Methyl methacrylate (MMA, 99%, Sigma-Aldrich, United States), oligo (ethylene glycol) methyl ether methacrylate (OEGMA, average MW = 500, Sigma-Aldrich, United States), acrylonitrile (AN, 99%, Sigma-Aldrich) and methyl acrylate (MA, 99%, Sigma-Aldrich, United States) were purified by passing the monomers through a column filled with basic alumina to remove the inhibitor. Deionized water (DI water) was obtained from Millipore-Sigma Milli-Q water purification system. Azobisisobutyronitrile (AIBN, 98%, Sigma-Aldrich, United States) was recrystallized in anisole and stored at 4°C in dark. Tris (2-pyridylmethyl)amine (TPMA) was synthesized based on previously reported methods (Xia and Matyjaszewski, 1999). Copper bromide (CuBr₂, 99%, Acros Organics, United States), ethyl α-bromoisobutyrate (EBiB, 98%, Sigma-Aldrich, United States), ethyl α-bromophenylacetate (EBPA, 97%, Sigma-Aldrich, United States), 2-hydroxyethyl 2-bromoisobutyrate (HO-EBiB, 95%, Sigma-Aldrich, United States), 2-bromopropionitrile (BPN, 97%, Sigma-Aldrich, United States), deuterated DMSO (DMSO-d₆, 99.9%, Cambridge Isotope Laboratories, United States), deuterium oxide (D₂O, 99.9%, Cambridge Isotope Laboratories), anisole (99%, Sigma-Aldrich, United States), dimethyl sulfoxide (DMSO, 99.7%, Fisher, United States), toluene (99%, Fisher, United States), and dimethylformamide (DMF, 99.8%, Fisher, United States) were used as received. Tetraethylammonium tetrafluoroborate (Et₄NBF₄, 99%, Alfa Aesar, United States), used as a supporting electrolyte for electrochemical analysis, was recrystallized from ethanol, and dried in a vacuum oven at 70°C for 48 h.

¹H nuclear magnetic resonance (NMR) was conducted on a Bruker 500 MHz AVANCE III NMR spectrometer. Polymerizations in organic solvents used DMSO-d₆ as the solvent for ¹H NMR, and D₂O for aqueous ATRP. DMF GPC was equipped with a refractive index (RI) detector, an Agilent 1260 Infinity II pump and a Wyatt Optilab T-rEX RI detector, with a PSS GRAM analytical column set (10 µm particle size) and LiBr-containing HPLC grade DMF as the eluent (LiBr: 0.05 M). Agilent EasiVial poly (methyl methacrylate) (PMMA) calibration kit (part number: PL 2020-0200) was used to calibrate the RI detector of GPC. Blue (λ _max = 450 nm, 5.0 mW.cm⁻²) and green (λ _max = 520 nm, 4.7 mW.cm⁻²) LED lights were purchased from aspectLED. The light intensity was measured by a ThorLabs PM100D compact power and energy meter console. The photoreactor for polymerization was prepared by mounting LED strips inside a round glass container, and air flow was applied to limit temperature increase during the polymerization. Solvent evaporation was performed on a Biotage V10 rapid solvent removal system. UV-vis spectra were taken on an Agilent Cary 60 UV-vis spectrometer and recorded by Cary WinUV software. Linear sweep voltammetries were measured by an Autolab PGSTAT302N potentiostat/galvanostat (Metrohm) run by a computer with NOVA 2.0 software. Dynamic light scattering (DLS) was measured by a Malvern Zetasizer Ultra light scattering system. Transmission electron microscopy (TEM) was performed using a JEOL JEM-2000EX TEM.

ht-PLP_PAN, was synthesized via previously reported procedures (Sun et al., 2020). PAN₁₆₅ (subscript defines the degree of polymerization) was used to prepare ht-PLP_PAN and was synthesized by initiators for continuous activator regeneration (ICAR) ATRP using BPN as the initiator (Lamson et al., 2016; Kopeć et al., 2017). In a typical procedure, a 10 mg/ml suspension of ball-milled PAN₁₆₅ in DI water was prepared. 10 ml suspension was stirred for 10 min and then transferred to a 25 ml autoclave reactor with high-temperature resistant liner. The autoclave reactor was securely sealed and placed in a pre-heated oven (180°C). The heating was turned off after 12 h and the oven was let to cool down to room temperature naturally. The dark brown solution was passed through a 0.22 µm syringe filter with a polyethersulfone membrane. The solution was directly dried by vacuum to yield brown solid ht-PLP_PAN.

In a typical procedure, 5 mg ht-PLP_PAN was added to a 10 ml Schlenk flask containing 0.3 ml DMF and 7.2 ml DI water with a magnetic stir bar. HO-EBiB (3.9 µL, 0.027 mmol, 1 eq), CuBr₂ (0.60 mg, 2.7 µmol, 0.1 eq), TPMA (2.35 mg, 8.1 µmol, 0.3 eq), NaBr (41.17 mg, 0.4 mmol), OEGMA (2.5 ml, 5.4 mmol, 200 eq) were subsequently added to the Schlenk flask. The Schlenk flask was then purged with N₂ for approx. 25 min, and 0.1 ml of the reaction was withdrawn and was used as the “t = 0” sample. Finally, the Schlenk flask was placed in the photoreactor, and light was turned on to start the polymerization. The conversion of OEGMA was monitored by ¹H NMR by withdrawing samples (∼0.1 ml each time) from the reaction mixture at different time points.

Linear sweep voltammetries were carried out in a 5-neck electrochemical cell placed inside the photoreactor, equipped with a 3-electrode system and connected to an Autolab PGSTAT302N potentiostat/galvanostat (Metrohm) controlled by NOVA 2.0 software. The 3-electrode system was composed by: 1) a Pt foil counter electrode; 2) a homemade quasi-reference electrode: Ag|AgI|(0.1 M n-Bu₄NI in DMF); 3) a glassy carbon (GC) disk tip (3 mm dia, Metrohm), connected to a rotating disk electrode (RDE) system, as working electrode. Before each experiment, the GC disk was cleaned by polishing with a 0.25-μm diamond paste, followed by ultrasonic rinsing in ethanol for 5 min. Ferrocene (Fc) was added at the end of each experiment as an internal standard, to refer all potentials to the saturated calomel electrode [SCE, E°(Fc⁺|Fc) = 0.475 V vs SCE in DMF]. A steady air flow was applied to limit the temperature increase caused by light irradiation, and all experiments were performed under inert atmosphere (N₂).

ht-PLP_PAN was synthesized based on a previously reported hydrothermal reaction (Sun et al., 2020). Dynamic light scattering (DLS) measurement suggested that ht-PLP_PAN formed aggregates in water (Supplementary Figure 1). Transmission electron microscopy (TEM) revealed that the average diameter of the quasi-spherical assemblies was ca. 20 nm (Supplementary Figure 2). At first, using ht-PLP_PAN as a photoinitiator in FRP was attempted under blue or green light irradiation. However, no polymerization was observed under N₂ atmosphere in the absence of additional initiators (Table 1, entries 5 and 9, discussed in detail later). Thus, our focus shifted to using ht-PLP_PAN as a reducing agent for the CuBr₂ complex with a common ATRP ligand, tris(2-pyridylmethyl)amine (TPMA). (Xia and Matyjaszewski, 1999).

To evaluate the capability of ht-PLP_PAN in photoreducing the CuBr₂/TPMA complex, photophysical properties of ht-PLP_PAN (Figure 1) were analyzed. Our previous study reported a broad UV-vis absorption profile (Figure 1A) and a short lifetime of the excited state ht-PLP_PAN (<5 ns), suggesting the singlet nature (Sun et al., 2020). Putative photophysical properties of ht-PLP_PAN were modeled using density functional theory (DFT) calculations with model oligoimine-based structures of different conjugation lengths oI_N (where N = 3–10 denotes the number of nitrogen atoms along the backbone), which were deemed likely to arise in the course of hydrothermal treatment (Figure 1B, see Supplementary Material for Cartesian coordinates). (Sun et al., 2020) In all instances, the energies of the lowest unoccupied molecular orbital (LUMO) of oI_N were higher than the LUMO (β) of [Br-Cu^II/TPMA] ⁺ , indicating conditions favorable for reduction of the Cu^II complex upon photoexcitation (Figure 1C). The LUMO _oIN-LUMO _[Br-Cu ^II _/TPMA] ⁺ gap increased, and the wavelength of the lowest-energy transition determined by time-dependent DFT (TDDFT) decreased with the decrease of N (Supplementary Figure 3), suggesting stronger “reducing power” of blue light absorbing species, in agreement with experimental observations. Experimental studies on the photoreduction of the Cu^II complex were performed using linear sweep voltammetry (LSV), which are discussed in Photocatalytic Mechanism and Comparison Between Activation by Cu ^I Complex and by ht-PLP _PAN section.

Blue (λ _max = 450 nm) and green light sources (λ _max = 520 nm) were chosen for photoinduced ATRP using ht-PLP_PAN. Oligo (ethylene glycol) methyl ether methacrylate with an average molecular weight of 500 (OEGMA₅₀₀) was polymerized using ATRP catalyzed by a CuBr₂-complex with TPMA as the ligand (molar ratio: CuBr₂/TPMA = 1/3) (Table 1). 100 ppm or 500 ppm (relative to the monomer concentration) of CuBr₂/TPMA were used, with 2-hydroxyethyl 2-bromoisobutyrate (HO-EBiB) as the ATRP initiator and 0.5 mg/ml ht-PLP_PAN as the photo-cocatalyst. Additionally, 40 mM of NaBr was added to suppress the dissociation of the weak Cu^II-Br bond in the ATRP deactivator (Simakova et al., 2012; Fu et al., 2018).

First, polymerizations were performed under blue light irradiation. Linear semilogarithmic kinetic plots for ATRP using both 500 ppm (entry 1, Table 1) and 100 ppm (entry 2, Table 1) of the Cu^II complex are shown in Figure 2A. Polymer molecular weight and dispersity (Đ) were measured by gel permeation chromatography (GPC) using dimethylformamide (DMF) as the mobile phase (Figure 2B). Increasing the loading of CuBr₂/TPMA from 100 ppm (entry 2) to 500 ppm (entry 1) resulted in a decreased Đ from 1.69 (entry 2) to 1.35 (entry 1), due to the higher equilibrium concentration of the deactivator. Clean shifts of the molecular weight distribution (MWD) traces are shown in Supplementary Figure 4. It is worth mentioning that the difference in the molecular weight measured from the light scattering (LS) detector of GPC (M _n,GPC) and the theoretical molecular weight (M _n,theo) was likely due to different polymer-column interactions between poly (OEGMA) and the calibration standard (PMMA) used for the calibration of the LS detector. Figure 2A also illustrates that the rate of polymerization for entry 1 (500 ppm of Cu complex) was faster than that for entry 2 (100 ppm of Cu complex). This difference was attributed to the higher concentration of propagating radicals in entry 1 and faster reduction of the Cu^II complex. Similar kinetic results were previously reported by several ATRP methods based on activator regeneration (Simakova et al., 2012; Mendonça et al., 2014). Despite the lower monomer conversion determined by ¹H NMR, entry 2 showed a higher M _n,GPC than entry 1 (Table 1). This difference indicated that the initiation efficiency of HO-EBiB was lower when a lower loading of CuBr₂/TPMA was used in entry 1.

In order to confirm the role of ht-PLP_PAN, a control experiment was performed in its absence (entry 3, Table 1). No substantial monomer conversion (<5% after 8 h) was observed, indicating that the ht-PLP_PAN was required to generate the Cu^I activator complex. In addition, the possibility of direct activation of the alkyl bromide (R-Br) initiator by ht-PLP_PAN was considered. A control experiment (entry 4, Table 1) showed that under blue light irradiation HO-EBiB was indeed activated in the presence of ht-PLP_PAN and in the absence of the Cu^II complex. The polymerization was uncontrolled (Đ = 2.74) due to the absence of Cu^II deactivators. In comparison, no polymerization occurred when no HO-EBiB was added, indicating that ht-PLP_PAN did not generate radicals directly from the monomer. A quantitative comparison between the activation of R-Br by the Cu^I complex and by ht-PLP_PAN is presented in the Photocatalytic Mechanism section.

Similar polymerizations were conducted under green light irradiation (entries 6–7, Table 1 and Figure 2). In general, polymerization rates under green light irradiation were slower than those under blue light irradiation, and proceeded only after an induction period (Figure 2A). Finally, similar control experiments (entries 8 and 9, Table 1) were performed under green light. In contrast with the control experiments performed under blue light, no polymerization was observed under green light irradiation in the absence of Cu^II, indicating that ht-PLP_PAN luminophores excited by this range of wavelengths were not capable to activate the alkyl bromide initiator under green light, even after prolonged irradiation (48 h, entry 8, Table 1).

In addition to the already discussed lower “reducing power” of longer-wavelength luminophores inferred from DFT model calculations, two other factors that could be responsible for the slower polymerization and absence of ht-PLP_PAN-driven activation of R-Br under green light are: 1) the lower light absorption of ht-PLP_PAN at 520 nm (Figure 1), and 2) the slightly lower intensity of the green light photoreactor (4.7 mW.cm⁻², vs 5.0 mW.cm⁻² of the blue light photoreactor). Nevertheless, green light irradiation still resulted in reasonably well-controlled polymerization of OEGMA₅₀₀ (Đ = 1.38) using 500 ppm of Cu^II/TPMA (entry 6, Table 1).

To demonstrate the versatility of ht-PLP_PAN as OPCs, a mixed solvent containing 1:1 (v/v) of DMSO and DMF was first used to polymerize MMA with ethyl α-bromophenylacetate (EBPA) as the initiator. It is worth mentioning that ht-PLP_PAN only partially dissolved when using DMSO/DMF due to the lower polarity of the reaction media, especially after the addition of MMA. Despite the weaker solubility of ht-PLP_PAN, well-controlled polymerizations were obtained with ht-PLP_PAN dispersed in the solvent using 100 ppm or 25 ppm of CuBr₂/TPMA as the catalyst (Table 2).

Similar to the case of aqueous media, better control was obtained for the polymerization of MMA with a higher loading of CuBr₂/TPMA (entries 1 and 5, Table 2). However, 500 ppm of the Cu complex were required to control the process in aqueous media (entries 1 and 6, Table 1), while concentrations of Cu^II complex as low as 25 ppm were sufficient in organic solvents (entry 6, Table 2). Linear semilogarithmic kinetic plots and the GPC traces observed under those conditions are shown in Figure 3. Clean shifts of GPC traces were also observed (Supplementary Figure 5). Additionally, the direct activation of EBPA by ht-PLP_PAN was evaluated by control experiments in the absence of CuBr₂/TPMA (entries 3 and 7, Table 1). Similar to results in aqueous media, ht-PLP_PAN activated EBPA only under blue light irradiation. The supplemental activation of EBPA under blue light irradiation likely contributed to the higher dispersity when 25 ppm loading of CuBr₂/TPMA was used (entry 2). The uncontrolled polymerization in the absence of any Cu complexes indicated that no deactivations occurred due to the absence of deactivators.

Finally, ATRP using ht-PLP_PAN was extended to other monomers and solvents (Supplementary Table 1). For example, methyl acrylate (MA) was polymerized using 100 ppm of the CuBr₂/TPMA complex and ethyl α-bromoisobutyrate (EBiB) as the initiator (entry 1, Supplementary Table 1). Good control over the polymerization was illustrated by the low Đ = 1.16 of the polymer (Supplementary Figures 6A,B). Additionally, ATRP of MMA was performed using 100 ppm of the CuBr₂/TPMA complex and anisole as a less polar solvent (entry 2, Supplementary Table 1). The rate of polymerization was slower in anisole compared to DMSO/DMF, as expected from the variation of the ATRP equilibrium constant, K _ATRP, with solvent polarity (Ribelli et al., 2019). Nonetheless, well-controlled polymerization (Đ = 1.19) was still observed (Supplementary Figures 6C,D).

In some ATRP methods with regeneration of the Cu^I/L activator, the agent used for the activator regeneration can also contribute to the activation of dormant alkyl bromides to form propagating radicals, which generally occurs by reaction between the agent and the alkyl bromide (Konkolewicz et al., 2013; Konkolewicz et al., 2014a; Konkolewicz et al., 2014b; Zhang et al., 2011). To better understand the polymerization mechanism of the current system, the photoreduction of the Cu^II complex by ht-PLP_PAN was first monitored by linear sweep voltammetry (LSV). As shown in Figure 4, LSV was scanned from ca. 0.5 V (vs SCE) where the Cu^I complex is oxidized if it exists. Indeed, the increased intensity of the anodic current under blue or green light irradiation confirmed the formation of Cu^I/TPMA complex in DMSO/DMF (volume ratio: 1/1) (Dadashi-Silab et al., 2021). The role of ht-PLP_PAN as the primary factor behind the observed reduction of Cu^II/TPMA complexes after 3 h of blue light irradiation was confirmed by comparing the fraction of Cu^I complexes estimated from the current values in the presence of ht-PLP_PAN (ca. 2.5%) and in its absence (<0.5%, Supplementary Figure 7A). The small extent of photoreduction in the latter case was likely caused by the presence of excess ligand that can act as the electron donor. Notably, the LSV traces acquired in water did not show a consistent increase of Cu^I/L concentration (Supplementary Figure 7B), in contrast with polymerization results (Table 1) that indicated ht-PLP_PAN was capable of photoreducing the Cu^II complex in water. This apparent discrepancy indicates that the amount of Cu^I produced by ht-PLP_PAN in water was below the reliable LSV detectability threshold but was still sufficient to effectively initiate polymerization owing to the higher K _ATRP of Cu catalysts in water (Tang et al., 2008; Fantin et al., 2015). It should be pointed out that the apparent lower reducing efficiency of ht-PLP_PAN in water in comparison with organic solvents could be caused by its nano-assembly, which, while important for allowing these mostly hydrophobic species to function in water, would inevitably decrease the number of accessible photocatalytic sites.

To further understand the composition of the catalytic system for aqueous ATRP under blue light irradiation, the following calculations were performed to quantify the concentrations of Cu^I and Cu^II species in polymerizations corresponding to conditions in Table 1. The activation rates of alkyl bromides by Cu^I/L and ht-PLP_PAN were estimated for the polymerization of OEGMA (entries 1, 2 and 4 of Table 1). First, the rate of activation by the Cu^I complex ( R a 1 ) was given by Eq. (1), (Matyjaszewski 2012; Krys et al. 2016; Krys and Matyjaszewski 2017) thus the concentration of the Cu^I complex ( [ Cu I / L ] ) was calculated. Eq. 2 shows the relationship between [ Cu I / L ] and the rate of polymerization, R p : R a 1 = k a 1 [ Cu I / L ] [ RX ] (1) R p = k p   K ATRP [ RX ] [ Cu I / L ] [ Br − Cu II / L ] [ M ] (2) where, [RX], [M], [Cu^I/L] and [Br-Cu^II/L] correspond to the concentration of the alkyl bromide initiator, monomer, Cu^I complex and Cu^II complex, respectively. k a 1 is the activation rate constant of HO-EBiB by Cu^I/TPMA. K ATRP is the ATRP equilibrium constant for Cu^I/TPMA with HO-EBiB, and k p is the rate coefficient of propagation for OEGMA. (Smolne et al., 2016)

On the other hand, the activation rate of alkyl bromides by ht-PLP_PAN ( R a − pc ) was determined from Eq. 3: R p ,  without Cu = k p [ M ] [ R ] = k p [ M ] R a − pc k t (3) where, R p ,  without Cu refers to the rate of polymerization in the absence of Cu^II complex. k t is the termination rate constant of OEGMA, (Smolne et al., 2016) and [R·] is the concentration of propagating radicals.

Details of the calculation, including the scaling of kinetic parameters, are included in the Supplementary Material (Supplementary Tables 2–4) (Fantin et al., 2015; Fantin et al., 2017). Specifically, R p ,  without Cu was monitored by ¹H NMR (entry 4 of Table 1, Supplementary Figure 8). The polymerization was slower when no Cu complex was present, and polymers with high molecular weight (over seven times higher than M _n,theo) were observed by GPC, indicating a low initiation efficiency when ht-PLP_PAN was the only activator.

As shown in Table 3, the calculated R p ,  without Cu was 6.34 × 10⁻⁶ M.s⁻¹, which was ca. one order of magnitude smaller than R p [ R p (100 ppm) = 5.46 × 10⁻⁵ M.s⁻¹, R p (500 ppm) = 7.22 × 10⁻⁵ M.s⁻¹]. Based on these values, the calculated [ Cu I / L ] was 5.5 × 10⁻⁷ M for ATRP using 100 ppm of CuBr₂/TPMA, and 3.6 × 10⁻⁶ M for ATRP using 500 ppm of CuBr₂/TPMA. Although Cu^I/L only consisted of 1.0% (entry 1, Table 3) and 1.4 (entry 2, Table 3) of the total Cu species, the calculated values of R a 1 (Table 3) were five to six orders of magnitude higher than R a − pc (9.3 × 10⁻¹⁰ M.s⁻¹). This significant difference in the activation rate was due to the high k a 1 value of [Cu^I/TPMA]⁺ in aqueous media (Fantin et al., 2017), making Cu^I/L the predominant activator over ht-PLP_PAN in this system.