Rational Design of Photocatalysts for Controlled Polymerization: Effect of Structures on Photocatalytic Activities

Abstract

Over the past decade, the use of photocatalysts (PCs) in controlled polymerization has brought new opportunities in sophisticated macromolecular synthesis. However, the selection of PCs in these systems has been typically based on laborious trial-and-error strategies. To tackle this limitation, computer-guided rational design of PCs based on knowledge of structure-property-performance relationships has emerged. These rational strategies provide rapid and economic methodologies for tuning the performance and functionality of a polymerization system, thus providing further opportunities for polymer science. This review provides an overview of PCs employed in photocontrolled polymerization systems and summarizes their progression from early systems to the current state-of-the-art. Background theories on electronic transitions are also introduced to establish the structure-property-performance relationships from a perspective of quantum chemistry. Typical examples for each type of structure-property relationships are then presented to enlighten future design of PCs for photocontrolled polymerization.

Graphical Abstract

1. INTRODUCTION

1.1. Photocatalysts for Chemical Transformations

Catalysts are effective in flattening the energy profile of a chemical reaction. That is, by virtue of an appropriate catalyst, a reaction with an insurmountable energy barrier can take place with appreciable yield under milder conditions. For example, biological systems rely on enzymes as catalysts to catalyze complex biosynthetic reactions that produce the chemicals needed to sustain life.^1–3 By mimicking this superior functionality, artificial catalysts have been progressively developed that have enabled a myriad of industrial reactions.^4–7 This catalytic strategy has become commonplace and has broken new ground for chemical synthesis and chemical engineering, facilitating the production of commodity chemicals, pharmaceuticals, and finding use in minerals and petroleum refining, and energy generation, among many others.^8–10

In particular, photocatalysts (PCs) are a special class of catalyst which harness light and convert solar energy into chemical energy.¹¹ A PC operates by first absorbing light to generate an electronically excited state (PC*), which can then undergo an electron or energy transfer reaction. While many photosensitized chemical reactions exist, PCs provide greater specificity for chemical reaction activation and thus limit unwanted side reactions.¹² With PCs, the use of light as an external stimulus provides temporal/spatial control over the reaction and allows chemical transformations over a wider temperature range, including at ambient temperatures. These merits have led to a revolution in organic^11,13,14 and macromolecular^7,15–17 synthesis.

Macmillan, Yoon, Stephenson, and others led pioneering discoveries of new light-driven organic transformations catalyzed by PCs.^{11,14,18–20} Though highly efficient, many of the earlier PCs only work with ultraviolet (UV) or violet/blue light. Another limitation is their reliance on precious transition metal centers such as iridium and ruthenium.^11,14 To circumvent the toxicity and natural scarcity associated with these rare transition metals,²¹ fully organic PCs were carefully developed with the aid of a highly sophisticated organic synthesis tool-kit,¹³ allowing for organic PCs with diverse functional groups and readily customizable properties. As such, by rational design, specific and targeted functionalities can be judiciously incorporated into organic PCs, which greatly increases the versatility of these PCs for applications in chemical and material syntheses.²²

Nevertheless, a rational strategy for constructing PCs is still needed. Due to sophistication and versatility of organic chromophore cores and substituents, specification of an “ideal” PC has been usually preceded by synthesis of a library of premodified compounds coupled with laborious screening. Without a priori knowledge, this lab-based strategy can be rather costly and poorly transferable between different areas and applications. In this regard, further efforts are needed to unravel the structure-property-performance relationships between the chemical structure of PCs and performance of photocontrolled polymerization. Indeed, by understanding the underlying mechanisms, the catalytic performance of PCs can be correlated with a range of properties including optical, photophysical, and electrochemical properties that govern the consecutive light absorption, excited state evolution, and electron/energy transfer processes in photocatalytic cycles.^13,23 These predictive relationships are the key to enabling the rational design of PCs.²⁴ Notably, PC design aided by theoretical calculations can reduce labor and time spent in discovering new functional PCs. In this vein, and because of a concomitant surge in the number of new PC structures and their applications, exceptional advances have been seen in the rational design of PCs over the past decade.

1.2. Photocatalysts for Photocontrolled Polymerization

Extending beyond small molecule transformations, photochemical transformations have endowed macromolecular synthesis with exceptional reaction control under mild and ambient conditions for the fabrication of functional materials, finding applications, for example, in electronics, coatings, and dentistry.¹⁵ Indeed, using photochemical reactions for regulating polymerization can allow for many external control approaches in materials processing, with a typical example being photoresist technology for the preparation of microprocessors.^25–27 If photochemical reactions undergo a catalytic cycle to reversibly activate and deactivate the polymerization, such a system is termed photocontrolled polymerization. Coupled with various modern techniques for polymer synthesis, which allow fine control over chain growth processes, photocontrolled polymerization catalyzed by PCs has provided a versatile toolbox for manufacturing functional polymer materials. The merits of photocontrolled polymerization systems include temporal and spatial control imparted by light,¹⁵ reaction orthogonality arising from light absorptions at discrete wavelengths,^28–34 and precise molecular weight control aided by modern polymerization techniques.¹⁷ Over the past decade, photocontrolled polymerizations have found broad applications, such as producing antimicrobial polymers,^35,36 decorating biocompounds or organisms with functional polymers,^37,38 synthesizing polymeric micelles with complex architectures,^39–43 surface patterning,^44–47 and 3D/4D additive manufacturing.^48–50 In addition, within a photocontrolled polymerization system, PCs uniquely regulate the activation/deactivation cycle; propagating species are generated under irradiation to allow chain growth and reversibly recombine with desired chain-end groups to become dormant upon cessation of irradiation. This is in contrast with photoinitiated polymerization systems, which may also effectively generate propagating species, but where removal of the irradiation stimulus does not lead to rapid and reversible recombination of the propagating species.⁵¹

Given the commanding role of PCs in controlling the activation/deactivation cycles in photocontrolled polymerizations, the performance of photocontrolled polymerizations, i.e., the efficiency,^52–54 robustness/tolerance to reaction conditions,^55–58 and orthogonality to different systems,^31,59 are highly dependent on the properties of PCs. However, early PCs for photocontrolled polymerization systems were usually selected by screening transition-metal-based PCs^60–63 and organic PCs^64–67 that were reported successful in organic synthesis.^13,14,68 Although some of the early PCs for polymerization were selected based on the similarity of chemical structures with respect to established examples,^69–72 such approaches are still essentially trial-and-error and do not properly take into account structure-property relationships. It was not until recently that the rational design of PCs was achieved by utilizing the established structure-property relationships to selectively tailor properties and functionalities of chromophores by the aid of computational analysis.^52,73–75

In this regard, the rational design of PCs with application-specific features has become imperative for addressing the central challenges confronted in various photocontrolled polymerization systems and expanding their application scope, such as decreasing the catalyst loading,^52,54 ultrafast polymerization for rapid materials manufacturing,⁴⁹ polymerization in ultralow volume for high-throughput systems,³⁵ and oxygen-tolerant polymerization for open-air production of polymers.⁵⁵ Apart from these, one of the most appealing advantages of rational PC design lies in its potential for developing orthogonal polymerization systems,^28,29 where carefully considered combinations of PCs and photosystems consequently result in numerous possibilities for selectivity and orthogonality to enable precise control over complex hybrid polymerization systems.

1.3. Scope of the Review

This review aims to summarize the guiding principles (structure-property-performance relationships), theories, and methodologies which can be applied for the design of PCs in photocontrolled polymerizations. Particular focus will be given to photoredox catalysts as a major and most useful type of PC for catalyzing photocontrolled polymerization. Photoredox catalysts operate via a specific mechanism where the excited state PC undergoes electron transfer to activate the initiator and subsequent back electron transfer to deactivate the initiator. In the following context, “PC” exclusively refers to photoredox catalysts. This review serves as a guide to researchers who seek to rapidly design a PC for a photocontrolled polymerization system with specific functionalities and demand for certain application scenarios. Accordingly, this review will not make an exhaustive list of existing PCs but will instead provide methodologies and tools allowing rational design of PCs. We will begin with a brief outline of existing photocontrolled polymerization techniques and commonly employed PC series. Background theories and knowledge required for rational PC design are addressed in detail, including those for light absorption and excitation of PCs and rate constants of electronic transitions and electron transfer reactions. Correlation between performance of photocontrolled polymerization systems and key PC properties of concern are summarized. As the main body of the review, we will then successively present the knowledge established on PC structure-property-performance relationships and their use in the control of polymerizations.

2. PHOTOCONTROLLED POLYMERIZATION

Photocontrolled polymerization encompasses a broad range of polymerization techniques, including radical, cationic, ring opening, and ring-opening metathesis processes. In this section, these polymerizations are analyzed and classified according to their photocatalytic mechanisms.

2.1. Photocatalysis in Photocontrolled Polymerization

Photocontrolled polymerization is a particular kind of polymerization controlled by photocatalysis in which the PC reversibly activates and deactivates the polymerization. As shown in Scheme 1, photocatalysis is governed by photophysical processes (1), activation reactions (2), and deactivation reactions (6), whereas polymerization is performed via initiation (3), chain growth (4), and chain transfer reactions (5). The first step (1) of photocatalysis involves several photophysical processes such as photon absorption, fluorescence, phosphorescence, internal conversion (IC), and intersystem crossing (ISC; ¹PC* → ³PC*). In the second step (2, activation step), the excited PC* reacts with an initiator through electron or energy transfer reactions. As a consequence, the initiator reaches a new electronic state, which decomposes in step (3) to form a propagating species and an end group species. The former can react with monomers, thereby initiating the polymerization. Upon sequential addition of monomers, the propagating species becomes polymeric and increases its chain length. During this chain growth (4), the propagating polymer chain can in some cases be deactivated by chain transfer (5) or by reversible recombination (6) to form a dormant chain, thereby completing the catalytic cycle. Since the photophysical (1), activation (2), and deactivation (6) steps are the key steps of photocontrolled polymerization, they are the focus of this review.

Scheme 1.

Key Steps in Photo-Controlled Polymerization

To illustrate the various photophysical processes in step (1), we can invoke a Jablonski diagram (Scheme 2), where the electronic states and associated vibrational levels of a molecule are shown. Upon absorbing a photon of appropriate frequency, the molecule initially at the vibrational ground state (ν = 0) of the electronic ground state S₀ (i.e., S_0,0) is vertically excited to some vibrational excited state (ν > 0) of S₁ (denoted as S_1,ν) or higher states in view of the Franck–Condon principle.^76,77 Thereafter, the molecule undergoes rapid vibrational relaxation to S_1,0. From S_1,0, the excited molecule can vertically return to S_0,ν via fluorescence by emitting a photon (Scheme 2, green arrow), which is also governed by the Franck–Condon principle. Alternatively, the excited molecule can undergo a nonradiative internal conversion (IC S_1,0-S_0,ν), with the extra electronic energy being converted into vibrational energy to be released through vibrational relaxation. When the spin–orbit coupling is appreciable, the molecule at S_1,0 can also access a triplet manifold (e.g., ISC S_1,0-T_2,ν) and then rapidly release the vibrational energy. Similarly, the T_1,0 molecule can reach S_0,ν through phosphorescence by emitting a photon, or via a nonradiative ISC T_1,0-S_0,ν process, followed by the release of vibrational energy. Generally, vibrational relaxation and IC between S_n,0 and S_1,ν and between T_n,0 and T_1,ν are fast events with a time scale at the picosecond level, whereas fluorescence, phosphorescence, IC S_1,0-S_0,ν, and ISCs T_1,0-S_0,ν and S_1,0-T_n,ν are slower events characteristic of a time scale between nanoseconds and seconds.

Scheme 2. Jablonski Diagram of Excited States^a.

^aNote: A, absorption; F, fluorescence; P, phosphorescence; IC, internal conversion; ISC, intersystem crossing; and VR, vibrational relaxation.

In principle, both ¹PC* and ³PC* generated from step (1) can serve as the reactant in the activation step (2). However, in photocontrolled polymerization systems, the PC is present at low concentration (typically 10–1000 ppm relative to the initiator). Therefore, PC* must be sufficiently long-lived to diffuse and interact with the initiator. Due to the spin-forbidden feature of the ISC and phosphorescence, the lifetime of ³PC* (microseconds to milliseconds) is usually much longer than that of ¹PC* (nanoseconds to microseconds). Therefore, ³PC* rather than ¹PC* is usually the major active species in step (2) of photocontrolled polymerization, unless the concentration of PC or the initiator is high or PC and the initiator form a complex before photoexcitation. For this reason, this review mainly focuses on activation via the ³PC* species.

The collision of ³PC* with an initiator R-X forms an exciplex ³(PC/R-X)*, which can then provide different forms which are symbolized by PC^•+/(R-X)^•−, PC^•−/(R-X)^•+, or PC/³(R-X)*. The first two forms (PC^•+/(R-X)^•−, PC^•−/(R-X)^•+) are obtained by electron transfer (ET), whereas the last one is obtained by triplet energy transfer (TET). The three forms correspond to three different catalytic cycles as shown in Scheme 3. In the oxidative quenching of ³PC* (Scheme 3A), an electron is transferred from ³PC* to the initiator R-X, resulting in a cationic P^•+ and an anionic (R-X)^•−. The latter is further decomposed to an end group anion X⁻ and a propagating radical R^• for radical polymerization. PC^•+ and X⁻ may form a PC^•+/X⁻ complex in the solution, thereby facilitating the deactivation process where PC^•+/X⁻ and another propagating species R′^• recombine into PC and R′-X to complete the catalytic cycle. In contrast, in the reductive quenching of ³PC* (Scheme 3B), an electron is transferred from R-X to ³PC*, leading to an anionic PC^•−, an end group radical X^•, and a propagating cation R⁺ which can be employed for cationic polymerization. Energy transfer from ³PC* to R-X via TET mechanism (Scheme 3C) results in the formation of homolytic cleavage of ³(R-X)*, with an end group radical X^• and a propagating radical R^• for radical polymerization.

Scheme 3. Catalytic Cycles of (a) ET via Oxidative Quenching Pathway, (b) ET via Reductive Quenching Pathway, and (c) TET in Photocontrolled Polymerization^a.

^aNote: R-X, initiator; P_n-X, polymer chain capped with an end group X; X⁻ or X^•, end group in (a)/(b,c); R^• and P_n^•, propagating radicals for radical polymerization in (a,c); R⁺ and P_n⁺, propagating cations for cationic polymerization in (b); and M, monomer.

2.2. Classification of Photocontrolled Polymerization Systems

The distinct catalytic cycles depicted in Scheme 3 allow for an easy classification of photocontrolled polymerization systems. Both the oxidative quenching (Scheme 3A) and TET (Scheme 3C) pathways produce propagating radicals R^•, which are suitable for photoreversible deactivation radical polymerization (photo-RDRP, Scheme 4). Hence, photo-RDRP is compatible with a broad range of common vinyl monomers including acrylates, methacrylates, acrylamides, methacrylamides, styrenics, and others. In contrast, the reductive quenching pathway (Scheme 3B) produces propagating cationic species (R⁺), which are suitable for photoliving cationic polymerization (photo-LCP) and photoring opening metathesis polymerization (photo-ROMP).^78,79 As shown in Scheme 4, photo-LCP is compatible with vinyl ethers, styrenics, and heterocyclic monomers (lactone and others), whereas photo-ROMP can polymerize many cyclic olefins.

Scheme 4. Classification of Photocontrolled Polymerizations with Their Corresponding Monomers (Left) and Compatible Catalytic Cycles (Right)^a.

^aNote: ET, electron transfer; OQP, oxidative quenching pathway; RQP, reductive quenching pathway; and TET, triplet energy transfer.

Major types of photo-RDRP techniques include photocontrolled atom transfer radical polymerization (photo-ATRP)^{51,60,64,72,80–85} and photoinduced electron transfer-reversible addition-fragmentation chain transfer (PET-RAFT) polymerization,^{55,56,61,73,75,86} while other types include photocontrolled-nitroxide-mediated radical polymerization (photo-NMP),^87,88 photocontrolled-cobalt-mediated radical polymerization (photo-CMRP),^89,90 photocontrolled-iodine transfer polymerization (photo-ITP),⁹¹ photocontrolled-reversible complexation mediated living radical polymerization (photo-RCMP),⁹² and photocontrolled-organotellurium-mediated radical polymerization (photo-TERP),⁹³ as listed in Scheme 4. Major types of photo-LCP techniques include photocationic RAFT polymerization^31,65,94 and photocationic NMP (Scheme 4).⁹⁵ In the absence of additional cocatalysts, photo-RDRP systems follow an oxidative quenching pathway. On the other hand, photo-LCP systems and photo-ROMP follow a reductive quenching pathway. Interestingly, PET-RAFT polymerization and photo-ATRP can also proceed through a reductive quenching pathway in the presence of additional electron-donating cocatalysts like tertiary amines.^66,67,96,97 While there is evidence for TET in some mechanistic pathways for photocontrolled polymerization (such as photo-NMP,⁸⁸ photo-ITP,⁹¹ and isolated cases of PET-RAFT polymerization^98–100 and photo-TERP⁹³), ET reactions are mostly accepted as the dominant mechanism. As such, this review will only focus on ET pathways and indicate any possible involvement of TET where needed. It should also be noted that many of these polymerization techniques can also proceed through different catalytic mechanisms in the presence of additives,^66,67,96 which is beyond the scope of this review.

2.2.1. Photocontrolled Atom Transfer Radical Polymerization (ATRP).

Photo-ATRP typically proceeds via an oxidative quenching mechanism, as shown in Scheme 5A. Reductive quenching mechanisms are also possible, but the addition of sacrificial reductants results in polymers with higher dispersity.^101,102 By absorption of a photon, the ground state PC is first excited to the S_n state subsequently relaxing to S₁ (¹PC*); it can then undergo ISC to reach a long-lived T₁ state (³PC*). Either ¹PC* or ³PC* can reduce the ATRP initiator R-X (activation of photo-ATRP) to yield a propagating carbon-centered radical capable of initiating radical polymerization (monomer propagation), a halide ion X⁻ (a bromide ion Br⁻, or a chloride ion Cl⁻), and the radical cation PC^•+. After a certain number of monomer additions to the carbon-centered radical, the PC^•+/X⁻ pair encounters the propagating radical R^• to oxidize it and afford a dormant halogen capped polymer chain (P_n-X) and the regenerated ground state PC (deactivation of photo-ATRP). Because of the presence of the end group in P_n-X, the polymer chain can be repetitively reactivated by ³PC* for continued monomer additions followed by recapping of the chain carrier with the X⁻. The greatest advantage of photo-ATRP systems over the traditional thermally mediated copper-catalyzed ATRP systems is the ability to externally and reversibly activate and deactivate the ATRP process with visible light to control chain growth. As presented in Scheme 5B, common initiators proven suitable for photo-ATRP include ethyl-α-bromophenylacetate (EBPA), ethyl-α-chlorophenylacetate (EClPA), ethyl α-bromoisobutyrate (EBiB), and benzyl α-bromoisobutyrate (BnBiB).

Scheme 5. (A) Proposed Mechanism of Photo-ATRP via an Oxidative Quenching Pathway and (B) Chemical Structures of ATRP Initiators Commonly Used in Photo-ATRP^a.

^aNote: ISC, intersystem crossing; M, monomer; X, bromine (Br) and chlorine (Cl) are the most commonly employed end groups (in some specific cases, iodine (I) was also employed); EBPA, ethyl-α-bromophenylacetate; EClPA, ethyl-α-chlorophenylacetate; EBiB, ethyl α-bromoisobutyrate; and BnBiB, benzyl α-bromoisobutyrate.

An important factor in the activation and deactivation of photo-ATRP is the ET reaction, specifically whether ET occurs via inner-sphere (ISET) or outer-sphere (OSET) electron transfer. For the activation process, the most accepted mechanism is OSET.^72,103 The corresponding products are PC^•+ and a negatively charged initiator (R-X)^•− which decomposes into a propagating radical species (R^•) and a catalyst-halide complex, i.e., PC^•+/X⁻ (OSET-2-body in Scheme 6). While this complex can reversibly dissociate to form free ions (i.e., PC^•+ and X⁻) in solution, this is an equilibrium process and is highly dependent on the nature of the PC, the halide anion X⁻, and the solvent. Indeed, the solvent polarity will have a significant effect, influencing the complex binding energy and lifetime of these species, and correspondingly, the ability for efficient deactivation of the propagating radical species to occur.¹⁰³

Scheme 6. Activation Pathways Proposed in Photo-RDRP^a.

^aNote: OSET-2-body, outer sphere electron transfer (OSET) that generates 2 bodies including the PC^•+/X⁻ ion pair and R^•; OSET-3-body, OSET followed by decomposition of (R-X)^•− into X⁻ and R^•.

Another activation pathway (OSET-3-body in Scheme 6) is also possible but results in three separate species (R^•, PC^•+, and X⁻).⁷² In this case, deactivation of photo-ATRP requires the recombination of all three species to provide a dormant polymeric species and return the catalyst to the ground state (i.e., 3-body-OSET in Scheme 7). Since such a three-body collision is entropically improbable, especially given the low concentrations of these species in solution, a two-body collision is a more likely pathway for deactivation in photo-ATRP (i.e., 2-body-OSET in Scheme 7).^72,103 This two-body collision thus relies on the persistence of the PC^•+/X⁻ complex in solution (produced upon the activation process via OSET-2-body in Scheme 6). Miyake and co-workers calculated the standard complexation Gibbs free energy for the formation ΔG⁰_complex) of complexes featuring oxidized N,N-diaryl dihydrophenazines coupled with bromide anions, i.e., PC^•+/Br⁻.¹⁰³ For the PC, 5,10-di(2-naphthyl)-5,10-dihydrophenazine, the ΔG⁰_complex in DMF was found to be −4.2 kcal/mol, while the value in THF was −11.1 kcal/mol, suggesting that the formation of these complexes is exergonic in both these solvents. Notably, the more exergonic ΔG⁰_complex in THF was ascribed to the lower solvent polarity, which favored complex association. Interestingly, the deactivation behavior of O-ATRP using this and similar PCs with charge transfer (CT) character was found to be tunable by adjusting the solvent polarity.¹⁰³

Scheme 7. Deactivation Pathways Proposed in Photo-RDRP^a.

^aNote: 2-body-OSET, two bodies (the PC^•+/X⁻ ion pair and R^•) undergo OSET to recover PC and R-X; 3-body-OSET, three bodies (PC^•+, X⁻,and R^•) undergo OSET to recover PC and R-X; and 3-body-ISET, three bodies (PC^•+, X⁻ and R^•) undergo ISET to recover PC and R-X.

Matyjaszewski and co-workers have also studied O-ATRP mechanisms, finding that the addition of Br⁻(using tetra-n-butylammonium bromide salt) provided more efficient deactivation during the polymerization of MMA catalyzed by N-phenyl phenothiazine under 365 nm irradiation.⁷² In this case, the addition of a large excess of free Br⁻ pushed the complex-free-ion equilibrium (i.e., PC^•+/X⁻ ← → PC^•+ + X⁻) toward the formation of the complex (PC^•+/X⁻), and thus provided a higher concentration of effective deactivator, i.e., the catalyst complex, rather than the free ions. This data is consistent with Miyake and co-workers’ previous works and supports the deactivation occurring via a two-body collision event between the propagating radical species and the associated PC^•+/X⁻ complex. These observations clearly supported the OSET-2-body (Scheme 6) as the dominant activation pathway and 2-body-OSET (Scheme 7) as the dominant deactivation pathway.

Further questions regarding the activation and deactivation mechanisms can be answered by studying reaction paths of these systems along with their energy barriers via quantum chemical calculations. The transition state theory¹⁰⁴ (TST) or Rice-Ramsperger-Kassel-Marcus theory¹⁰⁵ (RRKM) can be used to evaluate kinetic aspects of the minimum energy reaction paths. Alternatively, nonadiabatic molecular dynamics¹⁰⁶ may be performed to simulate the activation/deactivation processes.

Initial photo-ATRP systems were catalyzed by transition-metal-based PCs (Scheme 8A). The first example of a photo-ATRP was reported by Fors and Hawker,⁶⁰ who utilized fac-[Ir(ppy)₃] (Scheme 8A, 1) as a PC to reversibly activate and deactivate the ATRP process. Following this initial work, more transition-metal-based PCs were exploited. For example, Le Lagadec, Alexandrova, and co-workers explored the use of Ru-based complexes (Scheme 8A, 2,3) to effectively catalyze photo-ATRP of methyl methacrylate (MMA), n-butyl acrylate (BA), and styrene (St) under visible light in the presence of EBiB as the initiator.⁶² Lalevée, Fensterbank, Goddard, Ollivier, and co-workers discovered a novel Au-based complex (Scheme 8A, 4) as PC for photo-ATRP of various methacrylates with EBPA as the initiator, under sunlight or visible light.¹⁰⁷ Similarly, Lalevée and co-workers further demonstrated a class of Fe-based PCs (Scheme 8A, 5–7) for photo-ATRP of MMA and BA initiated by EBPA.¹⁰⁸ Poly, Lalevée, and co-workers reported a Cu-based PC¹⁰⁹ (Scheme 8A, 8) that also involves an oxidative quenching pathway in photo-ATRP.⁸⁵

Scheme 8.

Chemical Structures of Representative PCs in Photo-ATRP Systems, Including (A) Transition-Metal-Based PCs, (B) Perylene, (C) Phenothiazine Derivatives, (D) Dihydrophenazine Derivatives, and (E) Phenoxazine Derivatives

In an effort to eliminate transition metal residues for more widespread applications, researchers started to seek metal-free alternatives to catalyze photo-ATRP. In this regard, Fors, Hawker, and co-workers demonstrated the use of an organic PC N-phenyl phenothiazine (Scheme 8C, 10) to effectively catalyze organocatalyzed photo-ATRP (O-ATRP) of methacrylates with EBPA as the initiator via an oxidative quenching pathway.⁶⁴ Matyjaszewski and co-workers further expanded the monomer scope of O-ATRP to acrylonitrile.⁸¹ At the same time, independently, Miyake and Theriot demonstrated the successful implementation of perylene (Scheme 8B, 9) to catalyze O-ATRP of MMA with EBPA as the initiator in a controlled manner.¹¹⁰ Matyjaszewski and co-workers revisited the proposed mechanism behind O-ATRP and expanded the scope of organic PCs to more phenothiazine-based PCs including π-extended derivatives (Scheme 8C).⁷²

Since 2016, aided by computationally directed discovery, Miyake and co-workers consecutively developed and explored broad libraries of dihydrophenazine (Scheme 8D),⁸⁰ phenoxazine (Scheme 8E),^74,111 and dimethyl-dihydroacridine derivatives (Scheme 9A)¹¹² as efficient organic PCs for photo-ATRP. Remarkably, reduced catalyst loadings were achieved by pursuing core-extended organic PCs with CT excited states, which broadened absorption peaks to facilitate visible light absorption at longer wavelengths.^54,111 The importance of the N-aryl substituent was highlighted with phenoxazines, where changing the N-aryl group from phenyl to 1-naphthalene greatly enhanced the ISC process to increase the triplet quantum yield (Φ_T) to long-lived T₁ states.¹¹³ Kim et al. reported a computer-aided molecular design/screening strategy to directly construct organic PCs with donor–acceptor (D–A) scaffolds from a large library of common organic donor and acceptor moieties.⁵² In addition, highly conjugated thienothiophene derivatives were investigated by Yagci and co-workers as PCs for application in O-ATRP (Scheme 9B).¹¹⁴ In photo-ATRP (or O-ATRP), the initiator efficiency is defined as the number of polymer chains with the “living” end group divided by the total number of initiators employed (i.e., the portion of initiators that are activated). High initiator efficiency can be achieved by using PCs with the following critical characteristics: (i) sufficiently reducing excited states, (ii) maximized population of long-lived T₁ states, (iii) highly oxidizing PC^•+ that can effectively deactivate the propagating radical, and (iv) PCs that are chemically inert to radicals.

Scheme 9.

Chemical Structures of Representative PCs in O-ATRP Systems, Including (A) Dimethyl-Dihydroacridine Derivatives and (B) Organic Donor–Acceptor Scaffolds

2.2.2. Photo-Induced Reversible Addition–Fragmentation Chain Transfer (RAFT) Polymerization.

In the case of additive-free PET-RAFT polymerization, an oxidative quenching cycle is followed for the generation of carbon-centered propagating radicals, i.e., via an ET reaction^61,115 (Scheme 10A). Alternatively, a TET (Scheme 10B) mechanism can also be followed in certain cases. The ET mechanism is characterized by an electron transfer from the excited state PC* to the RAFT agent, which subsequently fragments to form a propagating radical species, an anionic thiocarbonylthio species, and a P^•+ species. Recombination closes the catalytic cycles and generates a dormant thiocarbonylthio-capped polymer chain.

Scheme 10. Proposed Mechanisms for PET-RAFT Polymerization Activated by (A) ET via the Oxidative Quenching Pathway or (B) TET and (C) Chemical Structures of Thiocarbonylthio RAFT Agents Commonly Used in PET-RAFT Polymerizations^a.

^aTrithiocarbonates: BTPA, BSTP, CDTPA, and CPDTC; dithiobenzoates: CPADB and CDB; xanthate. Note: M, monomer; Z, Z group of the RAFT agent.

For the activation step in PET-RAFT polymerization, an outer-sphere mechanism (e.g., OSET) should proceed regardless of an electron or energy transfer mechanism. However, the products of these reactions do differ considerably; for PET-RAFT polymerization via an ET mechanism, the products are a radical-cation PC species (P^•+), an anionic radical thiocarbonylthio species (X^•−), and the standard propagating radical (R^•). Hence, the ET mechanism will require recombination of three species, which will be entropically improbable, similar to the photo-ATRP deactivation. Thus, it is possible that some PC^•+X^•− species in PET-RAFT polymerization display exergonic ΔG⁰_complex values and thus avoid the three-body collision requirement. It must be noted that there is limited information regarding the deactivation in PET-RAFT polymerization; further studies are required to fully elucidate this behavior.

As in any chemical process, both the thermodynamics and kinetics of the system will determine the resulting reaction behavior. For both PET-RAFT and photo-ATRP polymerization systems, removal of the light source results in rapid cessation of monomer propagation.¹¹⁶ This behavior indicates a rapid deactivation process, which is consistent with retention of the associated P^•+X^•− complex, allowing bimolecular recombination to close the catalytic cycles. Again, these systems are highly dependent on the reaction conditions as well as the structure of the chemical species involved, and investigation of specific systems should be performed to uncover the reaction behavior more clearly.

Alternatively, for a TET mechanism, the propagating radical species is formed via cleavage of the RAFT agent weak C–S bond, in a similar fashion to the photoiniferter process proposed by Otsu.¹¹⁷ This results in the formation of a radical thiocarbonylthio species (X^•) and a propagating radical species (R^•). In this case, a TET mechanism will provide only two products that require recombination to close the catalytic cycle. The only TET-activated systems discovered so far are PET-RAFT catalyzed by Ir- and Ru-based catalysts, reported by Allonas, Boyer, and co-workers using transient absorption spectroscopy and computational studies.^98,100 As recently reported by Falvey and co-workers,¹¹⁸ it should be noted that higher triplet states such as T₂ of the RAFT agent may be responsible for the TET-activated systems as T₁ does not possess sufficient excitation energy to cleave the C–S bond but T₂ presented the capability in some cases. Specifically, they found that some dithioester RAFT agents exhibited low-energy T₁ states with adiabatic values calculated at ~33 kcal mol⁻¹, leading to no (or negligible) decomposition in laser flash photolysis observations. On the other hand, the trithiocarbonate RAFT agents were found with ~44 kcal/mol T₁ excitation energies, which are higher but still could not lead to effective decomposition. On the contrary, anionic RAFT^•− resulting from ET reactions were found to undergo rapid decomposition to generate radicals. On the basis of these results, Falvey and co-workers drew a conclusion that PET-RAFT polymerization catalyzed by dyes with low-energy T₁ cannot activate polymerization by TET but should by ET reactions instead.¹¹⁸ Further evidence for an ET mechanism in longer wavelength absorbing dyes characteristic of low-energy T₁ has also been provided. For example, Smith, Seal, and co-workers combined experimental and computational evidence to conclude that the commonly used ZnTPP catalyzes PET-RAFT via an ET process.¹¹⁵ Hence, at least for PCs with longer-wavelength absorption,⁵⁶ an ET mechanism is most likely.

Similar to that mentioned in section 2.2.1, accurate identification of an ET/TET mechanism in activation/deactivation of PET-RAFT polymerization should be conducted by calculating the minimum energy reaction paths, whereupon TST or RRKM methods based on the energy barrier can solve the kinetic problem.

Interestingly, the robustness of PET-RAFT polymerization has been demonstrated by Konkolewicz et al.,¹¹⁹ who showed that changes to reaction conditions such as light intensities and reactor geometries result in only marginal changes to polymerization control. While factors such as the light intensity and chemical concentrations will change the polymerization kinetics, polymer materials still display low dispersities under a range of conditions.

Inspired by photo-ATRP, Boyer and co-workers introduced Ir-based PC fac-[Ir(ppy)₃] and Ru-based PC Ru(bpy)₃²⁺ (Scheme 11A, 57) to RAFT polymerization.⁶¹ These PCs were shown to be suitable for dithiobenzoate, trithiocarbonate, and xanthate RAFT agents. With dependence on the selection of the RAFT agent, a range of monomer families were polymerizable through PET-RAFT polymerization with fac-[Ir(ppy)₃] as the PC, including (meth)acrylates, (meth)acrylamides, styrenics, isoprene, vinyl pivalate (VP), N-vinyl pyrrolidinone (NVP), and dimethyl vinylphosphonate (DVP). Boyer and co-workers explored the use of ZnTPP as a PC for PET-RAFT polymerization of acrylates, acrylamides, and styrene regulated by trithiocarbonates, as well as methacrylates regulated by dithiobenzoates.⁸⁶ Notably, although ZnTPP can be regarded as a transition-metal-based complex, it does not exhibit any metal-to-ligand charge transfer (MLCT) excited states because of the inert nature of Zn; instead, ZnTPP mainly exhibits local excitation upon exposure to light.¹¹⁵ The oxygen tolerance featured in PET-RAFT polymerization catalyzed by ZnTPP was discovered based on the capability of ZnTPP to photosensitize oxygen (O₂) via triplet–triplet annihilation (TTA) to yield singlet oxygen (¹O₂), which can be scavenged by monomers or the solvent.^6,55 Boyer and co-workers further reported water-soluble derivatives for aqueous PET-RAFT polymerization.^43,120

Scheme 11.

Chemical Structures of Representative PCs Employed in PET-RAFT Polymerization Systems, Including (A) Transition-Metal-Based PCs fac-[Ir(ppy)₃] (1) and Ru(bpy)₃²⁺ (57), (B) Porphyrin Derivatives, (C) Naturally Derived Chlorophyll a (62), Bacteriochlorophyll a (63), and Pheophorbide a (64) and (D) Xanthene Derivatives

The naturally derived chlorophyll a (62),⁷⁰ bacteriochlorophyll a (63),⁶⁹ and pheophorbide a,⁵⁹ (64) shown in Scheme 11C have also been explored for mediating PET-RAFT polymerization. While the former two natural PCs are generally successful, pheophorbide a (64) was found to only be compatible with dithiobenzoates for the polymerization of methacrylates and unsuccessful for the polymerization of acrylates in the presence of trithiocarbonates. The metal-free tetraphenyl porphyrin (TPP, Scheme 11B, 60) as a red-light-absorbing organic PC displayed similar performance to pheophorbide a and is active for dithiobenzoates but inactive with trithiocarbonates.⁸⁶ Zhang and co-workers discovered a reduced TPP with a bacteriochlorin core (RTPP, Scheme 11B, 61) as an effective organic PC for PET-RAFT polymerization of both acrylates and methacrylates in the presence of CDTPA or CDB as RAFT agents.¹²¹ Most encouragingly, the bacteriochlorin-based RTPP has strong absorption in the far-red to near-infrared (NIR) range of the absorption spectrum, which makes it an unprecedented NIR-absorbing organic PC for efficient PET-RAFT polymerization. To explore more candidates as organic PCs for PET-RAFT polymerization, the Boyer group screened other common organic dyes and identified a Br-substituted fluorescein derivative Eosin Y (Scheme 11D, 65) as an effective PC for PET-RAFT polymerization of MMA in the presence of CPADB as the RAFT agent.⁶⁷

In collaboration with the Miyake group, the Boyer group demonstrated the use of other halogenated derivatives of this dye class (Scheme 11D) as efficient PET-RAFT PCs and compared their performance in PET-RAFT polymerization in relation to their structure and properties.^73,75 In particular, a newly synthesized Br-substituted xanthene dye tetra-bromofluorescein (Scheme 11D, 67) was discovered as a highly sensitive and efficient pH/light dual-responsive organic PC for PET-RAFT polymerization of acylates, acrylamides, methacrylates, and methacrylamides in the presence of BTPA, CPADB, and CPDTC (Scheme 10C).⁷⁵ The merits of halogenated xanthene dyes are their exceptionally strong green-to-orange light absorption and efficient ISC to yield considerable long-lived T₁ states. These properties make them ideal organic PCs for PET-RAFT polymerization, requiring only ~20 ppm catalyst loadings to yield an excellent polymerization efficiency.⁷³ Good oxygen tolerance was also achieved in these xanthene dye-catalyzed systems due to their high Φ_T, which enables scavenging of O₂ species via TTA. To maximize the oxygen tolerance performance in organic PET-RAFT polymerization, another xanthene dye substituted with four heavy bromine atoms and four iodine atoms (Scheme 11D, 70) was judiciously designed and synthesized, exhibiting the best oxygen tolerance among organic PCs for PET-RAFT polymerization.⁷³ Additionally, a few effective candidates developed for O-ATRP were also implemented for PET-RAFT polymerization, which were demonstrably successful.^53,122,123

2.2.3. Photocontrolled Living Cationic Polymerization (LCP).

A typical photo-LCP system is the photocationic RAFT polymerization developed by Fors et al., which is characteristic of a reductive quenching pathway.⁶⁵ Photocationic RAFT polymerization is activated by a photoexcited PC abstracting an electron from the RAFT agent IBDTC or IBTTC, forming a propagating cation and a leaving thiocarbonylthio anion (Scheme 12). The propagating cation can then undergo a cationic RAFT process to polymerize monomers in a controlled manner. By occasional diffusive collision, the negatively charged PC^•− can reduce the propagating cation to close the catalytic cycle and restore the ground state PC and provide a polymer chain capped with a RAFT agent end group (deactivation). Photocationic RAFT polymerization is capable of effectively polymerizing vinyl ether monomers because of the electron-donating alkyloxy ligands attached to the vinyl group, which facilitates the propagation of cations. Common RAFT agents that can regulate photocationic RAFT polymerization include IBDTC and IBTTC (Scheme 12B), whose R group is an alkyloxy-substituted secondary carbon that mimics the poly-(vinyl ether) chain to optimize initiation of photocationic RAFT polymerization. Although ³PC* is denoted as the active species for the subsequent reductive quenching PET (Scheme 12), it still remains unclear whether ³PC* or ¹PC* or both should be the dominant active excited state for photocationic RAFT polymerization,¹²⁴ despite the fact that ³PC* appears to be more likely considering its much longer lifetime.

Scheme 12. (A) Proposed Mechanism for Photocationic RAFT Polymerization and (B) Common RAFT Agents and (C) Proposed Mechanism for Photocationic NMP Polymerization, (D) Common Monomers, and (E) Common Alkoxyamine Initiators^a.

^aNote: IBDTC, S-1-isobutoxyethyl N,N-diethyl dithiocarbamate; IBTTC, S-1-isobutoxylethyl S′-ethyl trithiocarbonate. M, monomer; Z, RAFT agent Z group.

In photocationic RAFT polymerization, the activation/deactivation mechanisms have been much less studied. Commonly, a cationic ¹PC⁺ is used to ensure sufficient oxidizing capability in excited states, and hence ³PC⁺* is populated. Accordingly, in an OSET scheme (reductive quenching pathway), PC^• and (R-X)^•+ are formed and the latter decomposes to X^• and R⁺ for cationic polymerization. Although current reports assumed that PC^• and X^• were formed, existing evidence may not rule out the possibility of forming ¹PC⁺X⁻ ion pairs from the activation process, especially considering the tendency of the RAFT end group to be anionic (X⁻, as with those in PET-RAFT) and the tendency of the PC being restored to the original cationic ¹PC⁺ ground state. Indeed, the formation of the ¹PC⁺X⁻ ion pair may facilitate a two-body collision with R⁺ which may effectively deactivate the polymerization to recover ¹PC⁺ and R-X. Overall, the exact activation/deactivation mechanisms of photocationic RAFT polymerization are still under debate and further investigations are needed, including computational study of the ISET mechanism, to uncover the reaction paths.

The first class of PCs that were proven to be effective for photocationic RAFT polymerization were triphenylpyrylium derivatives (Scheme 13A).¹²⁴ However, these organic dye-catalyzed systems still showed some dark polymerization, i.e., continued polymerization upon cessation of irradiation.^65,124 To circumvent this problem and obtain ideal temporal control for photocationic RAFT polymerization, Fors and co-workers further introduced a series of modified Ir-based complexes (Scheme 13C), which has resulted in greatly improved temporal control for photocationic RAFT polymerization, despite some reduction in polymerization rates.⁹⁴ More recently, Liao and co-workers have introduced a new metal-free cationic polymerization of vinyl ethers utilizing bisphosphonium salt doped anthanthrenes as organic PCs (Scheme 13D). These PCs present very strong visible light absorption and tunable redox potentials. The authors demonstrated temporal control as well as control of the molecular weights of a broad range of vinyl ether polymers.¹²⁵ In particular, compared with other existing organic PCs for photocationic RAFT polymerization, these bisphosphonium salts display better temporal control with no dark polymerization after ceasing light irradiation.

Scheme 13.

Chemical Structures of Representative PCs in Photo-LCP and Photo-ROMP Systems, Including (A) Triphenylpyrylium Derivatives, (B) Triphenylthiopyrylium Derivatives, (C) Substituted Ir-Based Complexes, and (D) Bisphosphonium Salt Derivatives

Another photo-LCP system is photocationic NMP developed by Kamigaito and co-workers,⁹⁵ whose photocatalytic mechanism (Scheme 12C) is similar to its RAFT counterpart (Scheme 12A), albeit with replacement of RAFT agents with alkoxyamines activated by mesolytic cleavage (Scheme 12E). With regard to the polymerization mechanism, it should be noted that alkoxyamines do not undergo chain transfer like the RAFT equilibrium and thus rely on the redox cycle to retain controllable chain growth. Apart from the vinyl ether monomers applied in cationic RAFT, photocationic NMP is also compatible with styrene derivatives (Scheme 12D).

2.2.4. Photocontrolled Ring-Opening Metathesis Polymerization (ROMP).

Photo-ROMP, also known as metal-free ROMP, was established by Boydston and co-workers.^79,126 Photo-ROMP is mechanistically distinct from traditional ROMP systems that are typically initiated using organometallic Ru, Mo, and W alkylidene complexes.^127,128 As shown in Scheme 14A, photo-ROMP operates via a reductive quenching pathway in which an excited state PC oxidizes an enol ether initiator to produce an enol ether radical cation chain carrier. This chain carrier can react with strained cyclic olefins to provide a cyclobutane radical cation intermediate species, which then undergoes rapid ring opening to form the propagating species. Subsequent reduction of the propagating species by the reduced PC^•− closes the catalytic cycle and provides an enol ether capped polymer chain. Under this mechanism, the ring opening of the cyclobutane intermediate must be sufficiently fast to overcome reduction of the radical cation intermediate to form stable cyclobutane side-products.¹²⁶

Scheme 14. (A) Proposed Mechanism for Photo-ROMP via a Reductive Quenching Pathway and (B–C) Chemical Structures of the (B) Initiators and (C) Monomers Used in Photo-ROMP^a.

^aNote: ISC, intersystem crossing; DCPD, dicyclopentadiene.

In much the same way as early PC selection for photo-ATRP and PET-RAFT was based on PCs from organic synthetic transformations, the first PCs for photo-ROMP were selected based on prior work on the photoredox-mediated synthesis of cyclobutanes under visible light by Nicewicz et al.,¹²⁹ namely triphenyl pyrylium derivatives. Furthermore, because of its reductive quenching pathway, the PC scope for photo-ROMP is similar to that for photo-LCP and includes other blue light absorbing pyrylium and thiopyrylium derivatives.⁷⁸ By systematically studying these varied pyrylium (Scheme 13A) and thiopyrylium (Scheme 13B) PCs, Boydston et al. observed higher efficiency of thiopyrylium derivatives over their corresponding pyrylium counterparts.⁷⁸ Critically, these PCs must have sufficient reduction potential in the excited state to oxidize the enol ether initiators, which typically have reduction potentials in the range of 1.4 V versus SCE.¹²⁶ Notably, however, it was shown that using PCs with higher excited state reduction potential led to slower polymerizations under analogous conditions.⁷⁸ This effect has been suggested to be a result of overoxidation of the enol ether initiators and demonstrates the need for carefully selected PC-initiator pairs. Regardless, a range of enol ether initiators have been successfully employed in photo-ROMP, as shown in Scheme 14B.

Photo-ROMP has expanded its monomer scope to a range of functional norbornene and dicyclopentadiene (DCPD) monomers (Scheme 14C) for the preparation of block (co)-polymers.^130–135 Interestingly, Boydston et al. showed that endo-DCPD, exo-DCPD monomers can be used to prepare linear polyDCPD polymers; these monomers readily form cross-linked thermoset materials when polymerized via traditional ROMP, and the linear forms have been previously inaccessible. These linear polymers could be readily cross-linked via a secondary thiol–ene reaction if required.¹³⁰ In addition, ether, silyl, and chloro-functionalized norbornenes, among others, were all shown to be suitable monomers or comonomers for photo-ROMP, leading to a broad range of functional polymers.^131,132,134 The high functional group tolerance shows the versatility of this technique and potential applicability for orthogonal chemistries.^133,134 Furthermore, photo-ROMP has been successfully performed under ambient conditions, which increases its appeal from an industrial perspective.^126,131

More recent work in photo-ROMP has shown that the counterion for the pyrylium PC can have a marked influence on the photo-ROMP.^132,136 In particular, it was shown that strongly coordinating counterions chloride, bromide, and nitrate precluded the photo-ROMP of norbornene. Conversely, noncoordinating counterions such as tetrafluoroborate (BF₄⁻), hexafluorophosphate (PF₆⁻), and tetrakis(3,5-bis-(trifluoromethyl)phenyl)borate (BAr^F₄⁻) all provided effective photo-ROMP of norbornene. More interestingly, as the size of the counterion increased from BF₄⁻ to PF₆⁻ to BAr^F₄⁻, the trans alkene content of the resulting polynorbornene decreased from 79% to 74% to 58%.¹³⁶ Moreover, solvent and temperature were also shown to affect the stereochemistry of polynorbornene produced through photo-ROMP, with >98% trans alkene content achievable using dichloromethane at low temperatures (−76 °C). Quantum chemical calculations showed that the smaller counterions stabilized a pro-trans isomer of the radical cation species formed after ring opening, leading to a higher trans-content in the final polynorbornene.¹³⁶

3. GUIDING THE DESIGN OF PHOTOCATALYSTS FOR PHOTOCONTROLLED POLYMERIZATION

3.1. Qualitative Structure-Property-Performance Relationships

Irrespective of other functionalities, by assuming the same polymerization formulation (monomer type, initiator, and solvent, etc.), key performance parameters of a photocontrolled polymerization include the initiator efficiency, dispersity of polymers, and the apparent polymerization (propagation) rate $k_{\text{p}}^{\text{app}}$. As mentioned before, the photocatalytic process initiating a photocontrolled polymerization is divided into four main stages: (i) light absorption, (ii) populating the excited state species responsible for catalysis, (iii) electron/energy transfer to generate propagating species, and (iv) the deactivation reaction to complete the catalytic cycle. Tuning key properties of the PC with respect to each of these stages (Scheme 15) will result in variations in the performance of photocontrolled polymerization systems.

Scheme 15. Roadmap of the Photocatalytic Pathway of a Typical Photocontrolled Polymerization^a.

^aNote: λ_max, maximum absorption wavelength of a given absorption peak; ε_max, maximum molar extinction coefficient of a given absorption peak; E_ex,S0–S1, excitation energy of the S₁ state; E_ex,S0-T1, excitation energy of the T₁ state; f, oscillator strength of excitation to an excited state; τ_S1, lifetime of the S₁ state; τ_T1, lifetime of the T₁ state; k_F, fluorescence rate constant; k_IC, internal conversion rate constant; k_ISC, intersystem crossing rate constant; ϕ_T, triplet quantum yield; IP, ionization potential; EA, electron affinity; E*_ox, excited state oxidation potential; E*_red, excited state reduction potential; E_USOMO, energy level of the upper singly occupied molecular orbital; and E_LSOMO, energy level of the lower singly occupied molecular orbital.

The properties of PCs depend on their electronic structures and are thus strongly correlated with their chemical structures. As a PC can be seen as a chromophore core with functional substituents attached, the properties of PCs can be manipulated by systematically changing their substituents. These functional substituents can include but are not limited to electron-donating groups, neutral groups, electron-withdrawing groups, heavy atom groups, and a myriad of conjugated groups with different electronic structures (as shown in Scheme 16A). Commonly used organic chromophore cores for a PC (which this review mainly focuses on) vary from highly reducing/oxidizing variants to large conjugated variants and water-soluble variants (Scheme 16B). By the aid of existing and new chromophore cores and substituents, PCs can be rationally designed according to the required performance of photocontrolled polymerization for a given application.

Scheme 16.

Various (A) Functional Substituents and (B) Organic Chromophore Cores Commonly Used in Designing a PC in Photocontrolled Polymerization

To this end, it is important to determine the guiding principles for rational design of a PC to attain better performance of photocontrolled polymerization (e.g., higher polymerization rates, higher initiator efficiency, and lower dispersity polymers) as well as tunable absorption wavelengths. Such guiding principles can be summarized as structure-property-performance relationships. Scheme 17 provides an overview of key relationships known to date with regard to variations of chemical groups listed in Scheme 16. These structure-property-performance relationships are discussed in section 3.1.1 for tuning light absorption, section 3.1.2 for increasing Φ_T, and section 3.1.3 for enhancing activation/deactivation efficiencies, respectively. On the other hand, existing examples which reflected or made use of these structure-property-performance relationships are demonstrated in detail in section 3.2.

Scheme 17.

Overview of Key Structure-Property-Performance Relationships Known to Date in Designing a PC

3.1.1. Tuning Light Absorption.

3.1.1.1. Tuning Absorption Wavelengths.

The absorbed photon must possess energy hv close to the energy gap between the final state S_n and the initial state S₀,^137,138 i.e., the corresponding excitation energy. Accordingly, substituents that can tune the excitation energy by moving up/down relevant orbitals should in turn tune the absorption wavelength $\lambda =\frac{c}{v}$ (vide infra).

For a typical π → π* excitation, which is useful and commonly seen in organic PCs, increasing the dimensions of the π-conjugation can stabilize the lowest π* orbital or destabilize the highest π orbital, thereby extending λ of the maximum of the absorption peak (λ_max). By contrast, directly attaching atoms with nonbonded (n) electrons (i.e., alkyloxy, alkylthio, or (di)alkylamino substituents) on the chromophore core will mix the high-lying n orbital into the highest π orbital of the chromophore core and hence extend λ_max.^139,140 Similarly, heavy halogen substituents can also mix their n orbitals into the highest π orbital, while a heavier halogen substituent with a higher-lying n orbital leads to longer λ_max. On the other hand, more electron-withdrawing substituents extend λ_max mostly by stabilizing the lowest π* orbital.

3.1.1.2. Maximizing the Molar Extinction Coefficient.

With respect to a specific absorption peak associated with an electronic transition i → f, the molar extinction coefficient ε has the maximum value ε_max at λ_max of an absorption peak and has a relation:

$${\epsilon }_{\max }\propto \frac{f_{i\to f}}{\text{Δ}{\overline{\nu }}_{\text{FWHM}}}$$ (1)

$$f_{i\to f}=\frac{4m_e\pi c}{3e^2\hslash \lambda }{|\langle {\text{Φ}}_f|\widehat{\mu }|{\text{Φ}}_i\rangle |}^2$$ (2)

$$\langle {\text{Φ}}_f|\widehat{\mu }|{\text{Φ}}_i\rangle ={\int }^{\text{}}{\text{Φ}}_f(r)\widehat{\mu }{\text{Φ}}_i(r)\text{d}r$$ (3)

where f_{i → f} is the probability of the electronic transition defining the relative strength of the electronic transition, $\text{Δ}{\overline{\nu }}_{\text{fwhm}}$ is the full width at half-maximum of the absorption peak, m_e is the weight of an electron and e is the charge of an electron, and r is the position vector representing any point in the space. $\langle {\text{Φ}}_f|\widehat{\mu }|{\text{Φ}}_i\rangle$ is the electronic transition dipole moment, which is an integral in the whole space. Usually, for a typical π → π* excitation, larger π conjugation of the chromophore scales up the distribution of the transition density ${\rho }_{i\to f}(r)={\text{Φ}}_f(r){\text{Φ}}_i(r)$, rendering the amplitude of its dipole moment $|\langle {\text{Φ}}_f|\widehat{\mu }|{\text{Φ}}_i\rangle |$ larger and hence enhancing ε_max.

3.1.1.3. Broadening the Absorption Peak.

Under the quantum-mechanical Franck–Condon principle^76,77 and the displaced harmonic oscillator (DHO) model,^137,138 higher Huang–Rhys factor S corresponds to a larger displacement between the potential energy surfaces of the two states in the electronic transition, which results in a broadening effect in the absorption peak. For example, flexible molecules usually exhibit a larger difference between the equilibrium geometries of S_n and S₀, which leads to higher S and broader absorption peaks.⁵⁶ This phenomenon is also commonly seen in the rather broad emission peaks of charge transfer states, where large deviation in equilibrium geometries is attributed to the vastly different electronic wave function of the charge transfer S_n as compared to S₀.^80,141

3.1.2. Increasing the Triplet Quantum Yield.

On the basis of the rate constants (k_F for fluorescence, k_IC for IC, and k_ISC for ISC), the quantum yield Φ_T1 of ³PC* can be calculated according to eq 4 (Scheme 18).⁷⁵ Accordingly, approaches to enhance Φ_T can be practically categorized as (i) increasing k_ISC, (ii) reducing k_F, or (iii) reducing k_IC.⁷⁵

$$\begin{array}{l}{\text{Φ}}_{{\text{T}}_1}=\left[k_{\text{ISC},{\text{S}}_1-{\text{T}}_2}\left(1-\frac{k_{\text{ISC},{\text{T}}_2-{\text{S}}_1}^{\prime }}{k_{\text{ISC},{\text{T}}_2-{\text{S}}_1}^{\prime }+k_{\text{IC},{\text{T}}_2-{\text{T}}_1}}\right)+k_{\text{ISC},{\text{S}}_1-{\text{T}}_1}-k_{\text{ISC},{\text{T}}_1-{\text{S}}_1}^{\prime }\right]\\ /[k_{\text{F}}+k_{\text{IC},{\text{S}}_1-{\text{S}}_0}+k_{\text{ISC},{\text{S}}_1-{\text{T}}_2}\left(1-\frac{k_{\text{ISC},{\text{T}}_2-{\text{S}}_1}^{\prime }}{k_{\text{ISC},{\text{T}}_2-{\text{S}}_1}^{\prime }+k_{\text{IC},{\text{T}}_2-{\text{T}}_1}}\right)\\ +k_{\text{ISC},{\text{S}}_1-{\text{T}}_1}-k_{\text{ISC},{\text{T}}_1-{\text{S}}_1}^{\prime }]\end{array}$$ (4)

Scheme 18. Jablonski Diagram for Rate Constants of Typical Excited State Decay Pathways^a.

^aNote: k_F, fluorescence rate constant; k_P, phosphorescence rate constant; k_IC, internal conversion rate constant between two electronic states of the same multiplicity; k_ISC, intersystem crossing rate constant from a singlet state to a triplet state; and k′_ISC, “reversed” intersystem crossing rate constant from a triplet state to a singlet state.

3.1.2.1. Increasing k_ISC.

In accordance with Fermi’s golden rule expression regarding ISC,^142–144 k_ISC is proportional to the squared amplitude of spin–orbit coupling matrix elements (SOCME), i.e., ${|\langle {\text{Φ}}_f|{\widehat{H}}_{\mathrm{SO}}|{\text{Φ}}_i\rangle |}^2$, whereas SOCME scales as the fourth power of the atomic charge Z_N manifesting the heavy-atom effect and also becomes appreciable when the initial and final states differ in an orbital of different angular momentum (e.g., ¹ (π,π*) → ³(n,π*)) manifesting the El-Sayed rule.¹⁴⁵ Also, an energy gap law exists such that narrower energy gap between the initial and final states is responsible for exponentially higher k_ISC.¹⁴⁶

Briefly, the heavy atom effect can be introduced by heavy substituents such as heavy halogens,⁷³ whereas the El-Sayed rule can be implemented by including ¹ (π,π*) → ³(n,π*) transition characters or ligand-to-ligand charge transfer characters in the ISC process. By minimizing the energy gap between S₁ and a final state in the triplet manifold whose energy is close to, but lower than S₁, the energy gap law takes effect.⁵²

In addition, by considering the second-order perturbation theory and the first- and second-order dependence of SOCME on nuclear motions, there is a spin-vibronic coupling effect¹⁴² commonly seen in chromophore core-twisted molecules, whose mechanism is beyond the present scope for qualitative discussions. For a detailed outline of spin-vibronic coupling, the interested reader is directed to a recent review.¹⁴²

3.1.2.2. Decreasing k_F and k_IC.

According to Fermi’s golden rule expression for fluorescence,^143,147 k_F is proportional to the squared transition dipole moment ${|\langle {\text{Φ}}_f|\widehat{\mu }|{\text{Φ}}_i\rangle |}^2$, which is greatly diminished with poor overlap between Φ_f and Φ_i. Hence, S₁ with charge transfer characters should largely retard fluorescence due to poor overlap between wave functions of S₁ and S₀. On the other hand, the IC process results from the coupling of nuclear motions into the electronic transition. Accordingly, k_IC can be effectively reduced by using more rigid and planar molecules with less nuclear mobility.

3.1.3. Completing the Catalytic Cycle by Efficient Electron Transfer Reactions.

As addressed in section 2.1, photocontrolled polymerization needs a complete photocatalytic cycle in both oxidative (Scheme 19A) and reductive (Scheme 19B) quenching pathways. Given the same monomer and reaction formulation, efficient polymerization and good control over the polymerization process is attained by the combination of effective activation and effective reversible deactivation. Hence, the redox properties of the PC corresponding to activation (³PC* → PC^•+ for oxidative quenching and ³PC* → PC^•− for reductive quenching) and deactivation (PC^•+ → PC for oxidative quenching and PC^•− → PC for reductive quenching) are the key to effective control over the polymerization process.

Scheme 19. (A–B) Illustration of the Photocatalytic Cycles via (A) the Oxidative Quenching Pathway and (B) the Reductive Quenching Pathway and (C) Derivation of Redox Potentials from Gibbs Free Energies^a.

^aNote: E⁰(B/A): the standard potential for the A → B half reaction of the A + C → B + D electron transfer reaction. E⁰(triplet): the triplet excitation energy in eV.

In accordance with the Marcus formula for the electron/energy transfer rate constant k_ET (i.e., Marcus theory),¹⁴⁸ it was concluded that the electron transfer rate k_ET scales exponentially with more negative ΔG₀ of the electron transfer reaction (³PC* + R-X → PC^•+ + (R-X)^•− for oxidative quenching and ³PC* + R-X → PC^•− + (R-X)^•+ for reductive quenching). Within the scope of this review, we focus on the half reaction of PC in the electron transfer reaction and only deal with the redox property $\text{Δ}{G}_0^{\text{PC}}$ of PC in $\text{Δ}{G}_0=\text{Δ}{G}_0^{\text{PC}}+\text{Δ}{G}_0^{\text{initiator}}$ for both the activation and deactivation reaction. On the other hand, the experimental convention for quantifying the redox properties of PCs is most commonly the redox potentials obtained using cyclic voltammetry. In this context, $\text{Δ}{G}_0^{\text{PC}}$ is expressed as $\text{Δ}{G}_0^{\text{PC}}=-F\left[E^{\text{PC}}\right]$, where F is the Faraday constant (F = 23.06 kcal V⁻¹ mol⁻¹) and E^PC is the redox potential for PC. By implementing the standard redox potential E⁰ with reference to the saturated calomel electrode (SCE), the relation between E⁰ and $\text{Δ}{G}_0^{\text{PC}}$ is given in Scheme 19C. Consequently, we can employ the E⁰ metric convention to reinterpret the exponential dependence of k_ET on ΔG₀.

Scheme 20 illustrates the property-performance relationships which associate the energy levels of frontier orbitals to redox potentials and the efficiencies of activation/deactivation processes in photocontrolled polymerization. For the oxidative quenching pathway (Scheme 20A), a higher-lying upper-SOMO of ³PC* leads to smaller ionization potential (IP) and hence more negative E⁰(PC^•+/³PC*) dictating more efficient activation, whereas a lower-lying HOMO of PC leads to larger electron affinity (EA) and thereby more positive E⁰(PC^•+/PC) for more efficient deactivation. For the reductive quenching pathway (Scheme 20B), a lower-lying lower-SOMO of ³PC* and correspondingly larger EA result in more positive E⁰(³PC*/PC^•−) for efficient activation while a higher-lying LUMO of PC and a smaller IP give rise to more negative E⁰(PC/PC^•−) for efficient deactivation. In the following context, we discuss the cases for activation and deactivation, respectively, regarding the oxidative quenching pathway (Scheme 20A) and the reductive quenching pathway (Scheme 20B).

Scheme 20.

(A–B) Illustration of the Relationships among Activation/Deactivation Efficiencies, Redox Potentials, and Energy Levels of Frontier Molecular Orbitals for (A) the Oxidative Quenching Pathway and (B) the Reductive Quenching Pathway

3.1.3.1. Effective Activation.

With respect to oxidative quenching, more negative E⁰(PC^•+/³PC*) signifies a less negative $\text{Δ}{G}_0^{\text{PC}}$ (Scheme 19C) which indicates a higher-lying upper SOMO (Scheme 20A). Because in a typical organic PC, the upper SOMO mostly consists of contributions from the π* orbital of the chromophore core, methods that can increase the π* energy level contribute to more negative E⁰(PC^•+/³PC*) and enhance the reducing ability of ³PC*, thus facilitating more effective electron transfer via oxidative quenching. These methods include introducing less electron-withdrawing (i.e., more electron-donating) substituents or smaller dimensions of the π conjugation to destabilize the π* orbital. Alternatively, higher-lying upper SOMO can also be achieved by mixing the high-lying n orbitals (where the n electrons occupy) into the π* orbital; substituents with n electrons (e.g., alkyloxy, alkylthio, or (di)alkylamino groups) directly connected to the π-conjugated chromophore core usually work well.⁸⁰

By contrast, more efficient reductive quenching of ³PC* is represented by more positive E⁰(³PC*/ PC^•−) and lower-lying lower SOMO, where the vacancy (or the “hole”) lies after electron excitation (Scheme 20B). In organic PCs, the lower SOMO of ³PC* is usually π orbitals of the chromophore where more electron-withdrawing substituents decreases the π orbital energy level and leads to a more positive E⁰(³PC*/PC^•−) and thus a higher oxidizing ability of ³PC*.

3.1.3.2. Reversible Deactivation.

In all photocontrolled polymerization mechanisms, as shown in Scheme 19, reversible deactivation is attained through the last step of a catalytic cycle, where the propagating species is oxidized by PC^•+ (in an oxidative quenching catalytic cycle) or reduced by PC^•− (in a reductive quenching catalytic cycle). Effective and reversible deactivation is highly important in three aspects: (i) Upon activation of an initiator, propagation of the disassociated active species (radical/cation) through additions of monomers is continued until the propagating radical is deactivated by PC^•+ or PC^•− leading to a dormant chain. Efficient deactivation thus minimizes dark polymerization and imparts precise temporal/spatial control by switching the light on/off. (ii) In photocontrolled polymerization, successful deactivation plays a major role in controlling the average polymer chain length. After activation of the initiator, the propagating species must be rapidly deactivated so as to allow only a few additions of monomers per activation cycle. This increases the chance of uniform growth of different chains to attain narrower molecular weight distributions. Although this may not be obvious in RAFT-based techniques where chain transfer exists, it can be a major factor in photo-ATRP techniques. (iii) Effective deactivation also keeps radical concentration low to prevent radical–radical quenching leading to nonunity initiator efficiency or nonuniform molecular weight distribution.

Consequently, the deactivation process usually needs to be highly exothermic for photocontrolled polymerization, so that the dormant polymer chain can be generated for the next photoactivation event. In this case with regard to oxidative quenching, the PC^•+ species should be highly oxidizing with E⁰(PC^•+/PC) excessively more positive than E⁰(initiator/initiator^•−), contributed by lower-lying HOMO of PC (Scheme 20A); with regard to reductive quenching, the PC^•− species should be highly reducing with E⁰(PC/PC^•−) excessively more negative than E⁰(initiator^•+/initiator), contributed by higher-lying LUMO of PC (Scheme 20B).

3.2. Examples for Catalyst Design in Photocontrolled Polymerization

3.2.1. Tuning Absorption: Bathochromic and Hyperchromic Effect.

3.2.1.1. Heavier Halogen Substitution.

Perhaps the most well-known strategy to tune the λ_max and ε_max of a chromophore is the introduction of halogen substituents. While it is generally accepted that heavier atom substitution results in longer λ_max and slightly higher ε_max,¹⁴⁹ heavy halogen substitution is the most effective and readily applicable method. For example, recent reports on modifying iridium complexes with heavier halogens attributed the observed bathochromic shift and hyperchromic shift to the heavy halogen substituents, which stabilizes the lowest unoccupied molecular orbital (LUMO) while leaving the highest occupied molecular orbital (HOMO) mostly unperturbed.^150,151

Similarly, halogenated xanthene dyes (XAN-1, XAN-2, XAN-3, and XAN-4 as examples in Figure 1) were implemented in PET-RAFT polymerization to explore how the variation of halogen substitution tunes λ_max and ε_max of the chromophore.⁷³ Theoretical calculations showed that the HOMO → LUMO transition has more than 98% contribution to the S₀ → S₁ vertical excitation of these halogenated xanthene dyes and indicated that the evolution of λ_max by halogen substitution can be exclusively described by the nature and energy levels of the HOMO and LUMO. Specifically, because the nonbonded (n) electrons of the halogen substituents have a significant contribution to the population of the HOMO, heavier halogens that naturally exhibit higher energy n electrons (i.e., I > Br) lead to higher HOMO energy levels. This phenomenon was shown by comparing the xanthene dyes XAN-1 with iodine substituents (Figure 1A) and XAN-2 with bromine substituents (Figure 1B). On the other hand, these xanthene core halogens do not contribute appreciably to the LUMO, and thus LUMO energy levels are less effected when comparing XAN-1 (Figure 1A) and XAN-2 (Figure 1B). Consequently, the heavier I-substituted XAN-1 has narrower HOMO/LUMO energy gap and longer λ_max = 548 nm compared to Br-substituted XAN-2 with λ_max = 540 nm. A similar trend was observed for I-substituted XAN-3 with longer λ_max = 563 nm compared to Br-substituted XAN-4 with λ_max = 555 nm.⁷³ It was also found that heavier halogen substitution also leads to an increase in ε_max (Figure 1), which had contributions from both narrower S₁/S₀ energy difference and higher transition electric dipole moment by the heavy atom effect.⁷³ Because of the notable bathochromic shift brought by introduction of heavier halogen, XAN-3 (Rose Bengal, λ_max = 563 nm) was reported to enable PET-RAFT polymerization mediated by yellow (560 nm) LED light, in comparison to the more commonly known XAN-2 (Eosin Y, λ_max = 540 nm) which favors PET-RAFT polymerization under green (530 nm) LED light.⁷³

Figure 1.

Chemical structures, frontier molecular orbitals, and their energy levels of the ground states of (A) XAN-1 (Erythrosine B), (B) XAN-2 (Eosin Y), (C) XAN-3 (Rose Bengal), and (D) XAN-4 (Phloxine B). q: percentage of the MO localized on the specified element. Reprinted from ref 73. Copyright 2019 American Chemical Society.

3.2.1.2. Chromophore Core-Twisting.

By introducing bulky substituents to the chromophore that produce significant steric hindrance, an originally planar chromophore core can be twisted, which leads to large variation in its molecular properties. Researchers have found that core-twisting of some common organic chromophores can lead to a bathochromic shift and broadening of the absorption peaks. These core-twisted dyes include nanocarbon (e.g., acene) derivatives,¹⁵² perylene diimide (PDI) derivatives,¹⁵³ and porphyrin derivatives.^154,155

Gidron and co-workers recently reviewed the helically locked twisted acene derivatives, focusing on the relationship between their twist angles and the resultant effect on their optical and electronic properties.¹⁵² Overall, with the increase in the twist angle, the twisted acene derivatives exhibited slight destabilization of the HOMO as well as slight stabilization of the LUMO (Figure 2).^152,156 This is caused by orbital rehybridization where greater σ character is mixed into the π orbitals as a result of core-twisting.^157,158 Consequently, the narrowed HOMO/LUMO energy gap led to a slight bathochromic shift as observed in the absorption spectra.^152,156 This general trend is applicable to most twisting nanocarbons.¹⁵²

Figure 2.

(A) Planar, bent, and twisted conformations of two adjacent p-orbitals. (B) Calculated HOMO and LUMO energy levels of anthracene (as a simplified example of twisted acene derivatives) with different twist angles. Reprinted from ref 152. Copyright 2019 American Chemical Society.

Similarly, Holten and co-workers investigated conformationally distorted porphyrin derivatives and discovered that the nonplanarity of the macrocycle induced by the steric hindrance of adjacent bulky peripheral substituents largely affects the S₁ state.¹⁵⁹ Specifically, the saddle deformations of twisted porphyrins tend to destabilize their HOMOs and stabilize their LUMOs, thereby yielding smaller HOMO/LUMO energy gaps and a bathochromic shift in the absorption spectra.¹⁵⁹ Senge more generally reviewed a large library of core-twisted porphyrins and concluded that macrocycle distortion of porphyrin derivatives can destabilize the π system.¹⁵⁴ For example, zinc(II) 2,3,7,8,12,13,17,18-octaethyl-5,10,15,20-tetraphenylporphyrin (POR-2, Figure 3) is an excellent core-twisting porphyrin dye whose planar porphyrin counterpart is zinc(II) 5,10,15,20-tetraphenylporphyrin (POR-1, Figure 3). While POR-2 only has eight additional peripheral ethyl groups in comparison to 9, these ethyl groups mostly impose geometrical impact on the chromophore without any notable electron-donating/withdrawing effect or any conjugated effect. As reported in the literature, the experimentally resolved structure of POR-2 indeed exhibits a saddle-distorted nonplanar macrocycle because of the steric hindrance formed between ethyl groups and phenyl groups.^154,155 This distortion led to a significant ~85 nm bathochromic shift in λ_max from 600 nm of ZnTPP⁸⁶ to 685 nm of ZnOETPP.¹⁶⁰ While POR-1 was reported in PET-RAFT polymerization to mediate one of the first orange (~600 nm) light-mediated polymerizations,^55,86 the large bathochromic shift of λ_max from POR-1 to POR-2 induced by a core-twisting effect has further enabled the implementation of deep-red (~700 nm) light in PET-RAFT polymerization.

Figure 3.

Chemical structures of POR-1 and POR-2.

3.2.1.3. Introduction of Atoms with Nonbonded Electrons Attached to the Chromophore.

Recently, Zhao and co-workers synthesized a series of PDI derivatives and reported the strong bathochromic shift in absorption caused by alkylthio substitution on the bay positions of PDI derivatives.¹⁶¹ As shown in Figure 4, because of the bay substituents, PDI-2, PDI-3, PDI-4, PDI-5, and PDI-6 all exhibit slight core-twisting characters as compared to the planar PDI-1. However, this nonplanarity is not significant and apparently the core-twisting halogen-substituted PDI-4 and PDI-5 exhibited comparable λ_max to PDI-1 (λ_max = 523 nm). Therefore, the alkyl/aryl sulfide substituents of PDI-2 (λ_max = 560 nm) and PDI-3 (λ_max = 552 nm) must have played an important role in the observed ~30 nm bathochromic shift in λ_max as compared to PDI-1. The authors observed that the presence of higher energy n electrons in the S atoms increased the HOMO energy level while having a reduced effect on the LUMO energy level, leading to lower HOMO/LUMO energy gaps (Figure 4). Consequently, these n → π* characters in S₀ → S₁ excitation (~99% contribution from HOMO → LUMO transition) result in the notable bathochromic shift of λ_max in PDI-2 and PDI-3 with respect to PDI-1. On the other hand, the bathochromic shift in PDI-6 is smaller, which is due to the strong electron-withdrawing effect of F substituents that greatly withdraw the n electrons on S and reduce their impact on the chromophore.¹⁶¹

Figure 4.

Chemical structures, molecular geometries, and visualized frontier molecular orbitals (isovalue = 0.03) of PDI derivatives PDI-1, PDI-2, PDI-3, PDI-4, PDI-5, and PDI-6. Reprinted with permission from ref 161. Copyright 2019 The Royal Society of Chemistry.

Würthner and co-workers observed similar bathochromic shifts for PDI derivatives substituted with aryl ethers at the bay positions, which led to a 56 nm bathochromic shift in λ_max.¹³⁹ More strikingly, the authors observed 181 nm bathochromic shift in λ_max by introducing pyrrolidinyl substituents on the bay positions,¹³⁹ which is likely also due to the high energy n electrons in the N atoms contributing to n → π* characters of the S₀ → S₁ excitation.

Recently, Martynov and co-workers employed (TD)DFT methods and spectroscopic methods to theoretically interpret the spectral properties across a library of unsubstituted and alkoxy-substituted free-base or zinc phthalocyanine derivatives PHC-1, PHC-2, PHC-3, PHC-4, PHC-5, and PHC-6 (Figure 5).¹⁴⁰ Two types of alkoxy substitutions are possible for phthalocyanine derivatives, i.e., the peripheral β-alkoxy-substitution (e.g., PHC-2 and PHC-5, Figure 5B,E) and the nonperipheral α-alkoxy-substitution (e.g., PHC-3 and PHC-6, Figure 5C,F). Interestingly, the two types perform very differently in affecting the spectral properties of the phthalocyanine chromophores. In the context of the peripheral β-alkoxy-substituted PHC-2 and PHC-5, λ_max of their Q-bands are located at 703 and 679 nm, respectively, almost the same as their unsubstituted counterparts PHC-1 and PHC-4 (Figure 5A,D). In contrast, the nonperipheral α-alkoxy-substituted PHC-3 (λ_max = 772 nm) and PHC-6 (λ_max = 751 nm) exhibit pronounced bathochromic shift in λ_max as large as ~70 nm (Figure 5C,F). The authors concluded that the nonperipheral α-alkoxy-substitution can strongly destabilize HOMO because of the larger HOMO population on α-carbons resulting in stronger mesomeric interactions with n electrons of the O atoms, which consequently leads to a narrowed HOMO/LUMO energy gap and the observed large Q-band bathochromic shift.¹⁴⁰ Another report discovered that the Q-band bathochromic shift is enhanced with an increasing number of nonperipheral α-alkoxy-substituents.¹⁶² Similarly, Martin and co-workers observed the same nonperipheral α-alkoxy-substitution effect in generating a large Q-band bathochromic for naphthalocyanines.¹⁶³

Figure 5.

Chemical structures and molecular geometries of (A) free-base phthalocyanine PHC-1, (B) free-base β-alkyloxy-substituted phthalocyanine PHC-2, (C) free-base α-alkyloxy-substituted phthalocyanine PHC-3, (D) zinc(II) phthalocyanine PHC-4, (E) zinc(II) β-alkyloxy-substituted phthalocyanine PHC-5, and (F) zinc(II) α-alkyloxy-substituted phthalocyanine PHC-6. Reprinted from ref 140. Copyright 2019 American Chemical Society.

Boyer, Liu, and co-workers implemented PHC-4 and PHC-6 in PET-RAFT polymerization via an oxygen-mediated reductive quenching pathway and performed (TD)DFT calculations to gain a deeper insight into the structure-property relationships (Figure 6). UV–vis spectroscopy was performed in dimethyl sulfoxide for PHC-4 (Figure 6C, λ_max = 661 nm) and PHC-6 (Figure 6G, λ_max = 723 nm). For both dyes, the S₀ → S₁ vertical excitation has over 95% contribution from the HOMO → LUMO transitions. The installation of highly electron-donating alkoxy groups⁸⁰ in PHC-6 significantly increases the HOMO and LUMO energy levels compared to PHC-4 but increases the HOMO to a larger extent.⁷³ This effect was coupled with excitation from n orbitals of the oxygen atom to π* orbitals of the tetraaza-prophyrin core, in addition to the excitation from the partial phenyl-π orbital to the π* orbital of the tetraaza-prophyrin (Figure 6F), thus narrowing the energy gap (Figure 6E) and leading to a 62 nm bathochromic shift with respect to PHC-4 (Figure 6A,B). Consequently, while PHC-4 enabled PET-RAFT polymerization under red (660 nm) LED light (Figure 6D), the implementation of PHC-6 in PET-RAFT polymerization not only enabled the use of deep-red (~700 nm) LED light but also enabled the use of near-infrared (NIR, 780 nm) light for efficient PET-RAFT polymerization via the oxygen-mediated reductive quenching pathway (Figure 6H).

Figure 6.

(A,E) Chemical structures and energy level diagrams, (B,F) visualization of HOMO, LUMO/LUMO+1 (isovalue = 0.3), (C,G) UV–vis spectra with λ_max denoted, and (D,H) plots of ln([M]₀/[M]_t) versus time revealing k_p^app and temporal control for model PET-RAFT polymerization via the oxygen-mediated reductive quenching pathway energy, corresponding to (A–D) zinc(II) phthalocyanine PHC-4 and (E–H) the zinc(II) β-alkyloxy-substituted phthalocyanine PHC-6, respectively. Percentage contribution of the dominant molecular pairs contributing to the S₀ → S₁ electronic transition is denoted in red. Reprinted with permission from ref 56. Copyright 2021 Nature Publishing Group.

3.2.1.4. Extended π-Conjugation.

Extension of π-conjugated systems is one of the most applied guiding principles for designing long-wavelength absorbing dyes.¹⁶⁴ This approach is based on the conventional recognition that larger π-conjugation contributes to bathochromic shifts. Although there are exceptions,^164,165 this statement is still useful in most cases given sufficient understanding of the core chromophore before modification.¹⁶⁴ The most commonly seen π-expanded dyes include polyaromatic dyes¹⁶⁴ and porphyrin derivatives.¹⁶⁶ It should be noted the statement that larger π-conjugation chromophores can be excited by longer-wavelength irradiation is only valid for the dimension of π-conjugation in the chromophore rather than the fused rings in the substituents, which was also mentioned by Ruiz-Morales.¹⁶⁷

Adachi and co-workers reported the effect of π-conjugation extension on polynaphthalenetetracarboxylic dianhydride diimide dyes PDI-1, PDI-7, and PDI-8 (Figure 7).¹⁶⁴ From PDI-1 (λ_max = 526 nm, ε_max = 81500 M⁻¹ cm⁻¹) to PDI-7 (λ_max = 650 nm, ε_max = 93300 M⁻¹ cm⁻¹) and PDI-8 (λ_max = 764 nm, ε_max = 162000 M⁻¹ cm⁻¹), with the π-conjugation extension along the long molecular axis, significant bathochromic shifts in λ_max and notable hyperchromic shifts in ε_max were observed. As shown in Figure 6, the extended π-conjugation leads to destabilization of HOMO and increases the HOMO energy level while scarcely affecting the LUMO energy level. This overall effect contributes to a narrowed HOMO/LUMO energy gap and therefore contributes to notable bathochromic shifts. The authors further explained that bridges connecting each naphthalene fragment are characterized as antibonding connections in HOMOs of these derivatives. Due to the increasing content of these antibonding connections from PDI-1 to PDI-7 and PDI-8, the HOMO is more destabilized and exhibits a higher energy level, while the LUMO always has bonding connections in all bridges and is thus not affected (Figure 7).¹⁶⁴ Meanwhile, the larger π-conjugation leads to an increase in the transition dipole moment and consequently yields enhanced ε_max.¹⁶⁴

Figure 7.

Chemical structures and HOMOs/LUMOs (and corresponding energy levels) of polynaphthalenetetracarboxylic dianhydride diimide dyes PDI-1, PDI-7, and PDI-8. Reprinted from ref 164. Copyright 2001 American Chemical Society.

On the other hand, Sessler and co-workers reviewed a large library of π-extended boron dipyrromethenes (BODIPYs), π-extended sapphyrins, π-extended porphycenes, π-extended corroles, π-extended rosarins, π-extended cyclo[n]pyrroles, and other types of expanded porphyrins with π-extension.¹⁶⁶ All of these π-extended dye classes exhibit larger bathochromic shifts in λ_max with larger π-extension.¹⁶⁶ Accordingly, Boyer, Liu, and co-workers investigated a series of tetraazaporphyrin/phthalocyanine derivatives and revealed λ_max = 582 nm for PHC-7, λ_max = 661 nm for PHC-4, and λ_max = 749 nm for PHC-8, which covers wavelengths from orange to near-infrared light (Figure 8).⁵⁶ For all three dyes, HOMO → LUMO transitions exhibit dominant (88–99%) contribution to the S₁ vertical excitation. As the tetraaza-porphyrin core is very electron-withdrawing,¹⁶⁸ overall, the perphenyl rings in PHC-4 exhibit a more notable electron-donating conjugated effect rather than electron-withdrawing inductive effect. Hence, the addition of four phenyl rings moves both the HOMO and LUMO energy levels of PHC-4 up but moves up the LUMO to a much lesser extent (because of larger conjugation stabilizing the π* orbital), compared with PHC-7, thus narrowing the energy gap and leading to a 79 nm bathochromic shift (Figure 8A,B). Indeed, as seen in Figure 8B, bottom, the HOMO → LUMO transition exhibits partial phenyl-π to tetraazaporphyrin-π* excitation. Similarly, larger aryl rings in PHC-8 yield an even more notable electron-donating conjugation effect, increasing both the HOMO and LUMO but moving up the latter to a lesser extent (larger conjugation stabilizing the π* orbital), thus leading to a much narrower energy gap (Figure 8C) and 88 nm bathochromic shift as compared to PHC-4 (Figure 8B,C). These dyes were implemented in PET-RAFT polymerization via the oxygen-mediated reductive quenching pathway, which achieved efficient polymerization under orange (~600 nm, PHC-7) and red (660 nm, PHC-4) lights.⁵⁶

Figure 8.

(A–C) Chemical structures, UV–vis spectra with λ_max denoted, energy level diagrams, and visualization of HOMO and LUMO/LUMO+1 (isovalue = 0.3) of (A) zinc tetramethyl tetraazaporphyrin PHC-7, (B) zinc phthalocyanine PHC-4, and (C) zinc naphthalocyanine PHC-8. Percentage contribution of the dominant molecular pairs contributing to the S₀ → S₁ electronic transition is denoted in red. Reprinted with permission from ref 56. Copyright 2021 Nature Publishing Group.

3.2.2. Tuning Excited State Evolutions: Enhancing the Triplet Quantum Yield.

3.2.2.1. Constructing Charge Transfer States.

As discussed earlier, it is essential to populate T₁ states in photocontrolled polymerization systems as the sufficiently long-lived T₁ states are more likely to encounter the initiators for electron transfer reactions, while simultaneously suppressing undesirable back electron transfer.^13,23,52 However, most purely organic chromophores lack efficient spin–orbit coupling and hence exhibit rather slow ISC and low Φ_T.⁵² In organic PCs, the use of donor–acceptor scaffolds with charge transfer excited states can in some cases result in high Φ_T. Typically, these charge-transfer PCs are vertically photoexcited to a Franck–Condon singlet state (S_FC, a locally excited higher singlet state S_n above S₁) and rapidly relax to the charge transfer S₁ state (S_CT). From here, the ISC process occurs to reach the corresponding charge transfer triplet state (T_CT′) which stays in the long-lived triplet manifold awaiting electron transfer with appropriate substrates. Upon reaching the S_CT state, the fluorescence process is greatly suppressed due to the minimal transition dipole moment induced by the spatially segregated molecular orbitals of the “excited electron” and the “hole” (see section 3.1.2.2). Similarly, the IC process is also greatly suppressed due to poor overlap of orbitals (see section 3.1.2.2). Consequently, a reasonable k_ISC can more likely compete with the suppressed k_F and k_IC, rendering high Φ_T of the PC. On the other hand, by rationally constructing multiple charge transfer states where ISC occurs between a S_CT and a T_CT′, characteristic of different molecular orbital types, this S_CT → T_CT′ transition should bear a significant SOCME according to the El-Sayed rule and hence an enhanced k_ISC.^52,169 While the T_CT′ state is constructed in a way that displays an energy level slightly lower than that of S_CT, the resultant narrow singlet–triplet energy gap ΔE_ST should also contribute to promoting spin–orbit coupling and k_ISC (the energy gap law).^22,52 By combination of these considerations, a high Φ_T can be obtained by simultaneous suppressing k_F + k_IC and promoting k_ISC through rationally constructing charge transfer states.

By attaching aryl groups with low-lying π* (either by extended conjugation or electron-withdrawing substitution), Miyake and co-workers constructed a library of dihydrophenazine derivatives,⁸⁰ phenoxazine derivatives,¹¹¹ and dimethyl-dihydroacridine derivatives,¹¹² bearing donor–acceptor motifs and characteristically displaying charge transfer excited states as effective O-ATRP PCs with high Φ_T and highly reducing T₁ states. As an example, Damrauer, Miyake, and co-workers used two comparable π-extended phenoxazine derivatives to demonstrate the effect of charge transfer characters of excited states on photophysical properties of a PC.¹¹³ They discovered that the N-phenyl-substituted phenoxazine derivative PXZ-1 (Figure 9A) possesses a Φ_T = 0.11. As shown in Figure 9C, after local excitation in the phenoxazine core of PXZ-1 reaching the Franck–Condon state S_FC, the electron density is rapidly shifted to one of the biphenyl substituents reaching the charge transfer state S_CT-Biph within 10 ps after solvent reorganization and conformational relaxation. However, the subsequent ISC process occurs as a S_CT-Biph→ T_CT-Biph transition does not have a narrow ΔE_st nor does it exhibit orthogonality in its orbital angular momentum modes. Consequently, the k_isc of S_CT-Biph → T_CT-Biph transition is limited and cannot effectively compete with k_f + k_ic of the S_CT-Biph → S₀ transitions, resulting in a low Φ_T of 0.11 (Figure 9A). In stark contrast, by replacing the N-phenyl group of PXZ-1 with the N-1-naphthyl group, the π-extended derivative PXZ-2 (Figure 9B) exhibits a remarkably high Φ_T of 0.91. As presented in the energy diagram (Figure 9C), upon local excitation in the phenoxazine core of PXZ-2 reaching S_FC and further S_CT-Biph, the electron density is further shifted to the N-naphthyl group, reaching S_CT-Naph within ~20 ps. Subsequently, two ISC pathways can occur (respectively denoted as A and B in Figure 9C). The ISC pathway A has an S_CT-Naph → T_CT-Naph transition as the rate-limiting step, which bears a very small ΔE_ST; however, although possible, the pathway A was not observed likely because of a lack of change in molecular orbital types.¹¹³ On the other hand, the ISC pathway B directly couples S^CT-Naph with T^CT-Biph. The k_isc of S_CT-Naph → T_CT-Biph transition is greatly enhanced because of the notable change in orbital angular momentum which greatly facilitates spin–orbit coupling. Consequently, PXZ-2 exhibits a high Φ_T of 0.91 arising from the excited state charge transfer characters.¹¹³ The high Φ_T = 0.91 of PXZ-2 has endowed the PXZ-2-catalyzed O-ATRP polymerization of MMA with an excellent initiator efficiency of close to unity, as well as good control of the polymerization, exhibiting a resultant dispersity of 1.1–1.2. This is lower than its other derivatives in the phenoxazine dye class.

Figure 9.

(A,B) Chemical structures and molecular geometries of (A) PXZ-1 and (B) PXZ-2, with Φ_T denoted below. (C) Diagram of excited and ground state energy levels with alternative ISC pathways presented for PXZ-1 and PXZ-2. Less likely pathways are indicated in dashed lines.¹¹³ Reprinted from ref 113. Copyright 2018 American Chemical Society.

Following this study, Damrauer, Miyake, and co-workers investigated a class of N-aryl 3,7-diphenyl phenoxazine derivatives PXZ-3, PXZ-4, and PXZ-5 (Figure 10A–C) with varied N-aryl substitution to establish more detailed relationships between Φ_T and the charge transfer character of excited states.¹⁶⁹ Unlike the previous report, the 3,7-diphenyl group chosen to replace the 3,7-dibiphenyl group isolated the impact of N-aryl substitution on the charge transfer character and Φ_T. As shown in Figure 10C for PXZ-3, upon photoexcitation to the Franck–Condon excited state S_FC, it rapidly relaxed to an emitting state S_deloc characteristic of an excited electron delocalized to the whole chromophore. Subsequently, S_deloc can either relax to S₀ by fluorescence/IC or T_{CT-Phen(core)} which exhibits partial charge transfer to a molecular orbital spanning over the phenoxazine core and its one-side adjoining phenyl group. Because of the difference in molecular orbital types between S_deloc and T_{CT-Phen(core)}, the spin–orbit coupling of the S_deloc → T_{CT-Phen(core)} transition is enhanced, leading to a reasonable k_ISC capable of competing over k_F and k_IC and resulting in a Φ_T = 0.30 (Figure 10A). On the other hand, by replacing the N-phenyl group with a N-1-naphthyl group, the resultant PXZ-4 exhibited an additional intermediate singlet state S_CT-Naph with charge transfer character to the N-1-naphthyl group, which is relaxed from S_deloc (Figure 10C). Due to the charge transfer character of the emitting state S_CT-Naph, the irradiative S_CT-Naph → S₀ transition is greatly impeded, leading to a low k_F = 3.0 × 10⁶ s⁻¹ of PXZ-4 as compared to PXZ-3 (k_F = 2.1 × 10⁸ s⁻¹). Meanwhile, the more pronounced change in molecular orbital types of the S_CT-Naph → T_{CT-Phen(core)} transition led to enhanced spin–orbit coupling according to the El-Sayed rule. Correspondingly, PXZ-4 bears a higher k_ISC = 1.5 × 10⁸ s⁻¹ compared with PXZ-3 (k_isc = 9.3 × 10⁷ s⁻¹). The overall effect of retarded fluorescence and enhanced ISC, contributed to the high Φ_T = 0.95 of PXZ-4 as compared to PXZ-3 (Φ_T = 0.30). On the other hand, the N-2-naphthyl counterpart PXZ-5 exhibits a naphthyl-to-phenoxazine center–center distance of 0.533 nm larger than that of PXZ-4 (0.437 nm). Consequently, PXZ-5 exhibits less Coulombic stabilization than PXZ-4 and thus PXZ-5 possesses a slightly higher energy S_CT-Naph compared to PXZ-4, resulting in a larger ΔE_ST of PXZ-5 than that of PXZ-4. Therefore, PXZ-5 exhibits a slightly lower Φ_T = 0.92 as compared to PXZ-4 (Φ_T = 0.95), arising from the variation in the N-naphthyl connectivity.¹⁶⁹

Figure 10.

(A–C) Chemical structures of (A) PXZ-3, (B) PXZ-4, and (C) PXZ-5 with k_F, k_IC, k_ISC, and Φ_T denoted below. (D) Energy level diagram of PXZ-3, PXZ-4, and PXZ-5 showing dominant decay pathways of excited states. Reprinted from ref 169. Copyright 2019 American Chemical Society.

3.2.2.2. Heavy Atom Effect.

On the basis of the heavy atom effect,¹⁷⁰ Φ_T of fluorescein dyes can be simply tuned by the halogenation strategy, where the spin–orbit coupling between S₁ and T_n is enhanced by halogen substituents with larger atomic numbers Z (see section 3.1.2.1), leading to higher k_ISC and Φ_T. Recently, Boyer and Liu reported a synthesized PC XAN-6 (Figure 11B), which possesses Φ_T = 0.58 and features four Br substituents; the Φ_T was significantly increased compared to its nonsubstituted counterpart fluorescein XAN-5 (Φ_T = 0.03, Figure 11A).⁷⁵ As denoted in Figure 11A,B, XAN-5 and XAN-6 bear highly comparable k_F and both exhibit negligible k_IC; however, heavier substituents of XAN-6 result in remarkably higher k_ISC by enhanced spin–orbit coupling compared to XAN-5.⁷⁵ Overall, this heavy atom effect leads to much higher Φ_T = 0.58 of XAN-6. Consequently, although XAN-5 has better redox properties favoring PET-RAFT polymerization, it was inefficient (k_p^app ≈ 0 min⁻¹) due to its low Φ_T.⁶⁷ By contrast, because of the considerable Φ_T = 0.58, the XAN-6-catalyzed model PET-RAFT polymerization exhibited a k_p^app of 0.016 min⁻¹, which is considered rather efficient.⁷⁵

Figure 11.

(A,B) Chemical structures of (A) XAN-5 and (B) XAN-6, with k_F, k_IC, k_ISC, and Φ_T denoted below. (C) Computationally derived Jablonski diagram of XAN-6 and its derivation of Φ_T from different excited state decay rate constants. Reprinted from ref 75. Copyright 2019 American Chemical Society.

Following this trend, heavier halogen substitution appears to result in higher Φ_T for the halogenated xanthene dyes XAN-1, XAN-2, XAN-3, and XAN-4 mentioned above (Figure 1).⁷³ XAN-1 substituted with heavier I atoms bears Φ_T = 0.62–0.69, which almost doubles that of the Br-substituted XAN-2 (Φ_T = 0.28–0.32). Although XAN-1 and XAN-2 possess similar absorption and redox properties, the high Φ_T = 0.62–0.69 has enabled much faster XAN-1-catalyzed model PET-RAFT polymerization with a k_p^app = 0.023 min⁻¹, as compared to that of the XAN-2-catalyzed system (k_p^app = 0.014 min⁻¹) with low T₁ population (Φ_T = 0.28–0.32), under the same polymerization conditions.⁷³

3.2.2.3. Chromophore Core-Twisting.

By state-of-the-art theoretical and experimental studies, researchers clarified the important role of core-twisting/nonplanarity in promoting ISC via enhanced spin–orbit coupling.^171,172 Hariharan and co-workers initially investigated a class of core-twisted PDI derivatives with multiple Br substitutions in the bay positions and discovered a significant enhancement in triplet generation.¹⁷³ However, to explore the isolated effect of core-twisting on ISC, the impact from heavy atoms must be excluded. With this consideration, Hariharan and co-workers further imparted light-atom-based twisting into the core chromophore of PDI derivatives forming PDI-10 and PDI-11 with the planar PDI-9 as a comparison (Figure 12). Indeed, the planar analogue PDI-9 possesses negligible Φ_T ~ 0 because of the weak spin–orbit coupling which diminishes ISC. By imparting core-twisting induced by steric hindrance at one side of the bay regions of the PDI core (PDI-10), it yields an enhanced Φ_T ~ 0.10 coupled with a k_isc = 1 × 10⁹ s⁻¹ (Figure 12B). By imparting this core-twisting at both sides of the bay regions, the resultant PDI-11 possesses a more twisted structure and exhibits an even higher Φ_T ~ 0.30 with a higher k_isc = 4 × 10¹⁰ s⁻¹ (Figure 12C). Quantum chemical calculations revealed that the C–H and C═C bending vibration modes in the twisted regions contribute to efficient ISC between ¹(ππ*) and ³(ππ*) driven by the Herzberg–Teller vibronic coupling.^153,174

Figure 12.

Chemical structures and representation of molecular geometries of PDI-9, PDI-10, and PDI-11, with Φ_T and k_ISC denoted below the dye labels. Reproduced with permission from ref 153. Copyright 2017 The Royal Society of Chemistry.

Very recently, Behrends and co-workers constructed a class of twisted acene derivatives Ant-Cn (n = 3–6) tethered with alkyl tethers (n = propyl-tether to hexyl-tether) to investigate the effect of core-twisting in Φ_T of chromophores (the strategy illustrated as Figure 13, top).¹⁷⁵ With shortened alkyl bridges, the twist angle is increased from 23° with a propyl bridge (Ant-C3), 30° with a butyl-tether (Ant-C4), 32° with a pentyl-tether (Ant-C5), to 38° with a hexyl-tether (Ant-C6). As a comparison, a planar open reference compound without a tether (open) featuring 0° twist angle is also studied. Time-resolved electron paramagnetic resonance (EPR) spectroscopy was performed to obtain the transient EPR spectra of all these compounds (Figure 13, bottom). The highly similar spectral shapes for open, Ant-C6, Ant-C5, and Ant-C4 allow direct comparison and clearly reveal increased Φ_T with shortened tethers. Despite a different spectral shape of Ant-C3 (because of significant stretching-induced variation in spin polarization¹⁷⁵), Ant-C3 bears the highest Φ_T as expected (Figure 13, bottom). Overall, a clear trend was observed by the authors toward increased Φ_T for more twisted compounds tethered with shorter alkyl tethers.¹⁷⁵

Figure 13.

(Top) Chemical structures of twisted acene derivatives Ant-Cn (n = 3–6) tethered with alkyl bridges (n = propyl bridge to hexyl bridge). (Bottom) Transient EPR spectra for Ant-Cn and the open (no alkyl bridges) reference compound. Reproduced with permission from ref 175. Copyright 2019 The Royal Society of Chemistry.

Gidron and co-workers broadly reviewed the consequences of core-twisting nanocarbon molecules on their properties (including Φ_T) and summarized the proposed mechanisms.¹⁵² For example, Brédas and co-workers demonstrated a direct correlation between the magnitude of spin–orbit coupling and the degree of nonplanarity in nonplanar aromatic heterocyclic compounds.¹⁷⁶ Combined with the above-mentioned findings, chromophore core-twisting can generally enhance spin–orbit coupling and accelerate ISC, thereby enhancing Φ_T of the chromophore.

Similarly, Holten and co-workers investigated a variety of core-twisted nonplanar metal-free porphyrin derivatives substituted with bulky alkyl groups that induced steric hindrance.¹⁵⁹ They observed an increase in k_ISC primarily from enhanced spin–orbit coupling in these nonplanar porphyrin complexes. To interpret this observation, they mentioned that ISC in porphyrins is commonly deemed to rely on the wave function overlap involving the central N atoms in the macrocycle and arising from out-of-plane distortion in S₁ and/or T_n.¹⁵⁹ Indeed, previous theoretical studies¹⁷⁷ and experimental studies¹⁷⁸ on chlorophylls argue that ISC can be modulated by the N-centered perturbations. Furthermore, Shchupak, Ivashin, and Sagun reported that common planar porphyrin derivatives exhibit small spin–orbit coupling matrix elements; however, the nonplanar porphyrin derivatives with more distortion of the macrocycle in T₁ exhibits a considerable increase in SOCME, which is induced by a mixture of s character to π-orbitals in T₁ compared to S₁ and hence enhances spin–orbit coupling.¹⁷⁹

3.2.3. Tuning Redox Properties.

3.2.3.1. Photo-ATRP via the Oxidative Quenching Pathway.

As shown in Figure 14A, the first photo-ATRP PC fac-Ir(ppy)₃ reported by Hawker and Fors ^60,84 exhibits a T₁ state that is strongly reducing, E⁰(PC^•+/³PC*) = E⁰[Ir(IV)/Ir(III)*] = −1.73 V, versus SCE (redox potential are all presented vs SCE herein). In light of the reduction potential of a typical ATRP initiator EBP (E⁰(EBP/EBP^•−) = −0.74 V), the reduction of the alkyl bromide initiator by fac-Ir(ppy)₃ is highly exothermic, supportive of efficient initiation.^72,80 On the other hand, the PC^•+ species of fac-Ir(ppy)₃ is a strong oxidant exhibiting E⁰(PC^•+/PC) = E⁰[Ir(IV)/Ir(III)] = 0.77 V, which is much higher than E⁰(EBP/EBP^•−) = −0.74 V and is thus capable of oxidizing the propagating radical and enabling reversible deactivation (Scheme 2A).⁷² A metal-free version of photo-ATRP was also presented by Hawker et al.⁶⁴ and Matyjaszewski et al.⁷² with PTZ-1 (Figure 14B) as the PC, which possesses highly reducing ¹PC* with E⁰(PC^•+/¹PC*) = −2.10 V (thus indicating highly reducing ³PC*) and highly oxidizing PC^•+ with E⁰(PC^•+/PC) = 0.68 V to complete the oxidative quenching cycle of photo-ATRP.⁷²

Figure 14.

Chemical structures of (A) fac-Ir(ppy)₃ and (B) PTZ-1 with E⁰(PC^•+/³PC*) and E⁰(PC^•+/PC) denoted below the dye label. Reproduced from ref 72. Copyright 2016 American Chemical Society.

The advantage of the organic scaffolds over the transition-metal-based compounds in photocatalysis have motivated the study of organic chromophores. As the chromophore core predetermines the benchmark redox properties, substituents attached to the chromophore provide the ability to rationally tune the redox properties based on the electronic nature of the substituting fragment. Thus, Miyake and co-workers compared the redox potentials of N,N-diphenyl dihydrophenazine (DPZ-1, Figure 15A), N-phenyl phenoxazine (PXZ-6, Figure 15B), and N-phenyl phenothiazine (PTZ-1, Figure 15C).¹¹¹ The three chromophore cores all exhibit highly reducing T₁ states with DFT-computed E⁰(PC^•+/³PC*) < −2.0 V, as computed under the same level of theory.¹¹¹ Specifically, among the three, DPZ-1 has the most negative E⁰(PC^•+/³PC*) of −2.25 V, followed by PXZ-6 and PTZ-1. On the other hand, E⁰(PC^•+/PC) of all the three chromophore cores are all much higher than E⁰(initiator/initiator^•−) of the EPA initiator (−0.74 V, Scheme 2B), and are sufficient for deactivation of O-ATRP.¹¹¹

Figure 15.

Chemical structures of the original chromophore core structures (A) DPZ-1 as a dihydrophenazine derivative, (B) PXZ-6 as a phenoxazine derivative, and (C) PTZ-1 as a phenothiazine derivative, with DFT-computed E⁰(PC^•+/3PC*) and E⁰(PC^•+/PC) denoted below the dye label. Reproduced from ref 111. Copyright 2016 American Chemical Society.

Miyake and co-workers reported dihydrophenazine derivatives DPZ-1, DPZ-2, DPZ-3, and DPZ-4 (Figure 15) as PCs for O-ATRP.⁸⁰ DFT predictions and experimental measurements displayed similar trends, where DPZ-1, DPZ-2, DPZ-3, and DPZ-4 all exhibit strong E⁰(PC^•+/³PC*) < −2.0 V (computed). The ATRP initiator ethyl α-bromophenylacetate (EBP, Scheme 2B) exhibits a computed E⁰(initiator/initiator^•−) = −0.74 V, thus all these PCs with computed E⁰(PC•+/³PC*) < −2.0 V are capable of efficiently activating EBP via PET.⁸⁰ In particular, they demonstrated that by installing electron-donating methoxy groups to DPZ-1, the resultant DPZ-2 possesses stronger E⁰(PC^•+/³PC*) = −2.36 V compared to DPZ-1 (−2.34 V). On the other hand, installation of more electron-withdrawing substituents leads to a lower E⁰(PC^•+/³PC*), i.e., −2.24 V for DPZ-3 and −2.06 V for DPZ-4 compared with −2.34 V for the reference DPZ-1 (Figure 16). Moreover, the relatively stable PC^•+ of these PCs formed from PET with computed E⁰(PC^•+/PC) = ~0.1 to 0.2 V are sufficiently oxidizing to deactivate the propagating radical. Indeed, the anion radical of the EPA initiator has E⁰(initiator/initiator^•−) of −0.74, −0.86, and −0.71 V after zero, one, and two monomer (methyl methacrylate) additions, which are much more negative than E⁰(PC^•+/PC) of PCs DPZ-1, DPZ-2, DPZ-3, and DPZ-4, allowing the O-ATRP oxidative quenching cycle to be completed.⁸⁰

Figure 16.

Chemical structures of dihydrophenazine derivatives (A) DPZ-1, (B) DPZ-2, (C) DPZ-3, and (D) DPZ-4, with DFT-computed E⁰(PC^•+/³PC*) denoted below the label. Reproduced with permission from ref 80. Copyright 2016 American Association for the Advancement of Science.

The authors further designed and synthesized DPZ-5 and DPZ-6 (Figure 17A,B). While exhibiting notable charge transfer characters in T₁ (Figure 17C,D), DPZ-5 and DPZ-6 still possess strong computed E⁰(PC^•+/³PC*) of −2.20 V and −2.12 V, respectively, which were comparable to the original DPZ-1 (Figure 16A). Indeed, in addition to the ³PC* being sufficiently reducing to activate O-ATRP, those with charge transfer T₁ states (i.e., DPZ-3, DPZ-5, and DPZ-6) yielded much more efficient O-ATRP polymerization with better control and high initiator efficiency.⁸⁰

Figure 17.

(A,B) Chemical structures of dihydrophenazine derivatives (A) DPZ-5 and (B) DPZ-6, with computed E⁰(PC^•+/³PC*) denoted below the label. (C,D) T₁ frontier molecular orbitals of (C) DPZ-5 and (D) DPZ-6 which visualizes the higher-lying SOMO (top) and the lower lying SOMO (bottom). Reproduced with permission from ref 80. Copyright 2016 American Association for the Advancement of Science.

Miyake and co-workers further discovered phenoxazine derivatives as O-ATRP PCs and sought to tune E⁰(PC^•+/³PC*) by modification of the chromophore core with substituents. As shown in Figure 18, installation of the electron-withdrawing trifluoromethyl substituent on the para position of the N-phenyl group (PXZ-7) leads to slightly less negative E⁰(PC^•+/³PC*) = 2.03 V, as compared to the original benchmark PXZ-6 exhibiting E⁰(PC^•+/³PC*) = 2.11 V. By replacing the N-phenyl group with a N-1-naphthyl group (PXZ-8), the resulting expanded conjugative effect leads to weaker E⁰(PC^•+/³PC*) = −1.84 V, apparently due to the electron-withdrawing conjugated effect of the N-1-naphthyl group with excited electron residing on its low-lying π* orbital (Figure 18G). Consequently, the higher SOMO, where the excited electron resides, is localized in the low-lying π* orbital of the N-1-naphthyl group, leading to a higher IP of the T₁ state and thus weakened E⁰(PC^•+/³PC*). The charge transfer character was incorporated into the T₁ state of 29 by this modification (Figure 18B), which is believed to improve the O-ATRP performance by enhancing Φ_T. 30 substituted with a N-2-naphthyl group is highly similar to 29, exhibiting computed E⁰(PC^•+/³PC*) = −1.90 V, versus SCE and a T₁ with the charge transfer character.¹¹¹

Figure 18.

(A–D) Chemical structures of phenoxazine derivatives (A) PXZ-6, (B) PXZ-7, (C) PXZ-8, and (D) PXZ-9 with computed E⁰(PC^•+/³PC*) denoted below the label. (E–H) T₁ frontier molecular orbitals of (E) PXZ-6, (F) PXZ-7, (G) PXZ-8, and (H) PXZ-9 which visualizes the higher-lying SOMO (top) and the lower lying SOMO (bottom). Reproduced from ref 111. Copyright 2016 American Chemical Society.

Miyake and co-workers also investigated the possible range of E⁰(PC^•+/³PC*) that the phenoxazine PC scaffold can access, with substituted derivatives possessing electron-withdrawing, electron-donating, or conjugation-extended substituents, which yielded a series of derivatives with E⁰(PC^•+/³PC*) ranging between −1.40 V and −2.20 V.⁷⁴ As shown in Figure 19A, more electron-donating substituents (e.g., the methoxy group) enhances the ³PC* reducing ability, rendering E⁰(PC^•+/³PC*) more negative, whereas more electron-withdrawing substituents (e.g., the trifluoromethyl group and the nitrile group) decreases the ³PC* reducing ability while rendering E⁰(PC^•+/³PC*) less negative. On the other hand, larger dimensions of the conjugation (or cross-conjugation) systems also diminishes the reducing capability of ³PC*. As shown in Figure 19B, PXZ-6 with the smallest conjugation/cross-conjugation bears the most negative E⁰(PC^•+/³PC*) and PXZ-16 with the largest conjugation/cross-conjugation bears the least negative E⁰(PC^•+/³PC*) among the four derivatives.

Figure 19.

Chemical structures of (A) PXZ-10, PXZ-11, PXZ-12, PXZ-13, PXZ-5, PXZ-14, and PXZ-15 for demonstration of the electron-donating/withdrawing effect on computed E⁰(PC^•+/³PC*) and (B) PXZ-6, PXZ-9, PXZ-5 and PXZ-16 for demonstration of the effect of extended conjugation on computed E⁰(PC^•+/³PC*), with E⁰(PC^•+/³PC*), E⁰(PC^•+/PC), and E_T denoted below the label. Reproduced from ref 74. Copyright 2018 American Chemical Society.

3.2.3.2. PET-RAFT Polymerization via the Oxidative Quenching Pathway.

One particular feature of PET-RAFT polymerization that is advantageous relative to photo-ATRP is that the initiator efficiency is easier to control because in the RAFT equilibrium the propagating species can react with a dormant RAFT agent or a dormant polymer chain with a RAFT end-group, thus inducing efficient chain transfer (Scheme 5A). Furthermore, PCs with less reducing T₁ states are also suitable for PET-RAFT polymerization because RAFT agents are thermodynamically more facile to reduce than ATRP initiators, under the proviso that they are still thermodynamically capable of activating some portion of the RAFT agents; the other RAFT agents can participate in chain growth by the chain transfer mechanism. Accordingly, for a PC capable of catalyzing PET-RAFT polymerization, its E⁰(PC^•+/³PC*) has to be stronger (i.e., more negative) than E⁰(RAFT/RAFT^•−) of the RAFT agents but need not be overwhelmingly exothermic.

Following the initial work where fac-Ir(ppy)₃ was implemented in PET-RAFT polymerization,⁶¹ Boyer and co-workers further reported the use of POR-1 and POR-3 (Figure 20) to catalyze PET-RAFT polymerization.⁸⁶ POR-1 bears a computed E⁰(PC^•+/³PC*) of −1.24 V and is thus capable of activating BTPA (Scheme 6C) to polymerize acrylates (e.g., MA, Figure 20C) as well as activating CPADB (Scheme 6C) to polymerize methacrylates.⁸⁶ By contrast, the replacement of the highly electron-donating Zn²⁺ cation with two H⁺ cations, renders POR-3 with a less reducing T₁, characterized by a computed E⁰(PC^•+/³PC*) of −1.00 V, which makes it unable to catalyze polymerization of acrylates (e.g., MA, Figure 20D) and only effective in activating CPADB to polymerize methacrylates. Although the weaker reducing capability of the T₁ state POR-3 is inefficient for the oxidative quenching pathway, the reductive quenching induced by addition of triethylamines can be enabled by POR-3 to polymerize acrylates. A recent study by Boyer, Liu, and co-workers more explicitly revisited this selectivity using thermodynamic studies (Figure 20E,F) where POR-1 was found to be thermodynamically favorable for completing the catalytic cycle.⁵⁶ Comparatively, POR-3 failed in the activation of the RAFT agent (i.e., a derivative of BTPA) and thus cannot catalyze polymerization of acrylates via oxidative quenching.

Figure 20.

(A,B) Chemical structures of (A) POR-1 and (B) POR-2. (C,D) Plot of ln([M]₀/[M]_t) versus time revealing k_p^app and temporal control for model PET-RAFT polymerization via oxidative quenching catalyzed by (C) POR-1 and (D) POR-2. (E,F) Proposed photocatalytic cycles with the Gibbs free energy change (ΔG) of key steps denoted in comparison to corresponding thresholds for PET-RAFT polymerization via the oxidative quenching pathway, catalyzed by (E) POR-1 and (F) POR-2, respectively. Green stands for favorable and red for inert. Colored values are ΔG of the corresponding process in kcal/mol, in conjunction with the derived thresholds. Reproduced with permission from ref 56. Copyright 2021 Nature Publishing Group.

Boyer, Miyake, Liu, and co-workers investigated the effect of varied halogen substituents (H, Cl, Br, and I) on E⁰(PC^•+/³PC*) of halogenated xanthene dyes XAN-1, XAN-6, XAN-2, XAN-7, XAN-3, and XAN-4 (Figure 20A–F) as PET-RAFT PCs.^73,75 As shown in Figure 21G, the series H, I, Br, and Cl are substituents with increasing electronegativity and electron-withdrawing character in halogenated xanthene dyes. With more halogen substituents of higher electronegativity, the overall more notable electron-withdrawing effect leads to less negative E⁰(PC^•+/³PC*) values (Figure 21H). Consequently, XAN-1 with four I substituents exhibits the strongest E⁰(PC^•+/³PC*) of −1.35 V, while XAN-4 possessing eight strongly electron-withdrawing halogens exhibits the weakest E⁰(PC^•+/³PC*) of −0.91 V. The polymerization rate k_p^app was reported to be highly dependent on E⁰(PC^•+/³PC*) of these halogenated xanthene dyes, where more negative E⁰(PC^•+/³PC*) in combination with higher Φ_T leads to higher k_p^app.^73,75

Figure 21.

(A–F) Chemical structures of (A) XAN-1, (B) XAN-6, (C) XAN-2, (D) XAN-7, (E) XAN-3, and (F) XAN-4. (G) Electronegativities of the halogen substituents versus H to represent their electron-withdrawing capability. (H) Computed E⁰(PC^•+/³PC*) of the PCs in comparison. Reproduced from ref 73 and ref 75. Copyright 2019 American Chemical Society.

3.2.3.3. Photocationic RAFT Polymerization via the Reductive Quenching Pathway.

With respect to photo-LCP systems as well as photo-ROMP, current studies mostly assume ¹PC* as the catalytic species without consideration of ³PC*, despite the lack of evidence. However, because the energy levels and electronic structures of ¹PC* are very similar to the corresponding ³PC* state, except that ³PC* is lower in energy, herein we follow existing studies and use properties of ¹PC* for the discussion in the following context.

A library of triphenylpyrylium and triphenylthiopyrylium derivatives (Figure 22A–E) was compared by Fors and co-workers as PCs for photocationic RAFT polymerization.¹²⁴ The pyrylium core bears a positive charge and is thus highly electron-withdrawing, which by contrast renders the methyl group in PY-2 and the methoxy group in PY-3 highly electron-donating. As a consequence, these electron-donating substituents increase the energies of both the lower SOMO of ¹PC* and LUMO of PC, leading to less positive E⁰(¹PC*/PC^•−) and more negative E⁰(PC/PC^•−) of PY-2 and PY-3 as compared to the unsubstituted PY-1. Since the methoxy group is a stronger electron-donating group, PY-3 displayed much less positive E⁰(¹PC*/PC^•−) and hence contributed to much slower polymerization of isobutyl vinyl ethers (6 monomer additions per photon, Figure 22C). On the other hand, PY-2 and PY-1 contributed to similar polymerization rates which were measured to be around 35 monomer additions per photon,⁹⁴ in spite of less positive E⁰(¹PC*/PC^•−) of PY-2 (Figure 22B) compared to PY-1 (Figure 22A).¹²⁴ Unexpectedly, using PY-4 (Figure 22D) as PC led to slow polymerization which the authors attributed to poor solubility, while TPY-1 (Figure 22F) also led to slow polymerization but was found to have a high Φ_T = 0.94 (underlying mechanism yet unclear).¹²⁴ Meanwhile, the authors discovered that the more negative E⁰(PC/PC^•−) of PY-3 resulted in rapid chain-end recapping and hence much better short-time temporal control where polymerization is effectively halted in the dark compared with PY-1.¹²⁴ However, further studies showed that, after dark periods lasting over several hours, notable monomer conversions still occurred with PY-3.⁹⁴ To tackle this problem, Fors and co-workers implemented a series of Ir-based complexes (Figure 22F–H) bearing more negative E⁰(PC/PC^•−).⁹⁴ Indeed, IR-1 (Figure 21F) and IR-2 (Figure 22G) bearing E⁰(PC/PC^•−) = −0.69 V and −0.79 V, respectively, displayed excellent temporal control where no monomer conversion was observed in >20 h dark periods after ceasing light irradiation. Interestingly, the authors observed ultraslow polymerization with IR-3 (Figure 22H) which was proposed to be due to a much higher rate of polymer recapping relative to propagation as a result of highly negative E⁰(PC/PC^•−) = −1.16 V; this led to sluggish polymerization with <1 monomer addition per photon. In general, IR-1 and IR-2 also contributed to relatively slow photocationic RAFT polymerization because of the less oxidizing ¹PC*, indicated by less positive E⁰(¹PC*/PC^•−) combined with more negative E⁰(PC/PC^•−), which results in faster polymer recapping compared with triphenylpyrylium and triphenylthiopyrylium derivatives.⁹⁴

Figure 22.

Chemical structures, redox potentials, and polymerization efficiency for PCs used in photocationic polymerization. Triphenylpyrylium derivatives¹²⁴ (A) PY-1, (B) PY-2, (C) PY-3, (D) PY-4, triphenylthiopyrylium derivative¹²⁴ (E) TPY-1, Ir-based complexes⁹⁴ (F) IR-1, (G) IR-2, (H) IR-3, and bisphosphonium salt derivatives¹²⁵ (I) BPP-1, (J) BPP-2, and (K) BPP-3.

More recently, Liao and co-workers demonstrated the use of bisphosphonium salt derivatives (Figure 22I–K) to achieve efficient photocationic RAFT polymerization at low catalyst loading of 10 ppm relative to the monomer isobutyl vinyl ether.¹²⁵ Significantly, this allowed achieved strict temporal control to be achieved with organic PCs. While BBP-1 (Figure 22I) suffered from poor temporal control at high monomer conversions, by installation of relatively electron-donating methyl groups, the resultant BBP-2 (Figure 22J) and BBP-3 (Figure 22K) exhibited more negative E⁰(PC/PC^•−) and strict temporal control with no monomer conversion observed over long dark periods lasting ~3 h.¹²⁵

3.2.3.4. Photocationic NMP via the Reductive Quenching Pathway.

Kamigaito et al. compared Ir-based complexes (Figure 23A–C) in photocationic NMP polymerization and discovered that IR-1 (Figure 23A) bearing the most positive E⁰(¹PC*/PC^•−) = 1.70 V exhibited the highest efficiency for photocationic NMP of isobutyl vinyl ether, whereas IR-4 (Figure 23B) with less positive E⁰(¹PC*/PC^•−) = 1.23 V yielded much slower polymerization (Figure 23D).⁹⁵ On the other hand, IR-5 (Figure 23B, a commonly used oxidative quenching PC) has a much less positive E⁰(¹PC*/PC^•−) = 0.33 V and exhibited almost no activity for photocationic NMP via the reductive quenching pathway (Figure 23D). By examining the chemical structures for the three Ir-based PCs, it is evident that the much more positive E⁰(¹PC*/PC^•−) of IR-1 and IR-4 comes from the positively charged and highly electron-withdrawing iridium core structure, compared to the neutral IR-5. On the other hand, the electron-donating (relative to the positively charged iridium core) tertbutyl groups renders IR-4 less positive E⁰(¹PC*/PC^•−) than IR-1.

Figure 23.

Chemical structures of Ir-based complexes (A) IR-1, (B) IR-4, and (C) IR-5 with E⁰(PC*/PC^•−) vs SCE denoted below the label. (D) Polymerization of isobutyl vinyl ether via photocationic NMP catalyzed by IR-1 (red), IR-4 (blue), and IR-5 (orange). Reproduced with permission from ref 95. Copyright 2019 Wiley-VCH.

3.2.3.5. Photo-ROMP via the Reductive Quenching Pathway.

Boydston and co-workers made comparisons between the pyrylium and thiopyrylium PCs in photo-ROMP. Apart from PY-1 (Figure 24A), PY-2 (Figure 24B), PY-3 (Figure 24D), and TPY-1 (Figure 24E) discussed in section 3.2.3.3, the additional phenyl groups in PY-5 (Figure 24C) are also electron-donating (due to conjugated effect) relative to the positively charged pyrylium core, leading to less positive E⁰(¹PC*/PC^•−).⁷⁸ It was found that norbornene monomers are easily oxidized and thus the use of PCs bearing more positive E⁰(¹PC*/PC^•−) tend to result in a larger portion of oxidized norbornene and correspondingly reduced polymerization after 150 min irradiation, i.e., after reaching the maximum conversions.⁷⁸ Indeed, as shown in Figure 24A,B, PY-1 with the most positive E⁰(¹PC*/PC^•−) yielded the least polymerization, with only 9% monomer conversion and also the most oxidation (91%); by contrast, PY-3 with much less positive E⁰(¹PC*/PC^•−) led to 75% monomer conversion and only 25% oxidization of norbornene monomers. Interestingly, the thiopyrylium derivatives generally possessed improved ratio of polymerization given each type of substitution compared with their pyrylium counterparts (Figure 24). This could be due to higher Φ_T of thiopyrylium compounds relative to pyrylium compounds (because of the heavy atom effect see section 3.1.2.1), which led to a population of less energized ³PC*. As aforementioned, the populated ³PC* is characteristic of less positive E⁰(¹PC*/PC^•−) compared to ¹PC* with similar electronic structure. For example, PY-1 exhibits Φ_T = 0.42 and TPY-1 exhibits Φ_T = 0.94.¹²⁴

Figure 24.

Chemical structures of PCs used in photo-ROMP, their redox potentials, and portions of oxidized norbornene vs polymerization. Triphenylpyrylium derivatives (A) PY-1, (B) PY-2, (C) PY-5, and (D) PY-3 and triphenylthiopyrylium derivatives (D) TPY-1, (E) TPY-2, (F) TPY-3, and (G) TPY-4. A portion of the norbornene monomers being polymerized and being oxidized after full conversion was achieved within 150 min in photo-ROMP catalyzed by each PC, respectively.

4. CONCLUDING REMARKS AND OUTLOOK

Considering the ever-growing versatility of functional polymers in various applications, discovery of PCs via a trial-and-error strategy represents a rate-determining step in finding polymerization systems that meet the demand for sophisticated macromolecular syntheses. To tackle this limitation, rational computer-guided strategies for PC design have emerged, providing rapid and economic methodologies for discovering new PCs. This rational strategy is based on the acknowledgment of structure-property-performance relationships. As shown in Scheme 21, the performance of the photocontrolled polymerization can be correlated with the properties of the PC, which can be tuned by rational design of their chemical structure.

Scheme 21.

Rational Design of a PC to Control the Performance of Photocontrolled Polymerization

The underlying logic of the structure-property-performance relationships has been introduced through theoretical derivations and augmented by physical and chemical observations. Accordingly, there are many examples of applications of these guiding principles promoting polymerization efficiency by enhancing absorption, triplet yield, redox capabilities, as well as choice of activation wavelengths through structural variation of the PC chromophores. To date, a number of useful PCs have been discovered for different photocontrolled polymerization, which have resulted from implementation of the structure-property-performance relationships outlined in this review. Scheme 22 lists some typical examples^{52,54,73,86,125,180,181} that outcompete their counterparts with respect to certain aspects of performance in photocontrolled polymerization of acrylates, methacrylates, acrylamides, methacrylamides, and vinyl ethers. Nevertheless, a rational approach in PC design for photocontrolled polymerization has tremendous potential in unrealized applications.

Scheme 22.

A Few Examples of the PC-Initiator Pairs Designed with Better Performance in Photocontrolled Polymerization Systems to Date

To provoke future exploration on this topic, we propose the following future directions. (i) Developing NIR-light-regulated polymerization systems. Although great effort has been made to red-shift the maximum absorption wavelengths of PCs, the majority of current PCs do not absorb light beyond 800 nm. There are, however, great benefits to developing PCs with longer wavelength absorption. Indeed, far-red light with wavelengths above 800 nm and infrared light show increased depth penetration through human tissues and other materials.¹⁸² The ability to harness these longer wavelengths will open the possibility to perform photocontrolled polymerization and other organic transformations within tissues or vessels and in a spatiotemporally controlled manner. This goal can be achieved by collaboratively customizing PCs,¹⁸³ modifying initiators and designing new quenching pathways mediated by additives, in a rational manner. However, using lower energy light also introduces a thermodynamic boundary on the redox properties of photoexcited PCs. To overcome this boundary defined by the energetics of a single photon, synthetic strategies are emerging that combine the energetics of two photons into a single redox event,¹⁸⁴ coupling photochemistry with electrochemistry,¹⁸⁵ and direct excitation to T₁.¹⁸⁶ (ii) Tackling aqueous solubility issues of large π-conjugation PCs.¹⁸⁷ As mentioned earlier, the enhancement of photophysical properties in many chromophores is achieved through extending the π conjugation. This extended conjugation is usually accompanied by a large increase in hydrophobicity for large fused ring structures commonly used as PCs. As such, to enable broadened applications in 3D printing and biomedical scenarios, additional design parameters should be included in future PC design. The balance between photophysical properties and solubility of the new PCs should be considered depending on the intended application. (iii) Using a computer-guided approach to interactively tune properties of both the PC and the initiator. In order to design new highly selective reactions between PC-initiator pairs, the properties of both the PC and the initiator should be carefully designed. Advancements in computational capabilities such as machine learning will accelerate the discovery and development of new systems and enable the development of new cutting-edge orthogonal polymerization systems. Indeed, the selective reactivity of excited state PCs to the initiators can enable a much broader range of chemistries extending beyond polymerization. Complex multifunctional materials should be realized as a result. For example, by judiciously designing PCs to selectively match specific initiators, single-unit monomer-insertion or atom transfer radical addition can be achieved for the precise placement of functional groups within polymer chains.^59,188 (iv) Designing PCs with more isolated absorption wavelengths for orthogonal polymerization. Combining point (i) and point (iii), increasing the wavelength selectivity of PCs will enable multiple polymerization systems to be independently activated. PCs should be designed that show limited absorption over predesigned wavelengths ranges, while still showing appreciable absorption in other wavelength ranges.²⁸ Combinations of PCs with precisely tunable and nonoverlapping absorptions will enable new orthogonal photosystems in chemical and material syntheses.^28,29 As shown in Scheme 23A, orthogonal photosystems have the potential to synthesize polymers with complex block sequences or architectures by switching between different wavelengths. This can greatly alleviate the burden in polymer purification and could facilitate a facile approach in rapid synthesis of complex polymeric materials. On the other hand, orthogonal systems can enable surface patterning^{45–47,108,189,190} with different materials orthogonally controlled by different wavelengths, which may much increase the information stored per area.^{28,29,33,34,43,191} This direction is challenging and can only be achieved by delicate rational design of two or more PCs along with the initiator by probing into both the optical and chemical aspects, and with the help of computational studies. In addition, the design of these PCs should also consider the other design principles mentioned previously to ensure not only strong light absorption but also efficient photochemical processes thereafter. (v) Designing PCs for additive manufacturing. Photoinduced additive manufacturing uses a photosensitive resin where multifunctional monomers undergo rapid polymerization to form 3D polymer networks in the area of irradiation (Scheme 23B).⁵⁰ 3D objects can be printed in a layer-by-layer manner using a repetitive irradiation program. The use of light in additive manufacturing can offer higher resolution and faster build speeds to produce geometrically complex objects compared with their thermally mediated 3D printing counterparts.^50,192 By using photocontrolled polymerization, many additional merits can be imparted into additive manufacturing.^49,193–195 For example, the “reversible deactivation” feature in photo-RDRP and the “living” feature in photo-LCP result in well-defined polymeric frameworks and enable chain extension from an existing layer, forming covalent bonds and, hence, strengthening binding between layers. Indeed, the dormant polymer end-group at the surface of printed objects can facilitate continued printing by chain extension and surface processing by chemical modifications. Importantly, implementation of orthogonal photocontrolled polymerization systems could gain facile control over nanostructuration of 3D networkers,⁴⁸ hence enabling tunable mechanical properties.^196–198 Moreover’ the extensive range of PCs and corresponding photochemistries available to perform photocontrolled polymerization can provide versatile control over the polymerization process and contribute to materials with diverse physical and chemical properties. However, PCs used in photoinduced additive manufacturing must exhibit high Φ_T upon irradiation to populate ³PC* to deactivate oxygen during polymerization by photosensitization. More importantly, the PC should be very efficient so as to catalyze ultrafast polymerization for additive manufacturing. In addition, PCs absorbing longer wavelength light may be sought to increase light penetration into layers of resin. In this regard, it is both complicated and challenging to develop a PC for a particular photoinduced additive manufacturing system, and a computer-aided rational approach is certainly needed. (vi) Developing Biocompatible PCs for photocontrolled polymerization in bioapplications. With regard to the construction or modification of biorelated polymeric materials, thermally initiated polymerization systems are usually inaccessible as biomaterials such as proteins, cells, and living tissues are sensitive to high temperatures. On the other hand, visible light is benign to these biomaterials and controlled polymerization can facilitate production of polymers with well-defined molecular weights. Hence, photocontrolled polymerization has been used in a range of biorelated areas where maintaining bioactivities is important including fabricating protein–polymer bioconjugates^63,199 and engineering cell surfaces with polymers³⁷ (Scheme 23C). In these directions, the PCs should be carefully chosen/designed to be nontoxic, biocompatible, and water-soluble. Due to the exposure to oxygen in biosystems, high Φ_T of PC may also be needed as mentioned in point (v). With regard to in vivo applications, infrared (IR) absorbing PCs may be desirable for better light penetration. Additionally, biodegradability^200–202 or photo-bleaching⁹⁶ is another key feature desired in bioapplications, in order to more easily eliminate the impact of residual PCs after synthesis.

Scheme 23.

Some Examples of Future Directions for Photocontrolled Polymerization

We anticipate these future directions will further expand the current toolbox for photocontrolled polymerization systems and potentially other photocatalytic chemical transformations. On the other hand, we expect that the knowledge of structure-property-performance relationships presented in this review will inspire rational PC design for photocontrolled polymerization to facilitate chemical, polymer, and material syntheses. It may also promote future studies on these guiding principles and the theoretical backgrounds behind them.

GLOSSARY

¹PC*

the lowest singlet excited state photocatalyst

³PC*

the lowest triplet excited state photocatalyst

ATRP

atom transfer radical polymerization

BA

n-butyl acrylate

BnBiB

benzyl α-bromoisobutyrate

BSTP

3-benzylsulfanylthiocarbonylthiosulfanyl propionic acid

BTPA

2-(butylthiocarbonothioyl) propionic acid

CDB

cumyl dithiobenzoate

CDTPA

4-cyano-4-[(dodecylsulfanylthiocarbonyl)-sulfanyl]pentanoic acid

CPADB

4-cyanopentanoic acid dithiobenzoate

CPDTC

2-cyano-2-propyl dodecyl trithiocarbonate

CT

charge transfer

D–A

donor–acceptor

DCPD

dicyclopentadiene

DFT

density functional theory

DHO

displaced harmonic oscillator

DVP

dimethyl vinylphosphonate

E ⁰

standard redox potential

EA

electron affinity

EBiB

ethyl α-bromoisobutyrate

EBPA

ethyl-α-bromophenylacetate

EClPA

ethyl-α-chlorophenylacetate

EPR

electron paramagnetic resonance

ES

excited state

f

oscillator strength

GS

ground state

HOMO

highest occupied molecular orbital

IBDTC

S-1-isobutoxyethyl N,N-diethyl dithiocarbamate

IBTTC

S-1-isobutoxylethyl S′-ethyl trithiocarbonate

IC

internal conversion

IP

ionization potential

ISC

intersystem crossing

k _ET

electron transfer rate constant

k _F

fluorescence rate constant

k _IC

internal conversion rate constant

k _ISC

intersystem crossing rate constant

k_p ^app

apparent propagation rate

LED

light emitting diode

LUMO

lowest unoccupied molecular orbital

LSOMO

lower singly occupied molecular orbital

M

monomer

MLCT

metal-to-ligand charge transfer

MMA

methyl methacrylate

ms

millisecond

μs

microsecond

NIR

near-infrared

NMP

nitroxide-mediated radical polymerization

ns

nanosecond

NVP

N-vinyl pyrrolidinone

O-ATRP

organocatalyzed atom transfer radical polymerization

OQP

oxidative quenching pathway

P3T

photoinduced Dexter/triplet energy transfer

PC

photocatalyst

PC^•+

radical cation of photocatalyst

PC^•−

radical anion of photocatalyst

PDI

perylene diimide

PET

photoinduced electron transfer

PET-RAFT

photoinduced electron/energy transfer-reversible addition–fragmentation chain transfer

photo-ATRP

photocontrolled-atom transfer radical polymerization

photo-CMRP

photocontrolled-cobalt-mediated radical polymerization

photo-ITP

photocontrolled-iodine transfer polymerization

photo-LCP

photocontrolled-living cationic polymerization

photo-NMP

photocontrolled-nitroxide-mediated radical polymerization

photo-RCMP

photocontrolled-reversible complexation mediated living radical polymerization

photo-RDRP

photocontrolled-reversible deactivation radical polymerization

photo-ROMP

photocontrolled-ring-opening metathesis polymerization

photo-TERP

photocontrolled-organotellurium-mediated radical polymerization

ps

picosecond

RAFT

reversible addition–fragmentation chain transfer

ROHF

restricted open-shell Hartree–Fock

ROMP

ring-opening metathesis polymerization

RQP

reductive quenching pathway

RTPP

reduced tetraphenyl porphyrin

s

second

SCE

saturated calomel electrode

SOCME

spin–orbit coupling matrix element

SOMO

singly occupied molecular orbital

S₀

singlet ground state

S₁

lowest singlet excited state

S₂

Second lowest singlet excited state

S_CT

charge transfer singlet excited state

S_FC

Franck–Condon singlet excited state

S_n

nth singlet excited state

St

styrene

T₁

lowest triplet excited state

T₂

second triplet singlet excited state

T_CT

charge transfer triplet excited state

T_n

n^th triplet excited state

TDDFT

time-dependent density functional theory

TPP

tetraphenyl porphyrin

TS

transition state

TTA

triplet–triplet annihilation

USOMO

upper singly occupied molecular orbital

UV

ultraviolet

VP

vinyl pivalate

VR

vibrational relaxation

ZnTPP

Zn(II) tetraphenyl porphyrin

ε _max

maximum molar extinction coefficient

λ _max

maximum absorption wavelength

τ

excited state lifetime

Φ_T

triplet quantum yield

Biographies

Chenyu Wu received his Ph.D. in Chemical Engineering under the supervision of Prof Cyrille Boyer at University of New South Wales, Sydney, in 2020. He is currently a post doctorate at Shandong University where his research focuses on computer-aided rational design of photocatalysts for polymerization, mechanistic understanding of photocontrolled polymerization and development of algorithms for molecular property calculations.

Nathaniel Corrigan received his Ph.D. in Chemical Engineering under the supervision of Prof Cyrille Boyer at University of New South Wales, Sydney, Australia in 2019. His Ph.D. thesis focused on the combination of visible light mediated reversible deactivation radical polymerization (RDRP) and flow chemistry for advanced macromolecular synthesis. He is currently a research associate at UNSW Sydney where his research focuses on exploiting visible light for controlled radical polymerization, flow chemistry, and materials synthesis via 3D printing approaches.

Chern-Hooi Lim received his Ph.D. in Chemical Engineering (2015) from the University of Colorado Boulder under the supervision of Prof. Charles Musgrave. He later joined Prof. Garret Miyake as a postdoctoral researcher and was awarded NIH F32 Postdoctoral Fellowship to advance the development of photocatalysis and organic photocatalysts for efficient small molecule and macromolecule syntheses. He is now the cofounder and CEO of New Iridium Inc, developing photocatalytic solutions that focus on utilizing biobased and CO₂ feedstocks for significantly greener and more economical chemical manufacturing. Dr. Lim was recently named to the C&EN Talented 12 (Class of 2021) for his contribution in advancing the science of photocatalysis.

Wenjian Liu obtained his Ph.D. in 1995 from Peking University and was a Cheung Kong professor at Peking University between 2001 and 2018. He is now a chair professor at Shandong University and the inaugural director of Qingdao Institute for Theoretical and Computational Sciences. He is an elected member of the International Academy of Quantum Molecular Science, fellow of the Royal Society of Chemistry, and fellow of the Asia-Pacific Association of Theoretical and Computational Chemists. He was honored to receive the Annual Medal of the International Academy of Quantum Molecular Science, the Bessel Award of the Humboldt Foundation, and the People and Fukui Medals of the Asia-Pacific Association of Theoretical and Computational Chemists. His research focuses on relativistic theories and methods for molecular electronic structure and magnetic properties, as well as wave function-based methods for strongly correlated electrons.

Garret Miyake is an Associate Professor and the Dr. Robert Williams Professor of Organic Chemistry at Colorado State University. He earned a B.S. at Pacific University. He performed Ph.D. studies with Eugene Chen at Colorado State University before conducting postdoctoral research with Robert Grubbs at the California Institute of Technology. The Miyake group has research interests in the fields of photoredox catalysis, organocatalyzed atom-transfer radical polymerization, sustainable plastics, and the synthesis of block copolymers that self-assemble to photonic crystals. He has been recognized by the 2021 Journal of Polymer Science Innovation Award, the 2017 ACS Division of Polymer Chemistry Herman F. Mark Young Scholar Award, a Camille Dreyfus Teacher-Scholar Award, a Cottrell Scholar Award, and a Sloan Research Fellowship. He is also a cofounder of New Iridium and Cypris Materials.

Cyrille Boyer obtained his Ph.D. in 2006 from the University of Montpellier. He is a full professor and deputy head of school in the School of Chemical Engineering, at the University of New South Wales, Sydney, and codirector of Australian Centre for Nanomedicine. He is an elected member of European Academy of Sciences and has published over 340 research articles, which have generated over 25,000 citations, resulting in an H-index of 90. He was selected as a “Highly Cited Researcher in Chemistry” by Web of Science in 2018, 2019, and 2020, and that for “exceptional cross-field performance” in 2021. His research group works on the development of new polymerization techniques using photocatalysts, resulting in the discovery of PET-RAFT polymerization in 2014 and the development of photosingle unit monomer insertion (photo-SUMI). He uses these polymerization tools for the preparation of functional polymers which find applications in nanomedicine, energy, and materials science. He was the recipient of the 6th Polymer International-IUPAC Award for Creativity in Applied Polymer Science or Polymer Technology, 2018 Award of Excellence in Chemical Engineering awarded by ICHEM, 2018 Polymer Chemistry Lectureship awarded by Royal Society of Chemistry, 2015 Prime Minister Prize for Physical Science in Australia, 2016 ACS Biomacromolecules/Macromolecules Award, 2016 Le Fevre Memorial Award (Australia Academia of Science), and 2014 Scopus Young Research Award.