We selected commercially available BaTiO 3 nanoparticles as the piezoelectric transducer required for the ultrasound-mediated controlled radical polymerization. This choice was partly motivated by the findings of Li and co-workers, who showed that ultrasonic agitation of BaTiO 3 nanoparticles resulted in an electrochemical potential sufficiently large to overcome the water splitting potential of −1.23 V under standard temperature and pressure20. Consequently, we expected this localized potential obtained from ultrasonic agitation to be sufficient to reduce ligand-stabilized Cu(II) complexes. Specifically, we chose a Cu(II) precursor mixture that contained equimolar amounts of Cu(OTf) 2 (OTf, OSO 2 CF 3 ), N,N,N′,N′,N″,N″-hexamethyl(tris(aminoethyl)amine) (Me 6 TREN), and Bu 4 NBr (40 mM each). The expected major species in this mixture, Cu(II)Br 2 /Me 6 TREN, has a reduction potential (Cu(II)/Cu(I)) of −0.69 V (ref. 26) and should, therefore, be amenable to ultrasound-mediated reduction. In preliminary experiments, when a solution the Cu(II) precursor in dimethylformamide (DMF) was ultrasonically agitated with BaTiO 3 nanoparticles, we observed a reduction in the light absorption band at around 950 nm, usually attributed to the d–d electronic transitions of d9 Cu(II) complexes (Fig. 2a). We could not directly observe the formation of Cu(I) complexes because of overlapping absorption peaks below 500 nm. However, the obvious consumption of Cu(II)Br 2 /Me 6 TREN, presumably to form copper complexes of lower oxidation states, encouraged us to use ultrasonic agitation of BaTiO 3 nanoparticles to initiate a Cu-catalysed ATRP.

Figure 2: Analysis of the piezoelectric reaction and evidence for reduction and initial polymerization. a, Ultraviolet–visible–near-infrared spectroscopy of the copper(II) precursor solution before after 1 h and after 2 h of ultrasonic agitation. The mixture contained Cu(OTf) 2 /Me 6 TREN/Bu 4 NBr (40 mM in DMF), and BaTiO 3 nanoparticles (4.5 wt%). b, The evolution of the number average molecular weight (M n ) and polydispersity (M w /M n ) of polymer P n as a function of time in a representative polymerization reaction. The polymerization mixture contained 1 (4 M in DMF), 2 (40 mM), Cu(OTf) 2 /Me 6 TREN/Bu 4 NBr (40 mM) and BaTiO 3 nanoparticles (4.5 wt%). The mixture was sonicated at 20 kHz using an ultrasonic horn with a 4 s on/4 s off cycle. The time on the x-axis refers to clock time. Sonication was stopped after 2 h to evaluate if any side reactions had occurred. Aliquots from the reaction were analysed by GPC. M n and M w /M n were determined by GPC calibrated to polystyrene standards. Full size image

To test the feasibility of ultrasound-mediated ATRP, we sonicated a suspension of BaTiO 3 nanoparticles in a standard Suslick cell that contained n-butyl acrylate 1 (4 M in DMF) and 1 mol% each of the catalyst precursor Cu(OTf) 2 /Me 6 TREN/Bu 4 NBr and ethyl α-bromoisobutyrate 2 as the initiator. The ultrasound irradiation was applied to the mixture with an alternating four seconds on/four seconds off cycle to dissipate heat produced during the process. Additionally, the reaction cell was externally cooled using a water bath held at temperatures between 15 and 25 °C. Encouragingly, over time we observed the formation of poly(n-butyl acrylate). The time-dependent evolution of molecular weight depicted in Fig. 2b reveals first-order polymerization kinetics with low dispersities (M w /M n < 1.15) of the resulting polymers. The lack of any observable side reactions, as determined by 1H-NMR spectroscopy and gel permeation chromatography (GPC), added to the evidence that supports a controlled radical polymerization (Supplementary Figs 1 and 2). These observations suggest that a polymerization mechanism similar to ATRP may be operating under our polymerization conditions.

An alternative hypothesis is that the polymerization of the acrylate monomer could be mediated by the high temperatures encountered in ultrasound-mediated solvent cavitation. To test this hypothesis, we conducted several control experiments. To establish whether the particles were necessary to mediate the process, we sonicated a polymerization mixture of 1 (4 M in DMF), 2 (40 mM) and Cu(OTf) 2 /Me 6 TREN/Bu 4 NBr (40 mM) with no BaTiO 3 nanoparticles. We observed the formation of only short oligomers (M n < 750 g mol–1) as determined by GPC and 1H-NMR spectroscopy (Supplementary Figs 3–5). Second, we sonicated a polymerization mixture of 1 (4 M in DMF), 2 (40 mM) and BaTiO 3 (4.5 wt%) with the copper catalyst precursor. We observed an increase in the viscosity of the polymerization mixture, which suggests that localized high temperatures could be initiating uncontrolled polymerization. GPC analysis of the reaction mixture, however, revealed the absence of any chromatographic peaks that could be attributed to any well-defined polymers. Third, we replaced BaTiO 3 nanoparticles with carbon black nanoparticles in a control polymerization reaction. We reasoned that carbon nanoparticles (200 nm average diameter) lack any crystalline order and should, therefore, have no piezoelectric property. When the polymerization mixture was sonicated, we again observed, by GPC and 1H-NMR spectroscopy, the formation of very short oligomers (Supplementary Figs 6 and 7). This observation further supports the assertion that thermal background polymerization, if present, is not sufficient to explain the controlled growth of polymer chains observed when piezoelectric nanoparticles were used. Fourth, we stirred the polymerization mixture at elevated temperature (80 °C) for 40 hours. Although very high temperatures (1,800 °C) can occur during ultrasonic agitation, we chose 80 °C as the control temperature as it should be sufficient to sustain the propagation step in conventional ATRP15. Again, we did not detect any polymerization (Supplementary Fig. 8). These observations suggest that both sonication and the presence of BaTiO 3 nanoparticles are necessary for the polymerization to occur. Fifth, we reduced the loading of BaTiO 3 nanoparticles in our polymerization mixture to probe further the role of the BaTiO 3 nanoparticles. When we reduced the loading (by weight per cent) of BaTiO 3 nanoparticles in the polymerization mixture, there was a corresponding decrease in the rate of polymerization (Supplementary Fig. 9). This experiment suggests that the nanoparticles directly mediate polymerization. However, they do not have a one-to-one stoichiometric correlation with the polymerization rate. Similarly, when we reduced the sonication ‘on’ time by changing the sonication cycles from a four seconds on/four seconds off mode to a two seconds on/eight seconds off cycle, we observed a dramatic reduction in polymerization rate (Supplementary Fig. 10). The overall polymerization rate was not directly proportional to sonication ‘on’ time. These two observations suggest that the piezoelectric nanoparticles are necessary for the generation of the polymerization activator, but that our current model does not fully explain the mechanism of polymerization. The polymerization activator formed in situ can then sustain polymerization similar to ATRP. The apparent lack of direct correlation between the activation and the polymerization rate has also been observed in conceptually similar ATRP—supplemental activator and reducing agent ATRP and initiators for continuous activator regeneration ATRP27. With these control experiments, we established that BaTiO 3 nanoparticles initiate and control polymerization via the transduction of mechanical energy with a copper species to ensure an ATRP-like mechanism of polymerization.

To understand how the particles mediated the polymerization, we propose two plausible mechanisms by which ultrasound initiates controlled radical polymerization (Fig. 3). The first possible mechanism entails the piezocatalytic reduction of the Cu(II) precursor to form a Cu(I)-based ATRP activator either directly or indirectly through a series of reductions and subsequent comproportionation with Cu(0). The activator for ATRP can also be a Cu(0) species27. Irrespective of the exact nature of activation, an ATRP-like mechanism is set up by the piezocatalytic reduction of Cu(II) precursors. Similarly, the controlled reduction of Cu(II) salts under ultrasound enhanced electrochemical reactions that occur in aqueous solutions28. The second possible mechanism is the transfer of an electron from the particle to another component when the polymerization mixture is sonicated. The newly formed radical can then take part in the initiation of a conventional chain-growth radical polymerization. A similar initiation of radical polymerization under ultrasonic agitation has been reported in polymerizations of methyl methacrylate in aqueous emulsions29.

Figure 3: Potential mechanisms for the ultrasound-mediated generation of ATRP activator. a, Sonication induces piezocatalytic reduction of the Cu(II) complex for the generation of the polymerization activator. The activator, in turn, generates the radical that adds a monomer unit onto the growing polymer. b, Sonication induces the generation of radicals that initiate a radical polymerization. Full size image

To test the possibility that ultrasound generates the polymerization activator, we pre-sonicated a mixture of BaTiO 3 (10 wt%) and Cu(OTf) 2 /Me 6 TREN/Bu 4 NBr (93 mM in DMF) in the absence of the monomer and the initiator. We reasoned that if a polymerization activator capable of initiating polymerization is pre-generated in the reaction mixture, then the initiation process by ultrasonic agitation would be conceptually similar to adding a reductant to a ligand-bound Cu(II)-based ATRP deactivator and we would expect to see polymerization mediated by conventional ATRP15. However, if the polymerization activator were being generated continuously, then we would expect to see polymerization dependent on continuous mechanical activation. After a half-hour pre-sonication step, we added the monomer and the initiator and split the resulting mixture into three portions. One portion was continued to be sonicated, one was stirred at 80 °C in the absence of ultrasound and one was left unagitated at 23 °C. Aliquots from all three portions were analysed for the conversion of monomer into polymer. Figure 4 depicts the results of our experiment. As expected, ultrasound-mediated polymerization continued to proceed in the sonicated reaction mixture, whereas no polymerization occurred in the mixture that was not agitated. We observed a minimal amount of polymerization in the stirred reaction mixture, which suggests any polymerization activators produced by the ultrasonic agitation are not sufficient to sustain the polymerization, and so continuous ultrasonic irradiation is necessary to sustain the polymerization rates we observed previously.

Figure 4: Polymerization kinetics in control experiments reveal that continuous ultrasonic agitation is necessary for a growing polymer chain. The precursor solution containing Cu(OTf) 2 /Me 6 TREN/Bu 4 NBr (93 mM in DMF) and BaTiO 3 nanoparticles (10 wt%) was sonicated for 0.5 h followed by the addition of 1 and 2. The final concentrations of the reaction components were: [Cu(OTf) 2 ]:[Me 6 TREN]:[Bu 4 NBr]:[2]:[1] = 0.04:0.04:0.04:0.04:4 M with 4.5 wt% BaTiO 3 nanoparticles. The reaction mixture was split into three portions and they were sonicated (blue), stirred at 80 °C (red) or held at 23 °C (green). Aliquots from all the portions were analysed by 1H-NMR spectroscopy to determine the monomer conversion. Full size image

Based on these observation, we propose that the following set of events occur in the ultrasound-mediated controlled radical polymerization. The ultrasonic agitation of BaTiO 3 nanoparticles results in a rapid switching between the generation of polymerization activator (Cu(I)Br/L n ) and the polymerization deactivator (Cu(II)Br 2 /L n ). L n here refers to the ligand that stabilize these copper complexes. These two seemingly opposing reactions presumably happen on different surfaces of the nanoparticles. Such simultaneous oxidation and reduction reactions have been observed previously in the context of piezoelectric particles24. This switching then happens in tandem with the conventional ATRP activation/deactivation cycle, which should also occur on the surface of the nanoparticles. The net effect is that ATRP-like polymerization occurs when the activator species is being generated, whereas the polymerization is dormant when the deactivator is being generated—these two states of reaction happen in quick, cyclic successions.

To explore the surface phenomenon further, we performed scanning electron microscopy–energy-dispersive spectroscopy (SEM–EDS) experiments on the nanoparticles as suspensions from the polymerization mixture. The SEM–EDS data are depicted in Supplementary Fig. 10. The lower-magnification SEM images (Supplementary Fig. 10) reveal the tendency of the particles that were not subjected to sonication to cluster (left-hand image) as opposed to the sonicated sample that had a higher degree of particle dispersion. The higher magnification images (Supplementary Fig. 10) further highlight the change the BaTiO 3 nanoparticles undergo during ultrasonic agitation of the polymerization mixture. The particles that were subjected to sonication usually exhibited a darker flat region (right-hand image) that was normally absent from the particles that were not subjected to sonication (left-hand image). The EDS data show that both the samples have similar spectra (a representative trace is presented in Supplementary Fig. 10) with one difference—copper was only detected on particles that were subjected to sonication and that were not clustered. This further supports our hypothesis that the copper-mediated ATRP-like polymerization occurs on a few activated surfaces of the BaTiO 3 nanoparticles.