Organocatalyzed Atom Transfer Radical Polymerization: Perspectives on Catalyst Design and Performance

Abstract

The recent development of organocatalyzed atom transfer radical polymerization (O-ATRP) represents a significant advancement in the field of controlled radical polymerizations. A number of classes of photoredox catalysts have been employed thus far in O-ATRP. Analysis of the proposed mechanism gives insight into the relevant photophysical and chemical properties that determine catalyst performance. Discussion of each of the classes of O-ATRP catalysts highlights their previous uses, their roles in the development of O-ATRP, and the distinctive properties that govern their polymerization behavior, leading to a set of design principles for O-ATRP catalysts. Remaining challenges for O-ATRP are presented, as well as prospects for further improvement in the application scope of O-ATRP.

1. Introduction

Controlled radical polymerizations (CRPs) allow for the synthesis of polymers of predictable molecular weights (MWs), narrow MW distributions, and defined end-group functionality through accessible experimental conditions.^[1] CRPs are differentiated depending on the nature of the reversible-deactivation equilibrium, with some variants including nitroxide-mediated polymerization (NMP),^[2] reversible addition-fragmentation chain transfer (RAFT) polymerization,^[3] and atom transfer radical polymerization (ATRP).^[4] Historically, ATRP affects this reversible deactivation mechanism via transfer of a halogen atom between the polymer species and a transition metal catalyst (typically copper- or ruthenium-based) (Figure 1, top). In copper-catalyzed ATRP, the dormant halogen-capped polymer chains react with a Cu(I) species (the activator) to liberate a carbon-centered radical capable of propagating with monomer. Growth of polymer chains proceeds until a Cu(II) species (the deactivator) reinstalls the halogen. While ATRP does significantly suppress radical termination, these undesirable events lead to a buildup of Cu(II) species. Several ATRP methodologies were developed to avoid a buildup of deactivator via regeneration of Cu(I) species.^[5] One of these methods, termed photoATRP, uses light for the photoreduction of Cu(II).

Figure 1.

Mechanisms of ATRP (top) and O-ATRP (bottom).

The impact of light on ATRP is expansive. Irradiation of copper-catalyzed ATRP reactions with visible light leads to improved polymerization outcomes compared to ATRP reactions performed in the absence of light.^[6] Photocontrolled ATRP systems boast numerous additional advantages, including low-temperature operation and spatial and temporal control of the reaction. Light has been used to realize a number of photocontrolled ATRP processes which can be categorized based on the role that light plays in the generation of active Cu(I) species from more stable Cu(II) precursors. In the first subset, termed photoinitiated ATRP, light cleaves a labile bond on a photoinitiator to generate a radical species which can then react with Cu(II) complexes. Photoinitiated ATRP has been successfully carried out using several photoinitiators in the presence of UV light with or without alkyl halide initiators^[7] and in the presence of visible light with dyes.^[8] The use of light to induce redox reactions represents the second major class of photocontrolled ATRP systems. Photoreduction of Cu(II) complexes^[9] and photoinduced electron transfer from the ligand to Cu(II) complexes have been used to generate the active Cu(I) species in situ to lower catalyst loading to the parts per million regime.^[10] The use of light for photocontrol has been a subject of investigation in many other CRPs as well.^[11]

A resurgence of interest in photoredox catalysis for small molecule transformations^[12] inspired the use of these methods for polymerizations.^[13] ATRP systems that employ photoredox catalysts (PCs) to mediate the reversible deactivation equilibrium are mechanistically distinct from the previously mentioned photoinduced and photoredox ATRP systems and can operate via reductive or oxidative quenching pathways. While photoredox-catalyzed polymerization systems that operate via the reductive quenching pathway have been developed,^[14] the reductive quenching pathway requires a sacrificial electron donor, which can result in undesirable side reactions.^[15] Therefore, only systems that operate via an oxidative quenching cycle will be discussed herein. In the oxidative quenching pathway, the catalyst undergoes photoexcitation to an excited state from which it can reduce a substrate via single electron transfer. Initially, iridium-based PCs were explored for small molecule atom transfer radical addition (ATRA) reactions with olefins^[16] before being employed for ATRP. In the presence of a visible light source, fac-[Ir(ppy₃)] has successfully catalyzed the ATRP of methacrylates^[17] and acrylates^[18] via an oxidative quenching cycle.

While the development of this photoredox-catalyzed system represents a significant advance in the field of photocontrolled ATRP, researchers aimed to replace this precious metal catalyst with an organocatalyst.^[19] Catalysts are challenging to remove from the product polymer and commonly remain behind in the polymer matrix. Metal contamination could limit the use of these polymers in value-added applications such as medical devices, drug delivery systems, and electronics. Motivation to design organic PCs that could mediate ATRP via an oxidative quenching pathway led to the development of organocatalyzed atom transfer radical polymerization (O-ATRP). Although O-ATRP is still in its relative infancy, many significant contributions have already been made. This Feature Article will discuss the current mechanistic understanding of O-ATRP, the classes of organic PCs capable of mediating O-ATRP, unique features of each of these catalyst classes, and design principles gleaned from these catalysts' performance, as well as remaining challenges and future directions for O-ATRP itself.

2. Current Mechanistic Understanding of O-ATRP

Figure 1 (bottom) depicts the proposed mechanism of O-ATRP.^[20] This proposed catalytic cycle for O-ATRP retains the main mechanistic feature of ATRP, reversible deactivation of polymer chains through halogen atom transfer, but is mediated by a PC. The cycle begins with photoexcitation of a ground state PC into a singlet excited state (¹PC*) by absorption of UV or visible light. After photoexcitation, ¹PC* can react from the singlet manifold or undergo intersystem crossing (ISC) to form a triplet excited state (³PC*). ¹PC* is more reducing than ³PC* but shorter-lived (lifetime, τ, typically in the nanoseconds), whereas ³PC* is often much longer-lived (τ often in the microseconds) and are usually invoked as the reactive excited state species participating in photoredox reactions.^[21] Potentially, both ¹PC* and ³PC* could participate in the activation step, a reduction of the alkyl halide bond of a polymer chain to liberate a carbon-centered radical capable of propagation. Ideally, the polymer chains are extended by a minimal number of monomer units before being deactivated by the radical cation form of the PC (PC^•+) to minimize termination reactions. Deactivation returns the polymer chain to a dormant state and the PC to a ground state, completing the photoredox-mediated O-ATRP cycle.

By considering each step involved in the O-ATRP mechanism, the chemical and photophysical properties that are important for PC performance can be determined. First, photoexcitation can be performed with a variety of wavelengths of light. However, photoexcitation using UV light can lead to side reactions which complicate the synthesis of well-defined polymers.^[22] It is preferable, then, for a PC to absorb in the visible, ideally possessing a wavelength of maximum absorbance (λ_max) in the visible region with a high molar extinction coefficient (ε). After photoexcitation, the PC should efficiently cross over to the triplet excited state, indicated by a high rate of intersystem crossing (ISC). The PC must possess a sufficiently long excited state lifetime (τ) and be sufficiently reducing in the excited state to execute the electron transfer reaction required for polymer activation, approximately −0.8 V vs. SCE for the alkyl bromides commonly employed in ATRP.^[23]

The thermodynamic feasibility of activation and deactivation can be evaluated from the excited state reduction potential of PC* [E⁰* = E⁰(PC^•+/PC*)] and oxidation potential of PC^•+ [E⁰_ox = E⁰(PC^•+/PC)]. Of the PC parameters discussed so far, the requirement of a strong E⁰* is arguably the most stringent and greatly narrows the field of potential candidates.^[24] Although more remains to be learned about the precise mechanism of deactivation, it has been proposed that a termolecular reaction involving PC^•+, X⁻, and the propagating radical species is the most favorable pathway for deactivation.^[20c] However, in another study, it was argued that the termolecular pathway is most likely entropically improbable given the low concentrations of the species involved. Instead, it was suggested that ion pairing between PC^•+ and X⁻ to form a PC^•+ X⁻ complex reduces deactivation to a more feasible bimolecular reaction, and indeed the ion complexation strength between PC^•+ and X⁻ was shown to be an important factor in polymerization outcomes.^[25]

Analysis of the mechanism leads to a list of the chemical and photophysical properties that should be considered in the design of a PC for O-ATRP. It should not be a surprise that this list bears a striking resemblance to a similar list proposed for metal photoredox catalysts.^[21] However, this list has some modifications to address the specific needs of photoredox-mediated O-ATRP.

1. Strong visible light absorption (λ_max close to visible and high ε)

2. High rate of ISC/high triplet quantum yield

3. Sufficiently long τ for electron transfer to occur

4. E⁰(PC^•+/PC*) < −0.8 V vs. SCE

5. Oxidizing PC^•+ with E⁰(PC^•+/PC) > −0.8 V vs. SCE

As of this contribution, five major classes of organic PCs, including polycyclic aromatic hydrocarbons, phenothiazines, phenazines, phenoxazines, and carbazoles, have been explored for use in O-ATRP (Figure 2). While this article does not provide a comprehensive review of the field of O-ATRP, it strives to highlight the unique photophysical and chemical properties that make certain organic molecules suitable for O-ATRP and recognize the broad scope of applications for which these PCs have been previously used. The following sections will include, for each catalyst type, discussion of the historic uses of these PCs, their key photophysical features, and their performance as O-ATRP catalysts, followed by a comparison of PCs, with a focus on those containing phenothiazine, phenazine, and phenoxazine core structures.

Figure 2.

Structures of photocatalysts 1–8. PC 1 = perylene; PC 2 = 10-phenyl-phenothi-azine; PC 3 = 1-naphthalene-10-phenothiazine; PC 4 = 5,10-di(4-trifuoromethylphenyl)-5,10-dihydrophenazine; PC 5 = 5,10-di(1-naphthyl)-5,10-dihydrophenazine; PC 6 = 1-naphthalene-10-phenoxazine; PC 7 = 3,7-di(4-biphenyl) 1-naphthalene-10-phenoxazine; PC 8 = 1,2,3,5-tetrakis(carbazol-9-yl)-4,6-dicyanobenzene (4CzIPN).

3. Classes of O-ATRP Catalysts

3.1. Polycyclic Aromatic Hydrocarbons

Polycyclic aromatic hydrocarbons (PAHs) consist of multiple fused aromatic rings composed of only carbon and hydrogen. PAHs are known as terrestrial carcinogens that results primarily from incomplete combustion reactions,^[26] and PAHs are also abundant in space.^[27] Although the properties of PAHs vary significantly with MW, PAHs are typically excellent light absorbers that exhibit strong fluorescence. PAHs have received much attention in the field of organic electronics (OEs) and organic photovoltaics (OPVs) for their ability to effectively manipulate photonic energy.^[28] Additionally, perylene (PC 1, Figure 2) and related PAHs have been investigated as photosensitizers and photoinitiators for cationic and free radical polymerizations.^[29] Perylene derivatives have tunable absorptions throughout the visible and IR regions, with typically high molar extinction coefficients (ε > 10⁴ M⁻¹ cm⁻¹).^[30] Perylene derivatives have been recognized as strong excited state reductants relative to other organic dyes,^[31] and perylene itself has E⁰(PC^•+/PC*) = −1.87 V vs. SCE from the S₁ state and E⁰(PC^•+/PC*) = −0.58 V vs. SCE from the T₁ state.^[32]

PC 1 was investigated early on as a potential PC for the first demonstration of organocatalyzed ATRP to proceed through the oxidative quenching pathway using visible light.^[20a,33] Using methyl methacrylate (MMA) and butyl acrylate (BA) as demonstrative monomers, polymers with moderately low Đ (a measure of the broadness of the MW distribution), typically 1.3–1.8, were synthesized under white light irradiation. Matrix-assisted laser desorption ionization–time of fight (MALDI–TOF) mass spectrometry and chain extension experiments confirmed re-installation of the bromide end group; however, it was not quantitative. Temporal control allowed the polymerization to be completely halted with the removal of light, and resumed with re-introduction of light. However, the system exhibited low initiator efficiencies (I*, the theoretical MW divided by the measured MW). In addition to perylene, PAHs anthracene and pyrene were investigated as PCs for O-ATRP under 350 nm irradiation.^[34] In addition to styrene and tert-butyl acrylate, MMA was polymerized to low conversion (<30%) with Đ typically 1.4-1.5. However, as with the perylene system, I* was consistently low.

3.2. Phenothiazines

Phenothiazine derivatives have been investigated in medical and biological applications since the 1880s.^[35] For example, methylene blue is a popular histologic stain used in biological research that changes color due to electron transfer reactions involving the phenothiazine core structure.^[36] Due to their redox-active nature, phenothiazines have been applied in dye-sensitized solar cells,^[37] organic light-emitting diodes (OLEDs),^[38] and battery electrode applications.^[39] Recently, phenothiazine's ability to act as a photoreductant has been exploited in catalysis for applications such as dehalogenation^[40] and C—C bond formation on aryl halide substrates.^[41] Phenothiazine can be relatively easily oxidized to form a radical cation, whose salt is stable enough to be isolated and studied.^[42] Phenothiazines were also applied in cationic polymerization,^[43] RAFT polymerization,^[44] and the synthesis of conjugated polymers with donor-acceptor architecture.^[45]

Phenothiazine's use in photophysical applications made it an excellent candidate for O-ATRP PCs. 10-phenyl phenothiazine (PC 2, Figure 2) was found to have E⁰(PC^•+/PC*) of −1.97 V vs. SCE in DMA.^[20c] PC 2, along with other phenothiazine derivatives, was employed in the O-ATRP of MMA under 380 nm irradiation to give polymers of low dispersity (Đ = 1.18 to 1.32) and high initiator efficiency (I* = 90% to 116%).^[20b,c] Preservation of chain-end functionality was confirmed by successful chain extensions and copolymerizations, as well as MALDI-TOF analysis. Since its first demonstration, PC 2 has also been used to polymerize acrylonitrile^[46] and biomass-based monomers^[47] via O-ATRP. A mechanistic study of O-ATRP catalyzed by PC 2, PC 3, and other phenothiazines was performed.^[20c] In that study, along with a previous study,^[20b] the short-lived singlet state was proposed to be responsible for ET from phenothiazine PCs, while later work suggested that the longer-lived triplet state was the electronic state responsible for O-ATRP initiation.^[48]

3.3. Phenazines

Phenazine derivatives are well-known natural products, first recognized as colorful secondary metabolites produced by Pseudomonas.^[49] In nature, phenazine metabolites produced from bacteria serve many purposes ranging from protection against competitive microorganisms to influencing structural organization in bacterial populations. Synthetic and naturally-derived phenazines have demonstrated antibiotic, antitumor, antimalarial, and antiparasitic activity.^[50] In the field of OEs, researchers have capitalized on the electron-rich nature of the phenazine core to make hole injection materials and donor–acceptor molecules with charge transfer (CT) character.^[51] Phenazine-containing emissive materials demonstrate tunable emission profiles and were successfully incorporated into OLEDs.^[52] Radical cations of phenazine are exceptionally stable and have been explored as components in ferromagnetic organic materials.^[53] In catalysis, phenazines were employed as components in photoinitiating systems for cationic polymerization.^[54]

Guided by density functional theory (DFT) predictions, several substituted diaryl dihydrophenazines were investigated for use in O-ATRP due to their strong excited state reduction potentials, with predicted E⁰(PC^•+/PC*) ranging from −2.06 to −2.36 V vs. SCE and absorption stretching into the visible regime.^[23] Initial polymerization results indicated that catalysts with electron-withdrawing groups on the aryl substituent, in particular PC 4 (Figure 2), gave superior results for the polymerization of MMA under white light irradiation (Đ = 1.17), albeit with a moderate I* of 66%. Further computational investigation revealed that, unlike dihydrophenazines with electron-donating or neutral substituents on the aryl group, those with electron-withdrawing substituents possessed spatially separated singly-occupied molecular orbitals (SOMOs) in the triplet excited state (Figure 3), indicating intramolecular CT (see Section 4). Based on this observation, more substituted dihydrophenazine catalysts with computationally predicted spatially separated SOMOs were synthesized and tested, and PC 5 (Figure 2) was found to polymerize MMA to low dispersities (Đ < 1.3) with near-quantitative initiator efficiency. High end-group fidelity was confirmed by copolymerizations and MALDI-TOF analysis. This contribution placed diaryl dihydrophenazines as the first example of visible light O-ATRP catalysts capable of producing results on par with conventional ATRP.

Figure 3.

PCs 4 and 5 display spatially-separated SOMOs in the DFT-predicted triplet excited state, whereas diphenyl dihydrophenazine (left) does not. Adapted with permission.^[23] Copyright 2016, American Association for the Advancement of Science.

3.4. Phenoxazines

The phenoxazine core structure has been found in several naturally occurring, biologically active compounds isolated from bacteria.^[55] In medicine, phenoxazine derivatives have been investigated as antitumor agents,^[56] antifungal agents,^[57] antimalarial drugs,^[58] and for gene therapy.^[59] Due to their strong emission in the visible light regime, derivatives containing the phenoxazine core structure have also been investigated for biological imaging applications^[60] and biosensor applications.^[61] Within the chemical and materials science fields, phenoxazine derivatives have been employed as donors in small molecule, donor-acceptor species for OLEDs,^[62] dyes for dye-sensitized solar cells,^[63] and they have been investigated for other OPV applications^[64] due to the electron-rich nature of the phenoxazine core and stability of the phenoxazine radical cation.^[65] In addition, phenoxazine-based polymers have been used as p-channel semiconductors in organic field transistors.^[66]

Several N-aryl substituted phenoxazines were synthesized for use as PCs and found to have predicted E⁰(PC^•+/PC*) ranging from −1.9 V to −2.1 V vs. SCE.^[67] It was found that those bearing naphthyl substituents, such as PC 6 (Figure 2) were able to synthesize PMMA under 365 nm irradiation in a controlled fashion, as evidenced by a linear growth of polymer MW with increasing conversion, low Đ (1.07 to 1.38), and high I* (typically 80% to 100%). Interestingly, the trend observed with diaryl dihydrophenazines that PCs exhibiting computationally-predicted, spatially separated SOMOs in the triplet excited state—a hallmark of CT—performed better in O-ATRP compared to PCs that did not exhibit this feature, was also seen with the phenoxazine PCs (see Section 4). Encouraged by these results, it was sought to retain the catalytic power of the phenoxazine structure while pushing absorption into the visible regime. Synthetic modification of the phenoxazine core with biphenyl substituents at the 3 and 7 positions yielded a new catalyst (PC 7, Figure 2) with an absorption profile redshifted by 65 nm (λ_max = 323 nm for PC 6, λ_max = 388 nm for PC 7) and a dramatically increased molar absorptivity (ε = 7,848 M⁻¹ cm⁻¹ for PC 6, ε = 26,635 M^–1 cm⁻¹ for PC 7) (Figure 4). Polymerization of MMA by PC 7 irradiated with white light resulted in polymers with quantitative I* and low Đ (1.13 to 1.31).

Figure 4.

Change in absorption spectra upon substitution of PC 6 with biphenyl groups to yield PC 7. Adapted with permission.^[67] Copyright 2016, American Chemical Society.

3.5. Carbazoles

Various carbazole-based derivatives^[68] and alkaloids have been synthesized^[69] to study their effects and functions in biological systems.^[70] As an example, the antibiotic properties of murrayanine, a naturally occurring carbazole-derived alkaloid were first described in 1965.^[71] The electron-rich carbazole moiety, when paired with an electron-poor substituent, forms a donor–acceptor pair that exhibits CT character.^[72] Carbazole's ability to form CT complexes has been exploited in various applications, including OEs,^[73] OPVs,^[74] and OLEDs.^[75] However, in 2012, carbazole derivatives attracted significant attention when the donor-acceptor pair of carbazolyl dicyanobenzene (4CzIPN, PC 8) was shown to exhibit thermally activated delayed fluorescence for efficient OLED application.^[75c,d] Additionally, PC 8 was recently applied in small molecule synthesis for the formation of C-C bonds via dual photoredox/nickel catalysis.^[76]

The use of PC 8 [E⁰(PC^•+/PC*) = −1.04 V vs. SCE]^[75c,77] as a catalyst for O-ATRP was recently described.^[78] PC 8 has absorption features into the visible spectrum and thus polymerization was able to be carried out under blue light irradiation. PC 8 mediated the polymerization of MMA with exceptionally low ppm-level catalyst loading, but typically gave broad molecular weight distributions (Đ ≥ 1.50). Photocontrol was demonstrated with on–off irradiation experiments, and chain ends were confirmed through MALDI-TOF and chain extension experiments, although it was estimated that 11% of chains were “dead” or terminated. Two other carbazole derivatives, 9-phenylcarbazole and 4,4′-bis(N-carbazolyl)biphenyl, were also employed as O-ATRP catalysts, producing polymers with high dispersity (Đ ≥ 1.80). These catalysts were also found to exhibit irreversible CV curves, an indication that the radical cations were not stable in the polymerization solvent.^[20c,77]

4. Discussion

The available photophysical data for each of the highlighted PCs 1–8 are summarized in Table 1. Returning to the overall list of design principles (see Section 2), a general comparison can be made between the 5 classes of PCs. In terms of the first principle—a strong photon absorption in the visible regime—perylene is the strongest visible light absorber among the catalysts, with λ_max = 436 nm and ε_max = 38,500 M^–1 cm⁻¹. The remaining catalysts all have absorption maxima in the UV. However, it is worth noting that PCs 4, 5, 7, and 8, despite having λ_max values < 400 nm, have absorption profiles that extend into the visible region, which has enabled them to conduct O-ATRP using white LED light (PCs 4, 5, and 7) or blue LED light (PC 8). Importantly, it has been shown in the case of phenoxazines that synthetic modification can be employed to drastically alter absorption properties without necessarily affecting catalytic behavior (see Section 3.4).

Table 1.

Summary of Properties of Photocatalysts 1–8.

PC	Λmax [nm]	εmax [M−1 cm−1]	τ	E0(PC•+/PC*) [V vs. SCE]	E0(PC•+/PC) [V vs. SCE]	Rev. CV?	Refs.
PC 1	436	38,500	5.5 nsa), 5000 μsb)	−1.87a), −0.58b)	0.98	Y	[32]
PC 2	320	3,200	4.5 nsa), 420 nsb)	−1.97a,e), −2.1a,f)	0.82e), 0.68f)	Y	[20b,20c,48,81]
PC 3	317	3,163	7.6 nsa)	–2.23a)	0.83	Y	[20c,67]
PC 4	367	4,700	21 nssa,c), 1.20 sb,c)	−1.80g), (−2.17)h,i)	0.29 (0.21)h,i)	Y	[23,82b]
PC 5	366	5,500	–	−1.64g), (−2.04)h,i)	0.23 (0.10)h,i)	Y	[23]
PC 6	323	7,848	3.2 nsa,d), 2.31 sb,d)	−1.67g), (−1.84)h)	0.70 (0.55)h)	Y	[67,82a]
PC 7	388	26,635	–	−1.80g), (−1.70)h,j)	0.65 (0.42)h,j)	Y	[67]
PC 8	375	19,000	17.8 nsa)	−1.06g)	1.50	N	[75–77]

^a)Singlet state;

^b)Triplet state;

^c)Determined for dihydrophenazine derivative of 5,10-diphenyl-5,10-dihydrophenazine. Singlet lifetime was determined at 298 K and triplet lifetime was determined at 77 K, both in 3-methylpentane;

^d)Determined for phenoxazine derivative of 10-phenyl phenoxazine. Singlet lifetime was determined at 298 K in cyclohexane and triplet lifetime was determined at 77 K in 3-methylpentane;

^e)Determined in ref. [20c] in DMA;

^f)Determined in ref. [20b] in MeCN;

^g)Determined from a broad and featureless emission peak that was attributed as emission from the charge transfer state. Charge transfer singlet and triplet states are close to isoenergetic such that singlet and triplet excited state reduction potentials are similar in values;

^h)Available DFT-predicted values are enclosed in parentheses;

ⁱ⁾Reported values here were computed at the improved M06/6-311+G(d,p)//M06/6-31+G(d,p) level of theory. Previously reported values in ref. [23] were computed at the M06/6-31+G(d,p) level of theory;

^j)Reported values here were computed at the improved M06/6-311+G(d,p)//M06/6-31+G(d,p) level of theory. Previously reported values in ref. [67] were computed at the M06/6-311+G(d,p)//M06/Lanl2dz level of theory.

Principles 2 and 3 state the need for efficient ISC to the triplet excited state and sufficiently long excited-state lifetime. Available singlet and triplet lifetimes are listed in Table 1. Singlet lifetimes typically are short (in the nanoseconds) and most likely decay to the ground state before a successful bimolecular encounter with the desired substrate can occur. Some exceptions do exist. For example, phenothiazine was shown to perform ET from its S₁ state to chloroalkanes because these two species associate into a complex and are therefore in close proximity prior to the electron transfer reaction.^[79] However, reactivity from the triplet state is typically proposed for photoredox catalysis due to its much longer lifetime (in the microseconds or more) because the triplet's transition to the singlet ground state is a spin-forbidden process.^[80] Triplet lifetimes and triplet quantum yield for some related phenothiazines, dihydrophenazines, and phenoxazines have been reported. For example, the triplet lifetime of 10-phenyl phenothiazine (PC 2) was determined to be ≈420 ns in acetonitrile,^[81] while the related 10-methyl phenothiazine has a triplet lifetime and quantum yield of 40 μs and 60% in DMA, respectively^[48] Similarly 5,10-diphenyl-5,10-dihydrophenazine's triplet is long-lived (1.20 s) and its quantum yield is at least 26% in 3-methylpentane at 77 K.^[82] Moreover, 10-phenyl phenoxazine was determined to have a long triplet lifetime of 2.3 s and a triplet quantum yield of at least 94% in 3-methylpentane at 77 K.^[82] Future work should include measurement of triplet lifetimes and quantum yields of the remaining PCs featured in Figure 2 in order to provide important photophysical information pertinent to O-ATRP.

After the successful generation of the active exited state, the remainder of the catalytic cycle consists of electron transfer reactions to activate and deactivate the alkyl halide bond (≈−0.8 V vs. SCE). Principles 4 and 5 are related to the thermodynamics of activation and deactivation, respectively. Table 1 lists E⁰(PC^•+/PC*) and E⁰(PC^•+/PC) of PCs 1–8; available DFT-predicted values are enclosed in the parenthesis. PC 1's singlet excited state is much more reducing than its corresponding triplet excited state with E⁰(PC^•+/PC*) values of −1.87 V and −0.58 V vs. SCE, respectively. PC 2 and 3′s singlet excited state has reported E⁰(PC^•+/PC*) values of ≈−2 V vs SCE. PC 4, 5, 6, 7 and 8′s E⁰(PC^•+/PC*) values were estimated from emission containing a broad and featureless peak, which suggests emission from a CT state. Singlet and triplet CT states are close to isoenergetic (<0.2 eV), thus the obtained E⁰(PC^•+/PC*) values from these emission data can represent both the singlet and triplet excited state reduction potential.^75c,83 PC 4, 5, 6, and 7 have E⁰(PC^•+/PC*) values of ≈−2 V while PC 8 has E⁰(PC^•+/PC*) = −1.06 V vs SCE.

Generally, PCs employed in O-ATRP (Figure 2 and Table 1) are more reducing than is thermodynamically necessary for activation. It is expected, however, that some amount of overpotential will be necessary to overcome electron transfer activation barriers (e.g., to overcome energetic costs for structural reorganization in the donor and/or acceptor), which makes a more negative E⁰(PC^•+/PC*) desirable.^[84] With respect to O-ATRP deactivation, all PCs in Table 1 have sufficiently oxidizing radical cations to deactivate the growing polymer chains, where their E⁰(PC^•+/PC) values are more positive than ≈−0.8 V vs SCE. Dihydrophenazine-derived radical cations are the most stable (≈0.2 V vs SCE for PC 4 and 5), followed by phenoxazines (≈0.7 V vs SCE for PC 6 and 7), phenothiazines (≈0.8 V vs SCE for PC 2 and 3), perylene (0.98 V vs SCE for PC 1), and 4CzIPN (1.50 V vs SCE for PC 8). Dihydrophenazines, phenoxazines, and phenothiazines' radical cations are sufficiently stable that they exhibit reversible cyclic voltammograms, and many can be isolated (see Section 3). The radical cations should be sufficiently stable that they should not degrade prior to deactivation by the propagating radical. On the contrary, they also should not be too stable that the rate of deactivation is too slow. In another extreme, the radical cation of 4CzIPN is so reactive that it does not exhibit a reversible cyclic voltammogram.

The O-ATRP PCs that have demonstrated the most success thus far are derivatives of phenothiazine, phenazine, and phenoxazine core structures, due to their balance of strong excited state reduction potentials with stable and sufficiently oxidizing radical cations. However, their catalytic outcomes are not equivalent, which indicates there is more to be learned about catalyst design from a closer examination of this series. The most striking difference between these three groups is the proposed ability of phenazines and phenoxazines, upon excitation, to more efficiently enter an intramolecular CT state, where the tri-cyclic core behaves as the electron donor and the N-aryl substituent behaves as the acceptor. The existence of the role CT plays in O-ATRP was first evidenced by computationally-observed spatially-separated SOMOs in PCs 4–6, among others (see Sections 3.3 and 3.4). Further, electrostatic potential mapped electron density shows high localization of negative charge on the N-aryl substituent in the excited state, along with the observation of strong emission solvatochromism, provides additional support for the formation of CT states in dihydrophenazine catalysts (Figure 5).^[25]

Figure 5.

DFT-predicted electrostatic potential mapped electron density indicates intramolecular CT as the photocatalysts transition from the ground state ¹PC to the triplet state ³PC* in naphthyl-substituted dihydrophenazine and phenoxazines, but not phenothiazine. High electron density region is indicated by “red” color while low electron density region is indicated by “blue” color. Partial charges (δ) in the unit of electron (e) of the N-aryl substituent, catalyst core, and core-substituent are indicated. All PC 3, 5, 6, 7 were optimized at the unrestricted M06/6-31+G(d,p) level of theory. Adapted with permission.^[67] Copyright 2016, American Chemical Society.

Direct comparison of the crystal structures of isoelectronic PCs 3 and 6 reveals that the core of PC 3 is bent into a boat shape whereas the core of PC 6 is planar (Figure 6).^[67] The bent geometry of phenothiazine was attributed to the larger van der Waals radius of sulfur (1.80 Å) relative to oxygen (1.52 Å). Previous work posited that the bent core structure reduces electronic coupling between the core and the N-aryl substituent, which reduces the extent of CT in phenothiazine in comparison to phenoxazine.^[85] To date, CT is empirically determined to be a major contributing factor to the successful polymerization results seen with phenazine and phenoxazine catalysts. While further investigation of the exact role of CT in O-ATRP is warranted, it is interesting to note that CT in these organic PCs is reminiscent to the photoredox-active metal-to-ligand charge transfer (MLCT) state observed for Ir and Ru polypyridyl PCs.^[86]

Figure 6.

Crystal structures of PCs 3 and 6 show that PC 6 has a planar core whereas PC 3 has a bent core. Adapted with permission.^[67] Copyright 2016, American Chemical Society.

Another factor to consider is the structural reorganization cost of the PCs during the photoredox O-ATRP cycle. Computations predict phenazines and phenoxazines (PCs 4–7) adopt close-to-planar geometries in their triplet, radical cation, and singlet ground states. As a result, these PCs have low structural reorganization energy and thus small kinetic barriers to electron transfer during the activation and deactivation steps.^[67] In contrast, phenothiazine is bent in the triplet and singlet ground states, while it is planar in the radical cation state. This represents more significant structural reorganization and thus kinetic cost during electron transfer events in the O-ATRP cycle. The effect that this additional energetic toll has on the PC performance for phenothiazine PCs versus phenoxazine and phenazine PCs is unclear, but it is hypothesized that this contributed to the poorer control over molecular weight distribution observed for phenothiazine PCs 2 and 3.^[67]

What arises from this deeper comparison of PCs, then, are additional design principles for O-ATRP PCs that are not as evidently presented by the mechanism as the first five. First is the ability to form CT states (see earlier Discussion).^[25] Second is a consistent geometry in the PC's ground state, excited state, and radical cation, as it limits the reorganization energy necessary for changing the redox state of the PC during an O-ATRP catalytic cycle.^[67] While future mechanistic studies will undoubtedly uncover more design principles, these two (in addition to the five presented earlier) represent the current level of understanding in O-ATRP PC development, while the modular design of derivatives that allow for the tuning of properties of the PC and their continued study will undoubtedly refine these principles and reveal new insight.

5. Future Outlook

Despite its status as a relatively new methodology, O-ATRP is already establishing itself as a robust method for the metal-free production of specialty polymers. For example, PC 2 has been used for metal-free, surface-initiated ATRP to functionalize silicon^[87] and nanodiamond surfaces.^[88] However, in order for O-ATRP to continue to expand its utility, some significant challenges remain to be addressed. Foremost among these is the need for increased mechanistic understanding. Factors such as the importance of CT and PC geometry make it clear that the success of O-ATRP is more detailed than simply matching reduction potentials. As more is learned into the intricacies of the mechanism, for example the kinetics of activation and mechanism of deactivation, additional design principles will be revealed, evolving PCs for more efficient catalysis. It will also be important to extend the application of O-ATRP to a broader monomer scope. Successful controlled polymerization of monomers, including various (meth)acrylates and acrylonitrile, has already been reported for O-ATRP. However, an increased monomer scope as well as expansion to additional classes of monomers, in particular unconjugated monomers such as vinyl acetate, is desirable.^[19] Additionally, the ability to produce high-MW polymers by O-ATRP must be demonstrated. Only phenazine PCs 4 and 5 have been shown to synthesize PMMA much larger than 20 kDa in a controlled manner.^[23] Finally, the ability to use O-ATRP to synthesize more advanced copolymeric architectures beyond linear diblocks, such as multiblock copolymers, dendrimers, and graft copolymers.

Biographies

Jordan C. Theriot is a Ph.D. candidate in the Department of Chemistry and Biochemistry at the University of Colorado Boulder. She received her B.S. in chemistry from the California Institute of Technology in 2012, where she completed a thesis under Prof. Robert Grubbs. In 2014, she moved to the University of Colorado to earn her PhD under the direction of Prof. Garret Miyake. Her research interests are focused on the design of organic photocatalysts for controlled radical polymerizations.

Blaine G. McCarthy received her bachelor's degree in chemistry from Clark University in May of 2015. As an undergraduate she worked under Professor Charles Jakobsche on the synthesis of unnatural amino acids. In the fall of 2015 she joined the group of Dr. Miyake at the University of Colorado at Boulder in pursuit of a doctoral degree. Her current research is on the development of photoredox catalysts for organocatalyzed atom transfer radical polymerization.

Chern-Hooi Lim is a postdoctoral researcher in the group of Garret Miyake in the Chemistry and Biochemistry Department at the University of Colorado, Boulder. He earned his Ph.D. with Charles Musgrave in the Chemical and Biological Engineering Department at the University of Colorado, Boulder in 2015. His current research interests include small molecule and polymer synthesis via visible light organo-catalyzed photoredox catalysis. He applies combined quantum mechanical modeling and organic synthesis approaches to organic photoredox catalyst design and elucidates mechanisms in catalysis and photochemistry.