A well-defined, versatile photoinitiator (salen)Co–CO2CH3 for visible light-initiated living/controlled radical polymerization

A well-defined organocobalt salen complex (salen)Co–CO2CH3 is used as a versatile photoinitiator for visible light-initiated living radical polymerization.


Introduction
Precise control over macromolecular compositions and architectures has long been a challenge in polymer science. The advent of living radical polymerization (LRP) has signicantly contributed to this eld over the last two decades, and gives the possibility to obtain well-dened functional polymeric materials with precisely controlled structures and desired morphologies. So far, based on various strategies to regulate the equilibrium between dormant and active species in LRP, many powerful living radical polymerization technologies have been developed, such as nitroxide-mediated radical polymerization (NMP), 1 atom transfer radical polymerization (ATRP), 2 reversible addition-fragmentation chain transfer polymerization (RAFT), 3 organoiodine-mediated radical polymerization (IRP), 4 and organometallic-mediated radical polymerization (OMRP), 5 etc.
Recently, the modulation of external stimuli to control radical polymerization through regulation of the activation and deactivation processes has been employed as an effective method to synthesize polymers with controlled molecular weights, narrow polydispersities and end group functionality. 6 Of the various physical and chemical stimuli used, including heat, metal catalysts, and electrochemical, 7 mechanical 6a and photochemical stimuli, 6 light-induced LRPs 8-10 are receiving growing attention due to their abundance, renewability and simple operation process. Moreover, light-induced LRPs provide opportunities to prepare novel materials through spatial and temporal control of the polymerization. 11 Development of a versatile LRP catalyst suitable for a broader range of monomers, including both conjugated monomers (acrylates, acrylamides) and unconjugated monomers (vinyl esters), under mild conditions, would be desirable and represent a breakthrough in polymeric material synthesis. 3d, 12 Very recently, we reported the visible light-induced LRP of acrylates and acrylamides mediated by organocobalt porphyrins at room temperature, based on the photo-induced reversible cleavage of the Co-C bond. 13 Structurally closely-related organocobalt salen (salen is a common abbreviation for N 2 O 2 bis-Schiff base bis-phenolates) complexes also have relatively weak Co-C bonds. These complexes are supposed to undergo a similar photolysis process to generate an organic radical that could initiate polymerization and salen cobalt(II), which functions as a persistent metal-centered radical to control polymerization. Compared with cobalt porphyrin, cobalt salen is more desirable for industrial applications due to its convenient synthesis, allowing facile adjustment of electronic and steric effects. 14 Thus, cobalt salen complexes appear to be an ideal candidate to achieve photo-induced LRP. In fact, cobalt salen complexes have been widely used in the hydrolytic kinetic resolution 15 and enantioselective ring opening polymerization of epoxides. 16 However, they have rarely been applied to catalyze living radical polymerization until recently. 17 To the best of our knowledge, there is no report on the success of photo-initiated LRP mediated by a cobalt salen complex.
In this article, we report the rst example of visible lightinitiated LRP of a diverse range of monomers (acrylates, acrylamides and vinyl acetate) (Scheme 1) using a novel, well-dened methoxylcarbonyl-cobalt salen complex [(salen)Co-CO 2 CH 3 , I, salen ¼ N,N 0 -bis(3,5-di-tert-butylsalicylidene)-1,2cyclohexanediamine]. Complex I functions as both initiator and mediator to control the polymerization without any additives, under visible light or sunlight irradiation. Furthermore, the addition of the traditional photoinitiator 2,4,6-trimethylbenzoyl diphenylphosphine oxide (TPO) (Scheme 1) dramatically increases the polymerization efficiency, while maintaining good controllability. Visible light is essential in the initiation step, albeit dispensable during the chain propagation process, which differs from most photo-LRP processes that require continuous irradiation aer the initiation step. The polymer structures were carefully investigated and were demonstrated to maintain the starting organocobalt salen segments at the chain ends. Wellcontrolled diblock copolymers were also easily prepared under mild conditions.
temperature. The photolysis occurred accompanied by a color change of the solution from dark green to red, indicating the formation of (salen)Co(II) (ESI, Fig. 5S and 6S †). Fig. 2 shows the 1 H NMR spectrum before and aer visible light irradiation.
Sharp signals in the region of 1.30-2.10 ppm corresponding to the tert-butyl group of I disappeared aer irradiation for 1 h, which suggests the cleavage of the Co-C bond in (salen)Co-CO 2 CH 3 . Moreover, the resulting CH 3 CO 2 c radical was trapped by excess TEMPO to form TEMPO-CO 2 CH 3 with a yield of 95%. 21 This product is evidenced by the chemical shis at d ¼ 1.05-1.30 ppm corresponding to the methyl protons in TEMPO-CO 2 CH 3 (Fig. 2). The formation of TEMPO-CO 2 CH 3 was also conrmed by ESI-MS ( Fig. 7S †). Simultaneously, broad peaks at À0.49 ppm and downeld in the 1 H NMR (ESI, Fig. 8S †) aer photolysis clearly indicate the formation of the corresponding (salen)Co(II).

Visible light-initiated LRP mediated by (salen)Co-CO 2 CH 3
The polymerization of different acrylates and acrylamides using I as both initiator and mediator at ambient temperature was tested. Three different light sources were used, including a 500 W xenon lamp, a commercial household compact uorescent lamp and direct use of sunlight (Table 1). By using the xenon lamp or the uorescent lamp, irradiation of the dark green solution of complex I with acrylates in benzene at room temperature yielded the corresponding polyacrylates, as evidenced by a color change to yellow (ESI, Fig. 5S †). The polymerizations were all performed with excellent control over molecular weight, with very narrow polydispersities along with the monomer conversions (Table 1, entries 1-6). The experimental molecular weights were all very close to the theoretical values and the GPC traces were monomodal and symmetrical.
Thus, the visible light-initiated polymerization of acrylates mediated by I demonstrated very good control capability over acrylates using a xenon lamp or household light sources. Following the successful polymerization of acrylates, polymerization of acrylamides including DMA, DEA and AMO with I also resulted in polyacrylamides with narrow polydispersities (Table  1, entries 9-15). To avoid gel formation due to high viscosity, the ratio of AMO to I was reduced to 200, and this gave excellent control over molecular weight. Furthermore, direct use of sunlight without ltering has rarely been found in photo c,9j,10f,13b,22 However, in our system, the polymerization of DEA with sunlight irradiation gave PDEA, with a predetermined molecular weight and narrow polydispersity ( Table  1, entry 13). To conrm that polymerization was induced only by I under light, control experiments in the absence of light for the polymerization of tBA and DMA were conducted (Table 1, entries 7-8). No polymers were detected in each case and no color change in solution was observed, indicating no polymerization occurred. Thus, visible light was essential for successful polymerization. To the best of our knowledge, this is the rst example of a well-dened ve-coordinate cobalt salen complex being used to give well-controlled polymerization under visible light irradiation at room temperature.

Visible light-initiated LRP by addition of TPO
Although the polymerizations by direct irradiation of I and different acrylates or acrylamides all gave promising results, the polymerization rates were rather slow. Polymerization of MA required 92 h irradiation to achieve 74% conversion. Switching from reversible termination (RT) to degenerate transfer (DT), by addition of external radicals, could dramatically increase the rate of polymerization. 18b,23 Thus, addition of a traditional photoinitiator to generate excess radicals would be helpful to improve the polymerization efficiency. Indeed, the photopolymerization of MA (600 equiv.) with organocobalt complex I in the presence of TPO (1 equiv.) reached 76% conversion within 8 h at ambient temperature, while 92 h were required for similar monomer conversion without TPO (Table 1, entry 1 vs. Table 2, entry 1). Signicant rate enhancement was also seen for the polymerizations of nBA, tBA, DMA and AMO, and a high degree of controllability remained (Table 1, entries 3-15 vs. Table 2, entries 2-6). More importantly, the (salen)Co-CO 2 CH 3 complex not only could mediate the polymerization of conjugated acrylates and acrylamides under visible light irradiation but also was suitable for the polymerization of typical nonconjugated vinyl acetate (VAc). The bulk polymerization of VAc gave PVAc of narrow polydispersity with excellent control over molecular weight (Table 2, entry 7). However, the polymerization was a little bit slow and a gel was formed aer 39% conversion due to high viscosity. To achieve high conversion and avoid gel formation, the load of VAc relative to I was reduced to 200 equivalents and polymerization was conducted in DMSO-d 6 . In this case, PVAc with 78% conversion was obtained with moderate control over the molecular weight and polydispersity (Table 2, entry 8). Thus, (salen)Co-CO 2 CH 3 is an alternate versatile catalyst to control the polymerization of both "more active" monomers (acrylates and acrylamides) and "less activated" monomers (VAc) under visible light irradiation.

Kinetics and possible mechanism
Reversible termination (RT) and degenerate transfer (DT) are two competing mechanisms that occur in OMRP and cobaltmediated radical polymerization (Scheme 2). 5,18b,24 Dormant species (organocobalt complexes) are the exclusive source of radicals in RT, while the concentration of chain radicals in DT is mainly dependent on the concentration of an external radical source. The rate of polymerization via a DT process approaches the rate of a regular free radical polymerization, which is much faster than that in a RT process. 18b In order to investigate the mechanism, the kinetics of polymerization mediated by I under different conditions (irradiated by Xe lamp, CFL, or addition of TPO) were investigated (Fig. 3a). Mixing I with 1 equivalent of TPO in the photo-induced polymerization resulted in rapid rst order kinetics with a conversion of 58% aer only 4.5 h (Fig. 3a, blue dots). On the other hand, a much slower rate was obtained without addition of TPO (Fig. 3a, black and red dots). The radical concentration was calculated to be $8.8 Â 10 À10 M from the slope of 0.043 h À1 and k p (MA at 298 K) 25 of 13 500 M À1 S À1 when irradiated by the Xe lamp, which was similar to that using the CFL (1.0 Â 10 À9 M), and one order of magnitude smaller than that in the presence of TPO (6.2 Â 10 À9 M). The lower concentration of propagation chain radicals in the absence of TPO arose from the fact that photolysis of organocobalt salen was the sole pathway for the formation of chain radicals, and the resulting cobalt(II) reversibly grabbed the chain radicals.
[(salen)Co-R] eq and [(salen)Co(II)] eq are the stationary state concentrations of dormant and deactivated species during the polymerization. A certain amount of (salen)Co(II) (>5 mol% of the initial concentration of I) was added at the beginning of the polymerization in order to shi the activation-deactivation equilibrium towards the dormant species, so that only a tiny proportion of (salen)Co-CO 2 CH 3 was dissociated during the polymerization. Thus, the equilibrium concentration of [(salen) Co-R] eq and [(salen)Co(II)] eq was assumed to be equal to the initial concentration of [(salen)Co-CO 2 CH 3 ] 0 and [(salen) Co(II)] 0 . To determine the K eq , 8% and 16% excesses of (salen) Co(II) to (salen)Co-CO 2 CH 3 were added before the irradiation, and the polymerization kinetics are shown in Fig. 4. Both the polymerizations gave linear consumption of MA with a decreased polymerization rate due to the addition of excess (salen)Co(II). The radical concentrations were calculated to be 3.56 Â 10 À10 M and 1.83 Â 10 À10 M, respectively. The ratio of [(salen)Co-CO 2 CH 3 ] 0 /[(salen)Co(II)] 0 combined with the radical concentration gave the equilibrium constants K eq ¼ 3.5 Â 10 10 M À1 and 3.4 Â 10 10 M À1 , respectively. The K eq was three orders of magnitude larger than that measured for the thermal polymerization of MA at 60 C mediated by (salen)Co(II) (K eq $ 2.4 Â 10 7 M À1 ), 17b and close to the value for thermal polymerization by the cobalt porphyrin system (K eq $ 8.7 Â 10 9 M À1 ). 26 With such a large K eq , when TPO was added, the major radical source would mainly be external TPO rather than organocobalt (salen) Co-CO 2 CH 3 . Thus, the visible light-initiated polymerization mediated by complex I without TPO is suggested to mainly undergo a RT process, while a DT process is predominant in photopolymerization in the presence of TPO. Furthermore, the addition of TPO had a negligible inuence on a end delity because only a tiny amount of polymer chain derived from the TPO segment was found in the MALDI-TOF-MS detection (ESI, Fig. 9S & 10S †). The living character of the visible light-initiated radical polymerization of MA was further demonstrated by the evolution of molecular weight versus conversion under different conditions (Fig. 3b). Gel permeation chromatography analysis revealed a linear increase in molecular weight with respect to conversion of MA under all conditions. The polydispersity index was small (M w /M n < 1.25) from the early stages of polymerization up to high MA conversions in all cases. The GPC traces for photopolymerization under these conditions were all monomodal and symmetrical throughout the whole process (ESI, Fig. 11S-13S †). Furthermore, it is worth mentioning that in the thermal polymerization of MA mediated by (salen)Co(II), there was a signicant deviation between measured and calculated molecular weights, because the relatively small K eq (2.4 Â 10 7 M À1 ) under thermal conditions resulted in incomplete conversion of cobalt(II) complexes to organocobalt species. 17b However, the large K eq of this photosystem at room temperature maintained nearly all the cobalt species as organocobalt complexes and fullled the ideal assumption of one polymer chain per cobalt. Therefore, the experimental molecular weights are all close to theoretical values, indicating high initiation efficiency for this visible light-initiated polymerization.

Effect of visible light
Surprisingly, we found that aer the initiation stage, the rate of polymerization was approximately the same with/without continuing irradiation. For example, aer passing the initiation  stage, one sample was protected from light by covering with aluminum foil and the other was continually irradiated under a light intensity of 3 mW cm À2 , however, both of these two samples of DMA gave similar conversions (45% and 46%).
This observation denitely indicates that polymerization also occurs at a similar rate in the dark aer initiation. This is interesting since visible light is essential for successful polymerization (Table 1, entry 8).
To assess the inuence of visible light on the polymerization process at ambient temperature, MA was chosen as a model monomer and kinetic studies were conducted under three different sets of conditions. The molar ratio of MA to I was set at 200/1 each time, one sample was reacted under continuous Xe lamp irradiation and the second sample was subjected to a periodic light on-off process every 12 hours. The third sample was irradiated for the rst 12 hours and then stored in the dark. The polymerization rate under continuous irradiation was slightly faster than that in the dark, and no signicant rate difference was observed under continuous and periodic irradiation within 40 h (Fig. 5). So visible light of 3 mW cm À2 has a negligible inuence over the polymerization rate aer initiation. However, aer a fairly long reaction time (more than 40 h), the polymerization under visible light irradiation showed a little bit higher monomer conversion than that in the dark (Fig. 5 &  ESI Fig. 14S †). These results indicate that the Co-C bonds in (salen)Co-PMA and (salen)Co-PDMA could undergo homolysis to give propagating radicals even in the dark at ambient temperature, although I did not initiate polymerization in the dark (Table 1, entries 7-8). The bond dissociation energy of the Co-C bonds in Co-C(O)OR was much stronger than in Co-CHR 1 R 2 due to the sp 2 -hybridized carbon in I.

Polymer structure analysis
The chain growth process occurs through monomer insertion into the cobalt-carbon bond in the dormant organocobalt species. Thus, in the nal polymer structure, the a ends were anticipated to be CO 2 CH 3 from the (salen)Co-CO 2 CH 3 , while the u ends should be Co(salen). In order to investigate the a ends of the polymer chains, a both deuterated and 13 C-labeled organocobalt complex, (salen)Co-13 CO 2 CD 3 , was synthesized under similar conditions (ESI †). It was characterized by 1 H NMR, 13 C NMR, 2 D NMR and ESI-MS. The 2 D NMR spectrum gave a clear peak at À1.315 ppm, corresponding to the deuterium atoms of the methyl group in (salen)Co-13 CO 2 CD 3 (Fig. 6a). Aer polymerization of MA, the isotopic-labeled methyl group in PMA shied to À1.455 ppm, corresponding to the deuterium atoms at the a ends. The structure of PMA synthesized from (salen)Co-13 CO 2 CD 3 was further investigated by 13 C NMR and compared with PMA prepared from regular (salen)Co-CO 2 CH 3 (Fig. 7). The PMAs formed were highly linear, as no signal was detected in the ranges of d( 13 C) ¼ 37-40 and 47-49 ppm that are characteristic of acrylate branching. 27 The two spectra were nearly identical except for the signal at 172 ppm, which was assigned to the 13 C-labeled carbon atom in the ester group from the starting (salen)Co-13 CO 2 CD 3 . Both the 2 D NMR and 13 C NMR detections demonstrated that the a ends of the formed polymers mediated by organocobalt salen complexes were CO 2 CH 3 from the (salen)Co-CO 2 CH 3 . 13b To analyze the structure of the u chain ends, GPC traces of PMA were measured by both refractive index and UV-visible detectors ( Fig. 8 & ESI, Fig. 15S †). Polymers with the (salen)Co chromophore showed a strong UV-visible absorption band around 360 nm. The GPC traces of typical PMA samples measured by RID and UV-vis detectors were nearly the same, except for a delay in the RID signal due to the series connection of the equipment. The elution curves were narrow and symmetrical, indicating that each polymer chain contained one cobalt complex at the u end.
The PMA obtained in our system was further characterized by 1 H NMR (Fig. 9). The chemical shis at 6.9 ppm, 7.4 ppm, and 7.8-8.0 ppm clearly demonstrate the presence of the (salen)Co group in the polymer chain. Furthermore, the molecular weight calculated from the 1 H NMR analysis was 9530 g mol À1 , which ts very well with the theoretical value (9270 g mol À1 ) and experimental molar mass (10 300 g mol À1 ), indicating high initiator efficiency.

Polymer end group functionalization
Photo-induced dioxygen insertion into a Co-C bond has been reported previously 28 and Co(III)-alkylperoxo complexes are important intermediates in cobalt-catalyzed hydrocarbon oxidation reactions. 29 Our strategy for the modication of the u end of the polymers obtained is illustrated in Scheme 3. As investigated in a periodic irradiation experiment, the Co-C bond in (salen)Co-PMA can undergo homolysis to give PMAc and (salen)Co(II). Schiffbase cobalt(II) complexes are well-known to bind dioxygen reversibly. 30 Thus the photolysis of (salen)Co-PMA in the presence of dioxygen resulted in the formation of a Co ( Fig. 10a shows the MALDI-TOF-MS spectrum of the OH groupfunctionalized PMA sample (M n,GPC ¼ 3450, M w /M n ¼ 1.16) with a series of molecular ion peaks, regularly separated by the molar mass of the MA monomer. Experimental isotopic mass values of the main peak series were equal to those expected for the PMA with CH 3 CO 2 À at the a end and a hydroxyl group at the u end, plus a sodium ion from externally-added salt for ionization, as shown in the upper part of Fig. 10b. More specically, the observed isotopic pattern (m/z) of the main series at m/z ¼ 1561.8040 g mol À1 indicates the assigned species with a DP (degree of polymerization) of 17, and matches well with the simulated pattern, m/z ¼ 1561.6310 g mol À1 (Fig. 10c). The observed isotopic pattern also agrees well with the simulation result.
To gain more evidence for the modication process (Scheme 3), an isotopic labeling experiment was conducted by reaction of (salen)Co-PMA with 18 O 2 . The resulting PMA-18 OH was also characterized by MALDI-TOF-MS (Fig. 11). The end-capped sample also gave a series of molecular ion peaks separated by MA units. The absolute mass of each peak clearly indicates the 18 O-labeled structure. Thus, the isotopic labeling and MALDI-TOF-MS analysis both suggest that the end group has been successfully functionalized by an OH group, which would be highly useful for further application.

Synthesis of block copolymers
To further probe the living nature of this visible light-initiated system, as well as to provide additional evidence for the presence of cobalt salen complexes at the u chain ends, block copolymers were prepared by sequential polymerization of acrylates and acrylamides. MA and DMA were chosen as model monomers. Three (salen)Co-PMAs with narrow molecular weight distributions were prepared by photopolymerization under different conditions (irradiated by Xe lamp or CFL, or with addition of TPO) and used as macroinitiators (ESI, Fig. 17S-19S †). Aer addition of fresh DMA, all the polymerizations under these conditions proved to be well-behaved processes leading to the desired PMA-b-PDMA with predetermined molar masses and low polydispersities (Fig. 20S †). The GPC traces all showed substantial shi with negligible residual macroinitiators aer block copolymerization. The successful block copolymer synthesis further indicated efficient reversible activation of the cobalt-carbon chain end and minimal termination under all polymerization conditions.

Conclusions
In summary, we have demonstrated that (salen)Co-CO 2 CH 3 acts as a versatile photoinitiator to mediate the polymerization of acrylates, acrylamides and non-conjugated VAc. All the polymerizations showed controlled behavior, evidenced by the  formation of polymers with predetermined molecular weights and low polydispersities. The equilibrium constant ((salen) Co(II) + Rc # (salen)Co-R, K eq ¼ [(salen)Co-R] eq /([(salen) Co(II)] eq Â [Rc])) between the organo radical, cobalt(II) and the organocobalt(III) species was determined to be 3.5 Â 10 10 M À1 . Addition of 1 equivalent of TPO dramatically increased the polymerization rate while maintaining high controllability, indicating a switch from RT polymerization to a DT mechanism. Visible light was found to be essential for the initiation but showed a negligible effect during propagation, which might result from the difference in the cobalt-carbon bond energy in (salen)Co-C(sp 2 ) and (salen)Co-C(sp 3 ). Polymer structure analysis demonstrated the presence of (salen)Co segments in the polymer chain u end and -CO 2 CH 3 segments in the polymer chain a end. Efficient block copolymer synthesis under Xe lamp, CFL, and with addition of TPO, further conrmed the versatile capability of this novel complex in controlled/living radical polymerization. The convenient synthesis of salen complexes offers a unique and appreciable approach for the preparation of functional polymeric materials.