The extensive application of antibiotics across diverse sectors, encompassing animal husbandry, agriculture as well as medical practice, has resulted in the appearance of antibiotic resistant genes as well as antibiotic resistant bacteria in natural environment [1]. Due to their harmful impacts on both human health and ecosystem, World Health Organization has categorized them as a significant public health threat in 21^st century [2]. Tetracycline (TC) is a broadly utilized antibiotic in both medical practice and animal farming. Its potent antibacterial properties, coupled with notable ecological toxicity, potential carcinogenic effects, and resistance to natural biodegradation, render even minute concentration of TC in the environment an obvious threat to human [3]. Consequently, there is an immediate need for an efficacious strategy to manage TC, thereby mitigating associated environmental risks. Peroxymonosulfate (PMS)-involved advanced oxidation processes (AOPs) have demonstrated notable efficacy for recalcitrant organic pollutants removal, attributed to the high oxidative potential of reactive oxygen species (ROS), rapid reaction kinetics, and their applicability in a broad pH range [4]. For instance, Hu et al. [5] successfully prepared a bubble network catalyst, Co@NCNTs-T (T = 500, 600, 700 ℃), through a simple one-step calcination method. These catalysts were constructed by encapsulating small cobalt nanoparticles within N-doped carbon nanotubes and sewing them into a bubble-like structure. This unique design significantly enhanced the inactivation resistance during the degradation of TC through PMS activation. Conventional metal-based catalytic materials are extensively used to activate PMS. Nevertheless, the majority of metal catalysts unavoidably suffer from the drawback of secondary pollution, stemming from the leaching of metal ions [6].

Up to now, carbon materials have demonstrated extensive accessibility, ecological compatibility, along with straightforward preparation techniques. Owing to their non-metallic composition, excellent stability, as well as abundance of active sites, they are regarded as promising green catalysts and have found widespread application in effluent treatment [6-8]. The catalytic performance of carbon materials is contingent upon their structural configuration, functional moieties, defects, as well as inherent properties [9]. Usually, the advanced oxidation degradation of pollutants by non-metallic carbon materials can be dominated by both nonradical and free radical pathways [10]. In nonradical way, the degradation of pollutants can mainly be achieved through electron transfer. It is proposed that defective regions within carbon materials are reactive sites for activating PMS, leading to the formation of reactive complexes for nonradical degradation. This robust interplay between defects and contaminants enables PMS to degrade target compounds efficiently through the extraction of electrons from adsorbed substances, thereby bypassing the formation of free radicals [11]. Wang et al. [12] prepared ionic liquids (IL) derived porous C material as PMS activators for the naproxen oxidation and degradation. They demonstrated that abundant defect sites on the porous carbon material N-PC1000 served as active sites facilitating direct electron transfer. These sites were capable of restricting delocalized π system with a "localized state", thus exhibiting distinctly different chemical as well as electronic characteristics. They also confirmed that the efficiency of organic matter removal increased as the defect level of porous carbon materials rose. On the other hand, in the radical pathway, the introduction of N (such as graphic N and pyrrolic N) or C—OH in the material may affect persulfate activation and produce different active species [13-15]. For example, the graphitic N enriched metal-free N-rGO prepared by Kang et al. [13] exhibited efficient capability in activating persulfate to generate reactive species, thereby facilitating sulfachloropyridazine removal, while pyrrolic N possessed relatively lower thermal stability and was prone to conversion to graphitic nitrogen. Moreover, graphitic N possesses greater electronegativity along with smaller atom radius, which facilitates electron transfer from adjacent carbon atoms, thereby enhancing charge density of neighboring carbon [16]. In such instances, radical way prevails over electron transfer approach. Owing to their chemical inertia, pristine carbon typically displays limited reactivity in activating PMS [17]. As a consequence, formulating more appropriate modification strategies to augment the catalytic degradation efficacy of non-metallic carbon materials continues to pose significant challenges.

Plastics, particularly those that are naturally non-degradable like polyethylene terephthalate (PET), polyethylene (PE), as well as polypropylene (PP), are widely utilized in the packaging, construction, along with textile sectors due to their value and malleability [18]. Thousands tons of waste plastics exacerbate new environmental challenges and represent a considerable depletion of resources [19]. In light of this, significant endeavors are undertaken to repurpose plastics and enhance their conversion to higher-value products. Plastics are carbon rich materials, so converting waste plastics into carbon materials including graphene, carbon nanosheets, and porous carbon is currently one of the research hotspots [20,21]. These carbon materials obtained from plastics exhibit versatility in their applications, encompassing supercapacitors, batteries or separation processes [18,22], as well as water purification techniques [23,24]. Currently, the primary methods for converting waste plastics into carbon involve traditional pyrolysis techniques. However, microwave treatment has demonstrated superiority over traditional pyrolysis in aspects such as energy transfer efficiency, enhanced yield of material, and its capability for instant and precise temporal control [25-27]. Zhou et al. [26] employed cobalt nitrate as a microwave adsorber to prepare carbon material catalyst S_0.3nullCo@P₂C, which exhibited obvious porous properties, through microwave pyrolysis of waste plastic PET. Within just 15 min, thoroughly carbamazepine removal was accomplished using S_0.3nullCo@P₂C. The research introduces an innovative method for waste plastics application via microwave carbonization. However, due to the composition of non-polar molecules in plastics, they basically do not absorb or absorb very little microwave radiation. So far, the use of microwave treatment for metal-free carbonization of waste plastics is still quite limited.

Thereby in this study, for the first time, common waste plastics PET and high-density polyethylene (HDPE) were selected as carbon sources to fabricate metal-free porous carbon materials through a simple microwave pyrolysis method, with sodium hydroxide as microwave absorber along with pore-forming agent and sodium lignosulfonate as carbonization auxiliary. The P₁S₂ material derived from PET exhibited characteristics indicative of defect enrichment along with the formation of C═O bonds. In contrast, H₁S₂, which was prepared through the carbonization of HDPE, possessed a significant quantity of C—OH groups and defects. Waste plastics derived P₁S₂ and H₁S₂ demonstrated remarkable efficiency in degrading TC, with degradation rate constants as high as 0.303 min^-1 and 0.235 min^-1. In addition, influence of catalyst dose, PMS amount, TC concentration, pH, temperature, and anions on TC removal were also checked. By employing correlation fitting, quenching experiments, and electron paramagnetic resonance (EPR) analysis, TC degradation was predominantly facilitated by electron transfer in P₁S₂/PMS system, where defects served as the primary active sites. In the meantime, removal of TC in the H₁S₂/PMS was predominantly facilitated by free radicals governed pathway, with C—OH serving as the primary active sites. The removal mechanism of TC was deduced by utilizing the Fukui function calculation, which pinpointed the susceptible sites for attack by active species, along with the identification of reaction intermediates through LC-MS.

The preparation method of carbon materials was demonstrated in Fig. 1a. Typically, 0.5 g waste PET powder, 1.0 g sodium lignosulfonate, and 0.5 g sodium hydroxide were well grinded in ceramic mortar. Mixture was placed in a 25 mL crucible and pyrolyzed at 800 W for 5 min (Table S1 was the selection criteria for preparation conditions). Due to the inability to directly carbonize waste plastic powder by microwave treatment in the presence of sodium hydroxide, biomass lignosulfonate sodium was used to assist in the waste plastics carbonization, while sodium hydroxide was used as both a microwave absorber and a pore-forming agent in the carbon materials formation. The resulting material was washed with H₂O/EtOH 5 times. After vacuum drying at 60 ℃ for 12 h, P₁S₂ was obtained. Keeping the weight amount of sodium hydroxide at 0.5 g and the total mass of PET powder and sodium lignosulfonate at 1.5 g, when the weight ratios of PET powder to sodium lignosulfonate were 1:4, 1:3, 1:2 and 1:1, the prepared materials were named as P₁S₄, P₁S₃, P₁S₂ and P₁S₁, respectively. When PET powder was replaced by HDPE powder, H₁S₂ was prepared under the same conditions as P₁S₂. C was prepared by microwave pyrolysis of 1.5 g sodium lignosulfonate, 0.5 g sodium hydroxide under 800 W conditions for 5 min without waste plastic. For comparison, P₁S₂–600 and H₁S₂–600 were also prepared by carbonizing at 600 ℃ for 2 h under N₂.

The appearance alongside pore features of carbon materials were examined utilizing scanning electron microscopy (SEM) and transmission electron microscopy (TEM) in Figs. 1b-g [28]. The non-metallic porous carbon catalysts P₁S₂ and H₁S₂ derived from waste plastics PET and HDPE exhibited significant porous structural characteristics. The formation of porous structures was not only attributed to the gases emitted during the pyrolysis of plastics and sodium lignosulfonate [29,30], which facilitated the pore formation, but also ascribed to the fact that NaOH, the activation reagent encapsulated in plastics-derived carbon material, was washed out to produce porosity. TEM element mapping (Figs. 1d and g) showed well distribution of C, O, and S in P₁S₂ and H₁S₂ [26].

Crystal structure of prepared carbon material was inspected using X-ray diffraction (XRD) (Fig. 2a). No peak of waste plastic was detected in all samples, indicating completely carbonization of plastics (Fig. S1 in Supporting information). All as-prepared catalysts exhibited two broad signals around 23.1°−24.0° and 44.1°, ascribing to (002) along with (100) planes of amorphous carbon and crystalline carbon [15]. In contrast to the C without adding plastic, the (002) diffraction peak of amorphous carbon widened and shifted from 23.12° to 24.0° after adding PET. The broadening and displacement of diffraction (002) peak suggested defects along with amorphous carbon structures in catalytic materials [31]. Moreover, compared with traditional pyrolysis method, the (002) diffraction peak in carbon materials prepared by microwave was significantly broadened, implying that microwave treatment was also helpful for the generation of defects and amorphous structure. Further replacing PET with waste plastic HDPE (weight ratio of 2:1 between plastic and sodium lignosulfonate), the carbon material produced via microwave treatment also exhibited characteristic of defects and amorphous carbon structures. In Raman spectroscopy, D band around 1350 cm^-1 as well as G band around 1580 cm^-1 represent disordered carbon along with graphitized carbon [32]. I_D/I_G is generally applied to quantify defects level in catalysts. With the increase of PET plastic content, the defects of C, P₁S₄, P₁S₃, and P₁S₂ gradually increased (Fig. 2b), and the I_D/I_G value of P₁S₂ reached 1.446. The formation of porous structure in P₁S₂ might be the reason for the increase in defect degree. As PET further increased, the I_D/I_G value of P₁S₁ reduced, which may be due to excessive PET being unfavorable for defect formation. In HDPE derived carbon materials, the I_D/I_G value was 1.008, and the defect degree was equivalent to the graphitization degree, indicating that there may be differences in the mechanism of P₁S₂ and H₁S₂ during PMS activation for pollutant degradation. Compared with the microwave carbonized materials P₁S₂ and H₁S₂, the decrease in I_D/I_G values of P₁S₂–600 and H₁S₂–600 indicated that microwave was more conducive to the formation of defect carbon than the traditional pyrolysis method. In addition, EPR test further showed that the content of vacancy defects in P₁S₂ was more obvious than that in H₁S₂ (Fig. S2 in Supporting information) [33,34]. Nitrogen adsorption desorption isotherms were investigated to exam specific surface area (SSA) and pore characteristics. The isotherms of catalysts exhibited conformity with type Ⅳ isotherms. H3 hysteresis loop within the range of 0.45 to 0.98 P/P₀ was indicative of mesoporous structure (Fig. 2c) [35]. In comparison with waste plastics-free C (SSA: 130.84 m²/g), the SSA of H₁S₂ increased prominently to 264.57 m²/g; while, the SSA of P₁S₂ slightly decreased to 91.83 m²/g. However, when the traditional pyrolysis method was used, the SSA of P₁S₂–600 (34.64 m²/g) and H₁S₂–600 (38.27 m²/g) dramatically decreased (Fig. S3 in Supporting information). These results were explained that, compared to microwave calcination, tube furnace calcination was detrimental to increasing the SSA of carbon materials. Components along with valence states of catalysts were checked with X-ray photoelectron spectroscopy (XPS) (Tables S2-S5 in Supporting information) [36]. Fig. 2d, Figs. S5a and S6a (Supporting information) showed the full scan of P₁S₂, H₁S₂, C, P₁S₂–600 and H₁S₂–600, in which C, O and S were clearly visible. In order to study valence states of C, O as well as S, high-resolution XPS was utilized. The C 1s near 284.8, 286.2, 288.4, 290.8 eV were C═C, C—OH, C═O, π-π* (Fig. 2e) [16,37]. The C═O content in P₁S₂ exceeded that of H₁S₂ and C, while C—OH content in H₁S₂ was significantly greater than P₁S₂ and C. As C═O and C—OH were widely recognized as active sites for activating PMS to degrade pollutants [14,15,38], different types of waste plastic may modulate contaminants removal pathway through affecting the exposure of different active sites in carbon materials. C═O peak in traditional pyrolysis prepared P₁S₂–600 was lower than that in microwave-assisted P₁S₂, and the C—OH peak in H₁S₂–600 was also lower than that in microwave-assisted H₁S₂, indicating that traditional pyrolysis was not favorable for C═O and C—OH generation (Figs. S5b and S6b in Supporting information). O 1s around 531.7, 533.2, and 535.4 eV (Fig. 2f) were C—OH, C═O, and adsorbed oxygen. Similarly, characteristic peak of C═O in P₁S₂ was significantly higher than that in H₁S₂, and feature peak of C—OH in H₁S₂ was the most prominent, which were corresponded to the phenomenon observed in the C 1s spectrum. Compared with H₁S₂ and C, some C═O group in P₁S₂ should mainly directly came from PET. For S 2p spectra in the catalytic materials (Fig. S4, Figs. S5d and S6d in Supporting information), signals around 164.0 and 165.0 eV were attributed to thiophene S 2p_3/2 and 2p_1/2 [39]. Meanwhile, 168.5 and 169.7 eV was attributed to surface bound sulfites and sulfates. The S 2p spectrum indicated the presence of various states of sulfur in all materials. Elemental analysis of various carbon materials could be found in Table S6 (Supporting information). The combined results of XPS and elemental analysis indicated that the prepared carbon materials were organic carbon materials.

The TC removal efficiency over different materials was analyzed in Figs. 3a and b and Fig. S7 (Supporting information). As the low kinetic constant of PMS self-oxidation (k_obs = 0.018 min⁻¹), TC degradation of pure PMS was only 17.4%. Degradation ability of C/PMS towards pollutant was also very low, with TC degradation of only 67.8% within 20 min, suggesting that PMS activating ability over C alone without adding waste plastics was very limited. It could be evidently seen that, compared with P₁S₂–600/PMS (72.2%), the degradation performance of P₁S₂/PMS system (100%) was significantly improved, accompanied by a substantial increase in rate constant from 0.114 min⁻¹ to 0.303 min⁻¹. Excellent TC removal performance of P₁S₂ should be attributed to the use of PET plastic as well as the application of microwave. These factors facilitated the generation of C═O active sites and defects for PMS activation improvement along with charge transfer enhancement. These, in turn, might trigger a nonradical degradation pathway [6]. Further investigation was conducted on the effect of PET to sodium lignosulfonate mass ratio during the P₁S₂ preparation on the TC removal efficiency. Catalytic performance showed an upward trend at the beginning followed by a decline as plastic content gradually rose (Fig. 3b). The rate constants of P₁S₄, P₁S₃, P₁S₂, and P₁S₁ were 0.098, 0.125, 0.139, 0.303, and 0.159 min^-1, suggesting that P₁S₂ had the highest TC degradation efficiency (effect under different conditions on TC removal see Fig. S10 in Supporting information). After determining the optimal ratio of 1:2 between plastic and sodium lignosulfonate, further exploration was conducted to prepare H₁S₂ using HDPE instead of PET, and the degradation performance of HDPE derived carbon materials on TC was studied (Fig. 3a, Fig. S7a). As shown in Fig. 3a, the degradation performance of H₁S₂/PMS system (100%, k_obs = 0.235 min^-1) was significantly higher than that of H₁S₂–600/PMS (78.4%, k_obs = 0.135 min^-1). The remarkable degradation capabilities of H₁S₂ could be related to the use of HDPE plastic as well as microwave, which were more beneficial to C—OH active site generation for the free radical degradation pathway. The results of Figs. 3a and b showed that both waste plastics utilization and microwave pyrolysis were instrumental in improving pollutant degradation performance of carbon catalysts via active centers exposure. Table S7 showed the quantitative data on the advantages of microwave pyrolysis compared to traditional pyrolysis. Considering that the high specific surface area of carbon materials may adsorb TC, adsorption tests were carried out to elucidate the role of catalytic processes in TC removal (Fig. S8 in Supporting information). The experimental results displayed that TC adsorption capacities of C, P₁S₂, P₁S₂–600, H₁S₂, and H₁S₂–600 were 30.9%, 27.2%, 24.6%, 22.0%, and 27.2%, respectively. In contrast, the direct TC removal process exhibited a similar speed compared to the process involving a 20 min pre-adsorption step (Fig. S9 in Supporting information). Consequently, the TC removal process described here predominantly relies on catalysis rather than adsorption. Further cycling experiments, mineralization tests, and substrate expansion experiments collectively demonstrated the outstanding stability of catalysts, effective mineralization of TC, and universality (Figs. S10 and S11 in Supporting information).

To elucidate the main active sites in P₁S₂/PMS and H₁S₂/PMS catalytic systems, correlation fitting between possible active sites and degradation efficiency was performed in two systems (Figs. 3c-f). From Raman and XPS, it was not difficult to see that P₁S₂ prepared by microwave carbonization had obvious carbon defects and substantial C═O group. Therefore, linear fitting was performed to investigate the relationship between the carbon defects or C═O content in specific materials (including C, P₁S₂–600, P₁S₂) and the TC removal rate as well as the degradation kinetics constant within catalytic systems. The removal rate and k_obs of TC demonstrated a linear relationship with the increase of defects number (Fig. 3c), showing a good correlation, with correlation coefficients as high as 0.996 and 0.999, respectively. In contrast, the correlation between C═O content and TC removal rate along with k_obs was not high (Fig. 3d). Although it was widely reported in literature that C═O was the main active site for generating ¹O₂, considering the low oxidation ability and poor stability of ¹O₂, its ability in pollutant removal may be overestimated [40]. Therefore, electron transfer caused by defects should be the main process of pollutant degradation over P₁S₂/PMS, while C═O site induced ¹O₂ should play a minor role. On the other hand, Raman and XPS indicated that the most prominent feature of H₁S₂ material prepared by microwave carbonization of HDPE was its abundant hydroxyl functional groups and some carbon defects. Therefore, linear fitting was performed on the C—OH content and defects in C, H₁S₂–600, and H₁S₂ based on the degradation experiment results. The correlation between C—OH content and TC removal rate along with k_obs was pretty good (R² values: 0.999 and 0.995, Fig. 3e). According to relevant reports, C—OH could activate persulfate to form free radicals [14,15]. Therefore, free radical pathway should play an important role in activating PMS to degrade TC in HDPE derived carbon material H₁S₂. There was a certain correlation between defects amount and TC degradation efficiency (Fig. 3f), thus, the effect of defects induced charge transfer in TC removal should not be ignored.

To confirm the contribution of active species during degradation, TC removal was conducted in the presence of various scavengers. As illustrated in Fig. 4a, the presence of either 10 mmol/L MeOH (scavenger for SO₄^•− as well as ^•OH) or TBA (typical scavenger for detecting ^•OH) in P₁S₂/PMS exhibited no significant inhibition during TC removal with slightly decreased TC removal rate of 89.5% and 93.1% [26]. Consequently, it was confirmed that TC removal in the P₁S₂/PMS system did not rely on these two radicals. Moreover, p-benzoquinone (BQ) as well as curcumin (Cur) were utilized as scavengers of O₂^•- and singlet oxygen (¹O₂). The presence of BQ had no obvious inhibition with 89.4% TC degradation, indicating that O₂^•- radicals had a rather limited role. After adding Cur, the inhibition of TC degradation was quite limited with a degradation efficiency of 88.8%, which may be related to the low oxidation ability and poor stability of ¹O₂. After adding e^- scavenger K₂Cr₂O₇, 72.7% of TC degradation efficiency was suppressed, indicating that e^- played pretty important role in P₁S₂/PMS system [36]. In comparison, a similar rule was observed in the quenching experiment of P₁S₂–600/PMS system (Fig. S15a in Supporting information), and the relatively lower degradation efficiency in P₁S₂–600/PMS should be related to the lower defect content in P₁S₂–600, thereby hindering electron transfer from TC to PMS. Fig. 4b revealed an interesting finding that the H₁S₂/PMS system showed obviously different results in the quenching experiment compared to the P₁S₂/PMS system, that is, with MeOH or TBA, TC removal decreased sharply (the TC degradation rates only reached 40.0% and 64.9%, respectively), indicating that SO₄^•− and ^•OH had important contribution during TC removal. In combination with the XPS result, SO₄^•− and ^•OH should be generated through PMS activation by C—OH. On the other hand, electron quenching experiment showed that the electron transfer induced pollutant degradation in H₁S₂/PMS was clearly lower than in P₁S₂/PMS. Furthermore, BQ or Cur almost exhibited no inhibitory effect on TC removal in H₁S₂/PMS, indicating that O₂^•- and ¹O₂ had rather limited contribution. Similarly, SO₄^•−, ^•OH and electron transfer also played important roles in TC removal over H₁S₂–600/PMS system. Compared to the H₁S₂/PMS, the relatively lower degradation efficiency in the H₁S₂–600/PMS system should be attributed to the lower C—OH content and less defects in the H₁S₂–600 prepared by traditional pyrolysis. Finally, in the absence of waste plastics as a carbon source, the degradation efficiency of TC by material C was significantly reduced, and the role of SO₄^•−/^•OH and electron transfer in pollutant degradation was rather limited (Fig. 4c). Given that the majority of carbon-based catalysts typically demonstrate a nonradical pathway accompanied by a weak or virtually insignificant radical pathway [8,23], the interesting discovery of modulating pollutant degradation pathways by adjusting the types of waste plastics offered novel insights into the design of carbon-based catalysts.

To offer deeper insight into the role of ROS in PMS activation, EPR spectral analysis was performed. During EPR tests, specific spin trapping agents were utilized: 5,5-dimethyl-1-pyrroline-n-oxide (DMPO) for the detection of SO₄^•−, ^•OH, O₂^•- radicals; 2,2,6,6-tetramethyl-4-piperidone (TEMP) for ¹O₂ identification; and 2,2,6,6-tetramethylpiperidine-1-oxyl (TEMPO) for e^- [41]. As illustrated in Fig. 4d, relatively strong peaks of DMPO-SO₄^•− along with DMPO-^•OH were observed in H₁S₂/PMS after adding DMPO. In contrast, no signals indicative of DMPO-SO₄^•− as well as DMPO-^•OH were observed in C/PMS as well as P₁S₂/PMS systems. The result of DMPO trapping experiments was consistent with the MeOH and TBA quenching experiments. O₂^•- was captured by DMPO in the presence of methanol. No O₂^•- signal was detected in all catalytic systems, which contradicted the quenching experiment results (Figs. S15d and S16 in Supporting information). This contradiction may be due to the rapid conversion of O₂^•- intermediate to ¹O₂ [42]. Further detection of ¹O₂ in the catalytic system using TEMP revealed the presence of ¹O₂ in all catalytic systems (Fig. 4e and Fig. S15e in Supporting information). In combination with the ¹O₂ quenching experiment, the role of ¹O₂ in TC degradation was rather limited because of the low oxidation ability and poor stability of ¹O₂. As shown in Fig. 4f and Fig. S15f (Supporting information), TEMPO-e^- signals were detected in all catalytic systems, revealing that e^- was existed during TC removal process in these five systems. The order of TEMPO-e^- signal strength was P₁S₂ > H₁S₂ > P₁S₂–600 > H₁S₂–600 > C, aligning with the findings derived from electron quenching experiments. It was worth mentioning that the signal intensity of TEMPO-e^- in P₁S₂/PMS system was obviously stronger than in H₁S₂/PMS system, which corresponded to the transition from the electron transfer pathway in P₁S₂/PMS to the free radical process in H₁S₂/PMS. Moreover, electrochemical experiments further verified the electron transfer mechanism (Fig. S17 in Supporting information).

Based on the above discussions, the TC degradation mechanisms through radical and nonradical pathways, mediated through PMS activation with P₁S₂ and H₁S₂ catalysts, were proposed (Fig. 5 and Eqs. S3-S12 in Supporting information). The types of waste plastic had significant regulatory effect on the activation mechanism between electron transfer dominated nonradical process and SO₄^•−/^•OH dominated radical pathway. In P₁S₂/PMS system, defects were the main active sites and electron transfer was the dominant process for TC removal. The TC degradation process could be inferred as follows: (1) Thanks to the electrostatic interaction and van der Waals force, PMS was adsorbed onto defects, leading to the formation of carbon activated peroxymonosulfate complex (C-S₂O₈^2-). Meanwhile, electron-rich TC exhibited a tendency to donate electrons to C-S₂O₈^2-, facilitating PMS decomposition along with pollutants removal (Eq. S3) [15,37]. Notably, structure defects acted as electron shuttle channels, promoting electron transfer from TC to PMS. (2) PMS activation driven by C═O was another nonradical route for TC removal. In this scenario, C═O rendered PMS activation and ¹O₂ production (Eqs. S4-S7) [38]. In addition, electron-rich C═O groups could function as Lewis active sites, enabling electron transfer to produce O₂^•- or ¹O₂ (Eqs. S8 and S9) [43]. However, in the H₁S₂/PMS system, free radicals dominated the degradation of TC. Wherein, C—OH activated PMS to generate SO₄^•−/^•OH for TC degradation, while defects triggered nonradical process played an auxiliary role. In radical pathway, when PMS was interacted with the C—OH on the surface of H₁S₂, C—OH functioned as electron donors, facilitating PMS reduction. This reduction led to the generation of HO₂⁻, SO₄²⁻, along with C═O (Eq. S10) [14]. Following this, HO₂⁻ further interacted with S₂O₈²⁻ to form SO₄^•− and O₂^•- (Eq. S11). Notably, O₂^•- was primarily considered a precursor of ¹O₂ through self-recombination with H₂O, rather than directly oxidizing pollutants (Eq. S9). Additionally, ^•OH was generated through the reaction between SO₄^•− and H₂O (Eq. S12) [44]. Based on quenching experiments, electrochemical analysis as well as EPR tests, it could be deduced that SO₄^•− and ^•OH free radical pathway dominated the TC oxidation, whereas electron transfer had a lesser contribution. In combination with FTIR spectra of PMS, P₁S₂ + PMS and H₁S₂ + PMS in Fig. S18 (Supporting information), it was supposed that: upon interaction with P₁S₂, PMS could form chemical bonds with vacancy defects, resulting in a significant elongation of the O—O bond, which would facilitate the nonradical degradation of TC via electron transfer; meanwhile, when interacted with H₁S₂, PMS could react with the O on C—OH, leading to the abstraction of an H atom from the -OH groups and the subsequent formation of SO₄^•‒ and ^•OH radicals, which facilitated the removal of TC.

Until now, studies have shown that P₁S₂/PMS and H₁S₂/PMS generate distinct reactive species during TC removal. Subsequent intermediates along with TC degradation paths analysis were conducted to elucidate how varying ROS exerted different influences on TC removal mechanisms. The DFT-calculated condensed Fukui function (detailed in Text S4 in Supporting information) was employed to identify reactive sites vulnerable to ROS attack to confirm the TC degradation process (Fig. 6a, Table S8 in Supporting information). Specifically, f⁻ signifies electrophilic reaction connected with nonradical oxidation, f⁰ stands for radical attack and f⁺relates to nucleophilic reaction [45]. An increased value of f⁰, f⁻ or f⁺ within an atom correlates with a greater likelihood of it serving as the attack site for corresponding species [46]. Electron transfer, serving as the primary removal pathway for electron-rich TC in P₁S₂/PMS, is viewed as a process involving electrophilic attack [47]. Similarly, as another electrophilic species, ¹O₂ readily targets the electron-rich sites on TC [46]. Additionally, SO₄^•− and ^•OH conduct free radical attack. The LUMO (lowest unoccupied molecular orbital) and HOMO (highest occupied molecular orbital), unveiled locations prone to electrophilic attack and radical attack (Figs. 6b and c). Typically, the HOMO readily undergoes electron loss, making it highly susceptible to attack by electrophilic species [48]. As exhibited in Fig. 6c, the HOMO distributions of TC aligned with the active sites derived from Fukui index calculations. The sites exhibiting larger Fukui values signify a greater propensity for reactivity. As expected, SO₄^•−and ^•OH exhibited a propensity to target sites on benzene rings and oxygen-containing functionalities, whereas electron transfer processes and ¹O₂ were more inclined to attack positions on nitride-containing groups. To quantify the reactivity of individual atoms within TC, the data presented in Table S8 (Supporting information) exhibited the natural population analysis (NPA) charge distribution, in conjunction with the Fukui index. In the P₁S₂/PMS/TC system, N48 (f⁻ = 0.68776), O16 (f⁻ = 0.39696), N20 (f⁻ = 0.3818), C40 (f⁻ = 0.18607), C27 (f⁻ = 0.30047), C38 (f⁻ = 0.20305), O21 (f⁻ = 0.38394), O23 (f⁻ = 0.35539), and O24 (f⁻ = 0.39606) exhibited heightened reactivity and a propensity for electrophilic attack, aligning with the observations derived from HOMO as well as LUMO analyses. In the H₁S₂/PMS/TC system, C40 (f⁰ = 0.05519), C38 (f⁰ = 0.03907), N48 (f⁰ = 0.21846), C29 (f⁰ = 0.01111), C46 (f⁰ = 0.003715), C41 (f⁰ = 0.00491), C31 (f⁰ = 0.00145), H51 (f⁰ = 0.0314), O24 (f⁰ = 0.02882), O19 (f⁰ = 0.00555) characterized with higher f⁰ values were more vulnerable to be attacked by SO₄^•−/^•OH radicals.

Additionally, to gain deeper insights into TC removal mechanism, LC-MS was utilized to identify the degradation intermediates (Fig. S19, and Tables S9 and S10 in Supporting information). Utilizing both the Fukui function and intermediates analysis, plausible TC removal pathways were deduced. Specifically, within P₁S₂/PMS/TC, two potential nonradical routes were presented (Fig. 6d) [8,49,50]. In the pathway Ⅰ: A1 (m/z 416) was generated via dehydrogenation subsequent to elimination of hydrogen atom at C32 and C33 through electron transfer, along with the removal of aminomethyl group on N48. A1 then underwent further degradation to form A2 (m/z 361) via processes of dehydration as well as deamidation. Additionally, continuous oxidation, dehydroxylation, demethylation and ring cleavage reaction yielded A3 (m/z 232) and A4 (m/z 200). In pathway Ⅱ of the deamidation process, the N20 site on TC was attacked by ¹O₂, resulting in the loss of amide group, and resulting in A5 (m/z 403). Subsequently, A5 lost N-methyl moiety on N48 site form A6 (m/z 375) and simultaneously removed hydroxyl groups, conversing A6 to A7 (m/z 310). The subsequent demethylation and ring opening reaction formed A8 (m/z 220). Ultimately, the intermediary small molecules were further broken down into A9 (m/z 150), A10 (m/z 158), A11 (m/z 60), A12 (m/z 98), A13 (m/z 126), A14 (m/z 136). Then these small molecules underwent further mineralization into CO₂ and H₂O by the action of ROS.

The intermediates detected in the H₁S₂/PMS system varied from those observed in the P₁S₂-induced degradation of TC, further substantiating distinct degradation pathways. Utilizing the Fukui function and the characterized intermediates, Fig. 6e delineated plausible degradation pathways for TC [47,50]. It should be noted that in this system, free radicals and non-free radicals should work together. In hydroxylation pathway Ⅰ, TC underwent initial oxidation mediated by SO₄^•−/^•OH and then hydroxyl groups were introduced into C40 and C38 sites, leading to the formation of ketone B1 (m/z 478). As a result of additional oxidation of ROS, deamidation ensued, resulting in the formation of B2 (m/z 378). Subsequently, as the degradation process progressed, B2 underwent hydroxyl rearrangement and deamination, ultimately decomposing into B3 (m/z 314). B4 (m/z 249) was obtained through the loss of a carbon atom from a ring structure, hydroxylation at the C41 site, and then experienced demethylation and continuous oxidation, ultimately yielding B5 (m/z 166). In pathway Ⅱ, C38, C40, O24 sites were attacked firstly to generate B6 (m/z 405) via hydroxyl groups connection and disruption of the cyclic hydrocarbon structure along with ring-opening process, following with deamidation to produce B7 (m/z 310) and further transformed into B8 (m/z 200) as well as B9 (m/z 136) with oxidation by SO₄^•−/^•OH. The intermediates produced aforementioned underwent intense oxidation due to the persistent attack by ROS and decomposed by ring-opening reactions to B10 (m/z 126), B11 (m/z 114), B12 (m/z 114), B13 (m/z 112), B14 (m/z 98), and B15 (m/z 82). Ultimately, these compounds broke down further into H₂O, CO₂, and smaller organic molecules. Finally, the results of the toxicity assessment of the degradation by-products by Toxicity Estimation Software Tool indicated the environmental friendliness of the catalytic system (Fig. S20 in Supporting information).

To sum up, by using different waste plastics as precursors, metal-free porous carbon catalytic materials with different characteristics were successfully synthesized under microwave carbonization. The as prepared carbon materials P₁S₂ and H₁S₂ exhibited excellent performance in TC removal; P₁S₂ achieved 100% degradation with a rate constant of 0.303 min^-1 in 20 min, while H₁S₂ reached the same degradation level with a rate constant of 0.235 min^-1. The carbon materials derived from waste plastics possessed outstanding practicability, which could accommodate different water environment substrates under a broad pH range (2.6–10.8) and the interference of various inorganic anions, while maintaining good recyclability. Mechanism studies showed that, in P₁S₂/PMS system, the TC degradation was mainly achieved through electron transfer nonradical pathway, with massive defects in P₁S₂ serving as the main active sites. Meanwhile, H₁S₂, which was rich in C—OH groups, achieved TC degradation through a radical process dominated by SO₄^•− and ^•OH. The different degradation pathways of TC were identified using LC-MS and Fukui function calculation, with the intermediates toxicity accurately predicted by T.E.S.T. This substantiated the effectiveness of the P₁S₂/PMS and H₁S₂/PMS systems in mitigating the eco-risks caused by antibiotic contaminants. This study not only embodied the concept of using waste to treat waste, but more importantly, it modulated the degradation mechanism of pollutants by adjusting different precursors of metal-free carbon materials.