Introduction

Plastics have fundamentally transformed human life, particularly through the proliferation of single-use items such as packaging, utensils, and bottles. Their ubiquity stems from affordability, light weight, and remarkable durability, rendering them indispensable across countless applications. As demand has surged, global plastic production has skyrocketed, surpassing 400 megatons (Mt) in 2022 alone1—roughly equivalent to the mass of 1,200 Empire State Buildings—and is projected to reach 1.2 gigatons (Gt) annually by 20602. However, this rapid growth comes with substantial environmental costs: in 2019, 72% of polymer waste was landfilled or mismanaged, and another 19% was incinerated for energy recovery—both posing considerable environmental hazards2,3. Additionally, plastic production is anticipated to drive a steep increase in greenhouse gas emissions, reaching 6.5 Gt of CO2-equivalents by 2050—around 15% of the global total4. Achieving a net-zero future will require a plastic recycling rate above 70%5; yet currently, only 9% are recycled, with most undergoing mechanical recycling to produce downgraded materials that ultimately return to waste (Fig. 1A). This stark discrepancy underscores the urgent need to re-envision the plastic lifecycle, turning end-of-life products into value-added resources rather than waste6.

Fig. 1: Overview of chemical recycling and upcycling of plastics.
figure 1

A Global plastic production by polymer and the fate of plastic waste. B Key challenges in chemical recycling. C Summary of the current study. D Comparison with the state-of-the-art.

Full size image

To this end, chemical recycling transforms post-consumer plastics into monomers or higher-value small molecules, presenting substantial ecological and economic benefits7,8. Nonetheless, plastics produced through chemical recycling account for less than 0.1% of the total output1. The primary challenge lies in the fact that over 77% of plastics originate from vinyl polymers—such as PP, PE, PVC, and PS—whose backbones consist of C–C and C–H bonds with high bond dissociation energies (BDE)9. Recently, photocatalysis has proven especially effective for functionalizing inert bonds in small-molecule transformations10,11. Building on this success, leveraging photon energy offers a compelling approach to plastic upcycling, enabling distinctive selectivity under ambient reaction conditions—unlike traditional thermal cracking12,13,14,15. Specifically, unactivated aliphatic C–H bonds (BDE > 95 kcal·mol−1) within polymer backbones can be cleaved by photoinduced radical species via hydrogen atom transfer (HAT), thereby initiating degradation (Fig. 1B). Indeed, several photocatalytic HAT-based upcycling strategies, both heterogeneous and homogeneous, have been developed for vinyl polymers16,17. However, existing methods have primarily targeted polystyrene owing to its relatively weaker benzylic C–H bonds (BDE = ca. 85 kcal·mol−1)18,19,20,21,22,23,24. For example, the photooxidation of PS to benzoic acid can be catalyzed by Lewis acidic FeCl₃-based catalysts25,26,27, as well as strong Brønsted acids like p-toluenesulfonic acid28. The Das group also developed an efficient, acid-free photocatalytic process for PS, using N-bromosuccinimide and sodium triflinate29. Other heterogeneous photocatalysts have also been explored30; for instance, Li and coworkers used modified TiO₂ under UV light to convert PS into benzoic acid31. Despite recent advances, photochemical valorization of plastics still faces serious obstacles that limit its practical application, including high reagent loadings, reliance on metal catalysts and UV light, and the need for pure oxygen and acidic conditions. Additionally, these methods are often substrate-specific, restricting their applicability to the mixed waste typically found in real-world scenarios32. Developing a single, versatile organic photocatalyst capable of efficiently revalorizing diverse polymers would be transformative, filling a critical gap in the management of unsorted plastic waste33,34,35.

Consecutive photoinduced electron transfer (conPET) represents a unique approach for accessing potent redox potentials by harnessing the energy of two photons within a single catalytic cycle36,37,38. To date, however, conPET has been demonstrated only in small-molecule reactions. In this work, we report a conPET framework employing a readily available phenothiazine derivative PTH-3CN, which effectively deconstructs a wide array of commodity polymers (resin codes 1–7) into value-added small molecules with catalyst loadings as low as 500 ppm (Fig. 1C). This metal-free, streamlined protocol proceeds without acid additives and requires only ambient air as the oxidant, showcasing exceptional robustness towards real-world plastics and scaling readily in a flow system39,40. Notably, for PS, this system operates at a reagent loading that is orders of magnitude lower than those of previously reported methods by weight, while still maintaining excellent efficiency in producing benzoic acid (Fig. 1D, details in Supplementary Table S11)18,19,20,21,22,23,24,25,26,27,28,29,30,31.

Inert vinyl polymers such as PP, PE, and PVC can also be broken down into formic acid and/or acetic acid at considerably faster rates than prior photocatalytic techniques41. Moreover, the protocol extends to other common non-biodegradable polymers—including polyvinyl acetate (PVAc), polymethacrylate (PMA), poly(methyl methacrylate) (PMMA), PET, PU, and PC—as well as mixed waste examples. Remarkably, the broad range of polymers upcycled by a single photocatalyst—namely, PTH-3CN—encompasses over 80% of the plastic waste generated globally each year3. Our findings also provide mechanistic insights into polymer photodegradation, revealing that singlet oxygen—often considered a potent HAT reagent for polymer C–H bonds23,28—is not the primary HAT species under our conditions, as supported by detailed experiments and density functional theory (DFT) calculations. Furthermore, we discovered that phenothiazine PTH-3CN in fact functions as a precatalyst, gradually decomposing under visible-light in air to form multiple arylamine derivatives. Several synthesized triarylamine model catalysts exhibited catalytic activity comparable to PTH-3CN, highlighting their role as active species in this process. Beyond producing a single target molecule, this strategy provides access to a range of chemically and industrially relevant products depending on the polymer substrate and reaction conditions, underscoring its broad potential in chemical recycling frameworks.

Results

Mechanistic hypothesis and reaction optimizations

We commenced our study by exploring reductive photocatalysts for the conPET process, using the photooxidation of polystyrene (PS) to benzoic acid (BA) as the benchmark reaction. The proposed catalytic cycle based on conPET is illustrated in Fig. 2A. A key aspect of this process is the excitation of the photochemically oxidized photocatalyst PC•+ to its excited state [PC•+]*, which, in principle, exhibits a considerably enhanced oxidation potential. This short-lived species can then engage in single electron transfer (SET) with PS, giving rise to the radical cation PS•+ while simultaneously regenerating PC. Alternatively, an SET between [PC•+]* and the arene solvent molecule is also feasible, whereby the resulting arene radical cation may mediate the formation of PS•+. Upon deprotonation and subsequent oxidation, PS•+ yields BA via a benzylic radical intermediate PS•42. Additionally, superoxide O2•–, formed as a byproduct of the initial PC oxidation, may be converted into other reactive oxygen species (ROS) capable of abstracting the benzylic hydrogen atom in PS, further promoting the production of PS•.

Fig. 2: Photocatalyst selection and method development.
figure 2

A Proposed photocatalytic cycle involving two photon absorptions. B Our initial conditions for polystyrene upcycling. aThe yield was determined by HPLC analysis. C DFT-calculated properties of various phenothiazine derivatives, and experimental results of selected photocatalysts. D The optimization of reaction conditions using PTH-3CN as the catalyst. E Other PS substrates. b0.50 mol% PTH-3CN was used. c5.0 mol% PTH-3CN was used with HFIP as a co-solvent.

Full size image

Along this vein, we screened various common photoreductants (Supplementary Table S1). To our delight, we discovered that, in the presence of phenothiazine (PTH), polystyrene PS1 (ca. 50 mg, Mw = 350 kDa, Đ = 2.06) was converted into benzoic acid (44%) and acetophenone (1.1%) under 390 nm LED irradiation in an oxygen atmosphere (Fig. 2B). Of note, the addition of lithium salts or hexafluoroisopropanol (HFIP)—previously essential in conPET reactions to mitigate back electron transfer (BET) between O2•– and PTH•+—was unnecessary, suggesting concurrent consumption of O2•–43. A series of phenothiazine derivatives were then screened in silico, evaluating both their calculated absorption maxima (λmax) and the redox potential of their excited [PC•+]* states (Fig. 2C, details in Supplementary Table S13). To improve upon the parent PTH, we selected six PTH derivatives that can be prepared in one step from commercially available materials, predicted to exhibit improved visible light absorption and higher Ered for the [PC•+]* state, for experimental testing (Fig. 2C). Indeed, five of these derivatives outperformed the parent PTH in producing BA from PS1. Among them, the cyano-substituted analogs PTH-CN and PTH-3CN delivered the best yields (≥50%) and were thus chosen for further optimization (Supplementary Tables S2–S4).

Compared to PTH-CN, PTH-3CN manifested significantly enhanced light absorption in the visible range (Supplementary Fig. S11). As a result, we anticipated that PTH-3CN might retain strong photocatalytic activity at lower concentrations with a longer irradiation wavelength. Indeed, when the catalyst loading was reduced to 0.5 mol%, PTH-CN achieved only 34% yield of BA, whereas the yield with PTH-3CN remained unchanged (Fig. 2D). Control experiments confirmed that the presence of PS, the light source, air, and PTH-3CN were all essential for the formation of BA (Supplementary Table S5). Finally, under the optimized conditions, the catalyst loading was minimized to 0.05 mol%, and pure oxygen was replaced with merely ambient air, still furnishing 50% isolated yield of BA from 0.10 g of PS1 using a 405 nm LED (0.4 W/cm2). During workup, the chlorobenzene solvent can be easily extracted and in principle, reused. Additionally, formic acid (27% NMR yield) was identified as the other major product (Supplementary Fig. S4), while CO2 generation was estimated to account for 15% of the carbon in PS (Supplementary Fig. S7), indicating that overoxidation constitutes the main undesired pathway in this process.

Practical upcycling of PS and product functionalization

Other commercially available PS materials were explored, including those with different molecular weights (PS2 and PS3), high-impact PS (HIPS), syndiotactic PS (sPS), divinylbenzene-crosslinked PS (PS4), styrene-acrylonitrile (SAN), and acrylonitrile butadiene styrene (ABS) resins (Fig. 2E). The conversion to BA was generally efficient, except in the case of sPS, owing to its poor solubility. Intriguingly, SAN and ABS resins required higher catalyst loadings and longer reaction times, which we attributed to the susceptibility of their copolymerized units to chain scission under the standard conditions, potentially consuming the catalytic turnover (vide infra). To demonstrate the practical aspect of our protocol, we first tested various types of post-consumer PS samples (PS5), including lunch box, CD case, Styrofoam, etc., using 0.05 mol% PTH-3CN (Fig. 3A). The yogurt cup and coffee lid required 0.5 mol% PTH-3CN to achieve comparable yields of BA, likely due to the presence of higher levels of impurities, such as ethylene vinyl alcohol (EVOH). Still, the method proved effective across a wide range of real-life PS wastes, indicating its tolerance for varying amounts of additives and impurities. Additionally, we scaled up the upcycling of Styrofoam to over 10 g with only 17.5 mg of catalyst, delivering over 4.0 g of BA in both batch and flow reactors (Fig. 3B).

Fig. 3: Practical demonstrations of the PS upcycling process presented in this study.
figure 3

A The scope of post-consumer PS wastes. a0.50 mol% PTH-3CN was used. B 10-gram scale upcycling of Styrofoam utilizing a batch or flow setup. C Tandem functionalizations of crude benzoic acid derived from PS. bEpiandrosterone (1.0 equiv), BA (1.2 equiv), Diethyl diazenedicarboxylate (1.5 equiv), PPh3 (1.2 equiv), THF, rt, 24 h. c(L)-Phenylalanine ethyl ester·HCl (1.2 equiv), BA (1.0 equiv), PyBOP (1.1 equiv), DIPEA (3.5 equiv), DCM, rt, 12 h. dBA (1.0 equiv), o-phenylenediamine (1.0 equiv), polyphosphoric acid, 180 °C, 2 h. eBA (1.0 equiv), Benzoquinone (1.0 equiv), Pd(OAc)2 (10 mol%), KOAc (2.0 equiv), DMA, 115 °C, 15 h. fBA (1.0 equiv), 4-chlorobenzotrifluoride (5.0 equiv), nBuAd2P (10 mol%), Pd(OAc)2 (5.0 mol%), Cs2CO3 (2.2 equiv), 3Å MS, DMF, 145 °C, 24 h. Additional details in Supplementary Information.

Full size image

Besides its primary industrial use in phenol production, BA is also widely used as a food preservative, as a precursor for the synthesis of plasticizers, and as a versatile starting point for further functionalization. To highlight this versatility, we further demonstrated that the crude BA from PS upcycling could be directly used in subsequent post-functionalizations (Fig. 3C). For instance, bioactive molecules like epiandrosterone and phenylalanine ethyl ester underwent Mitsunobu and amidation reactions to afford benzoate 1a and benzamide 1b, respectively. Condensation with o-phenylenediamine produced benzoimidazole 1c. Moreover, because of the low catalyst loading, palladium-catalyzed C–H functionalizations were also compatible: ortho-hydroxylation and -arylations led to the formation of salicylic acid (1 d) and diaryl benzoic acid 1e, respectively.

Mechanistic investigations

To shed light on the proposed conPET mechanism, UV-vis absorption spectroelectrochemistry and EPR analysis were used to characterize the radical cation PTH-3CN•+ (Supplementary Figs. S13 and S26), determining the reduction potential of its excited state to be 2.73 V vs. NHE—0.42 V higher than that of [PTH•+]*44. This potential is more than sufficient to oxidize PS and even chlorobenzene [with Eonset = 2.31 V (Supplementary Fig. S8) and Eox = 2.46 V45 vs NHE, respectively]. Indeed, chlorobenzene underwent direct C–H amination when PTH-3CN was employed (Supplementary Information Section II.4)43. Light on-off experiments (Supplementary Fig. S16) further suggested that a radical-chain mechanism is deemed unlikely. An induction period of ca. 4 h for BA formation was observed in the full reaction profile (Supplementary Fig. S18), implying that the upcycling process likely involves an initial photocatalyst activation phase. Additionally, the effect of light intensity on the reaction rate was also evaluated using a PS model substrate, revealing a non-linear relationship (Supplementary Fig. S19).

We further monitored the reaction progress using in situ UV spectroscopy (Fig. 4A). Surprisingly, the signal corresponding to PTH-3CN rapidly disappeared within 30 min and was not regenerated thereafter. Subsequent 1H NMR and HPLC analyzes revealed the stepwise formation of two major byproducts: sulfoxide SO1, followed by sulfone SO2 (Fig. 4B and Supplementary Figs. S20–S24)46. However, neither pristine SO1 nor SO2 displayed catalytic reactivity similar to that of PTH-3CN (Supplementary Table S8). To discern the actual catalytic species, we first irradiated 0.05 mol% PTH-3CN alone, without PS, under standard conditions for 8 h; by this point, both SO1 and SO2 had nearly disappeared. HRMS analysis of the reaction mixture unveiled multiple major components, which we tentatively assigned to triarylamine derivatives (Fig. 4B and Supplementary Fig. S25)47. We speculated that these triarylamines might serve as the active photocatalyst. Supporting this, when 0.10 g of PS1 was then added in situ and irradiated for an additional 12 h, a 42% yield of BA was still obtained. To further validate this hypothesis, we synthesized a truncated model compound TCPA, and its dimeric analog TCPBA, with the latter capable of absorbing visible light (Fig. 4C). Indeed, photo-upcycling of PS1 with 500 ppm TCPA and TCPBA yielded BA at 43% and 46%, respectively. Furthermore, spectroelectrochemical analysis revealed that upon irradiating TCPA•+ and TCPBA•+ in the presence of PhCl, the peaks corresponding to the neutral forms were fully restored (Fig. 4D). This result implies that the excited states of TCPA•+ and TCPBA•+ possess sufficient oxidizing power to oxidize PhCl.

Fig. 4: Mechanistic insights into the photocatalytic upcycling process.
figure 4

A Time-lapsed UV-vis data of the PS photodegradation in O₂ over 12 h. B Identification of the active catalysts in the degradation process. C Photocatalytic upcycling of PS using triarylamine model catalysts TCPA and TCPBA. D Spectroelectrochemistry of TCPA●+ and TCPBA●+ and the spectral changes after irradiations in the presence of chlorobenzene. E The revised photocatalytic cycle. F DFT-computed energy profile of the HAT processes.

Full size image

As such, we propose that PTH-3CN acts as a precatalyst to the triarylamines, which presumably operate within the conPET framework as depicted in Fig. 4E. As previously noted, the large excess of PhCl likely establishes SET with PhCl as the primary quenching pathway for excited [PC•+]*48. In addition to SET with PS to generate PS•+, the radical cation of PhCl can enter other quenching pathways, giving rise to biaryls and phenols with concurrent generation of HO2• from oxygen. GC-MS analysis of the crude reaction mixture confirmed these byproducts (Supplementary Fig. S41). Meanwhile, the rapid dismutation of HO2• and O2•– produces hydrogen peroxide49,50, which, along with other peroxide species (Supplementary Fig. S29), can be reduced by PC* to yield hydroxyl and alkoxy radicals51. These oxygen-centered radicals, as evidenced by EPR spin-trapping experiments (Supplementary Fig. S28), can abstract the benzylic C–H in PS, further facilitating degradation52. The detected byproducts, chlorophenol and phenol, could also arise from reactions between hydroxyl radical and PhCl53. The process might also generate chlorine radical; however, its effect appears insignificant, since using benzene instead of PhCl led to only a minor decrease in yield (Supplementary Table S1). The remarkably low catalyst loading required for this upcycling process can therefore be attributed to the synergistic actions of SET and HAT, together driving the generation of PS• from PS.

Notably, EPR spin-trapping and UV-vis analysis (Supplementary Figs. S28 and S30) also indicated PTH-3CN acts as a photosensitizer for singlet oxygen (1O2), a species often cited as a major HAT reagent for polymer degradation23,28. However, using 1O2 quenchers could yield misleading results under such highly oxidative conditions, as azide, diphenylanthracene, and 1,4-diazabicyclo[2.2.2]octane (DABCO) are themselves prone to oxidation (Supplementary Table S9). To elucidate the favored HAT pathway, we performed DFT calculations on a model compound M1 at the PCM(MeCN)/DLPNO-CCSD(T)/cc-PVTZ//ωB97XD/6-311 G(d,p) level of theory. Our findings deviated from reported energy profiles28, which were based on a less rigorous theoretical approach (Fig. 4F). Specifically, the calculations showed that the HAT pathway involving 1O2 is considerably less favorable than that involving hydroxyl radical. To experimentally assess the role of 1O2, we conducted photodegradation of PS in the presence of various 1O2 photosensitizers, even at elevated loadings (2.0 mol%), none of which led to more than trace amounts of BA (Supplementary Table S10). Thus, it appears unlikely that HAT between PS and 1O2 is the primary pathway for generating the benzylic radical PS• under our conditions. Nevertheless, the higher reactivity of 1O2 compared to its triplet counterpart may contribute to reactions further downstream. Once PS• is generated, chain cleavage proceeds via oxygen trapping, followed by alkoxy radical formation and β-scission. A comprehensive energy profile for this sequence is available in Supplementary Figs. S90–S92.

Applications to diverse polymer types

Although PTH-3CN functions only as a precatalyst, we selected it over the model triarylamines TCPA and TCPBA for subsequent studies with other polymer substrates due to the lower synthetic accessibility and reduced upcycling yields associated with TCPA and TCPBA. For poly(4-tert-butylstyrene) (tBu-PS), we observed the formation of over-oxidized products—4-acetylbenzoic acid (2b) and terephthalic acid (2c)—alongside 4-tert-butylbenzoic acid (2a) (Fig. 5A). The product distribution was influenced by the amount of PTH-3CN used; higher catalyst loadings led to increased yields of 2b and 2c. This observation agrees with our DFT calculations, which suggest that even unactivated methyl C–H can readily undergo HAT with hydroxyl radical (Supplementary Fig. S93). Subsequently, we employed a tandem Friedel–Crafts and oxidation strategy to convert polystyrene into aromatic acids with higher value than benzoic acid54. Initially, PS2 was acetylated to give Ac-PS with 94% degree of functionalization. Compared to previously reported metal-catalyzed oxidation methods54, our photodegradation approach can proceed without requiring elevated temperatures and high pressures, while still delivering comparable product yields. Notably, the product selectivity towards either 4-acetylbenzoic acid (2b) or terephthalic acid (2c) can be readily controlled by adjusting the reaction conditions. This showcases the broad applicability of this photochemical upcycling strategy for accessing other valuable aromatic acids from polystyrene.

Fig. 5: Photodegradation of poly(4-tert-butylstyrene), poly(4-acetylstyrene), and PMMA.
figure 5

A Catalyst loading-dependent outcomes in the photocatalytic upcycling of poly(4-tert-butylstyrene). aYields were determined by HPLC analysis unless otherwise noted. B Tandem transformation of polystyrene to poly(4-acetylstyrene) to 4-acetylbenzoic acid or terephthalic acid. C Corresponding GPC traces for photodegradation of PMMA samples using 0.05 mol% PTH-3CN.

Full size image

Encouraged by these results, we shifted our focus to PMMA, a polymer that is particularly difficult to depolymerize via HAT-based protocols due to its lack of hydridic C–H bonds. Typically, activating groups need to be incorporated via end-group modifications or copolymerizations—strategies incompatible with most PMMA waste containing only MMA units55,56,57. Gratifyingly, both a test tube rack, PMMA1 (Mw = 696 kDa, Đ = 3.11) and a commercially available material, PMMA2 (Mw = 97.7 kDa, Đ = 1.69), were successfully degraded into oxidized oligomers (Mw ≈ 2 kDa) with only 500 ppm PTH-3CN (Fig. 5C), accompanied by minor formation of formic and acetic acids (Supplementary Information Section II.8). These findings instilled confidence in extending this protocol to other traditionally non-recyclable plastics.

We next evaluated the efficacy of PTH-3CN for upcycling other widely used commodity polymers (Fig. 6). Our focus was primarily on post-consumer materials, specifically targeting the top four most-produced polymers: PP, LDPE, PVC, and HDPE. Using PTH-3CN, samples including an HDPE bottle, PVC tubing, an LDPE bag, and a PP box were effectively deconstructed into formic acid (FA), acetic acid (AA). The product distribution was highly dependent on the polymer backbone structure (Supplementary Fig. S40); namely, PE and PVC primarily yielded FA, whereas PP predominantly produced AA. The compatibility with PVC is noteworthy, as the HCl generated during the process can poison many metal catalysts41, presenting a major challenge in chemical recycling58. The relatively modest yield observed for PVC was attributed to a lower catalyst concentration (0.5 vs. 2.0 mol%), as higher loadings led to overoxidation of FA in the presence of HCl. The upcycling process also proved effective for other vinyl polymers, such as PVAc, PMMA, and PMA, which afforded FA and AA as well. It should be noted that the chain scission rates for these vinyl polymers were significantly slower compared to PS (Supplementary Fig. S39), necessitating higher catalyst loadings and prolonged reaction times.

Fig. 6: Photocatalytic upcycling of other commodity plastics.
figure 6

aYields were determined by 1H NMR analysis and based on repeat unit unless otherwise noted. b2.0 mol% PTH-3CN, PhCl/MeCN (3/1, 6.0 mL), 72 h. c0.5 mol% PTH-3CN, PhCl/MeCN (1/1, 2.0 mL), 48 h. d2.0 wt% PTH-3CN, PhCl/MeCN (1/1, 4.0 mL), 48 h. e2.0 mol% PTH-3CN, PhCl/MeCN (1/1, 1.0 mL), 48 h. f2.0 mol% PTH-3CN, PhCl/MeCN (1/1, 2.0 mL), 48 h. g0.05 mol% PTH-3CN, PhCl/MeCN (1/1, 1.0 mL), 30 h. h5.0 mol% PTH-3CN, HFIP/PhCl/MeCN (4/1/1, 3.0 mL), 36 h. i2.0 mol% PTH-3CN, PhCl/MeCN (1/1, 4.0 mL), 48 h. j5.0 wt% PTH-3CN, PhCl/MeCN/HFIP (2/1/1, 4.0 mL), 48 h. See Supplementary Information for more details.

Full size image

Moreover, PET and PU—the 5th and 6th most-produced polymer types, respectively—were also investigated59,60. A single-use PET bottle was smoothly converted into terephthalic acid (2c) in 93% isolated yield using HFIP as a co-solvent19. Likewise, both recycled PU foam and PU tubing were successfully deconstructed, delivering primarily FA in good quantities despite their inherent insolubility61. In a parallel study, polycarbonate safety glasses (PC) were efficiently deconstructed into monomeric and oligomeric carbonates (Supplementary Fig. S85). After workup, hydroxybenzoic acid (3) and bisphenol A (BPA) were isolated in a 2:1 ratio with a combined yield of 69%. The former originated from a unique mode of chain scission through C–C bond cleavage (vide supra), contrasting with the typical C–O bond hydrolysis under acidic or basic conditions62. Additionally, thermosetting polymers with limited solubility in common organic solvents, such as latex gloves, rubber septa, and epoxy-fiber composites, were also effectively converted into FA using our protocol. We believe the upcycling of plastics into small-molecule carboxylic acids represents a complementary strategy to CO₂ reduction63,64,65, with both approaches contributing to the reduction of the carbon footprint.

Finally, we explored the robustness of our protocol in processing mixed post-consumer plastics, simulating unsorted waste materials commonly encountered in daily life (Fig. 7). Given that polystyrene requires only a minimal catalyst loading, we first explored a selective treatment of PS. In the presence of PP, LDPE, PET, and PVC, PS was selectively upcycled, producing 43% isolated yield of BA and 26% FA in 20 h with 0.05 mol% PTH-3CN, with all other real-life plastics remaining intact, except for the dissolution of PVC. We then expanded the protocol to other polymer combinations, increasing the catalyst loading to achieve global degradation. To ensure uniform dispersion, the substrates were pre-heated for dissolution and then cooled to form well-dispersed particles. Remarkably, universal upcycling was accomplished with mixtures of PP, LDPE, HDPE, PS, and PVAc, as well as PP, LDPE, HDPE, PS, and PU, yielding small-molecule carboxylic acids. The latter mixture includes five of the six most commonly discarded polymers by type3. The practical applicability was further exemplified by the gram-scale reaction shown in Fig. 7. These findings highlight the versatility and practicality of our system in managing a broad spectrum of polymers and tackling complex, mixed-waste scenarios.

Fig. 7: Application to mixed plastics scenarios.
figure 7

aYields were determined by 1H NMR analysis and based on repeat unit unless otherwise noted. b0.05 mol% PTH-3CN, PhCl/MeCN (1/1, 1.8 mL), 20 h. c1.0 mol% PTH-3CN, PhCl/MeCN (3/1, 6.0 mL), 48 h. d3.8 wt% PTH-3CN, PhCl/MeCN (3/1, 6.0 mL), 48 h. See Supplementary Information for more details.

Full size image

Discussion

In conclusion, we have developed a highly efficient and versatile organo-photocatalytic protocol for upcycling a broad range of commodity polymers, addressing a critical gap in sustainable plastic waste management. Utilizing PTH-3CN as the catalyst, this approach enables efficient transformation of polymers under mild, metal-free conditions requiring only visible light and ambient air. The method demonstrated robust performance across diverse post-consumer and mixed plastic waste, converting them into valuable small molecules such as benzoic acid, formic acid, and acetic acid. Importantly, the upcycling outcome of PS can be altered through tandem reactions, such as Friedel–Crafts functionalization, enabling access to more valuable derivatives like 4-acetylbenzoic acid and terephthalic acid. These features highlight the potential of this strategy not only for plastic deconstruction but also for future integration with downstream upgrading processes that transform these small molecules into higher-value materials or chemical building blocks. The low catalyst loadings and the modifiability of the organocatalyst also pave the way for potential adaptation to heterogeneous systems and applications in C–H functionalizations of polymers and complex molecules, avenues that are actively being explored in our laboratory. By providing a scalable, energy-efficient solution to convert plastic waste into high-value products, this strategy holds promise for advancing the transition to a circular economy.

Methods

General procedure for performing the photocatalytic upcycling of PS

To a test tube was added polystyrene (ca. 104 mg, 1.0 mmol styrene unit) and chlorobenzene (0.50 mL), and the mixture was sonicated for 20 min to reach homogeneity. A stock solution (1.0 mg/mL) of PTH-3CN in acetonitrile along with additional acetonitrile (0.50 mL overall) were added. The reaction tube was placed in the photochemical reactor for 20 h and irradiated with a 50 W 405 nm LED light, while kept sparging with a gentle flow of air (ca. 5 mL/min) through a Teflon tube.

The reaction mixture was quenched by addition of saturated NaHCO3 (ca. 15 mL) and stirred for 15 min, which was washed with EtOAc (3 × 10 mL). The aqueous layer was re-acidified by 1.0 M HCl and extracted with DCM (3 × 20 mL). The combined organic layer was dried over Na2SO4, filtered, and the filtrate was concentrated in vacuo to afford the crude product. Purification by running through a short plug (3.0 cm I.D. × 3.0 cm) of silica gel (50% EtOAc in PE to 100% EtOAc) gave benzoic acid.

General protocol for tandem transformations of upcycled BA

The oxidative depolymerizations of polystyrenes (Mw = 350000, Mn = 170000) were carried out under the optimized conditions. Crude benzoic acids were obtained by direct removal of the solvent and other volatile byproducts (e.g., formic acid) and used in subsequent tandem transformations without further purification. The yields of benzoic acid in the first step were assumed to be 50%.

General procedure for conducting the photocatalytic upcycling of polymers shown in Figs. 6 and 7

To a test tube was added the polymer sample and chlorobenzene. For reactions employed PP and/or PE, the tube was placed in an oil bath at 130 °C for 2 h to make the solution homogeneous, and the solution was cooled to form well-dispersed small particles. For other polymers, small pieces of the polymer were directly used without preheating. If more than 1 mg of PTH-3CN was required, it was weighed and directly added to the reaction, followed by addition of acetonitrile. Otherwise, a stock solution (1.0 mg/mL) of PTH-3CN in acetonitrile along with additional acetonitrile were added. The reaction tube was placed in the photochemical reactor and irradiated with a 50 W 405 nm LED light, while kept sparging with a gentle flow of air (ca. 5 mL/min) through a Teflon tube.