Photoreforming of Nonrecyclable Plastic Waste over a Carbon Nitride/Nickel Phosphide Catalyst

With over 8 billion tons of plastic produced since 1950, polymers represent one of the most widely used—and most widely discarded—materials. Ambient-temperature photoreforming offers a simple and low-energy means for transforming plastic waste into fuel and bulk chemicals but has previously only been reported using precious-metal- or Cd-based photocatalysts. Here, an inexpensive and nontoxic carbon nitride/nickel phosphide (CNx|Ni2P) photocatalyst is utilized to successfully reform poly(ethylene terephthalate) (PET) and poly(lactic acid) (PLA) to clean H2 fuel and a variety of organic chemicals under alkaline aqueous conditions. Ni2P synthesized on cyanamide-functionalized carbon nitride is shown to promote efficient charge separation and catalysis, with a photostability of at least 5 days. The real-world applicability of photoreforming is further verified by generating H2 and organics from a selection of nonrecyclable waste—including microplastics (polyester microfibers) and food-contaminated plastic—and upscaling the system from 2 to 120 mL while maintaining its efficiency for plastic conversion.


INTRODUCTION
The majority (86%) of plastic packaging accumulates in landfills or escapes into the environment. 1−4 Plastic pollution represents not only a global environmental crisis but also a loss of valuable resources. Most polymers are synthesized from fossil fuels, and it is predicted that 3.5 billion barrels of oil ($176 billion) could be saved each year if all global plastic waste were recycled. 4 However, the implementation of widespread recycling is limited by suboptimal waste management, lack of awareness, and the diverse range of chemistries, complexities, and sizes of polymer products. 1,2 Small polymer pieces known as microplastics (defined as ≤5 mm) are particularly problematic for recycling. 5−7 Microplastics are present in a variety of products and are also formed when plastic degrades over time. 6 Their small size and dilution make collection and reuse challenging, which contributes to their ubiquity in oceans, 5−7 drinking water, and salt 8−10 around the world. Even among plastics that can be reused, recycling has its limitations. Many polymers are downcycled into lower quality products. Only 7% of recycled poly(ethylene terephthalate) (PET) bottles, for example, are recast as bottles. 1 Existing management structures are currently incapable of sustainably and economically processing the vast variety of plastic waste, and new technologies for transforming end-of-use polymers into valuable products are urgently required.
One such technology is photoreforming (PR), in which sunlight and a photocatalyst generate H 2 from an organic substrate and water. The substrate acts as an electron donor and is oxidized by the excited photocatalyst to other organic molecules. The photogenerated electrons are then transferred from the photocatalyst to a cocatalyst and reduce water to H 2 ( Figure 1). H 2 is a particularly valuable product given its high demand for agricultural, pharmaceutical, chemical, and renewable energy applications. 11,12 Unlike existing H 2 production technologies like steam reforming of fossil fuels 13 or thermalbased approaches for converting plastic into oil, 14 PR can be conducted at ambient temperature and pressure, uses sunlight as its only energy input, and produces fuel-cell-grade H 2 . 15 While photocatalytic degradation of plastics (typically to microplastics and CO 2 ) has been researched for several decades, 16−18 PR offers a novel approach by not only mitigating plastic waste but also generating valuable chemical products. The thermodynamics of the overall PR process are also nearly energy neutral: 19 PR of ethylene glycol at room temperature requires ΔG°= 9.2 kJ mol −1 (E°c ell = −0.01 V, see the Supporting Information for details).
Although PR of simple molecules and biomass has been researched extensively, 19−21 plastic substrates have been largely overlooked. The same characteristics that make polymer recycling difficultcomplex structures, low water solubility, and poor biodegradabilityalso make PR more challenging. As a result, there are only two previous studies on PR of plastics: one employed an expensive and UV-absorbing TiO 2 | Pt photocatalyst, 22 whereas the other used toxic CdS/CdO x quantum dots. 23 We propose cyanamide-functionalized carbon nitride (CN x ) coupled with a nickel phosphide (Ni 2 P) H 2 evolution cocatalyst as a noble-metal-and Cd-free alternative for PR of plastic waste ( Figure 1). Carbon nitride is a nontoxic and inexpensive polymeric photocatalyst, 24−26 and the introduction of cyanamide defects improves its photocatalytic efficiency. 27 CN x features a band gap of 2.7 eV that allows for visible light absorption and band edges (conduction band −0.5 V vs NHE, valence band +2.2 V vs NHE at pH 6) 28 suitable for the PR reactions. CN x has also been used for PR of biomass with various cocatalysts under a wide pH range. 29 With key characteristics including visible-light absorption, alkaline stability, low cost, and nontoxicity, CN x is a competitive alternative to both CdS/CdO x and TiO 2 |Pt for polymer PR. Ni 2 P has previously been utilized with unfunctionalized carbon nitride ( H 2 N CN x ) and a soluble sacrificial electron donor (triethanolamine) for H 2 evolution 30−32 and has potential for plastic PR given its alkaline compatibility and relatively high H 2 evolution activity. 33 Here, we demonstrate that the CN x |Ni 2 P photocatalyst can be employed under alkaline conditions to produce H 2 and organic chemicals from PET and poly(lactic acid) (PLA, a biodegradable but not typically recycled alternative to PET). We further apply this system to real-world, nonrecyclable waste, including polyester microfibers and oil-contaminated PET, and show that it can be upscaled from 2 to 120 mL without efficiency losses. This proof-of-concept demonstration of noble-metal-free, Cd-free, and visible-light-driven plastic PR with CN x |Ni 2 P offers a sustainable and scalable route toward simultaneous plastic waste elimination and renewable fuel and chemical synthesis.

SYNTHESIS AND CHARACTERIZATION OF THE
PHOTOCATALYST CN x was prepared from melamine at 550°C, 34 followed by postsynthetic functionalization with potassium thiocyanate 27 according to slightly modified literature procedures. CN x |Ni 2 P was produced by adapting a literature synthesis: 30  , Table S1). This same trend is evident in the X-ray photoelectron spectroscopy (XPS) quantification results (Table S2) and can be attributed primarily to residual PO x from the cocatalyst synthesis that adheres to the surface of CN x ( Figure S1 and Table S1) as well as to PO x surface species on Ni 2 P ( Figure S2 and Table S2). 30,35 Diffuse-reflectance UV−vis spectroscopy shows that the CN x |Ni 2 P composite retains its visible light absorption (λ < 460 nm, Figure 2a). The increased baseline at λ > 460 nm can be attributed to scattering from Ni 2 P. The fluorescence emission of CN x is quenched upon Ni 2 P addition, which could suggest reduced charge recombination due to enhanced electron transfer to the cocatalyst (Figure 2b). 30,32 This effect stems from the close contact between CN x and Ni 2 P in the annealed photocatalyst, as Ni 2 P powder mixed with CN x does not display quenching. Fourier transform infrared (FTIR) spectroscopy confirms that the bulk properties of CN x  characterized by vibrations at 804 cm −1 (heptazine core), 1221 and 1311 cm −1 (secondary amine −C−N bending), and 2177 cm −1 (CN stretch)are unaffected by Ni 2 P (Figure 2c). 27 Similarly, powder X-ray diffraction (XRD) patterns show only a minor shift in the CN x lattice spacing from 3.25 to 3.30 Å upon Ni 2 P addition, which is likely caused by elemental doping of the CN x structure 36,37 ( Figure S3a). Although the cocatalyst peaks cannot be observed in XRD analysis due to the low loading (2 wt %), measurements of Ni 2 P alone confirm that hexagonal Ni 2 P has been successfully synthesized ( Figure S3b).
Transmission electron microscopy (TEM, Figure 2d), scanning electron microscopy (SEM, Figure S4), and energydispersive X-ray spectroscopy (EDX, Figure S4) show that Ni 2 P is uniformly distributed across CN x . As determined from TEM, the Ni 2 P nanoparticles are 9.4 ± 0.6 nm in diameter ( Figure 2d and Figure S5), which is slightly smaller than Ni 2 P synthesized alone (12.8 ± 1.1 nm, Figure S6). The nanoparticles also exhibit a lattice spacing of 0.22 nm, which corresponds to the (111) plane of hexagonal Ni 2 P. 30 XPS further verifies the binding of Ni 2 P to CN x (Figure 2e,f, Figure S2, and Table S2). The high-resolution C 1s (Figure 2e) and N 1s ( Figure S2a) spectra of CN x and CN x |Ni 2 P are nearly identical, confirming that the surface properties of CN x are largely unaffected by cocatalyst addition. Although the Ni 2p ( Figure 2f) and P 2p ( Figure S2b) edges of CN x |Ni 2 P are low in intensity due to the small quantity of cocatalyst (2 wt %), they still reveal similar spectra to those of bulk Ni 2 P. The Ni 2p 3/2 edge of both Ni 2 P and CN x |Ni 2 P can be deconvoluted into two peaks: Ni−P and NiO x (from surface oxidation). 30 Ni−P in particular shifts to a lower binding energy (from 853.08 to 852.18 eV) on CN x . The same trend is observed in the P 2p spectra. These results, combined with the slight shift in the CN peak to higher binding energies (from 288.18 to 288.28 eV in C 1s and from 398.58 to 398.78 eV in N 1s ), suggest a metal−support interaction in which electron density shifts from CN x to Ni 2 P. 27,38 This interaction should improve electron extraction and thereby enhance PR efficiency.

PHOTOCATALYSIS
Having established the synthesis protocol and characterization of CN x |Ni 2 P, we subsequently studied its photocatalytic performance. When compared to literature reports for H 2 N CN x |Ni 2 P under the same conditions, CN x |Ni 2 P offers comparable H 2 yields with triethanolamine as a sacrificial electron donor (Table S3). We therefore applied CN x |Ni 2 P to polymer PR. All conditionsincluding Ni 2 P loading, photocatalyst concentration, pH, and substrate treatmentwere optimized for maximal total H 2 production (Figure 3a,b and Table S4).
In a typical optimized experiment, the substrate was pretreated (24 h at 40°C with stirring in the dark) in aqueous KOH to initiate polymer breakdown and improve PR performance (Table S5). 23 Quantitative 1 H nuclear magnetic resonance (NMR) spectroscopy of the polymers shows that 72% of PLA is solubilized to lactate during pretreatment, whereas 62% of the ethylene glycol in PET is released (Table  S6 and Figure S7). Terephthalate and lactate can be detected by liquid chromatography−mass spectrometry (LC-MS analysis up to 1000 m/z) of the pretreated solutions of PET and PLA, respectively ( Figure S8). Only a few longer chain molecules are observed, suggesting that the polymers hydrolyze primarily to their monomers. CN x |Ni 2 P was then ultrasonicated in H 2 O for 10 min following a reported procedure. 29 This ultrasonication process is known to increase the surface area and activity of the photocatalyst (Table S7). 29 The photocatalyst and pretreated substrate mixture were added to a photoreactor and exposed to simulated solar light (AM 1.5G, 100 mW cm −2 ) at 25°C under a N 2 atmosphere. All H 2 evolution values are background-corrected by yield without substrates, which accounts for ∼6% of total H 2 yield and may be at least partially due to residual P precursor from the cocatalyst synthesis (Table S8). No H 2 is detected without the photocatalyst, light absorber, cocatalyst, or light (Table S8). Mass spectrometry of the headspace gas confirms that no CO 2 is released ( Figure S9a), and CO 3 2− is only produced from certain substrates (PLA) during PR, as determined by 13 C NMR spectroscopy ( Figure S10). Isotopic labeling experiments with D 2 O verify that H 2 originates from water rather than the substrate ( Figure S9b).
A Ni 2 P loading of 2 wt % was optimal ( Figure 3a and Table  S4) as too little cocatalyst is available for electron extraction at lower loadings, whereas parasitic light absorption prevents further improvement at higher loadings. 30−32 A variety of other noble-metal-free and alkaline-compatible cocatalystsincluding Ni, Fe, and Co salts, Ni(OH) 2 , Ni, Fe, and Cu oxides, and Fe x Pwere also tested with CN x for polymer PR (Table S9). All showed inferior performance, with the second-best cocatalyst (Ni(OH) 2 ) offering a H 2 yield half that of Ni 2 P.
Harsh conditions (e.g., high pH) are often required to solubilize plastic, and polymer PR with CN x |Ni 2 P improves and Table S4). However, this enhanced H 2 yield is likely not exclusively due to improved substrate solubility, as CN x |Pt with PET performs equally well at different molarities (black circles in Figure 3b). Instead, previous studies suggest that Ni 2 P forms a thin Ni(OH) 2 layer under alkaline conditions, which is thought to improve H 2 evolution activity via enhanced water dissociation. 30,39−41 To reduce the cost and corrosiveness of the system, all following experiments were conducted in 1 M KOH.
A variety of common polymerspolyethylene, PET, PLA, polypropylene, polystyrene, polyurethane, and polystyreneblock-polybutadiene (rubber)were photoreformed under these optimized conditions (Table S10). While all polymers produced small quantities of H 2 , PET and PLA offered the highest yields and were selected for further study. Both PET and PLA are polar polymers and contain esters, which facilitates hydrolysis in alkaline aqueous media and could account for their superior performance.
After 50 h of irradiation, 82.5 ± 7.3 and 178 ± 12 μmol H 2 g sub −1 were produced from PET and PLA, respectively ( Figure  3c and Table S11). The values correspond to turnover numbers of 7.8 ± 0.7 and 16.8 ± 1.1 mol H 2 mol Ni −1 for the respective polymers as well as external quantum yields at λ = 430 nm of 0.035 ± 0.005% for PET and 0.101 ± 0.018% for PLA (Table S12). H 2 conversionsdefined as the moles of H 2 detected divided by the theoretical H 2 yieldof 4.4 ± 0.6% and 1.6 ± 0.2% were achieved after 8 days of PR with PET and PLA, respectively (Table S13). Note that these calculations assume that only the aliphatic portion of PET is oxidized during PR, as is consistent with previous reports. 23 The system was still active after 8 days, suggesting that higher H 2 conversions could be achieved at longer time scales (for example, H 2 conversions of 50% are observed after 18 days of ethylene glycol PR, Figure S11). At higher pH (10 M KOH), H 2 conversions increase to 24.5 ± 3.3% for PET and 6.7 ± 0.8% for PLA (Table S13). Values reported for CdS/CdO x under the same conditions were 16.6 ± 1.0% for PET and 38.8 ± 4.0% for PLA. 23 CN x |Ni 2 P also maintains 17% of its efficiency under visible-light-only irradiation (λ > 420 nm, Table S8), showcasing its improved absorption range over both TiO 2 (0% retention) and H 2 N CN x (2−9% retention, Figure S12 and Table S14). The H 2 yield from PET over CN x |Ni 2 P is 4 times lower than that with CdS/CdO x (Table 1 and Table S15). CN x |Ni 2 P requires an electron transfer process from the light absorber to cocatalyst, which likely limits its photocatalytic efficiency in comparison to CdS/CdO x . For further comparison, CN x |Pt, H 2 N CN x |Ni 2 P, and TiO 2 |Ni 2 P were prepared and studied under identical conditions (Table 1 and Table S14). As expected, the benchmark Pt catalyst improves H 2 yield by 3 times. After 20 h of PR, H 2 N CN x |Ni 2 P offers H 2 yields similar to (or in some cases higher than) CN x |Ni 2 P. This indicates the applicability of Ni 2 P to a range of carbon nitrides. However, the activity of H 2 N CN x |Ni 2 P decreases over time, whereas that of CN x |Ni 2 P remains constant ( Figure S13 and Table S14). Future investigations will determine the reasons behind this variation, but CN x was selected for the current work due to its apparent high long-term stability and superior utilization of visible light. Finally, TiO 2 |Ni 2 P has lower yields at 1 M KOH and only absorbs ultraviolet light (no H 2 observed at λ > 420 nm, Table  S14). PR of lactic acid (the monomer of PLA) is also faster on h −1 ). 42 CN x |Ni 2 P is thus characterized by unique benefits including low expense, visible light absorption, and long-term stability, all of which are critical parameters for scalable PR of plastics.

POSTCATALYSIS CHARACTERIZATION
Characterization of the CN x |Ni 2 P catalyst after PR of PLA by TEM ( Figure S14), SEM/EDX ( Figure S4), and XPS ( Figure  S15 and  Figure S15). The absorption of the photocatalyst increases in the visible range after use ( Figure  S14), potentially due to agglomeration ( Figure S4), but emission remains constant before and after PR ( Figure S14). FTIR spectroscopy shows that the NCN functionality of CN x remains intact ( Figure S14), and XPS confirms that the surface properties of CN x are unchanged after catalysis (no shifts in the C 1s and N 1s edges detected, Figure S15). XPS also shows that the Ni−P and NiO x species in the Ni 2p spectrum are replaced by Ni(OH) 2 at 855.98 eV ( Figure S15), which is consistent with the literature 30,39,40 and results mentioned in the previous section. Because of its insolubility, the catalyst can be easily separated from solution by centrifugation and reused while maintaining half of its original photocatalytic activity (Table S17). This loss of activity can likely be attributed to agglomeration during the isolation/drying process or minor chemical changes in the CN x structure, rather than to an intrinsic photocatalytic instability. When CN x |Ni 2 P is left in

Journal of the American Chemical Society
Article solution, it remains active over 18 days of ethylene glycol PR until all substrate is consumed ( Figure S11).

SUBSTRATE OXIDATION
Ideally, PR would generate useful products not only during the reduction half-reaction (H 2 ) but also through polymer oxidation. Although CO 2 is the expected final oxidation product, gaseous CO 2 was not detected and CO 3 2− was only evident after PR of PLA ( Figures S9 and S10). We therefore used 1 H NMR spectroscopy to analyze the reaction mixtures and identify organic oxidation products. All peak assignments were verified by adding authentic samples ( Figure S16), and maleic acid was used as a standard.
The 1 H NMR spectra show that both PET and PLA form a variety of oxidation products after 5 days of PR ( Figure 4, Figure S10, and Table 2). During pretreatment under alkaline conditions, PET hydrolyzes to its monomersterephthalate (TPA, b) and ethylene glycol (EG, c, also see Figures S7 and S8)or soluble oligomer fragments. TPA is not oxidized during PR (no H 2 produced, Table S16, and no oxidation products observed, Figure S17), as aromatic moieties are challenging to photoreform. 23 Because of its insolubility under most aqueous conditions, TPA could potentially be isolated and reused for PET synthesis. Isophthalate (a), an isomer of TPA, is present in small quantities in the original polymer and is not an oxidation product.
The EG portion of PET is oxidized by CN x |Ni 2 P to formate (i), glyoxal (ii), glycolate (iii), acetate (iv), and other intermediates (unlabeled). Glyoxylate and glycoaldehyde cannot be unambiguously detected in the 1 H NMR spectrum due to their overlap with the D 2 O and EG peaks, respectively, but become visible in the 13 C NMR spectrum ( Figure S10a, labeled as v and vi). Signals marked with an asterisk ( * ) are already present in a control spectrum of CN x |Ni 2 P irradiated without a polymer substrate and can be attributed to residual solvents from the photocatalyst synthesis or other impurities. PR of EG alone offers high H 2 evolution activities (46 ± 6 μmol H 2 g cat −1 h −1 , Table S16) and generates the same array of oxidation products ( Figure S17). With the exception of acetate, all of these oxidation products can be formed according to the proposed mechanism in Figure  S18a. In brief, ethylene glycol is expected to oxidize to glycoaldehyde, followed by glyoxal and glycolate, glyoxylate, oxalate, formate, and finally CO 3 2− . PR of the oxidation products followed by 1 H NMR analysis at various time intervals ( Figure S17, Tables S16 and S18) also supports the proposed series of reactions. However, the use of highly alkaline conditions during PR initiates a variety of other reactions, such as aldol condensation. These processes add further complexity and likely explain the appearance of acetate and other unidentified products corresponding to the unlabeled peaks in the 1 H and 13 C NMR spectra. None of the more oxidized intermediatesformate, acetate, and CO 3 2− seem to accumulate over extended PR time scales ( Figure S11). This leads to a mass imbalance between the quantity of measurable oxidation products and H 2 , which can likely be attributed to the additional unidentified chemicals observed in 13 C NMR spectroscopy ( Figure S10a).
PLA offers a much simpler system. It hydrolyzes to lactate (d, e, also see Figures S7 and S8) during pretreatment, which is then oxidized primarily to CO 3 2− and small quantities of formate (i), acetate (iv), and other unidentified products (Figure 4b, Figures S10b and S18b). All of the above results are similar to those reported for PR of PET with CdS/CdO x 23 as well as for the oxidation of EG 43−45 and lactate 46,47 under a variety of conditions. The acidic products also slightly reduce the pH of the PR system, from 14.0 to 13.2 after 5 days.
Although the accumulation of organic products causes incomplete conversion of the plastic precursors to H 2 , it also  Maleic acid was used as a reference standard. Glycolate was not quantifiable as its peak overlaps with that of ethylene glycol (n.a. indicates not available).

Journal of the American Chemical Society
Article prevents the release of greenhouse gases like CO 2 and potentially allows for the extraction of additional chemicals. In the future, this system could be improved by tailoring the selectivity of the oxidation half-reaction toward a single highvalue product. For example, acetate had a global market of 13 million tons in 2015 and is widely used as a platform chemical. 48 The development of selective oxidation cocatalysts remains a key challenge for polymer PR.
PR is often believed to proceed via hydroxyl (OH • ) radicals. 19,49 To investigate this possibility, we performed PR with the OH • scavenger TPA. TPA reacts with OH • to form 2hydroxyterephthalic acid (TPA-OH), which fluoresces at λ = 430 nm when excited at λ = 315 nm. After 20 h of PR, no TPA-OH emission was detected ( Figure S19). This, combined with a previous proposal that holes photogenerated on CN x are not oxidizing enough to produce OH • , 29,50 suggests that OH • plays a minimal role in PR. Instead, PR likely proceeds via direct hole transfer between CN x and the substrate.

APPLICATION TO REAL-WORLD WASTE AND MICROPLASTICS
Finally, we evaluated the real-world efficacy of the CN x |Ni 2 P system through long-term PR of polyester microfibers and food-contaminated PET (Figure 5a). Polyester microfibers are known to shed from synthetic clothing and enter the environment and drinking water. 8,51 Despite their prevalence, their dilution and aquatic state make microplastics challenging to reuse. Another common recycling issue is food contamination, which congests equipment and reduces the quality of recovered plastic. 52 PR could therefore be an ideal vehicle for transforming these nonrecyclable plastic items into valuable products. The lowest possible microfiber loading was utilized (5 mg mL −1 ; no H 2 was detected at lower loadings after 20 h, Table  S19) to approach real-world conditions. Note that this concentration is still much higher than that seen in European tap water: 3.8 fibers L −1 . 8 In addition, a commercial PET bottle was ground into pieces ≤0.5 cm 2 and coated with soybean oil (25 mg mL −1 bottle, 5 mg mL −1 oil). All other parameters including catalyst concentration, pH, and pretreatmentwere kept consistent with PR of pure polymers.
After 5 days of illumination, yields of 104 ± 10, 22.0 ± 1.3, and 11.4 ± 1.2 μmol H 2 g sub −1 were achieved from microfibers, a PET bottle, and oil-contaminated PET, respectively ( Figure 5a and Table S20). The PR rate with microfibers increases over time as more surface area is exposed for hydrolysis; SEM shows that the sample develops cracks and pits after PR ( Figure S20). Note that the H 2 yield of microfibers is higher than that of the PET bottle because of the quantity of substrate utilized; the activities of the microfibers and PET bottle are equivalent (2.67 ± 0.25 and 2.87 ± 0.16 μmol H 2 g cat −1 h −1 , respectively). Oil limits access to the PET bottle, accounting for its lower performance in comparison to the bottle alone. These samples offer PR rates one-tenth that of pure PET, likely due to additional fillers and lower solubility (only 25% of the ethylene glycol in the microfibers was released after pretreatment in comparison to 62% from pure PET; Figure S7 and Table S6). 1 H NMR spectroscopy of the PR solutions after catalysis shows that the polyester microfibers oxidize primarily to glyoxal (2440 nmol after 24 h) and acetate (2100 nmol, Figure S21). The PET bottle, meanwhile, yields a wide range of oxidation products similar to those observed in pure PET ( Figure S21). This provides an initial demonstration of the transformation of real-world plastic waste into both H 2 and organics.

UPSCALING OF PHOTOREFORMING
Compatibility with upscaling is essential for any technology aimed at eliminating global plastic waste. Having demonstrated small-scale PR of nonrecyclable plastic waste, we therefore upscaled our setup from 2 to 120 mL (Figure 5b,c and Table  S20). With an irradiation area of 60 cm 2 and depth of 2 cm, the reactor was semioptimized for maximal light absorption. All concentrations (catalyst, substrate, and KOH) were kept constant. With this new setup, 53.5 μmol H 2 g sub −1 was generated from PR of microfibers over the course of 5 days.
In contrast to small-scale PR of microfibers, the upscaled H 2 production rate decreases gradually over time. This is likely due to inefficient stirring in the reactor rather than catalyst degradation (as we have shown that CN x |Ni 2 P is stable under PR conditions). When adjusted for area of irradiation, the upscaled H 2 production (0.53 μmol H 2 cm −2 ) is greater than that achieved at small scales (0.26 ± 0.03 μmol H 2 cm −2 ), which is a promising support for the scalability of PR. However, it should be noted that the quantity of H 2 generated is equivalent to 17.9 μW (0.00215 Wh), meaning that a 15 m 2 reactor would be required to charge a typical smartphone (5 Wh) 53 with the modest quantum yield currently offered by CN x |Ni 2 P.

Journal of the American Chemical Society
Article As these numbers suggest, polymer PR cannot currently compete with either established H 2 production technologies (steam reforming of fossil fuels, ∼80−90% conversion) 13 or gasification of plastic to H 2 (∼65−95% conversion and plastic consumption rates in the order of kg h −1 , depending on the specific technology). 54 To enhance the real-world applicability of plastic PR, future work must focus on key bottlenecks including catalyst efficacy, conversion rates and selectivity, substrate solubilization, reduction or reuse of KOH, and reactor design.

CONCLUSION
In this work, we have established a noble-metal-and Cd-free photocatalyst for visible-light-driven reforming of plastic waste. CN x |Ni 2 P functions due to the strong binding of the Ni 2 P cocatalyst to CN x , which promotes charge separation, catalytic efficiency, and stability. The CN x |Ni 2 P photoreforming system successfully generates H 2 by using PET and PLA as abundant and freely available waste feedstocks. The oxidation halfreaction is suggested to proceed via direct hole transfer from the photocatalyst to the substrate and yields valuable organic chemicals (e.g., acetate and formate) rather than CO 2 , thereby improving the sustainability and overall process value of the system. CN x |Ni 2 P can also reform real-world polymer samples, including polyester microfibers and oil-contaminated PET, at both small (2 mL) and larger (120 mL) scales. These results showcase a unique benefit of photoreforming: its applicability to waste materials that cannot otherwise be recycled or reused. Plastic is a valuable resource that contains stored energy and chemical feedstocks, yet much of its potential is lost to landfills and environmental pollution. Polymer photoreforming with CN x |Ni 2 P takes advantage of this underutilized resource to simultaneously reduce plastic pollution and generate H 2 and organics in an inexpensive, sustainable, and sunlight-driven process.
Synthesis of Carbon Nitride. Unfunctionalized carbon nitride ( H 2 N CN x ) was prepared by heating melamine to 550°C for 3 h under air according to a modified literature procedure. 34 The obtained powder was ground with a pestle and mortar. Cyanamide-functionalized carbon nitride (CN x ) was prepared by combining H 2 N CN x and KSCN (weight ratio 1:2) and heating first to 400°C for 1 h followed by 500°C for 30 min (ramp rate 30°C min −1 ) under Ar. 27 After cooling naturally, the powder was washed with H 2 O and dried under vacuum at 60°C.
Synthesis of Ni 2 P. NiCl 2 ·6H 2 O and NaH 2 PO 2 ·H 2 O (ratio of 1:5) in a minimum amount of water were first stirred for 1 h and then sonicated for 1 h. The mixture was dried under vacuum at 60°C. The dry solid was then heated for 1 h at 200°C under Ar (ramp rate 5°C min −1 ). After cooling to room temperature, the black powder was washed with water (2×) and ethanol (1×) and dried under vacuum at 60°C.
Synthesis of Ni 2 P with Light Absorber. Analogous to a previously reported procedure, 30 CN x , H 2 N CN x , or TiO 2 nanoparticles (300 mg) and NiCl 2 ·6H 2 O (20 mg for 2 wt %) were combined in a minimum of water (1 mL), stirred first for 1 h, and then sonicated for 1 h. NaH 2 PO 2 ·H 2 O (100 mg for 2 wt %) was subsequently added to the Ni mixture and again stirred and sonicated for 1 h each. The mixture was dried under vacuum at 60°C. The dry solid was heated for 1 h at 200°C under Ar (ramp rate 5°C min −1 ). After cooling to room temperature, the powder was washed with water (3×) and ethanol (3×) and dried under vacuum at 60°C.
Physical Characterization. Emission spectra (λ ex = 360 nm, λ em = 450 nm) were recorded on an Edinburgh Instruments FS5 spectrofluorometer equipped with a Xe lamp and integrating sphere. All samples were prepared at a concentration of 1.6 mg mL −1 in 1 M aqueous KOH in a quartz glass cuvette (1 cm path length). UV−vis spectra were recorded on a Varian Cary 50 UV−vis spectrophotometer using a diffuse reflectance accessory. Fourier transform infrared spectroscopy (FTIR) spectra were collected on a Thermo Scientific Nicolet iS50 FTIR spectrometer (ATR mode). Powder Xray diffraction (XRD) was conducted on a PANalytical Empyrean Series 2 instrument using Cu Kα irradiation. Scanning electron microscopy (SEM) and energy dispersive X-ray spectroscopy (EDX) were conducted on a TESCAN MIRA3 FEG-SEM. Samples were sputter-coated with a 10 nm layer of either Pt or Cr prior to microscopy. Transmission electron microscopy (TEM) was conducted on a Thermo Scientific (FEI) Talos F200X G2 TEM. All samples were dispersed in ethanol at low concentrations (∼4 μg mL −1 ) and drop-cast on carbon-coated Cu grids.
Samples for X-ray photoelectron spectroscopy (XPS) were dispersed in ethanol (concentration of 5 mg mL −1 ) and drop-cast (50 μL, 7×) onto clean FTO glass slides and dried. XPS was performed on a Thermo Fisher Scientific K-alpha + spectrometer. Samples were analyzed using a microfocused monochromatic Al X-ray source (72 W) over an area of ∼400 μm. Data were recorded at pass energies of 150 eV for survey scans and 40 eV for high-resolution scans with 1 and 0.1 eV step sizes, respectively. Charge neutralization of the sample was achieved through a combination of both low-energy electrons and argon ions. Three well-separated areas were selected on each sample for analysis to examine any surface heterogeneity. Data analysis was performed in CasaXPS using a Shirley type background and Scofield cross sections, with an energy dependence of −0.6.
Inductively coupled plasma optical emission spectrometry (ICP-OES) measurements were completed by the Microanalysis Service at the University of Cambridge (Department of Chemistry) on a Thermo Scientific iCAP 700 spectrometer. For quantification of the bulk Ni and P content of the catalyst before and after PR, the catalyst was digested in 2:1 H 2 O 2 :H 2 SO 4 overnight before measurement.
Nuclear Magnetic Resonance (NMR) Spectroscopy. 1 H NMR and 13 C NMR spectra were collected on either a 400 or 500 MHz Bruker spectrometer. All samples, including polymers before and after PR and pure oxidation intermediates and products, were prepared in 1 M NaOD in D 2 O with sample concentrations of 25 mg mL −1 .
where I analyte is the integral of the analyte peak, N analyte the number of protons corresponding to the analyte peak, M analyte the molar mass of

Journal of the American Chemical Society
Article the analyte, and m standard the known mass of the standard in the sample. pH Measurements. pH was measured on a Mettler Toledo pH meter and probe. Samples in 1 M KOH were diluted to 0.1 M KOH prior to measuring to avoid damaging the probe, and the pH was back-calculated by using the assumption that a[OH − ] ≈ c [KOH].
Substrate Pretreatment. Following a slightly modified reported procedure, 23 polymers (50 mg mL −1 ) were soaked in 2 M aqueous semiconductor-grade KOH in a sealed vial for 24 h at 40°C with stirring at 300 rpm. The solutionincluding the undissolved pieces of polymerwas then used for PR as below.
Light Chromatography−Mass Spectrometry (LC-MS). Ten microliters of pretreated solution was added to 1 mL of methanol and submitted for analysis to the mass spectrometry team at the University of Cambridge (Department of Chemistry) on a ThermoFinnigan Orbitrap setup combined with a Dionex Ultimate 3000 HPLC.
Photocatalytic Generation of H 2 . A dispersion of the catalyst (CN x |Ni 2 P, H 2 N CN x |Ni 2 P, or TiO 2 |Ni 2 P) in H 2 O (5 mg mL −1 ) was ultrasonicated as described previously (10 min, pulses of 30 s at 100% amplitude followed by 5 s pauses). 29 The resulting mixture (0.65 mL), 1 mL of pretreated polymer in 2 M aqueous semiconductorgrade KOH, and 0.35 mL of H 2 O were used per sample. Final conditions were 2 mL of 1 M aqueous KOH, 1.6 mg mL −1 catalyst, 25 mg mL −1 polymer, or PET bottle (5 mg mL −1 used for polyester microfibers). CN x |Pt was made by ultrasonicating CN x and then adding H 2 PtCl 6 as a precursor (Pt forms via in situ photodeposition). The prepared samples were added to Pyrex glass photoreactor vials (internal volume 7.91 mL) and capped with rubber septa. After briefly vortexing, the samples were purged with N 2 (containing 2% CH 4 for gas chromatographic analysis; no CH 4 was observed in the samples postillumination without the addition of this internal standard) at ambient pressure for 10 min. The samples were then irradiated by a solar light simulator (Newport Oriel, 100 mW cm −2 ) equipped with an air mass 1.5 global (AM 1.5G) filter and a water filter to remove infrared radiation. Visible-light-only experiments were conducted by adding a λ > 420 nm cutoff filter. All samples were stirred at 600 rpm and kept at a constant temperature of 25°C during irradiation. H 2 generation was monitored by periodically analyzing samples of the reactor head space gas (50 μL) by gas chromatography (see below). Overpressure within the vial is minimal (an increase of 0.03 atm per 10 μmol of H 2 produced).
Gas Analysis. The accumulation of H 2 was measured via gas chromatography on an Agilent 7890A gas chromatograph equipped with a thermal conductivity detector and HP-5 molecular sieve column using N 2 as the carrier gas. Methane (2% CH 4 in N 2 ) was used as an internal standard after calibration with different mixtures of known amounts of H 2 /N 2 /CH 4 . CO 2 detection was performed with mass spectrometry on a Hiden Analytical HPR-20 benchtop gas analysis system fitted with a custom-designed 8-way microflow capillary inlet to a HAL 101 RC electron impact quadrupolar mass spectrometer with a Faraday detector.
Upscaled Photocatalytic Generation of H 2 . CN x |Ni 2 P (170 mg) in H 2 O (15 mL) was ultrasonicated as described above and added to a photoreactor. The utilized photoreactor (internal volume 190 mL) is constructed from PEEK and stainless steel and features a quartz window (11 cm diameter). 60 mL of pretreated polyester microfibers (600 mg) in 2 M aqueous KOH and 45 mL of H 2 O were also added to the photoreactor. The final experimental conditions were the following: 120 mL of 1 M aqueous KOH, 1.42 mg mL −1 CN x |Ni 2 P, and 5 mg mL −1 microfibers. The reactor was capped with a rubber septum and purged with N 2 (containing 2% CH 4 for gas chromatographic analysis) for 30 min. The sample was then irradiated at room temperature by a solar light simulator (LOT-Quantum Design, 100 mW cm −2 ) equipped with an air mass 1.5 global filter (AM 1.5G).
Treatment of Data. All analytical measurements were performed in triplicate, unless otherwise stated, and are given as the unweighted mean ± standard deviation (σ). All measurements are listed as H 2 yield per weight of substrate (μmol H 2 g sub −1 ) and activity per weight of catalyst (μmol H 2 g cat −1 h −1 ). σ was calculated via eq 2: where n is the number of repeated measurements, x the value of a single measurement, and x ̅ the unweighted mean of the measurements. σ was increased to 5% of x ̅ in the event that the calculated σ was below this threshold. External Quantum Yield (EQY) Determination. Ultrasonicated CN x |Ni 2 P 2 wt % (3.2 mg), pretreated polymer (50 mg), and 1 M aqueous KOH (2 mL) were added to a quartz cuvette (path length 1 cm), which was then sealed with a rubber septum. The sample was purged with N 2 containing 2% CH 4 for 10 min. The sample was next activated via 4 h of illumination in a solar light simulator (Newport Oriel, 100 mW cm −2 ) equipped with an air mass 1.5 global filter (AM 1.5G) and a water filter to remove infrared radiation. After a second round of N 2 purging, the sample was irradiated by a Xe lamp (LOT LSH302) fitted with a monochromator (LOT MSH300) focused at a single wavelength of λ = 430 nm (accurate to a full width at halfmaximum of 5 nm). The light intensity was adjusted to ∼1000 μW cm −2 , as measured with a power meter (ILT 1400, International Light Technologies). The cuvette was irradiated across an area of 0.28 cm 2 . The evolved headspace gas was analyzed by gas chromatography and the EQY (%) calculated via eq 3: where n H 2 ,exp is the H 2 (mol) measured in experiment, n substrate,exp the substrate (mol) used in experiment, and n H 2 ,ideal n substrate,ideal −1 the ideal ratio of moles H 2 to substrate, as determined from eqs S2 and S4 in the Supporting Information.
Power Calculations. Equation 5 was used to calculate the power output from the H 2 produced.
where V H 2 is the molar volume of H 2 at 25°C (24.47 L mol −1 ), n H 2 the moles of H 2 produced, ρ H 2 the density of H 2 at 25°C (8.235 × 10 −5 kg L −1 ), u H 2 the lower heating value of H 2 (120 × 10 6 J kg −1 ), and t irr the irradiation time (s).

* S Supporting Information
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/jacs.9b06872. Chemical equations and thermodynamic calculations for the photoreforming reactions; tables containing all photocatalytic experiment data; additional X-ray photoelectron spectroscopy, powder X-ray diffraction, scanning electron microscopy, transmission electron microscopy, mass spectrometry, and nuclear magnetic reso-