sonication.

Sonication Experiment of Polymer P-2 for Furan release
Sonication was performed using a modified version of the standard procedure: P2 (70 mg) was dissolved in acetonitrile-d3 (approximately 8 mL) and added to a modified Suslick cell and degassed with nitrogen for 10 minutes. No gas was purged through the solution as it was sonicated. The mixture was kept in ice bath for the whole duration of the process over the course of 450 minutes of total sonication time. After 450 min of sonication time, an aliquot of 0.6 mL was taken and filtered through Whatman Puradisc 10 mm syring filter with 0.2 μm PTFE membrane to remove metal impurities. The filtered solution was directly used for 1 H NMR analysis.
From the NMR analysis of the post-sonication sample, it was clear that the mechanophore within polymer P-2 had undergone a chemical change. The geometry of the adduct (endoexo-cis) was found to be mechanically susceptible to undergo a mechanochemically triggered process to yield retro-[3+2] and retro-[4+2][3+2] products; to release free furan, along polymer attached moieties containing acrylate and azido-benzyl derivative (Scheme S9). Indicative of the generation of furan was the two triplets at 7.51 ppm and 6.44 ppm in the crude 1 H NMR of the post-sonication sample a ( Figure S25, light blue dotted lines). There have been new signals appearing in the region between 5.8 to 6.8 ppm in the postsonication sample, which was attributed to the methyl acrylate generated throughout the sonication (see section 6.11). Figure S25. Partial 1 H NMR stack (500 MHz, acetonitrile-d3, 256 scans) of P-2, post-sonication sample a of P-2 without solvent removal, post-sonication sample b of P-2 after methanol wash, P-S14, P-S15, methyl acrylate c spiked post-sonication sample a of P-2 (top to bottom, respectively). a Post-sonication sample of P-2 with NMR spectrum recorded after discarding metal particulates without solvent removal. b Post-sonication sample of P-2 after methanol wash. c 0.5µL methyl acrylate was spiked into 10 mg of post-sonication sample of P2.

Sonication Experiment of PMA
Scheme S10. Sonication of PMA.
Sonication was performed in acetonitrile-d3 using the procedure described in section 6.10. Complete scission of PMA was confirmed by GPC. From the NMR analysis of the postsonication samples, it was clear that methyl acrylate was generated upon scission of PMA chains. The disappearance of the methyl acrylate peaks of post-sonication samples after methanol wash indicates that the peaks at 5.7 to 6.4 are small molecules that are not attached to the polymer chain. The methyl acrylate generated in the previous sonication of P2 (Figure S26) was attributed to the presence of PMA not the mechanophore. Activation of polymer samples (15 mg) was caried out in toluene (2 mL) at 90°C for 18h under inert nitrogen atmosphere for samples of P1 -P4 to yield samples P1b -P4b. Upon completion, solutions were allowed to reach room temperature and solvent was removed under reduced pressure. Residual solid (thin film) was washed with methanol and solid polymer sample dried on high vacuum for 2 days before being analysed by NMR and GPC. Sample of P-S14 was thermally activated to P-S14b as described but at temperature of 70°C and for 24h.

Thermal activation of P-1
Thermal activation of polymer P-1 led to its slow degradation as is indicative of the formation of the acrylate PMA peaks, as in reference of heated sample of P-S14b. Despite high temperature, the mechanophore appears relatively thermally stable, as conversion is only approximately 21% ( Figure S28). Figure S28. Partial 1 H NMR stack (500 MHz, acetone-d6, 1024 scans) of 1, P-1, P-1b and P-S14b (top to bottom, respectively).

Thermal activation of P-2
Thermal activation of polymer P-2 led to its slow degradation as is indicative of the formation of the acrylate PMA peaks, as in reference of P-S14b. Conversion was found to be approximately 21% ( Figure S29). Figure S29. Partial 1 H NMR stack (500 MHz, acetone-d6, 1024 scans) of 2, P-2, P-2b and P-S14b (top to bottom, respectively).

Thermal activation of P-3
Thermal activation of polymer P-3 led to its slow degradation as is indicative of the formation of the acrylate PMA peaks, as in reference of P-S15b. Conversion was found to be approximately 24% ( Figure S30). Figure S30. Partial 1 H NMR stack (500 MHz, acetone-d6, 1024 scans) of 3, P-3, P-3b and P-S15b (top to bottom, respectively).

Thermal activation of P-4
Thermal activation of polymer P-3 led to its slow degradation as is indicative of the formation of the acrylate PMA peaks, as in reference of P-S15b. Conversion was found to be approximately 24% ( Figure S31). Figure S31. Partial 1 H NMR stack (500 MHz, acetone-d6, 1024 scans) of 4, P-4, P-4b and P-S15b (top to bottom, respectively).

Crystal structures
Single crystals of compounds S8 -S11 were obtained by diffusion of hexanes into a concentrated solution of a given compound in ethyl acetate over 10 days.    CoGEF calculations were performed on Spartan'20 following Beyer's method. S4 The structure of the mechanophore was built in Spartan'20 and minimized using molecular mechanics (MMFF). The distance between the terminal methylene groups was constrained and increased by increments of 0.1 Å, and the energy was minimized by molecular mechanics (MMFF) at each step using the Energy Profile function implemented in Spartan'20. The geometry of the obtained structures was then minimized by DFT (ωB97X-D/6-31G*, gas) using the Equilibrium Geometry function implemented in Spartan'20. The relative energy of each intermediate was determined by setting the energy of the initial state at 0 kJ/mol. Fmax values were determined from the slope of the final 40% of the energy/elongation curve (i.e. from 0.6 Emax to Emax).