Introduction

To address the issue of “white pollution” caused by over 353 million tons of plastic waste released annually1, the concept of chemical upcycling emerges as a promising strategy2,3,4. In addition to conventional thermal processes such as pyrolysis and hydrogenolysis, photocatalytic processes have increasingly attracted academic attention for plastic conversion5. This is due to their relatively mild conditions and the possibility of utilizing renewable solar energy. Typically, polyesters such as poly(ethylene terephthalate) (PET) and polylactic acid (PLA) have shown potential as electron donors in the photocatalytic hydrogen evolution reaction6,7,8,9,10,11,12,13,14. However, this approach requires alkali pretreatment (typically 1–10 M solution of NaOH or KOH) and often results in mixtures of small molecules through non-selective photoreforming pathways. Alternatively, research has explored the utilization of carbon scaffolds in plastics for the photocatalytic production of valuable chemicals. Examples include the conversions of polyvinyl chloride into acetic acid15, polyethylene into formic acid16, PLA into pyruvic acid17, and polystyrene into aromatics18,19,20,21,22.

PLA, one of the most common plant-based plastics, has an annual global production rate of approximately 0.7 million tons and accounts for 31% of bioplastic production as of 202323. It is widely used in packaging, agriculture, medicine, and 3D printing. With the market projected to triple from 2022 to 2023, reaching $3.1 billion, the amount of end-of-life PLA is expected to increase in the plastic waste stream24. Fortunately, PLA can be easily sorted from other plastic waste with high purity using near-infrared sorting techniques25, making it suitable for chemical recycling into monomers or upcycling into value-added chemicals. Comprising lactic acid units with hydrolyzable ester bonds, PLA exhibits good recyclability in solvolysis processes. Common approaches include depolymerizing to lactic acid26,27,28,29,30,31, methyl lactate32,33, and lactide34 as precursors for PLA reproduction. Notably, Ma et al. reported the first instance of amination of PLA using Ru/TiO2 at 140 °C, yielding 77% alanine after 94 h35. Very recently, Liu et al.36 introduced a novel photocatalytic pathway to upcycle PLA into alanine over CoP/CdS catalysts, which can reach up to 2.4 mmol/g catalyst/h with high selectivity over 75% at 80 oC. These excellent works demonstrate the potential to convert plastic waste into nitrogen-containing chemicals, a valuable class of chemicals in pharmaceuticals, agrochemicals, materials, and other industries. While pathways to organonitrogens from biomass have been well-established37,38,39,40,41, those from plastics remain relatively limited42,43,44.

In a prior study, our research group discovered the exceptional photocatalytic activity of CdS nanosheets in converting biomass-derived lactic acid into alanine45. Under visible light irradiation, the catalyst exhibited a remarkable alanine production rate from lactic acid of 10.5 mmol/g catalyst/h. Afterward, several photocatalytic systems for alanine production have been introduced, including black phosphorus quantum dots anchored on sulfur-doped graphitic C3N4 hollow nanospheres46, flower-like CdS supported on Ti3C247, Ru single-atom loaded on CdS nanosheets48 and Mo-doped In2O349. Among these, CdS-based materials have emerged as particularly dominant due to their high redox activity under visible light. Given PLA’s ability to be hydrolyzed into lactic acid50, there exists a promising opportunity to incorporate PLA waste into our CdS-based photocatalytic system for alanine production. Anticipating that the photocatalytic synthesis of alanine from PLA may present greater challenges than from lactic acid due to its more complex pathway, we aimed to offset potential reductions in product yield by improving the efficiency of CdS-based catalysts. Several strategies exist for engineering high-performance CdS-based photocatalysts, including morphology control51, heterojunction construction52,53,54, element doping55,56, and defect engineering57,58,59,60. Surface defects, in particular, can modify the geometric and electronic structures of materials, potentially creating more active sites and enhancing charge separation and migration61,62. CdS-based materials with sulfur vacancies have demonstrated boosted photocatalytic activities in toluene oxidation57, H2O2 evolution58, CO2 reduction59, N2 reduction60, and water splitting63. Typically, CdS-based materials were fabricated by solvothermal reaction at high temperature, and anion vacancies were introduced by complex methods such as hydrogen heat treatment, plasma techniques, atomic layer deposition or ultrahigh vacuum62. Herein, we introduced a simple room temperature synthesis of sulfur vacancy-rich CdS nanocrystals (NCs). We further, for the first time, evaluated the catalytic activities of these CdS NCs in PLA conversion to alanine under mild conditions (Fig. 1). Encouragingly, we achieved a production rate of approximately 5 mmol/g catalyst/h at 70 °C, doubling the production rate reported in existing literature at a higher temperature of 80 °C36.

Fig. 1: Our approach to upcycle PLA to alanine.
Fig. 1: Our approach to upcycle PLA to alanine.The alternative text for this image may have been generated using AI.
Full size image

PLA is upcycled into alanine using sulfur vacancy-rich CdS NCs under mild conditions. Plastic waste and haystack icons were created by ©cgdeaw’s Images and ©Mojo, respectively, via Canva.com.

Results

Synthesis and characterization of sulfur vacancy-rich CdS

Since the ratio of the S precursor to the Cd precursor is crucial for the generation of sulfur vacancies in CdS, a series of CdS NCs were synthesized by dropwise adding different amounts of Na2S solution to Cd(OAc)2 solution at room temperature (Supplementary Table 1). The mixing process was completed within 1 h, followed by centrifugation, washing, and drying to yield CdS NCs as yellow powders. The obtained CdS samples were labeled as CdS-1, CdS-2, CdS-3, CdS-4, CdS-5, and CdS-6, corresponding to the molar ratio of Cd to S in the precursor at 5:1, 5:2, 5:3, 5:4, 5:5, and 5:6, respectively (Fig. 2a). From the TEM images (Fig. 2b and Supplementary Fig. 1), it is observed that CdS NCs are nanoclusters composed of nanoparticles with diameters ranging from 8 to 12 nm. Notably, CdS NCs synthesized with different precursor ratios exhibit similar morphologies with only slight variations in particle sizes. Specifically, CdS-6 shows a larger particle size compared to other samples. HRTEM image demonstrates clear and orderly lattice fringes with a lattice distance of 0.21 nm, corresponding to the interlayer spacing of cubic-structured CdS (c-CdS) (220) planes. In addition, crystal lattices with spacings of 0.33 and 0.13 nm were also observed, aligning well with (111) and (311) crystal planes, respectively (Fig. 2b). Further investigation using HAADF-STEM and BF-STEM highlighted a high level of crystallinity of the CdS NCs, with no evidence of amorphous layers observed (Supplementary Fig. 2). The XRD patterns (Fig. 2c) reveal three characteristic diffraction peaks at 26.5, 44.1 and 52.1°, corresponding to the (111), (220) and (311) planes of c-CdS (JCPDS No. 42-1411), respectively. Additionally, no other crystalline phases were observed in the XRD patterns. This indicates that the sample phase is relatively stable and remains unchanged with different precursor ratios. In addition, the sharper diffraction peaks of CdS-6 suggest an increased crystal size compared to the other samples64, aligning with the observation from the TEM images. The BET surface areas for CdS-1, CdS-2, CdS-3, CdS-4, CdS-5, and CdS-6 were 147.3, 144.1, 145.6, 146.0, 133.7, 113.1 m2/g, respectively (Supplementary Fig. 3 and Supplementary Table 2). For CdS-1 to CdS-4, as the sulfur content increases, the specific surface areas showed no significant differences. However, further increasing the amount of sulfur resulted in a decrease in specific surface area from CdS-5 to CdS-6, providing additional support for the observed larger particle size of CdS-6. Given that the introduction of sulfur vacancy could deform the cell structure and reduce the particle size in metallic sulfides61,65, it is possible that sulfur vacancies are present to some extent in CdS-1 through CdS-5.

Fig. 2: Synthesis and characterization of CdS NCs.
Fig. 2: Synthesis and characterization of CdS NCs.The alternative text for this image may have been generated using AI.
Full size image

a Schematic diagram of the synthesis process. b TEM and HRTEM (inset) images of CdS-3. c XRD patterns, d EPR spectra, and e Sulfur vacancy contents of different CdS samples.

We performed further characterizations to investigate the presence of sulfur vacancies among CdS NCs with varying Cd:S molar ratios. Electron paramagnetic resonance (EPR) spectra (Fig. 2d) were conducted to identify unpaired species in the catalyst, which can provide compelling evidence for the existence of sulfur vacancies59,60,66,67,68. The CdS-6 exhibits no obvious EPR signals. As the added S content decreased, a characteristic EPR signal of single-atom sulfur vacancies at g = 2.003 was detected59. As the ratio of S to Cd decreased, the intensity of the EPR signal gradually increased from CdS-5 to CdS-1, indicating the formation of more sulfur vacancies. Moreover, X-ray photoelectron spectroscopy (XPS) measurements were performed to investigate the elemental chemical states and the concentrations of surface sulfur vacancy (Supplementary Fig. 4). In the high-resolution Cd 3 d XPS spectrum of CdS-6, two major characteristic peaks at binding energies of 405.5 and 412.3 eV are assigned to the Cd 3d5/2 and Cd 3d3/2 signal, respectively, suggesting the Cd2+state of Cd ions58,60,69. Regarding the sulfur element, a doublet peak located at 161.9 and 163.1 eV, corresponding to S 2p3/2 and S 2p1/2, respectively, revealed the presence of S57,58,60,69. Compared to CdS-6 without sulfur vacancies, the sulfur peaks of CdS-5 to CdS-1 all shifted towards lower binding energies due to the existence of S vacancies58,70. This indicates that the increase in S vacancies markedly changes the chemical environment of Cd and S. The Cd/S ratio in each sample was subsequently calculated based on the signals from the Cd 3 d and S 2p regions (Supplementary Table 3) and then converted to the corresponding concentration of surface sulfur vacancies (Fig. 2e). The surface concentrations of the sulfur vacancy measured by XPS were 0, 10.7, 16.7, 20.0, 24.2, and 33.3% for CdS-6 to CdS-1, respectively. ICP-OES was employed to determine the bulk sulfur vacancy content in CdS, corresponding to 1, 12.3, 15.3, 23.1, 28.1, and 37.5% for CdS-6 to CdS-1, respectively. The continuous increase in sulfur vacancies concentration from CdS-6 to CdS-1 aligns with the EPR analysis. These collectively indicate that sulfur vacancies not only exist on the surface, but also disperse throughout the materials. As such, we successfully synthesized a series of CdS NCs containing different sulfur vacancy content for the subsequent photocatalytic conversion of PLA into alanine.

Photocatalytic amination of PLA to alanine

We investigated the photocatalytic activity of CdS-3 in the one-pot conversion of PLA into alanine. Under visible light irradiation at 50 °C for 4 h, the reaction produced alanine at a rate of 2.15 mmol/g/h, along with a significant amount of lactamide and ammonium lactate (Fig. 3a, entry 1 and Supplementary Fig. 5). It was previously reported that poly(3-hydroxybutyrate), a polyester of β-hydroxybutyric acid, can be converted into its monomers and corresponding amino acid through a simple thermal reaction due to its unique structure71. To confirm that amino acid synthesis involves a photocatalytic reaction rather than a thermal reaction, we conducted a series of control experiments. First, under a dark condition with no catalyst, only lactamide and ammonium lactate could be detected (Fig. 3a, entry 2). Even when the catalyst CdS-3 was added, alanine was not produced in the absence of light (Fig. 3a, entry 3). This highlights that the synthesis of amino acids requires the simultaneous participation of light and catalyst. Importantly, the comparison shows that light and catalyst have minimal effect on the yield of lactamide and lactic acid, indicating that the depolymerization of PLA into lactamide and ammonium lactate is a non-catalytical thermal reaction, while the subsequent synthesis of amino acids is a photocatalytic reaction.

Fig. 3: Product yields under different reaction conditions and reaction pathway.
Fig. 3: Product yields under different reaction conditions and reaction pathway.The alternative text for this image may have been generated using AI.
Full size image

a Product yield of amination reactions under different conditions. Reaction conditions for “As drawn” entry: 200 mg PLA, 10 mg CdS-3, 4 mL 25% NH3 solution, 16 mL deaerated water, 50 °C, 4 h, 1 bar N2, visible light. Other entries show variations in conditions compared to the “As drawn” entry. b Hydrolysis of lactamide in NH3 solution. Reaction conditions: 200 mg lactamide, 4 mL 25% NH3 solution, 16 mL deaerated water, 50 °C, 4 h, 1 bar N2. c Reaction pathway of PLA amination. d Alanine production over CdS-3 at different time (at 50 °C) and temperature (for 4 h). Reaction conditions: 200 mg PLA, 10 mg CdS-3, 4 mL 25% NH3 solution, 16 mL deaerated water, 1 bar N2, visible light. The plastic bottle icon was created by ©smartstartst via Canva.com.

Three possible pathways for converting PLA into lactamide and ammonium lactate were proposed: (1) PLA undergoes ammonolysis to form lactamide, followed by the partial hydrolysis of lactamide to form ammonium lactate; (2) PLA is initially hydrolyzed to ammonium lactate, and subsequently, part of ammonium lactate is further hydrolyzed to lactamide; (3) PLA is simultaneously depolymerized to lactamide and ammonium lactate. To verify pathway (1), we replaced PLA with an equivalent amount of lactamide, anticipating a comparable alanine yield (Fig. 3a, entry 4). However, the actual alanine yield was significantly lower than the expected value. At the same time, it was observed that lactamide can be partially converted into ammonium lactate, with the ratio of ammonium lactate/lactamide much lower than that in the PLA reaction solution. Therefore, at least a portion of the ammonium lactate in the PLA conversion comes from the direct hydrolysis of PLA. We then studied the hydrolysis of lactamide into ammonium lactate under identical reaction conditions, except without light and catalyst (Fig. 3b). The hydrolysis of lactamide to ammonium lactate proceeded relatively slowly, with conversion rates of 4, 8, 14, 17, and 22% at 0.5, 2, 4, 6, and 10 h, respectively. The ammonium lactate yield took up to 24 h to reach 38%, which further verified the direct production of ammonium lactate from PLA instead of from lactamide and fully ruled out the possibility of pathway 1. When lactic acid was used instead of PLA as the substrate, the yield of alanine was greatly improved (Fig. 3a, entry 5). Simultaneously, no lactamide was detected, confirming that the lactamide originates from the direct ammonolysis of PLA. Subsequently, we replaced NH3 with NaOH and maintained the same pH, keeping other conditions unchanged (Fig. 3a, entry 6). Not all PLA was converted, indicating that NH3 not only provides an alkaline environment to promote the hydrolysis of PLA but also participates in the ammonolysis of PLA. The above analysis excluded pathway (2) and supported path (3). Evidently, PLA undergoes simultaneous ammonolysis and hydrolysis in an ammonia solution to produce lactamide and ammonium lactate, respectively. At the same time, lactamide is partially converted into ammonium lactate, and under photocatalytic conditions, ammonium lactate is aminated into alanine (Fig. 3c).

In the next step, we investigated the reaction kinetics of the photocatalytic conversion of PLA into alanine (Fig. 3d and Supplementary Fig. 6a). The production rate of alanine exhibited a linear increase during the first 4 h (2.15 mmol/g/h, from 0 to 4 h), and slowed down in the following 6 h (0.57 mmol/g/h, from 4 to 10 h). The decrease in alanine yield after 4 h was attributed to the inhibitory effect of alanine on the conversion of lactate (Supplementary Fig. 7), primarily due to the coordination of alanine onto Cd sites72,73. There was a significant increase in alanine yield as temperature increased (Fig. 3d and Supplementary Fig. 6b). At 70 °C, the alanine yield reached 4.95 mmol/g/h, which is 2.3 times higher than that at 50 °C and two times higher than that over the reported CoP/CdS catalyst at 80 oC36. Since PLA has a glass transition temperature of around 60 oC74, it is plausible that temperatures above 60 oC have loosened its structure and facilitated the penetration of the NH3 solution during the depolymerization process. Remarkably, the reaction even occurred at almost room temperature (30 °C), although some PLA remained unconverted due to insufficient temperature and limited reaction time. This mild reaction condition (room temperature, visible light) holds a practical significance, as it avoids the harsh conditions (140 °C) previously reported in the thermo-catalytic conversion of PLA35. Since NH3 participates in both the depolymerization and amination steps, the effect of NH3 concentration was investigated (Supplementary Fig. 8). Increasing NH3 concentration shifted the depolymerization product toward lactamide rather than lactate due to the dominance of ammonolysis over hydrolysis (Supplementary Fig. 8a). However, the alanine yield did not show any significant change (Supplementary Fig. 8b), indicating that NH3 concentration is not the determining factor for the amination step within the concentration range of 2.5–10%.

The alanine production activity of various CdS catalysts with different sulfur vacancy content was evaluated (Fig. 4a). CdS-6 without sulfur vacancies exhibited no alanine production activity, while the others displayed detectable alanine signals in the reaction solution. This indicates the positive role of sulfur vacancies in promoting alanine production. The alanine production rate increased significantly as the sulfur vacancy content increased from CdS-5 to CdS-3 and reached its peak (2.15 mmol/g/h) at CdS-3. However, the alanine production rate decreased slightly with further increases in sulfur vacancies from CdS-3 to CdS-1. As mentioned earlier, the photocatalyst only catalyzes the amination step of ammonium lactate, so the alanine generation mainly reflects the CdS activity. To further verify this, we employ an excess amount of lactic acid instead of PLA as a substrate to test the activity among CdS catalysts (Fig. 4b). A similar volcano-shaped curve with CdS-3 at the peak was observed. This trend became more pronounced as the amount of substrate increased, supporting our observation of activity variation from CdS-6 to CdS-1. Importantly, the peak value of 29.2 mmol/g/h achieved with CdS-3 surpasses our previous records for alanine production rate from lactic acid (10.5 mmol/g/h) using CdS nanosheets under similar conditions45. This achievement has prompted us to explore possible explanations for the enhanced activity of the sulfur vacancy-rich CdS NCs.

Fig. 4: Catalytic activity of various CdS NCs.
Fig. 4: Catalytic activity of various CdS NCs.The alternative text for this image may have been generated using AI.
Full size image

Product yields with a PLA and b lactic acid as substrates. Reaction conditions: 200 mg PLA or 1.8 g lactic acid, 10 mg catalyst, 4 mL 25% NH3 solution, 16 mL deaerated water, 50 °C, 4 h, 1 bar N2, visible light.

Origin of the enhanced catalytic activity of sulfur vacancy-rich CdS

We investigated reasons for different catalytic activity across various CdS NCs, starting with a mechanism study of lactic acid to alanine reaction. A small amount of electron scavenger nitrobenzene added (0.02 M) can inhibit the generation of alanine, indicating that electrons are indispensable (Supplementary Fig. 9a). In contrast, a larger amount of the hole scavenger isopropanol was required (>1.0 M) to suppress the alanine production. This results from lactic acid itself also serving as a hole scavenger, so there is a competition effect45. Therefore, this reaction makes full use of photogenerated electrons and holes. The radical scavenger DMPO also completely hindered the amination of lactic acid, suggesting that radical intermediates were produced during the reaction. With DMPO as spin capture agent, the oxygen-centered radicals DMPO-OCR (aN = 1.53 mT, aH = 1.72 mT) and its ring-opening signal (aN = 1.5 mT) were detected under light conditions over CdS-3 by ESR (Supplementary Fig. 9c). This oxygen-centered radical originates from the dissociation of the α-hydroxy group of lactic acid over CdS, serving as an intermediate in the pathway to alanine45. Meanwhile, the ESR signals of CdS-1 and CdS-6 were weakened, which is consistent with the photocatalytic activity. Under dark conditions, there is no ESR signal. Therefore, its reaction mechanism should be the same as the catalytic mechanism of CdS nanosheets previously reported by our group45. Specifically, lactic acid is oxidized to pyruvic acid by the photogenerated holes. Subsequently, pyruvic acid reacts with NH3 to form imine, which is finally reduced to alanine with the photogenerated electrons. Hydrogen evolution is the main side reaction (Supplementary Fig. 9b). It was previously reported that hydrogen production directly affects alanine production activity45. However, when we performed hydrogen production over the CdS NCs, there was no obvious correlation between catalytic activity in lactic acid amination and H2 production (Fig. 5f).

Fig. 5: Characterization and explanation of activity differences among different CdS NCs.
Fig. 5: Characterization and explanation of activity differences among different CdS NCs.The alternative text for this image may have been generated using AI.
Full size image

a UV-Vis, b photocurrent, c EIS, d time-resolved TAS, e photoluminescence spectra of CdS-1 to CdS-6. f H2 evolution over CdS NCs. Reaction conditions: 20 mmol lactic acid, 10 mg CdS, 18 mL deaerated water, 50 oC, 4 h, 1 bar N2, visible light. g FT-IR of lactic acid and the CdS NCs mixed with lactic acid. h Adsorption energy of lactic acid on non-defective and sulfur-vacancy sites of the CdS (110) surface. i Comparison of structure, properties, and alanine yield among CdS NCs.

Generally speaking, morphology and crystal phase are important factors affecting photocatalytic activity45,54,75. As mentioned earlier, different CdS NCs in this work have the same cluster morphology and cubic phase, thus excluding their influence on the photocatalytic activity. In addition, BET analysis shows that with the increase of sulfur vacancies, the specific surface area only increases less than 30% from CdS-6 to CdS-4 and then remains unchanged from CdS-4 to CdS-1 (Supplementary Fig. 3 and Supplementary Table 2). Therefore, the difference in catalytic activity caused by the specific surface area can also be ignored.

We hypothesized that the primary factor leading to the differences in activity among the CdS NCs is the sulfur vacancy, which can affect light absorption, charge transfer, charge separation, or surface reactions76. To compare the light harvesting among different CdS NCs, UV-vis spectroscopy was conducted (Fig. 5a). All samples exhibited visible light absorption with wavelengths shorter than 700 nm. Compared to CdS-6, the absorption edges of the other CdS NCs exhibit a blue shift with the introduction of sulfur vacancies, which is detrimental to light harvesting due to the narrower absorption spectrum ranges. In addition, Tauc plots showed that CdS NCs have band gaps in the range of 2.22–2.28 eV with no clear correlation to alanine yield (Supplementary Fig. 10). Therefore, we believe that light absorption and band gap may not be determining factors for alanine production.

To investigate the charge transfer efficiency in the CdS NCs, transient photocurrent was measured through light on-off runs (Fig. 5b). A larger photocurrent response indicates better transportation of photogenerated carriers, which enhances photocatalytic activity. Nonetheless, the photocurrent response exhibits an opposite trend to the sulfur vacancy content. This aligns well with the EIS spectra, where an increase in sulfur vacancy content leads to higher charge resistivity, as evidenced by larger semicircle diameters in the Nyquist plots (Fig. 5c). This suggests that sulfur vacancies are not conducive to charge transfer because they act as traps for photogenerated carriers62.

It has been reported that sulfur vacancies can enhance the separation of photogenerated charges77. However, time-resolved TAS spectra showed a longer average decay time for CdS-6 (Fig. 5d, Supplementary Fig. 11, and Supplementary Table 4), suggesting that sulfur vacancies act as trapping sites for charge carriers78. In addition, CdS-1 to CdS-5 showed higher peaks at 530 nm in the photoluminescence spectra (Fig. 5e)79, which originated from the recombination of electron–hole pairs. This confirmed the role of sulfur vacancies as recombination centers rather than as enhancers of charge separation.

In addition to side reactions and the physical and optical properties of the catalyst, the interaction between the substrate and the catalyst at the active site is an important factor59,64. Herein, FT-IR was conducted to reveal the binding of lactic acid on the surface of CdS NCs (Fig. 5g and Supplementary Fig. 12). Generally, lactic acid bonds to the Cd atoms on the surface of CdS NCs through carboxyl groups80,81,82. Due to its weaker bonding ability, CdS-bound carboxylates engage in an equilibrium exchange with free carboxylic acid80,82. This dynamic exchange allows the catalytic reaction to proceed efficiently by preventing the permanent occupation of catalyst-active sites by a layer of substrate. The peak at 1716 cm−1 is attributed to the asymmetrical stretching vibration νas(OCO) of lactic acid80,81, which shows a slight red-shift due to environmental changes when in full contact with CdS. Importantly, a new characteristic peak appeared at 1573 cm−1, corresponding to the symmetrical stretch νs(OCO)80,81. The energy difference between the asymmetric and symmetric stretch bands (Δν = νas(OCO) − νs(OCO) = 143 cm−1) suggests that lactic acid bonds to the CdS surface via a bridging mode81. The peak at 1716 cm−1 resulting from free lactic acid gradually decreases, while the characteristic absorption at 1573 cm−1 attributed to bonded lactic acid gradually increases with sulfur vacancies. This binding mode is consistent regardless of whether light or dark conditions are applied (Supplementary Fig. 13). This demonstrates that CdS with abundant S vacancies can provide more catalytically active sites, as more Cd atoms are exposed on the surface. DFT calculations were conducted to compare the adsorption energy of lactic acid on non-defective sites and sulfur vacancies (Fig. 5h). The (110) surface was chosen for the calculations since it has been reported to be the only exposed facet in the equilibrium morphology of cubic CdS83. The more negative adsorption energy value of lactic acid on two Cd atoms via a bridging mode on a vacancy (−0.84 eV) compared to that on a non-defective site (−0.69 eV) further supports our hypothesis.

We believe that the availability of more adsorption sites is the main reason why the catalytic activity of CdS increases with the increase in sulfur vacancies. However, as sulfur vacancies are further increased, the catalytic activity of CdS-2 and CdS-1 decreases due to a decrease in the charge transfer rate (Fig. 5i). Taken together, this explains the volcano-shaped trend of alanine yield as sulfur vacancy content increases (Supplementary Fig. 14). In addition, a similar trend was observed with pyruvic acid as the substrate, further supporting this argument (Supplementary Fig. 15). This enhancement mechanism rather than the common explanations related to light absorption or charge transportation may come from the presence of excess sulfur vacancies both in the bulk and on the surface. On one hand, the key role lies in surface vacancies that act as adsorption sites for lactic acid. On the other hand, more than 10% vacancy in the bulk seems to hinder the charge transportation and separation. This insight can guide future studies in designing effective sulfide photocatalysts with vacancies concentrated on the surface for specific transformations.

Demonstration in converting real PLA products

To investigate the stability of CdS-3 for photocatalytic alanine production, recycling tests were performed (Fig. 6a). The CdS-3 maintained 88% catalytic activity after five cycles, with its morphology remaining unchanged as a cluster (Supplementary Fig. 16), indicating its robust stability under the photocatalytic reaction conditions. Furthermore, we employed the defective CdS in the conversion of a commercial PLA cup (Fig. 6b). Since commercial PLA products may contain additives or have a more compact structure than pure PLA, the temperature was elevated to 70 °C to offset any reduced rate. Starting with the PLA flake cut from the cup, we observed an alanine production rate of 1.23 mmol/g catalyst/h. Hypothesizing that reducing the particle size could enhance the depolymerization rate, we ground the PLA cups into powder (90–180 µm) before the reaction. This led to an increased yield of alanine of 2.35 mmol/g/catalyst/h. Notably, this number is nearly seven times higher than that obtained from lignocellulosic biomass-derived glucose (0.34 mmol/g/h) under photocatalytic reaction conditions45.

Fig. 6: Recyclability, real-life plastic conversion, and process diagram of the CdS photocatalytic system.
Fig. 6: Recyclability, real-life plastic conversion, and process diagram of the CdS photocatalytic system.The alternative text for this image may have been generated using AI.
Full size image

a Alanine yield over CdS-3 in five recycling cycles. b Upcycling of a commercial PLA cup into alanine. Reaction conditions: 200 mg PLA, 10 mg CdS-3, 4 mL aqueous NH3 25% solution, 16 mL deaerated water, 50 °C (for a) or 70 oC (for b), 4 h, 1 bar N2, visible light. c Conceptual process diagram for the production of alanine from PLA. The plastic waste icon was created by ©cgdeaw’s Images via Canva.com.

To further expand the substrate scope, we investigated the amination of other polyesters to produce the corresponding amino acids in our photocatalytic system (Supplementary Table 5). Poly(3-hydroxybutyric acid) remained almost solid after the reaction, with a very low monomer yield (<5%). The resulting product, β-aminobutyric acid, was detected at a formation rate of 0.4 mmol/g/h. In contrast, polyglycolic acid exhibited 60% conversion with a 27% yield of glycolic acid, but the formation rate of glycine was only 0.1 mmol/g/h. This is likely due to the high Cα-H bond energy of glycolic acid, which hinders the amination step45. No amino acid product was formed when polycaprolactone was employed. These results demonstrate the potential of our catalytic system for tackling different polyesters, although further optimization is required.

Finally, we propose a conceptual process for upcycling PLA waste into alanine, which includes a photoreactor, a membrane distillation unit, and a crystallizer (Fig. 6c). In our previous study, membrane distillation was shown to facilitate the recycling of ammonia and concentrate the reaction mixture at temperatures of 40–46 °C, offering energy-saving benefits compared to conventional distillation84. Since the reaction mixture consists of only three components, alanine can be separated by crystallization, while lactamide and lactate are recycled back to the photoreactor. Future studies aim to optimize reactor configuration and separation conditions to maximize PLA conversion into alanine in this proposed system.

Discussion

We successfully synthesized a series of CdS NCs catalysts rich in sulfur vacancies for the photocatalytic transformation of PLA into alanine. This one-pot pathway involves the ammonolysis and hydrolysis of PLA to ammonium lactate, followed by CdS-catalyzed amination of ammonium lactate to yield alanine. Interestingly, the CdS NCs with an intermediate degree of sulfur vacancy defects demonstrated the highest photocatalytic performance in the conversion of PLA into alanine. This volcano-shaped trend resulted from the increase in binding sites with increasing sulfur vacancy content, and the decrease in charge transfer as the sulfur vacancy content further increased. The sulfur vacancy-rich CdS NCs maintained a stable alanine production rate of approximately 2 mmol/g catalyst/h over several catalytic cycles, which could be boosted up to around 5 mmol/g catalyst/h at 70 oC. In addition, our photocatalytic system demonstrated potential for the amination of other polyesters. The salient features of our approach include: (i) facile synthesis procedure of the catalysts at room temperature, (ii) conversion of plastic waste driven by visible light possibly at near room temperature, and (iii) high activity, easy separation and stability of the catalyst over multiple cycles. This work represents a promising strategy for vacancy defect engineering of common photocatalysts for the amination of waste polymers.

Methods

Materials

Cadmium acetate dihydrate (98%) and lactic acid solution (85%) were purchased from Sigma-Aldrich. Sodium sulfide (90%) was purchased from Shanghai Macklin Biochemical. Polylactic acid (Mw ∼60,000), polyglycolide, poly(3-hydroxybutyric acid), and polycaprolactone (Mw ∼50,000) was purchased from BLDpharm. Aqueous ammonia solution (25%) was purchased from VWR. Lactamide (98%) was purchased from TCI.

Catalyst preparation

Typically, Cd(OAc)2·2H2O was dissolved in water in a 100 mL beaker with magnetic stirring to prepare a 15 mL solution of 0.67 M. Subsequently, a certain amount of freshly prepared 0.67 M Na2S solution was steadily introduced into the beaker using a syringe pump while stirring continuously. The drip time was maintained for 60 min, and the reaction was terminated upon completion of the drip. The resulting yellow samples were collected through centrifugation at 2375×g for 3 min, followed by three washes with water and a final wash with ethanol. The washed samples were then dried in a 60 °C oven overnight. The CdS samples obtained were labeled as CdS-1, CdS-2, CdS-3, CdS-4, CdS-5, and CdS-6, respectively, denoting the molar ratio of Cd to S (5:1, 5:2, 5:3, 5:4, 5:5, and 5:6) in the initial input. The amount and speed of dropping the sulfur precursors are shown in Supplementary Table 1.

Catalytic activity evaluation

The photocatalytic reaction was conducted in a 100 mL stainless-steel autoclave equipped with a magnetic stirrer and top irradiation. Illumination was provided by a 300-W Xe lamp (PLSSXE300/300UV, Perfect Light) fitted with a UV cutoff filter (420–780 nm). Typically, 200 mg of PLA and 10 mg of CdS were dispersed in 16 mL water and bubbled with nitrogen for 1 h to remove oxygen. Subsequently, 4 mL of a 25% NH3 solution was introduced. The autoclave was sealed and purged with nitrogen ten times. The autoclave was then placed in a steel holder and heated to 50 °C, with temperature controlled by a hotplate and a water-cooling system connected to the steel holder’s jacket. Upon reaching the desired temperature of 50 °C, the photocatalytic reaction was initiated by activating the light source. After 4 h, the liquid products were filtered through a PES membrane with a 0.45 μm pore size and analyzed using 1H nuclear magnetic resonance (NMR) spectroscopy. The NMR analysis was carried out on a Bruker Ascend™ 400 (400 MHz) instrument, with the sodium salt of 2,2-dimethyl-2-silapentane-5-sulfonic acid (DSS) as the internal standard and deuterium oxide as the solvent. For the experiment with real-life plastic, a commercial PLA cup was cut into flakes of 0.5 × 0.5 cm and followed the same procedure, with the reaction temperature being elevated to 70 °C.

For recycling experiments, the CdS NCs were collected after the photocatalytic reaction by centrifugation and ultrasonically washed with ethanol and deionized water several times to fully remove the reactants adsorbed on the surface. The washed catalyst was freeze-dried and then directly used for the next cycle.

Characterization

Transmission electron microscopy (TEM) images were recorded on a JEOL JEM2100F with 200 kV acceleration voltage. High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) and bright-field scanning transmission electron microscopy (BF-STEM) were conducted on an aberration-corrected JEOL ARM 200CF operating at 200 kV. The crystal structures of the samples were determined by X-ray powder diffractometer (XRD, Bruker D8 ENDEAVOR) with Cu Kα radiation (λ = 0.1540 nm). The chemical states and sulfur vacancy concentrations of the samples were detected by ESCALAB 250Xi X-ray photoelectron spectrometry (XPS). The ASAP 2020 Brunauer–Emmett–Teller (BET) analyzer was employed for surface area analysis. Element molar ratios were determined by an inductively coupled plasma optical emission spectrometer (ICP-OES) on a Thermo Scientific iCAP 6000 series ICP spectrometer. Ultraviolet-visible (UV−vis) diffuse reflectance spectra were acquired with a Varian-Cary 5000 spectrophotometer. Photoluminescence spectra were recorded using a Tecan Infinite 200 Pro with an excited wavelength at 450 nm with CdS dispersed in ultrapure water (1 g/mL) as samples.

Transient absorption spectroscopy (TAS) was conducted on a HELIOS femtosecond transient absorption spectrometer (Ultrafast Systems, LLC). The 400 nm pump pulses were generated by frequency doubling of the 800 nm fundamental output from the regenerative amplifier (Libra, 1 kHz, 50 fs) using a β-barium borate crystal. The probe pulse used was a visible white light continuum (420–850 nm) generated by focusing the 800 nm fundamental output from the regenerative amplifier into a 2 mm sapphire crystal. The probe white light beam was passed through a 750 nm short-pass filter to eliminate any residual 800 nm fundamental components to prevent any strong secondary photoexcitation of the sample. The time-dependent absorption data extracted as an average at 510–520 nm was used for curve fitting with a three-phase exponential decay function:

$$y={A}_{1}{e}^{\frac{-x}{{\tau }_{1}}}+{A}_{2}{e}^{\frac{-x}{{\tau }_{2}}}+{A}_{3}{e}^{\frac{-x}{{\tau }_{3}}}+{y}_{0}$$
(1)

Where the component lifetimes, τ1, τ2, and τ3, corresponded to electron trapping, electron–trapped hole recombination, and electron–hole recombination at the valence band edge, respectively78.

The average decay time for each sample was calculated as follows:

$${\tau }_{{{{\rm{avg}}}}}=\frac{{A}_{1}{\tau }_{1}+{A}_{2}{\tau }_{2}+{A}_{3}{\tau }_{3}}{{A}_{1}+{A}_{2}+{A}_{3}}$$
(2)

Samples for electron spin-resonance (ESR) spectroscopy were analyzed at room temperature using a Bruker-A300 spectrometer with 5,5-dimethyl-l-pyrroline-N-oxide (DMPO) solution (50 mM) as the paramagnetic species spin-trap agent. For ESR measurements, 20 mmol of lactic acid was added into 16 mL of ultrapure water, 4 mL of ammonia and 120 μL of DMPO solution. Subsequently, 10 mg of CdS NCs were dispersed in the above mixture solution (200 μL) by ultrasonic treatment. The sealed glass capillary containing the above suspension was then placed in a glass tube. This glass tube was inserted into the ESR cavity and irradiated under a 300 W Xenon lamp to record the spectra at selected times.

The photocurrent measurements of the photocatalysts were performed on a CHI 760E electrochemical station with a three-electrode system. Specifically, a Pt wire served as the counter electrode, while an Ag/AgCl electrode acted as the reference electrode. About 10 mg of CdS was dispersed in 500 μL of ethanol containing 50 uL of Nafion by sonication to achieve a uniform suspension. Subsequently, the suspension was dropped onto the carbon paper several times to form a film. After drying at room temperature, the working electrode was obtained. The electrolyte used was a 0.2 M Na2SO4 and 1 M lactic acid solution. The photocurrent test was carried out under the irradiation of a 300 W Xenon lamp with a 420 nm UV cutoff filter.

The electrochemical impedance spectroscopy was conducted on a Gamry Interface 1010E Potentiostat with a Pt plate as the counter electrode and an Ag/AgCl electrode as the reference electrode. For catalyst ink preparation, 3 mg of CdS was dispersed in 450 μL of ethanol and 50 uL of Nafion by sonication to achieve a uniform suspension. The working electrode preparation, electrolyte and irradiation used were similar to those of photocurrent measurement. Fourier-transform infrared (FT-IR) spectra were recorded on a Bruker VERTEX 70 with the range of 400–4000 cm−1. Briefly, 10 mg of CdS NCs were mixed with 750 uL of lactic acid solution (20 uL lactic acid in a mixed solution of 0.1 mL methanol and 4.9 mL dichloromethane) and sonicated for 30 min. The mixture was then vacuum-dried to obtain the powder for testing. A 300 W Xenon lamp with a 420 nm UV cutoff filter was used for tests with irradiation.

For the H2 production experiment, 20 mmol of lactic acid and 10 mg of CdS were dispersed in 18 mL of water and bubbled with nitrogen for 1 h to remove oxygen. The mixture was heated to 50 oC and then irradiated by a 300 W Xenon lamp equipped with a 420 nm filter as the visible light source. An online gas chromatograph (GC D7900P, TCD detector, N2 carrier, 5 Å molecular sieve column, Shanghai Fechcomp) was used to quantify the H2 production amount for 4 h.

Computational methods

The structural data of cubic CdS (a = b = c = 5.89 Å) was obtained from the website of Materials Project85 before crystal optimization. A 2 × 2 CdS (110) with five Cd-S layers was constructed as slab models to represent the non-vacancy CdS(110) model. The vacancy CdS(001) was constructed by removing an S atom on adsorbed Cd−S−Cd sites where lactic acid would interact with the CdS(110) surface. The bottom three layers were fixed to simulate the bulk properties, while the remaining layers and molecules were allowed to relax. The vacuum thickness of all slab models was 15 Å. The structure of the lattice acid molecule in the vacuum were calculated using a 15 × 15 × 15 Å unit cell.

All periodic density functional theory (DFT) calculations were carried out using the Vienna ab initio simulation package (VASP)86,87,88,89 with the Perdew−Burke−Ernzerhof (PBE) exchange-correlation functional90. A plane wave basis set with an energy cutoff of 400 eV was used. Brillouin zone integration was performed using Γ-only k-point mesh and Gaussian smearing of 0.05 eV. Grimme’s dispersion correction was employed during each geometry optimization step91. All structures were refined until the Hellman-Feynman forces on each ion were lower than 0.05 eV/Å.

The adsorption energies (ΔEads) of lactic acid on the surface was calculated as follows:

$${\Delta E}_{{{{\rm{ads}}}}}={E}_{{{{\rm{slab}}}}+{{{\rm{La}}}}}-{E}_{{{{\rm{sla}}}}{{{\rm{b}}}}}-{E}_{{{{\rm{La}}}}}$$
(3)

Where Eslab+i, Eslab, and ELa are DFT-calculated energies of the adsorption complex, the clean slab, and the gas-phase adsorbate lattice acid molecule, respectively.