As a side effect of urbanization, environmental pollution is a pressing worldwide problem and it has attracted much attention in research field [1, 2]. Among the persistent organic pollutants, antibiotics are widely used to treat infectious diseases. In recent years, these antibiotics have been detected in both natural water substrates and urine due to their overuse and release [3, 4]. Tetracycline (TC) with broad-spectrum sterilization was the second-largest antibiotic generated and consumed in the world. Owing to the effect in disruption of DNA synthesis, both reproductive and developmental risks from TC were reported widely [5]. Consequently, it is necessary to eliminate these harmful pollutants efficiently. However, the traditional wastewater treatment was useless to remove hazard TC, because of its antibacterial properties and chemical stabilities of conjunct structures [6]. The photocatalytic technique using solar energy was supposed as an ideal option to handle this intractable problem, because of the strong oxidability and energy saving [7-10].

According to previous literature reports, some narrow band-gap semiconductors that possessed with visible light response were commonly utilized in photocatalytic field. Since Ye's report in 2010, as a narrow band-gap (E ~ 2.1 eV) semiconductor, silver phosphonate (Ag₃PO₄) has presented with a wide range of light absorbance (to visible-light range) because the involved non-metallic element phosphorus could contribute to charge separation by inhibiting the hybridization of Ag 4d and O 2p orbitals [11, 12]. The merit endows Ag₃PO₄ as an outstanding photocatalyst for TC degradation while there is no or a few other organics in the systems [13, 14]. However, like other direct band-gap excitation semiconductors based on the oxidation of photo-induced holes and ‧OH free-radicals, Ag₃PO₄ can also not be used to degrade TC in the actual wastewater since a lot of coexisting NOM seriously inhibit the probability of aimed TC degradation [15-17]. Despite the fact that the positive potential at the conduction band (CB, 0.18 V vs. NHE) of single Ag₃PO₄ commonly blocks the generation of reactive oxygen species (E⁰(O₂/·O₂^-), −0.23 V), there have been many efforts to reconstruct Ag₃PO₄-centrated heterojunction through optimization of the electronic structure for realization of MOA process mainly because of its both excellent visible lights harvesting and h_vb⁺/e_cb^- separating capacity. For this purpose, Xu et al. had introduced g-C₃N₄ to Ag₃PO₄ to constructure Z-scheme heterostructures, and they presented with enhanced photocatalytic performance in TC degradation under visible light irradiation [17]. Moreover, Yin et al. prepared another Z-scheme Ag₃PO₄/g-C₃N₅ composite for TC degradation. As-prepared photocatalyst displayed obvious oxygen activation thereby promoting the improvement of degradation performance [18]. It is suggested that both g-C₃N₄ and g-C₃N₅ could elevate the photocatalytic efficiency of Ag₃PO₄ since these non-metal components achieved electronegativity to actuate separation of charge carriers through MOA [19].

Different from traditional Type-Ⅱ heterostructures, Z-scheme or S-scheme charge transfer mechanism in heterostructures could overcome the defects that the CB potential of Ag₃PO₄ is too positive to activate O₂. Especially, S-junction contained two n-type semiconductors results in the small potential barriers for the Fermi level homogenization between oxidative and reductive photocatalysts which promote the charge separation and transportation in an internal field [20, 21]. Although Ag₃PO₄, g-C₃N₅ and g-C₃N₄ were all classified as n-type semiconductors and they had been as suitable components to build S-scheme junctions in previous literature [22-25], to our knowledge, most of the as-proposal dual-S-scheme junctions had been designed to be single path of MOA and two routes of ‧OH radical formation in which its charge transfer pathway likes an inverted "V" letter [25]. A newly dual S-scheme junction Ag₃PO₄-based photocatalyst to enable two paths of MOA, that is, a positive "V" letter, hasn't been tried yet. Obviously, two paths to run MOA is more efficient than the single path for degradation of TC under identical incident irradiation. Herein, we prepare a new g-C₃N₄/Ag₃PO₄/g-C₃N₅ dual S-scheme junction with a desired "V" type band structure by a post-interposition of in-situ precipitated Ag₃PO₄ into prepared g-C₃N₄/g-C₃N₅ composite precursor, which just has the ability to build two paths of MOA and successfully apply to degrade TC in actual wastewaters under visible light. All experimental details were both enlisted in Supporting information.

The N₄/APO/N₅ samples were synthesized by post-interposition of in-situ formation APO into the N₄/N₅ precursor (Detailed in experimental section in Supporting information). Firstly, N₄ and N₅ precursor presented with the shape of nanosheets and nanorods, respectively (Fig. 1a, Figs. S1a and b in Supporting information) [26]. On the other hand, Ag₃PO₄ nanoparticles (Fig. 1b), N₄/APO (c) and APO/N₅ (Figs. S1c and S1d in Supporting information) with high crystallinity could be observed. Due to the post-interposition of N₄/N₅ samples, Ag₃PO₄ nanoparticles contacted with both the N₄ nanosheets and N₅ nanorods in N₄/APO/N₅ sample (Fig. 1c). This indicates that the as-synthesized sample is N₄/APO/N₅ with "V" type conduction band structure arrangement, instead of N₄/N₅/APO. This structure is also confirmed by TEM images in Figs. S1e and f (Supporting information). Moreover, the elemental mapping of C and N also pointed out the uniform distribution of N₄ and N₅ on the surface of Ag₃PO₄ with two interfaces (Figs. 1d-i). These large contact areas between the two kinds of carbon nitride and Ag₃PO₄ boost the generation of heterojunctions and can provide two channels to stimulate the charge separation from Ag₃PO₄ to N₄ or N₅.

As shown in X-ray diffraction (XRD) patterns (Fig. 1j), the cubic phase Ag₃PO₄ (JCPDS No. 006–0505) with good crystallinity is detected [27]. Moreover, (100) and (002) planes in both g-C₃N₄ and g-C₃N₅ could be confirmed from the stacking of special triazole and triazine units [26]. According to the weighing and atomic ratio in X-ray photoelectron spectroscopy (XPS) in Table S1 (Supporting information), the ratio between N4 and N5 could be confirmed average 1.5. On the other hand, the peak position of Ag₃PO₄ on (210) plane is also moving from 33.6° to 33.4° in the composite sample. It validated the extension of lattice distance in Ag₃PO₄ sample, owing to the coordinative gravity from the Ag and the conjunction rings [28]. Furthermore, the low crystallinity for Ag₃PO₄ was demonstrated in N₄/APO/N₅ sample from negative charges at N₄/N₅ [29]. In summary, XRD patterns pointed out the successful fabrication of g-C₃N₄, g-C₃N₅, and Ag₃PO₄ moiety in the N₄/APO/N₅ sample.

In addition, in the XPS measurements (Figs. 1k-m and Fig. S2 in Supporting information), three peaks in the N 1s spectra could be fitted at 398.8,401.2 and 405.8 eV, former two peaks indicated C—N=C and C—N-H units, the peak located at 405.8 eV suggested the activation of conjugated electrons (Fig. 1l) [26, 30]. In high-resolution spectra of Ag 3d, two characteristic peaks at 374.2 and 368.2 eV were presented, which could be assigned to Ag element in Ag₃PO₄ (Fig. 1m) [31]. Furthermore, by compared to N₄/N₅ composite sample, the negative shift in C and N spectra resulted from the modification of Ag₃PO₄ [32], it proved that electrons might leave from Ag₃PO₄ into N₄ and N₅, respectively. On the other hand, opposite trends in P, O, and Ag peaks also revealed Ag₃PO₄ could play as electron donor to give electrons to N₄ and N₅ [33]. This result confirmed that the fabrication of dual S-scheme N₄/APO/N₅ is successful for improvement of electron separation in two channels. Moreover, both BET and FT-IR measurements are operated. As shown in Fig. S3a (Supporting information), conjunction ring (808 cm^-1), residue amino group (3200–3600 cm^-1) and C—N (1200–1600 cm^-1) structure in C₃N₄ and C₃N₅ could be confirmed. In the Ag₃PO₄ sample, peak at 928 and 541 cm^–1 could be attributed to the PO₄^3– in the Ag₃PO₄. The specific surface area (SSA) was measured by BET and the SAA of N₅/APO/N₄ was 11.6 m²/g (Fig. S3b in Supporting information).

The band-gap properties of as-prepared N₄/APO/N₅ sample with "V" type conduction band structure arrangement are presented in Fig. 2. It is obvious that the involvement of both N₄ and N₅ component contribute to the absorbance of incident light in composite sample (Fig. 2a and Fig. S4 in supporting information), because of the typical conjunct structure between their triazine and triazole units. Among all three composite samples, ternary N₄/APO/N₅ heterojunction achieves the strongest optical absorbance in visible light region (420–520 nm) [34]. This is partly due to the d-π interaction between Ag and conjunct rings in N₄ and N₅ ingredient that boosts charge carriers' absorption [35-37]. More importantly, we quantified the positions of CB of both N₄ and N₅ component of as-prepared samples to confirm whether O₂ could be allowed to trap their electrons into ^•O₂^-. Based on Tauc plot in Fig. 2b, the band gaps were calculated as 2.43, 2.80, and 2.08 eV corresponding to APO, N₄ and N₅, respectively. Through VB-XPS in Fig. 2c and the empirical equation: E_CB = E_VB - E_g, the potential of the VB could be fitted as 1.53 V (for N₄), 1.32 V (for N₅), and 2.61 V (for APO). Thus, the potential of CB was counted as −1.27 V, −0.76 V, and 0.18 V of N₄, N₅, and APO, respectively. The features of CB very agreed with that excited electrons in N₃/APO/N₅ might be not concentrated at the CB of APO because its positive potential was not able to transfer electrons to O₂ (E⁰(O₂/^•O₂^-), −0.23 V vs. NHE) [38]. Instead, both CBs of N₄ and N₅ component in N₄/APO/N₅ were significantly shifted to more negative position, which was very in favor of O₂ capturing electrons to form ^•O₂^-. Meanwhile, owing to its potential of VB of APO slightly lower than that of ^•OH free-radical formation (~2.7 V vs. NHE), the corresponding oxidation reactions mediated by the VBs of N₄/APO/N₅ mainly consist of NOM oxidations or intermediates H₂O₂ oxidation that is easy to carry out, by which can sustain ^•O₂^- continuous production on the CB side. These ^•O₂^- also reacted with the photo-generated holes to from the strong oxidizing ¹O₂ (E⁰(O₂/¹O₂), ~1.99 V vs. NHE). Thereby, the dual S-scheme heterojunction in N₄/APO/N₅ sample was definitively established through our proper arrangement of connection among N₄, N₅ and APO in the prepared N₄/APO/N₅.

In Mott-Schottky plot of N₄ (Fig. 2d), N₅ (Fig. 2e) and APO (Fig. 2f) based on the fitting, three positive slops indicated that N₄, APO, and N₅ are n-type semiconductors. Therefore, the contact between Ag₃PO₄ and carbon nitride-based materials were n-n interface. This is remarkable evidence for S-scheme junction [20]. Specifically, the Fermi levels of APO, N₄ and N₅ were calculated as 0.23 V, −0.78 V, and −0.72 V, respectively. This narrow potential gap between APO and N₄ (or N₅) delivers the internal electrons field to develop the charge transfer from APO to N₄ (or N₅) in two channels with low energy barriers [39]. It could be guessed that the N₄/APO/N₅ junction is a S-scheme junction (Fig. 2g), because ^•O₂^- could be generated at the CB of N₄ and N₅ with the negative enough potential. That is, a "V" alignment of CBs in N₄/APO/N₅ composite photocatalyst had distinct advantages of "dual-channel" activating O₂, while the oxidation capacity of the VBs will remain basically unchanged. This is opposite with the traditional type Ⅱ junction mechanism (Fig. 2h), in which the most of photo-induced electrons are sunk in the CB of APO and they cannot be trapped by O₂.

TC·HCl was used as target pollutant to assess the photocatalytic performance of as-prepared N₄/APO/N₅. Compared with the low efficiency of N₄/N₅ and APO in Fig. 3a and Fig. S5 (Supporting information), N₄/APO/N₅ (with Ag₃PO₄ mass of 0.1) indeed showed the best result that 94.7% TC was removed after light irradiation for 1 h. Based on the pseudo-first-order model as follow (Eq. 1) [40]:

$ \ln \left(C / C_0\right)=-k t $

(1)

where t (min) is the reaction time, k (min^-1) refers to the kinetic constant for photocatalysis, C and C represent the initial and instantaneous concentration of TC after irradiation for time t, respectively. The kinetic constant of N₄/APO/N₅ (0.0483 min^-1) was enhanced 3.0-times and 2.5-times for N₄/N₅ (0.0161 min^-1) and APO (0.0193 min^-1) (Table S2 in Supporting information). These binary composites, including N₄/APO and APO/N₅, also display improved photocatalytic performance than single APO. Moreover, the physic mixture of N₄/N₅/APO sample show the kinetic constant of 0.0312 min^-1, it is less than that of N₄/APO/N₅ sample. In analysis of synergistic effect in this ternary photocatalysts, the synergistic index was accounted as k_N/k_APO+k_N, where k_N, k_APO, and k_N are the kinetic constant of N₄/APO/N₅, APO, and N₄/N₅ in photodegradation process, respectively [41]. The value of the synergy index was computed as 1.49, it is a remarkable signal to affirm the cooperation effect of each component in dual S-scheme N₄/APO/N₅ photocatalyst. It is impressive that this photocatalytic result is better than that in other previous literature reports (Table S3 in Supporting information).

Stability and durability are important for photocatalysts in a long-term application, especially for silver-based materials. As we all know, photo-corrosion caused by photo-excited electrons was the main factor for recession of noble metal photocatalysts. As shown in Fig. 3b, the photocatalytic activity of ternary N₄/APO/N₅ sample indeed decreased from 94.7% to 86.2% after 3 cycles. However, in XRD pattern (Fig. 3c) and XPS spectra (Fig. S6 in Supporting information), the metallic Ag as the product of photo-corrosion is not detected in used samples. This anti-corrosion property also suggests that some electrons might migrate away from Ag₃PO₄ to carbon nitride materials [27]. Thus, we attributed the slight decrease of activity of N₄/APO/N₅ to the accumulation of intermediates of degradation TC on the surface. Moreover, TEM images of used sample in Fig. S7 (Supporting information) also displayed for testing its morphology, which revealed the strong connection and stability in the N₄/APO/N₅ composite. In a contrast, the photocatalytic activity was significantly decayed from 70.2% to 53.1% employing single Ag₃PO₄ sample. Moreover, the influence of water matrix and inorganic ions were presented in Text 3 and Fig. S8 (Supporting information).

Reactive oxygen species (ROS) were measured by adding trappings agent and ESR experiments, respectively. As showed in Fig. 4a, there all existed the inhibition effect on the degradation of TC when we added FFA, p-BQ, IPA and EDTA for capturing ¹O₂, ^•O₂^-, ^•OH radical and photo-induced h_vb⁺, respectively, but p-BQ has the biggest effect among the three. This undoubtedly indicated ^•O₂^- and ¹O₂ as the lead role to take part in the degradation. What the features of inhibition was consistent with was the direct observations of ^•O₂^- formation for three photocatalysts by DMPO-trapping ESR measures (Figs. 4b-e). N₄/APO/N₅ case showed the biggest ^•O₂^- yield (Fig. 4e), while APO case showed little (Fig. 4b), indicating the existence of enhanced MOA process in our N₄/APO/N₅ composite samples. In contrast, the inconspicuous peak of ^•O₂^- radicals in ESR indicate the poor MOA process of pure APO samples. It was coincident with the previous conclusion that excited electrons at CB of Ag₃PO₄ with the very weak reductive potential and could not complete the ^•O₂^- production [42, 43]. However, the composition of N₄ and N₅ could trigger charge separation in MOA process by construction of heterojunction with S-scheme mechanisms [38, 44]. Obvious ^•O₂^- signals in the Figs. 4c and d revealed the moderate MOA process between APO and N₄ (or N₅). Furthermore, the strong peak intensity in ESR (Fig. S9 in Supporting information) and hindrance from FA both indicated the vital role of ¹O₂ in TC degradation, this radical was generated from the further oxidation of ^•O₂^- by h_vb⁺. Additionally, the hindrance from both EDTA case and IPA case revealed the non-negligible role of photo-generated h_vb⁺ in the direct oxidation process of TC followed by the hydrolysis. According to these results, ^•O₂^- and h_vb⁺ all took part in the photocatalytic degradation of TC, but the dual S-scheme heterojunctions N₄/APO/N₅ predominated the active MOA process. Namely, N₄/APO/N₅ showed optimal photocatalytic performance to generate ^•O₂^- (Fig. 4f), and it was 1.5- and 1.7-times higher than that of N₄/APO and APO/N₅ in ^•O₂^- generation after light illumination for 10 min. This definitely indicated the extra pathway in a dual S-scheme charge transfer mechanism on N₄/APO/N₅ heterojunction.

The detection of intermediate products during mineralization process for organic pollutants could not only correlate the degradation paths with different ROS, but also evaluate the security of intermediate processes in order to prevent the production of more toxic products. To this end, we used HPLC-MS to determine the intermediates of TC degradation by N₄/APO/N₅ photocatalyst in different irradiation times and found there were 8 intermediates detectable (Fig. 5a and its MS in Fig. S10 in Supporting information). According to these intermediates as well as ESR results above (few ‧OH free-radical formation), there would be three possible pathways proposed to degrade TC pollutant in water. In the pathway I, the hydroxylation of benzene ring and N-methyl group of TC were resulted by direct h_vb⁺ oxidation followed by hydrolysis to generate TC-1, TC-3, and TC-2 [45]. Moreover, successive deamidation and N-demethylation reactions correspond to pathway Ⅱ and Ⅲ, respectively. In pathway Ⅱ, TC degradation was mainly attributed to the direct oxidation of photo-induced h_vb⁺ followed by ^•O₂^- reaction in terms of the intermediate products. After the deamidation to TC (TC-4 as the product), the second-order intermediate in the path as TC-5 was caused by the following N-demethylation of TC-4 [46]. In the last channel, the formation of TC-6 also displayed N-demethylation process at the initial step. Afterwards, deamination and N-demethylation reactions guide two different reaction ways to yield TC-7 and TC-8. According to previous researches in the transition state of organic pollutant degradation, deamination of benzene ring and N-demethylation were mostly caused by the ^•O₂^- [42]. In short, thanks to the unconventional MOA by our N₄/APO/N₅ photocatalyst under visible light irradiation, the degradation of TC was significantly enhanced at least in two degradation pathways (Fig. 5a).

The bioaccumulation factor and developmental toxicity of both TC parent molecule and its degradation intermediates were calculated by Toxicity Estimation Software (T.E.S.T.) [47]. As presented in Fig. 5b, most of these intermediates possessed with lower bioaccumulation factor than TC, especially for these intermediates such as TC-7, TC-4, TC-6 and TC-5 generated in ^•O₂^- dominated processes. It suggested that the detoxification effect was presented with the ^•O₂^--dominated processes of TC photodegradation. However, TC-1, as typical intermediate raised from hydroxylating on benzene ring of TC, was an exception. For the developmental toxicity index, most of these photocatalytic degradation products display relatively lower indexes of developmental toxicity (Fig. 5c), but the sequence was almost the opposite to the bioaccumulation factors, especially, TC-7 toxicity has increased slightly relative to TC parent. It meant that as a green technique for antibiotic detoxification and elimination in water environments, our N₄/APO/N₅ photocatalyst showed good prospects, but further improvement was needed in the future. The combined toxicity of reactant was measured by the microalgae (C. pyrenoidosa) cultivation in 24 h (Fig. S11 in supporting information). As shown in Fig. S8, the detoxification effect from the photocatalytic degradation was obvious. Comparing to the initial reactant, the final reactant decreased the inhabitation rate from 27.3% ± 3.2% into 13.4% ± 2.1% (P < 0.05). It was confirmed that the relatively low toxicity in the final reactant.

The dynamic of carriers should be explored for in-depth understanding the photocatalytic mechanism in our dual S-scheme. In comparison to other photocatalysts, the lower fluorescence intensity in N₄/APO/N₅ composite suggests the effective charge separation (Fig. 6a). More importantly, the weakest peak of N₄/APO/N₅ sample depicted that the dual modification of N₅ and N₄ with Ag₃PO₄ may provide an additional channel to transfer photo-generated carriers [48]. Furthermore, time-dependent PL spectra showed a long quantum lifetime for N₄/APO/N₅ sample. The calculation for the two-parameter equation of quantum lifetime is presented as follows (Eq. 2):

$ \tau_{\text {average }}=\Sigma_{i=1}^2 A_i \tau_i^2 / \Sigma_{i=1}^2 A_i \tau_i $

(2)

As depicted in Fig. 6b and Table S4 (Supporting information), the lifetime of APO, N₄/N₅ (result obtained from our previous research [26]), and N₄/APO/N₅ were corresponding to 43.36, 7.88, and 105.18 ns, respectively. Specifically, the distinctive prolongation of recombination lifetime (τ₂) certified that dual heterojunctions suppress the fast recombination of charge carriers in Ag₃PO₄ because N₄ and N₅ possess some active sites to shuttle photo-excited electrons and consume it in MOA process [49]. Because of charge lifetime extension, the dual S-scheme heterojunction might provide more active carriers to participate in photocatalytic reactions.

The kinetic of carriers was further analyzed by photo-electrochemical detection. It was evident that N₄/APO/N₅ sample acquire the best conductivity among various samples by the I-t response curve (Fig. 6c). On the other hand, N₄/APO and APO/N₅ samples also showed an enhanced current response than pure Ag₃PO₄. The order of the current response is also followed with a sequence of photocatalytic activity results of degradation TC. This suggested that the charge separation was activated by the construction of dual heterojunctions with N₄ and N₅. The small radian in Nyquist plots (Fig. S12 in Supporting information) also corroborated the excellent conductivity in the N₄/APO/N₅ [50].

To further calculate the separation efficiency of charge carriers, H₂O₂ activation method was operated in photo-current experiments and the transfer efficiency could be quantified as the ratio of $ I_{\mathrm{H}_2 \mathrm{O}} / I_{\mathrm{H}_2 \mathrm{O}_2}$ [51]. In Fig. S13 (Supporting information), the corresponding efficiency of APO, N₄/N₅, and N₄/APO/N₅ were calculated as 10.2%, 18.8%, and 31.1%, respectively. The obvious charge separation enhancement in N₄/APO/N₅ could attribute to double MOA sites of the "V" type dual S-scheme heterojunction. In this heterojunction, some amino groups, and lone pairs on tri-azo unit of N₅ and N₄ could be Lewis alkaline sites and molecular oxygen sites to consume excited electrons [52, 53]. This process could suppress the recombination of charge carriers and boost ROS generation, which was consisted with above ROS quantitation result and promotion of charge separation. In general, all photo-electrochemical measurements demonstrated the modification of reductive photocatalysts (g-C₃N₄ and g-C₃N₅) are beneficial for the charge transfer and MOA process.

Furthermore, we measured the surface zeta potential of N₄/APO/N₅ samples (Fig. 6d). The most negative surface potential of N₄/APO/N₅ sample is −13.4 mV, and it indicated that more excited electrons are concentrated on ternary sample than that on N₄/N₅ (−6.7 mV) and APO (−11.2 mV). The transient surface photovoltage (TR-SPV) responses were used to well investigate the charge separation and the dynamic processes of electron-hole pairs. In Fig. 6e, during the light irradiation, the negative surface photo-voltage could be validated in all N₄ and N₅-based samples, illustrating excited electrons might migrate to surface of photocatalysts [54]. Moreover, as electron acceptor, O₂ can trap electrons on conduction band of N₄ and N₅ simultaneously. Thus, the rapid separation of charge carriers can take place in short exciting time in the SPV (~10^–10 s) [55]. By compared with other two samples, the stronger voltage and longer response time in N₄/APO/N₅ certified that dual S-scheme might facilitate electron shift in photocatalyst [56-58].

According to our previous research and above statements, the interaction between donor and acceptor could affect the charge transfer, and it can be testified by a well-known acceptor MV²⁺. As shown in Fig. 6f and Fig. S14 (Supporting information), the addition of MV²⁺ suppressed the luminescence intensity of σ → σ* at Ag orbitals, based on the Gaussian fit for the luminescence peaks [59, 60]. Furthermore, the Stern-Volmer equation analyzed the dynamic of electron donation for the dual S-scheme and showed as follow (Eq. 3):

$ \frac{F_0}{F}-1=k \times C_{\mathrm{MV}^{2+}} $

(3)

In this equation, the ratio of peak intensity could reveal the kinetic constant for trapping from MV²⁺. As presented from the well-fitting curves in inset of Fig. 6f, the high constant as 39.99 in N₄/APO/N₅ suggested the significant donor trait, it also validated that excess excited electrons might be released from N₄ and N₅ for MOA to generate sufficient ROS in the dual S-scheme [61].

In this work, directed against TC degradation easier to carry out by ^•O₂^-, a dual S-scheme N₄/APO/N₅ photocatalyst was prepared through post-interposition of in-situ formed Ag₃PO₄ to the g-C₃N₄/g-C₃N₅ for remove of TC in waters by means of intensive molecular oxygen activation (MOA) into ^•O₂^- (Scheme 1). In comparison to g-C₃N₄, Ag₃PO₄, and g-C₃N₅, the tenantry composite revealed 4.3-, 3.3-, and 2.5-times improvement in the degradation of TC in waters. All trapping experiments and ESR measurements of ROS had pointed out significant traits of enhanced MOA process by N₄/APO/N₅ photocatalyst under visible-light irradiation. Based on the band structure calculation, the dual S-scheme construction with two paths for MOA could be confirmed like a "V" shape. The further investigation of dynamic of charge separation revealed that the rapid electrons transfer efficiency was presented due to the existence of the two paths for O₂ trapping electron of conduction bands. This novel dual S-scheme junction shape succeeded in degradation of TC under visible light irradiation, even in the presence of different NOM and inorganic ions.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

CRediT authorship contribution statement

Futao Yi: Writing – original draft, Methodology. Ying Liu: Methodology, Investigation. Yao Chen: Methodology, Data curation. Jiahao Zhu: Data curation. Quanguo He: Resources, Investigation. Chun Yang: Writing – review & editing, Funding acquisition, Formal analysis. Dongge Ma: Writing – review & editing, Resources. Jun Liu: Writing – review & editing, Funding acquisition, Conceptualization.

Acknowledgments

This research was funded by the National Natural Science Foundation of China (No. 22106042), Hunan Provincial Natural Science Foundation of China (Nos. 2024JJ5124, 2024JJ5126), the Scientific Research Foundation of Hunan Provincial Education Department (No. 23B0564). We thank for the Ting Yang in the experiment support of XRD, XPS and UV–vis absorption spectra from the www.shiyanjia.com.

Supplementary materials

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2024.110544.