Mechanistic Insight into the Photocontrolled Cationic Polymerization of Vinyl Ethers

Abstract

The mechanism of the recently reported photocontrolled cationic polymerization of vinyl ethers was investigated using a variety of catalysts and chain-transfer agents (CTAs) as well as diverse spectroscopic and electrochemical analytical techniques. Our study revealed a complex activation step characterized by one-electron oxidation of the CTA. This oxidation is followed by mesolytic cleavage of the resulting radical cation species, which leads to the generation of a reactive cation–this species initiates the polymerization of the vinyl ether monomer–and a dithiocarbamate radical that is likely in equilibrium with the corresponding thiuram disulfide dimer. Reversible addition–fragmentation type degenerative chain transfer contributes to the narrow dispersities and control over chain growth observed under these conditions. Finally, the deactivation step is contingent upon the oxidation of the reduced photocatalyst by the dithiocarbamate radical concomitant with the production of a dithiocarbamate anion that caps the polymer chain end. The fine-tuning of the electronic properties and redox potentials of the photocatalyst in both the excited and the ground states is necessary to obtain a photocontrolled system rather than simply a photoinitiated system. The elucidation of the elementary steps of this process will aid the design of new catalytic systems and their real-world applications.

Graphical Abstract

INTRODUCTION

The wealth of photoredox reactions developed during the past decade has offered a blank canvas on which to design “living” polymerizations in which polymer chain growth can be controlled at will by the intensity or wavelength of light.¹ The intrinsic resolution of light enables unparalleled spatial control over these polymerizations, a factor that may prove desirable in a wide array of settings. Photocontrolled variants of living radical polymerizations, including atom-transfer radical-polymerization (ATRP),² organotellurium-mediated radical polymerizations (TERP),³ and reversible addition–fragmentation chain transfer (RAFT) polymerizations,⁴ have already demonstrated usefulness with a variety of monomers. This unprecedented command over polymeric architectures inspires numerous applications from the complex patterning of surfaces⁵ to the synthesis of sequence-controlled polymers.⁶

In an effort to expand the range of monomers capable of light-regulated polymerization, we recently reported a photocontrolled cationic polymerization of vinyl ethers.⁷ As shown in Figure 1, our initial reaction conditions capitalized on the high potency of the oxidative photocatalyst (PC) 2,4,6-tris(4-methoxyphenyl)pyrylium tetrafluoroborate (1a) when combined with chain-transfer agents (CTAs) 2a or 2b as a means to control the cationic polymerization of various vinyl ethers (3a–e). Polymers with predictable number-average molar mass (M_n) and narrow dispersity (Đ) values were obtained by modulating the CTA-to-monomer ratio. These conditions reversibly generate cationic chain ends that propagate in the presence of light and are deactivated in the dark, thus enabling the temporal regulation of chain growth through light irradiation. In contrast to previously reported radical systems that rely on the chain end reduction, this cationic process requires oxidation of the terminal group (Figure 2), which suggests that the transformation proceeds through a mechanistic pathway that is fundamentally distinct from those of its radical counterparts. A deeper understanding of the mechanism of this unique polymerization is critical for applications to a broader range of monomers and, more important, the design of more intricate systems.

Figure 1.

Cationic polymerization of vinyl ethers regulated with blue light.

Figure 2.

Photocontrolled radical polymerization vs cationic polymerization: two mechanistically distinct pathways. PC = photocatalyst; Z = Br, S₂CR, S₂CSR′, etc.

Our current mechanistic hypothesis is depicted in Figure 3. Single-electron transfer (SET) from the CTA (or polymer chain end) to the excited PC generates a radical cation species (step I) that undergoes mesolytic cleavage, leading to the formation of an active cationic chain end and a stable dithiocarbamate or trithiocarbonate radical (step II). The resulting cationic species is then engaged in a RAFT-type degenerative chain transfer (step III). Finally, one-electron reduction of the stable radical by PC^• turns over the PC while producing a dithiocarbamate (or trithiocarbonate) anion, which caps the polymer chain end and deactivates chain growth (step IV).

Figure 3.

Proposed mechanism for the light-regulated polymerization of vinyl ethers via a photocatalyst (PC) and chain-transfer agent.

To probe the elementary steps, we used a combination of spectroscopies–namely, time-resolved photoluminescence, UV–visible absorption, and electron spin resonance (ESR)–and electrochemistry. The efficiencies of the polymerization were also explored with a number of PCs and CTAs. The body of data presented herein reveals much about the processes at play and should serve as a platform for the development of related transformations and translational technologies.

RESULTS AND DISCUSSION

Exploration of the Photocatalytic System

Building on our initial report, we investigated several factors that could influence the polymerization kinetics and control in these systems, such as the choice of PC, CTA, and solvent. All reactions were kept at ambient temperature under a constant stream of air and placed at a fixed distance from the light. Isobutyl vinyl ether (IBVE, 3a) was used as a reference monomer.

Influence of PCs on Polymerization Rates

Our mechanistic hypothesis states that the rates of activation (step I) and deactivation (step IV) for a given CTA should be governed primarily by the nature of the PC. In an attempt to pinpoint the key features of an efficient PC for this system, we prepared an array of pyrylium derivatives (1a–e; Figure 4; see Supporting Information [SI] for details).^8–11 The para substituent of each aryl group as well as the heteroatom of the pyrylium core are easily tuned through synthesis, which allows for simple modulation of the redox and photophysical properties of the catalyst (Table 1). Other strongly oxidizing photoredox catalysts were used for this study as well, including acridinium 4 and ruthenium or iridium complexes 5 and 6.^12–17

Figure 4.

Photocatalysts used for the study.

Table 1.

Redox Potentials and Photophysical Properties of Photocatalysts

PC	EPC+*/PC• (V vs SCE)	EPC+/PC• (V vs SCE)	ε450 (L·mol−1·cm−1)	Φf	ΦISC
1a	+1.84a,8	−0.508	67 300b	0.958	0.038
1b	+2.55a,8	−0.328	6350b	0.588	0.428
1c	+2.239	−0.559	32 100b	–	–
1d	+2.4910	−0.0310	30 100b	0.3311	0.6711
1e	+2.45a,8	−0.198	7330b	0.038	0.948
4	+2.18a,8	−0.498	403012	0.0358	0.388
5	+1.4513	−0.8013	900014	0.03415	0.6815
6	+1.2113	−1.3713	176016	0.6816	–

^aPotential of the singlet excited state.

^bExperimentally determined in dichloromethane (see Supporting Information). Reference numbers for each reported physical data are indicated.

The results from the polymerization of 3a using CTA 2a, a PC, and blue LEDs are summarized in Table 2. Notably, only pyrylium derivatives afforded polymers (Table 2, entries 1–7), all of which had M_n’s close to theoretical values and Đ’s of approximately 1.2. In the dark, no polymerization was observed (Table 2, entry 11).⁷ Phenyl and tolyl derivatives 1b and 1c (Table 2, entries 2–5) demonstrated the highest polymerization rates, reaching full conversion after less than 10 min compared with several hours for 1a. This outcome is likely due to the higher oxidation potentials of the PCs ( ${E°}_{\mathbf{1}{\mathbf{b}}^{\ast }/\mathbf{1}{\mathbf{b}}^{•}}=+2.55\text{ V and }{E°}_{\mathbf{1}{\mathbf{c}}^{\ast }/\mathbf{1}{\mathbf{c}}^{•}}=+2.23\text{ V}$, whereas ${E°}_{\mathbf{1}{\mathbf{a}}^{\ast }/\mathbf{1}{\mathbf{a}}^{•}}=+1.84\text{ V vs SCE}$). The relatively slower rate observed with thiopyrylium 1e compared with that of 1b despite the similar redox potentials and molar attenuation coefficients of these compounds may be linked to a low quantum yield of fluorescence (Φ_f = 0.03), which implies that the singlet excited state of the pyrylium contributes mainly to the electron transfer step. However, additional experiments are required to confirm the exact role of both the singlet and the triplet states in this reaction.¹⁸ Finally, the lower reactivity of tribromo congener 1d compared with that of the other pyryliums is likely due to poor solubility in dichloromethane (Table 2, entry 6).

Table 2.

Selected Results of Cationic Polymerization of Isobutyl Vinyl Ether (3a) with Various Photocatalysts

entrya	catalyst	time (min)	Mn (exp) (kg/mol)	Mn (theo) (kg/mol)f	Ð
1	1a	480	10.7	10.1	1.19
2	1b	10	10.5	10.1	1.23
3	1bb	10	10.2	10.1	1.19
4	1c	10	11.1	10.1	1.17
5	1cb	30	10.7	10.1	1.17
6	1d	480	10.5	10.1	1.18
7	1e	300	10.3	10.1	1.21
8	4	–e	–	10.1	–
9	5	–e	–	10.1	–
10	6	–e	–	10.1	–
11	1ad	–e	–	10.1	–

^aReaction conditions: 3a (1 equiv), photocatalyst (PC; 0.02 mol %), and 2a (0.01 equiv) at room temperature in dichloromethane with blue-light-emitting diode irradiation (9 W bulb).

^bUsing 0.01 mol % of PC.

^dNo irradiation.

^eNo conversion was observed after 720 min.

^fConversion was determined by ¹H NMR using benzene as an internal standard.

Influence of the CTA on Chain Growth Control

Because 2a and 2b play the dual role of initiator and CTA, we hypothesized that the structure of the CTA may profoundly affect the photocontrol of the process.¹⁹ We therefore synthesized a library of CTAs (2c–f) with various groups appended to the dithiocarboxylic core. We also synthesized one CTA (2g) prepared from 4-methoxystyrene rather than IBVE (Figure 5). We anticipated that xanthate 2c, dithioesters 2d and 2e, and pyrrolidinone dithiocarbamate 2f would show electronic properties distinct from those of 2a and 2b, a factor that should impact the rate of each step in the mechanism.

Figure 5.

Library of chain-transfer agents used for this study.

In our previous study,⁷ we found that, compared with 2a, CTA 2b was applicable to a wider monomer scope and delivered polymers with only a slightly broader Đ. However, the preparation of 2a with greater purity enabled the controlled polymerization of ethyl vinyl ether (EVE; Table 3, entry 2), n-propyl vinyl ether, and n-butyl vinyl ether (see SI). Polymerization attempts with 2-chloroethyl vinyl ether and 2a resulted in no conversion (see SI), and 2b is still required for this monomer. Consequently, the purity of the CTA is paramount to the polymerization of less reactive monomers.

Table 3.

Selected Results of the Cationic Polymerization of Isobutyl Vinyl Ether (3a) and Ethyl Vinyl Ether (3b) with Various Chain-Transfer Agents

entrya	CTA	monomer	Mn (exp) (kg/mol)	Mn (theo)b (kg/mol)	Ð
1	2a	3a	9.11	10.1	1.16
2	2a	3b	7.39	7.3	1.12
3	2b	3a	10.1	10.1	1.43
4	2b	3b	8.20	7.3	1.35
5	2c	3a	15.1	10.1	1.98
6	2c	3b	10.6	7.3	1.89
7	2d	3a	15.0	10.1	2.10
8	2d	3b	10.9	7.3	2.11
9	2e	3a	9.43	10.1	1.49
10	2e	3b	7.23	7.3	1.47
11	2f	3a	12.9	10.1	2.18
12	2f	3b	10.4	7.3	2.21
13	2g	3a	74.9	10.1	3.76
14	2g	3b	40.5	7.3	3.18
15	-	3a	52.2	–	3.96

^aReaction conditions: 3a (1 equiv), 1a (0.01 mol %), and chain-transfer agent (0.02 equiv) at room temperature in dichloromethane with blue-light-emitting diode irradiation (ca. 12 W LED strip) for 360 min.

^bConversion was determined by ¹H NMR using benzene as an internal standard.

The polymerization of IBVE (3a) or EVE (3b) using CTAs 2c, 2d, and 2f yielded macromolecules with M_n’s mostly in agreement with theoretical calculations but with broader Đ’s (1.9 < Đ < 2.2; Table 3, entries 5–8 and 11–12). The experimental M_n’s of xanthate 2c and dithioester 2d were higher than predicted with both monomers (Table 3, entries 5–8). By contrast, CTA 2e enabled control similar to that offered by 2b (Table 3, entries 9 and 10). Subtle increase of the electron donating character of the aryl group therefore has a crucial effect on the control over the chain growth. The CTA derived from 4-methoxystyrene (2g) resulted in uncontrolled polymerization with both monomers (Table 3, entries 13 and 14), which indicates that the monothioacetal structure is critical. Notably, uncontrolled polymerization occurs in the absence of CTA (Table 3, entry 15), which suggests that the direct activation of the monomer is a possible background pathway. Based on the aforementioned results, electron-rich X groups (see Figures 1 and 3 for the notation) such as nitrogen- and sulfur-containing moieties should be favored in the design of future CTAs. Moreover, only CTAs derived from vinyl ethers exhibited good control over chain growth.

Suitable Solvents for the Polymerization.²⁰

Pyrylium salts have poor solubility in hydrocarbon solvents; therefore, benzene, toluene, and small alkanes are unsuitable for this methodology. Similarly, the high oxidizing strength of excited pyryliums precludes the use of solvents such as tetrahydrofuran (THF). Polymerizations in acetonitrile yielded well-controlled polymers, but poly(IBVE) started to phase-separate at an M_n of ∼5 kg/mol. A mixture of dichloromethane and acetonitrile prevented this issue. Last, almost no polymerization was observed in nitromethane, an outcome that agrees with the reported low propagation rates of cationic polymerization of vinyl ethers in this solvent.²¹

Initial One-Electron Oxidation, a Critical Step for Activation

The elucidation of the electron transfers among species in solution is key to understanding the activation of the cationic process (step I). Our previous studies revealed that the fluorescence of 1a* in a steady-state experiment is quenched by both CTA 2a and IBVE (3a).⁷ However, this result does not distinguish between static and dynamic quenching (Figure 6), and the latter is the only quenching at play for an electron transfer.²² Therefore, time-resolved fluorescence spectroscopy was used in conjunction with ESR and electrochemical analysis to analyze electron transfer in these reactions.

Figure 6.

Static vs dynamic quenching of a photocatalyst (PC) by a quencher (Q).

Time-Resolved Fluorescence Spectroscopy

The fluorescence decay for PC 1a using 440 nm pulsed excitation was measured with various amounts of potential quenchers, CTA 2a, and monomer 3a (Figures 7, 8, and S14). A clear decrease in the PC* lifetime (τ) was observed with increasing concentrations of both 2a and 3a. More precisely, the relationship between τ and the concentration of both quenchers (Figure 8) follows eq 1, which is directly derived from the Stern–Volmer equation, where τ₀ is the fluorescence lifetime of catalyst without quencher, k_q is the bimolecular quenching constant, and [Q] is the concentration of quencher.²²

$$\frac{{\tau }_0}{\tau }=1+k_q{\tau }_0[Q]$$ (1)

Figure 7.

Fluorescence decay of 1a after a 440 nm pulsed excitation: (A) For various concentrations of chain-transfer agent 2a. (B) For various concentrations of isobutyl vinyl ether (3a).

Figure 8.

Linear relationship between the fluorescence lifetime of photocatalyst 1a and the concentrations of 2a (red) and 3a (blue).

This behavior is consistent with collisional quenching occurring with both the CTA and the monomer. However, calculations of bimolecular quenching constants (k_q’s) suggest that the former is a more efficient quencher (k_q = 7.52 × 10⁹ M⁻¹· s⁻¹) than 3a (k_q = 1.23 × 10⁸ M⁻¹·s⁻¹) by nearly 2 orders of magnitude. The quenching rate of CTA 2a is near 10¹⁰ M⁻¹·s⁻¹, which is characteristic of a diffusion-controlled process.²² These observations suggest that SET can indeed occur between the singlet excited PC and either the CTA or the monomer. Transfer to the CTA is the more favorable pathway (k_q = 7.52 × 10⁹ M⁻¹·s⁻¹); however, in the early stages of polymerization, the direct oxidation of the highly concentrated IBVE is feasible (Table 3, entry 15).

Electropolymerization of IBVE

The spectroscopic evidence discussed above agrees with the measured potentials corresponding to the onset of oxidation of both CTA 2a ( ${E°}_{\mathbf{2}\mathbf{a}/\mathbf{2}{\mathbf{a}}^{•+}}=+0.98\text{ V vs Ag}/{\text{Ag}}^{+}[+1.19\text{ V vs SCE}]$) and 3a ( ${E°}_{\mathbf{3}\mathbf{a}/\mathbf{3}{\mathbf{a}}^{•+}}=+1.25\text{ V vs Ag}/{\text{Ag}}^{+}[+1.46\text{ V vs SCE}]$) (Figure 9), the latter being more difficult to oxidize. Nevertheless, the oxidation of 3a could potentially lead to uncontrolled polymerization, as seen when no CTA is added (Table 3, entry 15).²³ The absence of this background reaction when a CTA is used and at low PC loading can be rationalized by several electronic considerations. First, compared to 3a, CTA 2a is oxidized more readily by catalysts 1. Second, if the oxidation of 3a still occurs, the 270 mV difference in potentials should allow for a second electron transfer from oxidized 3a to CTA 2a.

Figure 9.

Cyclic voltammogram of 2a (red), 3a (blue), and a combination of 2a and 3a (black) using a platinum microelectrode (12.5 μm).

To assess whether the direct oxidation of CTA 2a leads to polymerization, we obtained a cyclic voltammogram of 3a with and without 2a (Figure 9). The absence of a reduction wave in both cases likely reflects the irreversibility of the reaction and, potentially, the passivation at the electrode due to the growth of poly(IBVE). Compared to IBVE alone, the onset of oxidation with 2a lies ca. 300 mV lower, which suggests that the polymerization of 3a can be effected at a lower potential (ca. +0.8 V vs Ag/Ag⁺) when CTA 2a is present. Indeed, the polymerization of 3a could be initiated at an applied voltage of +0.8 V vs Ag/Ag⁺ in the presence of 2a (see SI), whereas no polymer was isolated in the absence of 2a under the same conditions. This result supports the conclusion that the direct oxidation of CTA 2a is the major pathway for step I and that a two-step oxidation via an oxidized monomer species is likely only a minor concurrent pathway.

ESR and UV–visible Characterization of Pyranyl Radicals 1a^• and 1b^•

ESR spectroscopy has proven to be a powerful tool for the observation of free radical species in similar polymerization systems;^18,24 moreover, both 1a^• and 1b^• have previously been characterized using this technique,^25,26 which prompted the use of ESR as a means to track our postulated electron transfers. Samples containing a mixture of CTA 2a and PCs 1a or 1b were monitored during steady-state irradiation with blue LEDs. The beginning of irradiation coincided with the appearance of an ESR absorption signal with both PCs (Figure 10), an observation characteristic of stable radical intermediates. With the combination of 1b and 2a, a hyperfine splitting structure was observed, which was expected due to the numerous aromatic protons adorning the pyrylium core. This hyperfine coupling (hfc) structure could be reproduced through simulation (Figure S15A) and was in complete agreement with previously reported hfc constants for that compound.²⁵ Moreover, the formation of 1b^• was confirmed with UV–visible absorption spectroscopy (Figure S21A).^25b By contrast, no hyperfine splitting structure was observed for the analogous system containing 1a and 2a, even when a low-amplitude modulation was used (M = 0.08 G).²⁷ However, this absence of hyperfine structure was noted by Kawata and colleagues²⁶ and could be due to a broadening induced by the Heisenberg exchange effect.²⁸ To unambiguously attribute the signal arising from the irradiation of 1a and CTA 2a to pyranyl radical 1a^•, we used THF, a known sacrificial electron donor,^25,26 in lieu of CTA 2a (Figure S16). A comparison of these results with the spectrum in Figure 10 corroborates the suggested formation of pyranyl radical 2a in the reaction conditions.

Figure 10.

Electron spin resonance spectrum of 1a with 2a (red) and 1b with 2a (black) under steady-state 450 nm irradiation.

Having confirmed the one-electron oxidation of CTA 2a by excited PCs 1a or 1b, we turned our attention to the putative electron transfer between 1a (or 1b) and vinyl ether 3a. Steady-state irradiation of a mixture of 1a and 3a or 1b and 3a revealed the formation of a long-lived radical species (Figure 11). A comparison of these results to the traces in Figure 10 revealed striking differences. In particular, these new radical species have broader signals and different hyperfine splitting structures. Furthermore, the ESR spectra showed a broadening of the line width over the course of the acquisition, which may be caused by the increase in viscosity during the polymerization.²⁹ Considered together, these results suggest that radical species other than 1a^• and 1b^• are created in the presence of 3a. However, the complex hyperfine splitting structure of these new radical species indicates that the adducts contain a pyrylium core structure.

Figure 11.

Electron spin resonance spectrum of 1a with IBVE (3a; red) and 1b with 3a (black) under steady-state 450 nm irradiation.

Several hypotheses may explain the discrepancies between the spectra in Figures 10 and 11. First, after the initial photoinduced electron transfer, radicals 1a^• or 1b^• may react with 3a to generate a new adduct. Second, the high concentration of IBVE may induce conformational change through noncovalent bonding, thereby leading to a nonequivalent set of hfc’s via the loss of C2 symmetry. To distinguish between these possibilities, we added 3a to a sample containing 1b^•, which was pregenerated using THF as an electron donor. No change was observed in either the hyperfine structure or the intensity of the signal after 3a addition (Figure S20), which indicates that no reaction occurred between 1b^• and 3a. Thus, we investigated the potential formation of a donor–acceptor complex between ground state 1a (or 1b) and the electron-rich monomer (3a). Equilibrium 1 illustrates this hypothesis in the case of 1a (Figure 12).

Figure 12.

Formation of a donor–acceptor complex between 1a and 3a: UV–visible spectra of 1a with various concentrations of 3a.

The results of careful UV–visible absorption spectroscopy of 1a and 1b with various amounts of 3a revealed a slight shift of the local absorption maximum above 450 nm (Figure 12). This new band at 487 nm was attributed to the absorption of [1···3a]⁺, and the equilibrium constants K_DA were estimated to be 0.06 M (for 1a) and 0.19 M (for 1b) according to the Benesi–Hildebrand method (Figure S22).³⁰ Nicewicz and co-workers have previously demonstrated the impact of similar donor–acceptor complexes on the dynamics of alkene oxidation by acridinium PCs,¹² and we expect that, similarly, the formation of [1···3a]⁺ significantly influences the kinetics of electron transfer. Consistent with this finding, a spectrum that has good correlation with the experimental ESR signal in Figure 11 could be simulated by adding two equivalent protons to the spin system of 1b^• (Figure S19).³¹ Finally, ESR analysis of the complete polymerization system–namely, 1a (or 1b), CTA 2a, and monomer 3a–produced spectra similar to those in Figure 11 (Figures S17 and S18).

Exploration of the Mesolytic Cleavage

The second key step of our mechanism involves the cleavage of the radical cation arising from CTA or chain end oxidation into a dithiocarbamate radical and a propagating cationic chain. Vinyl ethers are known to homopolymerize under cationic conditions, but the analogous process has not been observed under radical conditions,³² which substantiates the fragmentation pattern described above. Furthermore, when methyl methacrylate was added to the polymerization mixture, no poly(methyl methacrylate), a product expected from alkyl radical formation, was isolated.³³

Significantly, when blue light was shone on the reaction of CTA 2a and catalyst 1a in a ratio typical of the polymerization, the only isolated products were thiuram disulfide 7 and starting material 2a (see Figure 13A and SI). Compound 7 likely arises from radical recombination after mesolytic cleavage. Indeed, one-electron oxidation of sodium dithiocarbamate affords thiuram disulfide 7 in excellent yields with various oxidants such as iodine (see SI).³⁴

Figure 13.

(A) Isolation of thiuram disulfide 7 via putative mesolytic cleavage. (B) Crossover experiment with disulfides 7 and 8.

Thiuram disulfides are commonly used as CTAs for radical RAFT polymerization and as reagents for the vulcanization of rubber.³⁵ Notably, when substituted for CTA 2a, disulfide 7 affords no control over the polymerization (Table S7), a result that agrees with the proposed cationic process. Disulfide 7 was not detected during NMR analysis of aliquots obtained over the course of the polymerization reactions, which may be attributed to the detection limits of NMR spectroscopy or the rapidity of the reduction of the radical or disulfide dimer (step IV, vide infra). Moreover, a crossover experiment with disulfides 7 and 8 indicated that 7 is in equilibrium with the radical dithiocarbamate in the reaction conditions (Figure 13B), which agrees with the results of previous studies.¹²

Is a RAFT Equilibrium Occurring?

In their initial report of a cationic RAFT process, both Kamigaito^19b,36 and Sugihara³⁷ invoked a degenerative chain transfer to account for the observed control over chain growth. However, from a mechanistic standpoint, our photoredox process differs in that CTA 2a plays the role of both initiator and CTA, as shown by steps I, II, and IV. From this proposed catalytic cycle, it is unclear whether the RAFT equilibrium (step III) is truly necessary to achieve control over chain growth. Therefore, we devised experiments to ascertain the presence or absence of such equilibrium in this polymerization. Quantum yields of the polymerization were estimated through actinometry, and temporal control was investigated with PCs 1a and 1b.

Quantum Yields of the Polymerization

Using the well-studied potassium ferrioxalate actinometer (see SI, Figure S23, for more details),³⁸ the quantum yields of polymerization were estimated to be approximately 6 monomer additions per photon absorbed for PC 1a (0.02 mol %) and approximately 35 monomer additions/photon for 1b (0.01 mol %). This difference was anticipated based on the higher polymerization rate measured for 1b compared with that of 1a (Table 2). By contrast, photoinitiated cationic polymerizations are characterized by higher quantum yields such as 200 monomer additions/photon.³⁹ However, the fact that both quantum yields are well above unity suggests that a chain-degenerative mechanism, as shown in step III, is likely occurring, because narrow Đ’s and predictable M_n’s were obtained for all pyrylium PCs.

On/Off Experiments with PCs 1a and 1b

The stark differences in polymerization rates and quantum yields between PCs 1a and 1b led us to investigate whether temporal control over chain growth is possible with the latter. A reaction mixture containing monomer 3a, catalyst 1a (or 1b), and CTA 2a was exposed to light and then stirred in the dark for the same time period and re-exposed to light (Figure 14). Conversion and M_n were monitored at each switching point with NMR spectroscopy and size exclusion chromatography analyses of aliquots. Although the plot of conversion versus time clearly illustrates that polymerization proceeds only in the presence of light with 1a (Figure 14A), the corresponding plot for 1b shows that polymerization is not halted in the dark (Figure 14B). Consequently, pyrylium 1b should be considered a fine catalyst for photoinitiated, rather than photocontrolled, cationic polymerization, ^1f as it allows for predictable M_n and a Đ of ∼1.2 but offers no temporal control of chain growth. This outcome also suggests that the recapping step (step IV) is much slower for 1b than for 1a, which can be explained by the difference in ground state redox potentials ( ${E°}_{\mathbf{1}{\mathbf{a}}^{+}/\mathbf{1}{\mathbf{a}}^{•}}=-0.50\text{ V versus SCE},\text{ and }E{°}_{\mathbf{1}{\mathbf{b}}^{+}/\mathbf{1}{\mathbf{b}}^{•}}=-0.32\text{ V versus SCE}$).⁸ Indeed, the redox potential for the reduction of disulfide 7 has been measured at $E{°}_{7/7^{•-}}=-0.302\text{ V versus SCE}$ by Nichols and Grant,⁴⁰ which supports our conclusion that reduction of 7 is more likely with 1a^• than with 1b^•, therefore resulting in a more efficient deactivation of chain growth with the former. However, the fact that living characteristics are observed with both photocatalysts is a strong indicator that a RAFT equilibrium influences chain growth.

Figure 14.

Monomer conversion vs time with intermittent light exposure with photocatalysts (A) 1a and (B) 1b.

Catalyst Turnover and Chain End Capping

The last step of our proposed mechanism was interrogated via an ESR experiment coupled with UV–visible spectroscopy. The reduction of 1a with a stoichiometric amount of cobaltocene (CoCp₂)⁴¹ afforded radical 1a^•, as shown in Figure 15 (see SI for details). The spin concentration of this stable radical (lifetime >6 h) was estimated to be 250 μM under these conditions. The addition of an excess of thiuram disulfide 7 resulted in an instantaneous depletion of the signal to a spin concentration of approximately 0.8 μM (Figure 15A). This change indicated a SET from 1a^• to disulfide 7. The reduction of 7 concomitantly regenerates PC 1a, as shown on the UV–visible spectra (Figure 15B) and produces the dithiocarbamate anion likely responsible for capping the chain end and thus deactivating chain growth. In the polymerization process itself, this SET event likely happens with either 7 or the dithiocarbamate radical, which are postulated to be in equilibrium (Figure 13B).

Figure 15.

Reduction of 1a to 1a^• with CoCp₂, followed by oxidation back to 1a with disulfide 7. (A) Electron spin resonance spectra of 1a (black underneath green curve), 1a^• after the addition of CoCp₂ (red), and 1a after the addition of 7 (green). (B) UV–visible spectra of 1a (black), 1a^• after the addition of CoCp₂ (red), and 1a after the addition of 7 (green). Characteristic absorption bands are indicated.

Notably, the addition of CTA 2a to radical 1a^• generated by cobaltocene reduction resulted in no changes in the ESR and UV–visible spectra (Figure S26). This result corroborates the finding that the CTA does not interact with the reduced PC.

CONCLUSION

Using various spectroscopic techniques, we gained intimate knowledge of each elementary step in a novel cationic polymerization of vinyl ethers regulated by visible light. Fine-tuning of the electronic structure of both the PC and the CTA unveiled a number of factors that govern the rate of polymerization as well as the prerequisites for a well-behaved living process: while more oxidizing pyrylium salts generally engender higher polymerization rates, quantum yields of fluorescence and the solubility profile also play a role in the activity of the PCs. Interestingly, no other family of PCs has proven competent for the photocontrolled polymerization of vinyl ethers yet. CTAs synthesized from a vinyl ether derivative and containing an electron-rich heteroatom appended to the dithiocarboxylic core delivered the best control over chain growth.

Time-resolved photoluminescence and ESR spectroscopy revealed single-electron transfers from excited PCs 1a* and 1b* to CTA 2a and monomer 3a. Oxidation of 3a by PCs is responsible for the uncontrolled background polymerization. Comparison of bimolecular quenching constants and behaviors during cyclic voltammetry of 2a and 3a, however, suggests that oxidation of 2a is more facile than that of 3a, and that a second electron transfer from oxidized 3a to 2a might prevent the background polymerization. Moreover, polymerization of 3a can be mediated by 2a at a voltage lower than the onset of oxidation of 3a, which indicates that the direct oxidation of 2a by 1* is the major contributor to the activation step. Finally, the formation of a donor–acceptor complex between monomer and PC in the ground state was uncovered by meticulous ESR and UV–visible spectroscopic analyses of a mixture of 3a and PCs 1a and 1b. Such complexes presumably play a key role in the activation step. Isolation of thiuram disulfide 7 supports a mesolytic cleavage pathway following oxidation of the CTA. In the reaction conditions, 7 is in equilibrium with the radical dithiocarbamate species arising from homolytic breaking of the disulfide bond. Determination of the quantum yields of polymerization for various PCs, as well as on/off experiments, substantiate the existence of a degenerative chain transfer mechanism. This RAFT-type equilibrium is likely pivotal in the obtention of polymers with predictable M_n’s and narrow Đ’s. Last, chain end deactivation and PC regeneration via electron transfer between 1a^• and 7 were corroborated by ESR and UV–visible spectroscopies.

We are confident that the results of this study will serve as a starting point for explorations of similar photoredox-catalyzed polymerizations and that our findings will prove critical in the eventual adoption of this system in complex practical settings.