Effects of UV/PMS oxidation on the degradation of zidovudine: kinetics, degradation products, and reaction pathways

Abstract

Zidovudine (AZT), a persistent pharmaceutical contaminant detected in diverse biological and environmental matrices, raised significant concerns due to its ecological and health risks. This study systematically investigates the degradation kinetics, mechanisms, and toxicity evolution of AZT in a UV/peroxymonosulfate (UV/PMS) system. The UV/PMS process demonstrated superior performance with a degradation rate constant of 0.0384 min⁻¹, surpassing UV/H₂O₂ (0.0138 min⁻¹) and UV/NaClO (0.0300 min⁻¹), achieving 84.44% removal efficiency. Radical quenching experiments and kinetic modeling revealed synergistic contributions from direct photolysis (51.0%), hydroxyl radicals (18.1%), and sulfate radicals (30.9%). Degradation exhibited strong pH dependence, with optimal efficiency at pH 5.2–6.1 (k = 0.0486 min⁻¹, >92% removal), while alkaline conditions significantly inhibited the process. Coexisting substances differentially influenced degradation: HCO₃⁻ (10 mM) reduced efficiency to 68.6% (k = 0.0194 min⁻¹), NO₃⁻ (3 mM) slightly enhanced removal to 90.85% (k = 0.0414 min⁻¹), and NO₂⁻ (3 mM) and humic acid (10 mg L⁻¹) caused severe suppression (46.2% and 36.84% removal, respectively) through radical quenching and UV absorption. In real water matrices, Yellow River source reservoir water inhibits AZT degradation: under identical oxidant concentrations, UV/PMS, UV/NaClO, and UV/H₂O₂ systems showed 26.85%, 31.2%, and 32.9% lower efficiencies than in ultrapure water. Increasing PMS to 15 and 25 mg L⁻¹ enhanced UV/PMS removal to 70.04% and 81.03%. Inhibition is linked to inorganic ions, scavenging radicals, alkaline pH (8.27), high turbidity interfering with UV absorption, and organics competing for radicals. Three primary degradation pathways were identified, involving thymine formation, azide group elimination, demethylation, and double bond addition. Toxicity assessments using Vibrio fischeri bioluminescence indicated an initial increase followed by partial reduction in acute toxicity, though residual toxicity persistently exceeded baseline levels.

---

Water impact

This study addresses zidovudine (AZT), a persistent antiviral contaminant exhibiting limited removal efficiency (<68%) in conventional water treatment systems. The UV/peroxymonosulfate (UV/PMS) process demonstrated high efficacy, achieving >84% AZT degradation through synergistic contributions from direct photolysis (51%) and reactive radicals (SO₄˙⁻/OH˙). Optimal performance occurred at pH 5.2–6.1 and a low PMS dose (6 mg L⁻¹). While nitrate exerted minimal interference, bicarbonate (HCO₃⁻) and humic acid significantly inhibited degradation. Toxicity assays indicated partial detoxification during treatment. UV/PMS presents a scalable strategy for mitigating antiviral contamination risks in aquatic environments.

1. Introduction

Antiviral drugs (e.g., AZT for AIDS treatment) are widely used, with global antiretroviral therapy requiring ∼2.18 tons per day for 19.8 million patients.^1,2 Due to structural stability, these pharmaceuticals persist in medical effluents and are discharged into wastewater treatment plants (WWTPs). It was reported that 24 820 tons of ethylene glycol dimethyl ether are input into WWTPs worldwide every year.³ However, conventional WWTPs showed limited capacity to remove antiretroviral drugs.⁴ As a result, WWTPs release large quantities of effluent containing high concentrations of antiviral drugs, jeopardizing ecosystems and promoting the development of viral resistance in aquatic environments. Additionally, natural disasters can alter drug concentrations in groundwater.

AZT, a well-known antiretroviral drug, is frequently detected in aquatic environments due to its extensive use in AIDS treatment. The removal efficiency of AZT is less than 68% because its glucuronide conjugates cleave to release free AZT during wastewater treatment, which cannot be effectively degraded by conventional biological treatment processes.⁵ Once released into aquatic environments, AZT poses ecological risks due to its persistence. Studies have reported that the concentration of AZT in the surface water of the Nairobi River Basin, Kenya, can be as high as 7680 ng L⁻¹, which poses significant challenges to conventional treatment processes.⁶ Therefore, further research is still needed to find efficient technologies for removing antiviral drugs from water.

The UV/PMS system activates peroxymonosulfate via UV light to generate sulfate (SO₄˙⁻) and hydroxyl (OH˙) radicals, whose strong oxidizing capacity enables efficient degradation of various organic and recalcitrant pollutants.⁷ Simetić et al. summarized studies on 4-MBC and EHMC degradation by different AOPs (Table S1), noting that UV/PMS showed superior performance: with 0.3 mM PMS, EHMC degradation exceeded 90% at 1000 mJ cm⁻², and it had stronger tolerance to water matrix interference.⁸ Although advanced oxidation processes (AOPs) are critical for treating recalcitrant pollutants, research on AZT degradation via the UV/PMS system has three key gaps: existing studies note involvement of SO₄˙⁻ and OH˙ but fail to quantify their specific roles in AZT degradation through quenching experiments combined with probe compounds, leaving core oxidizing species ambiguous and lacking theoretical support for process optimization; the N₂ release mechanism of AZT's azido group and thymine ring-opening patterns remain unclear, with intermediate product formation sequences and transformation relationships unclarified via UPLC/Q-TOF-MS and DFT analysis, making it impossible to identify recalcitrant intermediates; existing work focuses only on AZT removal efficiency without tracking toxicity changes along the “AZT–intermediates–end products” pathway via Vibrio fischeri assays or ECOSAR predictions, failing to rule out risks of forming highly toxic intermediates despite pollutant removal.

This study comparatively investigates the degradation efficiency of AZT by the UV/PMS system, with the following objectives: (1) examine the impact of various reaction parameters and water quality conditions on AZT removal; (2) qualitatively and quantitatively analyze the contribution of different radicals in the UV/PMS system; (3) elucidate the degradation pathways of AZT through identification of its transformation products and evaluate the ecological risks of AZT and its degradation products using ECOSAR. This study is expected to provide valuable insights for the research on degrading antiviral drugs in drinking water via UV-based advanced oxidation technologies and offer theoretical support for practical water treatment engineering applications.

2. Materials and methods

2.1 Materials

The chemical reagents used in this study are shown in Tables S2 and S3, respectively, and the ultrapure water was produced by Milli-Q ultrapure water machine. The main instruments and equipment used in the experiments and their corresponding manufacturers and models are shown in Table S4.

2.2 Experimental procedures

The degradation experiments were conducted using a commercial parallel-beam apparatus (Trojan Technologies). As illustrated in Fig. S1, the system featured a low-pressure mercury lamp (70 W, 254 nm emission wavelength) positioned parallel to a reaction dish. The spectral diagram of the light source is shown in Fig. S2. UV irradiation was delivered through a quartz cylinder to 100 mL of the reaction solution (depth: 1.64 cm; light intensity: 0.141 mW cm⁻²). Prior to experiments, the lamp was preheated for ≥30 min to stabilize output.

Reaction solutions were prepared by sequentially adding ultrapure water, AZT stock solution (11.2 μmol L⁻¹), phosphate buffer (5 mM final concentration; pH adjusted with 100 mM NaH₂PO₄/Na₂HPO₄ solutions), and PMS stock solution (300 μmol L⁻¹) into the reaction dish. Additional components (e.g., Cl⁻, NO₃⁻) were introduced as required for specific experimental conditions. UV irradiation commenced immediately after positioning the dish under the quartz cylinder.

Samples (1 mL) were collected at UV doses of 0, 100, 200, 300, 400, and 500 mJ cm⁻² (corresponding to 0, 11.85, 23.68, 35.53, 47.38, and 59.23 min, respectively). Each aliquot was quenched with 0.1 mL Na₂S₂O₃ solution (0.1 M) in 2 mL amber vials, then stored at 4 °C until analysis within 24 h. All experiments were performed in triplicate to ensure reproducibility.

AZT molecular properties were calculated using density-functional theory (DFT) in Gaussian 16 with the 6-311G(d,p) basis set.⁹ Optimized geometries yielded electrostatic potential maps, frontier molecular orbitals (HOMO/LUMO), and charge distributions.

2.3 Analytical methods

For degradation product analysis, reaction aliquots collected at 0, 30, and 60 min were directly injected into a UPLC/Q-TOF-MS system (solid-phase extraction omitted to minimize analyte loss). Product structures were tentatively identified by matching exact masses (mass error <5 ppm) with theoretical values, aided by Mass Frontier 7.0 and literature-derived fragmentation patterns. The mass-to-charge ratio measurements, theoretical values, and deviations are recorded in Table S7. Reaction pathways were reconstructed using ChemDraw 20.0 and validated against experimental MS2 spectra. Detailed information regarding other analytical methods is provided in Test S1.

3. Results and discussion

3.1 Degradation effect of AZT by UV/PMS

3.1.1 Comparison of different oxidization systems. The degradation of AZT by different UV-based oxidation systems was investigated at pH 7.0. As shown in Fig. S3, the degradation rates of AZT by UV/PMS, UV/NaClO, UV/H₂O₂, UV alone, NaClO alone, and PMS alone increased with reaction time, achieving removal efficiencies of 93.95%, 89.0%, 72.1%, 37.8%, 50.9%, and 13.9% after 60 min, respectively. The degradation of AZT by all six distinct treatment systems followed pseudo-first-order kinetics, with rate constants (k) of 0.0474, 0.0374, 0.0215, 0.0080, 0.0120, and 0.0025 min⁻¹, respectively.

As illustrated in Fig. S3(a), the UV/PMS system achieved the highest AZT removal efficiency (93.95%), which can be attributed to the strong oxidative capacity of SO₄˙⁻, known for its high redox potential (2.5–3.1 V).¹⁰ The UV/NaClO system ranked second in degradation performance, demonstrating its potential applicability. This is because NaClO generates multiple oxidative radicals under UV irradiation, such as non-selective ˙OH and ˙Cl (highly reactive with electron-rich organics), as well as Cl₂˙⁻, ˙ClO, O₂˙⁻, and ˙ClOH,¹¹ which collectively promote AZT degradation. In contrast, the UV/H₂O₂ system exhibited the lowest efficiency under identical conditions, likely due to H₂O₂'s limited UV absorption coefficient, which restricts radical generation.¹² Notably, PMS alone (1–10 mg L⁻¹) showed minimal AZT degradation (Fig. S4), as uncatalyzed PMS fails to produce sufficient reactive species under ambient conditions.¹³ These findings highlight the UV/PMS system as an environmentally promising approach for antiviral drug removal.

Fig. S3(b) illustrates the effects of oxidant concentration on AZT degradation in the UV/PMS system, showing a linear relationship between reaction time and ln(C/C₀) (R² > 0.99). As the PMS dosage increased from 1 to 10 mg L⁻¹, the AZT removal efficiency increased from 58.9% to 93.0%, accompanied by an increase in the first-order reaction rate constant from 0.0159 to 0.0447 min⁻¹. Given the inherent UV-induced degradation of AZT, UV activation of HSO₅⁻ breaks its O–O peroxygen bond to generate SO₄˙⁻ and OH˙ (eqn (1)). Subsequently, SO₄˙⁻ reacts with H₂O/OH⁻ (eqn (2) and (3)),¹⁴ producing additional OH˙ that collectively promote AZT degradation. At a fixed light energy (500 mJ cm⁻²), higher PMS concentrations enhance the generation of strongly oxidizing radicals, thereby accelerating AZT degradation. Unlike previous observations of inhibition at high oxidant levels,¹⁵ this study found no such effect, likely because the PMS dosage remained below the inhibition threshold. Excessive PMS could lead to self-quenching of SO₄˙⁻ (eqn (4))¹⁶ or reaction with PMS to form weakly oxidizing SO₅˙⁻ (eqn (5) and (6)), reducing radical availability for AZT degradation.¹⁷ These findings highlight the need for optimizing PMS dosage in practical applications to balance radical generation and prevent rate-limiting side reactions.

HSO₅⁻ → SO₄˙⁻ + OH˙ (1)

SO₄˙⁻ + H₂O → OH˙ + HSO₄⁻, k = 6.6 × 10² M⁻¹ s⁻¹ (2)

SO₄˙⁻ + OH⁻ → SO₄²⁻ + OH˙, k = 1.4 × 10⁷ M⁻¹ s⁻¹ (3)

SO₄˙⁻ + SO₄˙⁻ → S₂O₈²⁻ (4)

SO₄˙⁻ + PMS → HSO₄⁻ + SO₅˙⁻, k < 10⁵ M⁻¹ s⁻¹ (5)

OH˙ + PMS → H₂O + SO₅˙⁻, k = 1.7 × 10⁷ M⁻¹ s⁻¹ (6)

Building on the experimental results of AZT degradation in the UV/PMS system, a four-level, two-factor orthogonal array (L16(4²)) was used to determine optimal PMS concentrations (1, 3, 6, 10 mg L⁻¹) and UV doses (200, 300, 400, 500 mJ cm⁻², 0.141 mW cm⁻² light intensity) for subsequent exploratory experiments, with the AZT degradation rate as the evaluation index. Table S5 summarizes the orthogonal design and results, revealing positive correlations between PMS concentration/UV dose and degradation efficiency. Among the two factors, PMS concentration exhibited a more significant impact than UV dose. To reduce resource waste and potential environmental risks, excessively high PMS concentrations and UV doses were deliberately avoided during parameter selection for follow-up experiments. Considering the marginal improvement in the degradation rate from 6 mg L⁻¹ to 10 mg L⁻¹ PMS and the standard UV dose range (400–500 mJ cm⁻²) in practical water treatment, the following parameters were selected: AZT concentration = 0.2 mg L⁻¹, PMS concentration = 6 mg L⁻¹, and UV dose = 500 mJ cm⁻². This configuration achieves a balance between degradation efficacy and practical feasibility, ensuring compatibility with operational constraints in real-world water treatment systems.

3.1.2 Influencing factors in the degradation of AZT by UV/PMS.
(1) pH values. The effect of pH on the degradation kinetics of AZT in the UV/PMS system was examined under controlled pH conditions (i.e., pH = 3.1, 4.0, 5.2, 6.1, 7.2, 8.1, 9.0, 10.2 and 11.2) using ultrapure water as the reaction medium (Fig. 1). Clearly, the results showed that acidic pH values were more conducive to the degradation of AZT, since the removal efficiencies remained above 90.5% in the pH range of 3.1–6.1, with the highest removal efficiency of 93.7% achieved at pH 5.2. The best degradation performance was observed at pH 5.2, where the rate constant k reached 0.0486 min⁻¹, followed by pH 4.0 (k = 0.0480 min⁻¹) and pH 6.1 (k = 0.0450 min⁻¹). In contrast, degradation efficiency gradually declined with increasing pH beyond 7.0: the removal efficiency dropped to 89.03% (k = 0.0384 min⁻¹) at neutral pH 7.2, further decreased to 82.69% (k = 0.0319 min⁻¹) at pH 9.0, and finally fell to 71.5% (k = 0.0220 min⁻¹) at pH 11.2. Moreover, the rate constant k decreased from 0.0486 min⁻¹ at pH 5.2 to 0.0220 min⁻¹ at pH 11.2, confirming that alkaline conditions significantly inhibited AZT degradation in the UV/PMS system.

Fig. 1 Semilogarithmic linear fitting of the degradation effect of AZT in the UV/PMS system versus the reaction time under different pH conditions (reaction conditions: PMS = 6 mg L⁻¹, AZT = 0.2 mg L⁻¹, UV = 500 mJ cm⁻²).

These results indicate that acidic pH enhances AZT degradation in the UV/PMS system, whereas alkaline conditions inhibit reaction kinetics, potentially attributed to pH-dependent radical speciation and PMS activation pathways.¹⁸

The observed pH-dependent degradation behavior of AZT in the UV/PMS system can be attributed to the following mechanistic considerations: (1) under acidic conditions, elevated redox potentials of HSO₅⁻ (2.6–3.1 V) and SO₄˙⁻,¹⁹ coupled with enhanced SO₄˙⁻ generation,²⁰ synergistically improve oxidative capacity.²¹ (2) Near-neutral conditions (pH 7.2) promote PMS self-decomposition and SO₄˙⁻-to-OH˙ conversion,²² reducing degradation efficiency. (3) Alkaline conditions (pH > 8) exhibit markedly reduced degradation efficiency compared to acidic environments, with even lower efficiency than neutral conditions. This phenomenon is caused by pH-dependent radical conversion: the redox potential decreases from 2.6–3.1 V (SO₄˙⁻ in acidic media) to 1.8 V (OH˙ in alkaline media), significantly reducing oxidative capability. Comparative analysis of UV/PMS and UV/H₂O₂ systems (Fig. S3) further supports this degradation pattern. (4) Radical selectivity significantly affects reaction kinetics. While OH˙ undergoes non-selective oxidation via dehydrogenation and addition pathways, SO₄˙⁻ primarily undergoes electron transfer reactions. The conversion of SO₄˙⁻ to OH˙ induces competitive consumption of oxidants by reaction intermediates, thereby impeding AZT degradation. This mechanistic interpretation aligned with persulfate-based antibiotic removal studies.²³ (5) Additionally, the quenching effect of phosphate buffer salts on free radicals should be considered. Although the second-order rate constants for H₂PO₄⁻/HPO₄²⁻ reacting with SO₄˙⁻ and OH˙ (10⁴–10⁶ M⁻¹ s⁻¹) are relatively low, their mM-level concentrations cause significant competitive radical consumption relative to μM-level AZT. Acidic conditions (pH < pK_a1 = 7.2) stabilize H₂PO₄⁻ as the dominant phosphate species. As evidenced by eqn (7)–(10),²⁴ H₂PO₄⁻ demonstrates lower reactivity toward radical quenching (k(SO₄˙⁻) = 1.2 × 10⁴ M⁻¹ s⁻¹) compared to HPO₄²⁻ (k(SO₄˙⁻) = 1.6 × 10⁶ M⁻¹ s⁻¹). This reduced quenching capacity under acidic conditions consequently minimizes phosphate-induced inhibition on AZT degradation.²⁴

H₂PO₄⁻ + SO₄˙⁻ → HSO₄²⁻ + HPO₄˙⁻, k = 5.0 × 10⁴ M⁻¹ s⁻¹ (7)

H₂PO₄⁻ + OH˙ → H₂O + HPO₄˙⁻, k = 2.0 × 10⁴ M⁻¹ s⁻¹ (8)

HPO₄²⁻ + SO₄˙⁻ → SO₄²⁻ + HPO₄˙⁻, k = 1.2 × 10⁵ M⁻¹ s⁻¹ (9)

HPO₄²⁻ + OH˙ → OH⁻ + HPO₄˙⁻, k = 1.5 × 10⁵ M⁻¹ s⁻¹ (10)

Furthermore, the phosphate buffer is used in conjunction with the parallel-beam apparatus for UV irradiation (254 nm), and it exhibits excellent thermal and photostability: under UV irradiation, it does not undergo photolysis (thus preventing the generation of unintended free radicals or introduction of contaminants), and at room temperature, it does not undergo non-target reactions with PMS or AZT.²⁵ This property avoids data deviations caused by fluctuations in the buffer's own characteristics and meets the reproducibility requirements of scientific research.

Fig. S5 reflects the change of AZT absorbance with wavelength λ at different pH values, and the absorbance of AZT at pH = 3.1, 4.0, 5.2, 6.1, 7.2, 8.1, 9.0, 10.2, 11.1 gradually decreases, so AZT at acidic pH is better in degradation.

(2) Inorganic ion. The inhibitory effect of HCO₃⁻ (0–10 mM) on AZT degradation in the UV/PMS system was systematically evaluated (Fig. 2a). AZT removal efficiency decreased progressively from 89.03% to 68.57% with increasing HCO₃⁻ concentrations (0–10 mM), accompanied by corresponding reductions in pseudo-first-order rate constants from 0.0382 to 0.0194 min⁻¹. HCO₃⁻ acts as a radical quencher by reacting with SO₄˙⁻ and OH˙ (eqn (11) and (12)), thereby suppressing AZT degradation. Kinetic analysis based on eqn (11) and (12) quantified the second-order rate constants as 1.6 × 10⁶ M⁻¹ s⁻¹ (SO₄˙⁻) and 8.5 × 10⁶ M⁻¹ s⁻¹ (OH˙), consistent with previous observations in sulfadimethoxine degradation systems.²⁶

SO₄˙⁻ + HCO₃⁻ → SO₄²⁻ + HCO₃˙ + H⁺, k = 1.6 × 10⁶ M⁻¹ s⁻¹ (11)

OH˙ + HCO₃⁻ → H₂O + CO₃˙⁻, k = 8.5 × 10⁶ M⁻¹ s⁻¹ (12)

Fig. 2 Semilogarithmic linear fitting of azidothymidine (AZT) degradation efficiency in the UV/peroxymonosulfate (UV/PMS) system as a function of reaction time at different concentrations of (a) HCO₃⁻, (b) NO₃⁻, (c) NO₂⁻, and (d) humic acid (HA). The symbols and lines in the figure represent the experimental data and the linear least-squares regression fittings of the experimental data, respectively.

NO₃⁻, a photosensitive ion widely present in natural waters,²⁷ was investigated for its influence on AZT degradation in the UV/PMS system through five concentration gradients (0, 0.1, 0.5, 1, and 3 mM). As illustrated in Fig. 2(b), the AZT degradation rate exhibited a modest concentration-dependent enhancement, increasing from 89.03% (0 mM NO₃⁻) to 90.85% (3 mM NO₃⁻). First-order rate constants showed minimal variation (0.0387–0.0414 min⁻¹) across tested concentrations, confirming the limited catalytic role of NO₃⁻. This behavior aligns with UV-induced generation of reactive species from NO₃⁻ (eqn (13)–(16)),²⁸ where derived radicals (e.g., OH, NO₂˙) may oxidize AZT through bond cleavage. Studies have reported that nitrate ions (NO₃⁻) exhibit a linear relationship between absorbance and concentration at 220 nm (R² > 0.999), with a limit of detection (LOD) of 1 μmol L⁻¹. This confirms that NO₃⁻ is capable of absorbing UV light under the experimental conditions, thereby forming the excited state [NO₃⁻]*—a critical prerequisite for the UV-induced generation of reactive species mentioned earlier (eqn (13)–(16)). However, the low rate constant for SO₄˙⁻ quenching by NO₃⁻ (5.0 × 10⁴ M⁻¹ s⁻¹) (eqn (17)) indicates negligible interference with dominant degradation pathways.²⁹ While the resulting NO₂⁻ (oxidation potential = 2.3 V) retains oxidative capacity, its contribution to AZT degradation remained negligible compared to dominant radical-mediated mechanisms.³⁰

NO₃⁻ + hv → [NO₃⁻]* + O(³P) (13)

[NO₃⁻] → NO₂⁻ + O(³P) (14)

[NO₃⁻] → NO₂˙ + O˙⁻ (15)

O˙⁻ + H₂O → OH⁻ + OH˙ (16)

SO₄˙⁻ + NO₃⁻ → SO₄²⁻ + NO₃˙, k = 5 × 10⁴ M⁻¹ s⁻¹ (17)

The inhibitory effect of nitrite (NO₂⁻) on AZT degradation in the UV/PMS system was evaluated by varying NO₂⁻ concentrations (0–3 mM). As shown in Fig. 2(c), AZT removal efficiency decreased progressively from 89.03% (0 mM NO₂⁻) to 46.21% (3 mM NO₂⁻), with corresponding first-order rate constants declining from 0.0387 to 0.0101 min⁻¹. Despite NO₂⁻ generating reactive species under UV irradiation (eqn (18)),³¹ their low redox potential (1.03 V) rendered them ineffective for AZT oxidation.³² The suppression primarily arose from NO₂⁻ quenching SO₄˙⁻ and OH˙ via rapid second-order reactions (eqn (19) and (20)),³³ with rate constants of 1.0 × 10¹⁰ M⁻¹ s⁻¹ and 8.8 × 10⁸ M⁻¹ s⁻¹, respectively, yielding weakly oxidative NO₃⁻. Additionally, NO₂⁻ exhibited a stronger internal filtration effect than NO₃⁻,²⁷ reducing UV absorption by PMS. These combined mechanisms—radical quenching and UV shielding—account for the observed inhibition of AZT degradation in NO₂⁻-amended systems.

NO₂⁻ + hv → NO₂˙ + e⁻ (18)

OH˙ + NO₂⁻ → NO₂˙ + OH⁻, k = 1.2 × 10¹⁰ M⁻¹ s⁻¹ (19)

SO₄˙⁻ + NO₂⁻ → SO₄²⁻ + NO₂˙, k = 9.8 × 10⁸ M⁻¹ s⁻¹ (20)

To elucidate the role of natural organic matter (NOM) in AZT degradation by the UV/PMS system, humic acid (HA) was selected as a representative NOM. As demonstrated in Fig. 2(d), increasing HA concentrations (0–10 mg L⁻¹) progressively inhibited AZT degradation, reducing the first-order rate constant from 0.0384 min⁻¹ (0 mg L⁻¹ HA) to 0.0073 min⁻¹ (10 mg L⁻¹ HA), accompanied by a decline in removal efficiency from 89.03% to 36.48%. This suppression arose from HA's competition with AZT for reactive species, with quenching capacity increasing proportionally to HA concentration.³⁴ HA also reduced UV transmittance through light absorption, inhibiting PMS activation and direct photolysis of AZT.

3.1.3 Degradation efficiency of AZT in water from the Yellow River diversion reservoir by UV/PMS technology. The degradation efficiency of AZT in Yellow River-derived reservoir water is compared in Fig. 3. When compared with ultrapure water, it can be observed that Yellow River-derived reservoir water exerts a certain inhibitory effect on AZT degradation. Under the same molar concentration of oxidants (PMS = 6 mg L⁻¹, NaClO = 2.98 mg L⁻¹, H₂O₂ = 1.36 mg L⁻¹), the degradation efficiencies of AZT in Yellow River-derived reservoir water by UV/PMS, UV/NaClO, and UV/H₂O₂ systems were reduced by 26.85%, 31.2%, and 32.9%, respectively, compared to those in ultrapure water.

Fig. 3 The degradation efficiency of AZT in Yellow River-derived reservoir water by the UV/PMS system under different PMS concentrations and by various oxidation systems (NaClO = 2.98 mg L⁻¹, H₂O₂ = 1.36 mg L⁻¹, corresponding to the molar concentration of PMS = 6 mg L⁻¹).

When the PMS concentration in Yellow River-derived reservoir water was increased to 15 and 25 mg L⁻¹, the removal efficiency of AZT by the UV/PMS technology improved to 70.04% and 81.03%, with the corresponding pseudo-first-order rate constants of 0.0208 and 0.0267 min⁻¹, respectively. The relatively poor degradation performance of AZT in Yellow River-derived reservoir water is attributed to multiple factors: as shown in Table S6, the presence of inorganic compounds such as chloride in Yellow River-derived reservoir water consumes strong oxidizing radicals, thereby reducing the degradation efficiency of AZT. Additionally, the pH of Yellow River-derived reservoir water is 8.27 (alkaline), and the conclusion from section 3.1.2 regarding the effect of pH indicates that alkaline conditions inhibit degradation, thus making the pH of Yellow River-derived reservoir water unfavorable for AZT removal by the UV/PMS technology. Furthermore, compared to ultrapure water, Yellow River-derived reservoir water contains more impurities, resulting in higher turbidity and lower light transmittance, which affects the absorption of UV radiation energy by both PMS and AZT, thereby reducing the activation efficiency of PMS and the photolysis of AZT.³⁵ Furthermore, the UV₂₅₄ value in Table S6 is indicative of aromatic compounds containing C=C double bonds and C=O double bonds, as well as humic-like macromolecular organic substances in the solution. The presence of these substances competes with AZT for free radicals in the system.

3.2 Kinetic study on the degradation of AZT by UV/PMS

AZT degradation in the UV/PMS system proceeds via UV photolysis and radical-mediated oxidation pathways. To identify dominant radicals, methanol (MeOH) and tert-butyl alcohol (TBA) were employed as selective quenchers. MeOH, an α-hydrogen-containing alcohol, scavenged OH˙ (k = 9.7 × 10⁸ M⁻¹ s⁻¹) and SO₄˙⁻ (k = 1.1 × 10⁷ M⁻¹ s⁻¹). In contrast, tert-butyl TBA preferentially quenched OH˙ (k = 3.8–7.6 × 10⁸ M⁻¹ s⁻¹) with minimal reactivity toward SO₄˙⁻ (k = 4.0–9.1 × 10⁵ M⁻¹ s⁻¹).³⁶ As shown in Fig. 4, increasing quencher concentrations (0–4 mM) progressively inhibited AZT degradation. This suppression reflects competition between quenchers and radicals for oxidative species. At 0.12 mM (3 × PMS concentration), MeOH and TBA reduced AZT degradation by 11.60% and 4.73%, respectively, confirming contributions from both OH˙ and SO₄˙⁻. At 4 mM (100 × PMS concentration), MeOH and TBA further suppressed degradation by 25.37% and 18.71%, respectively. The enhanced inhibition by TBA at higher concentrations arises from its partial quenching of SO₄˙⁻.³⁶

Fig. 4 Quasi-first-order kinetic analysis of AZT degradation: (a) fitting curves of AZT degradation by UV/PMS in the presence of MeOH, TBA, and IPA; (b) reaction rate comparison of AZT, NB, and ATZ in UV/H₂O₂ and UV/PMS systems. Symbols and lines in the figure represent experimental data and their linear regression fittings, respectively. This revision clearly ties each subplot label (a) and (b) to its specific content, ensuring the figure’s interpretation is both clear and academically rigorous.

To further elucidate reactive species roles, the contributions of superoxide radicals (O₂˙⁻) and singlet oxygen (¹O₂) were excluded. Given that O₂˙⁻ has a low redox potential,³⁷ its participation was deemed negligible. Isopropyl alcohol (IPA) quenching experiments confirmed OH˙ and SO₄˙⁻ as the dominant species, with IPA exhibiting high reactivity toward these radicals (k = 2.8 × 10⁹ and 6.0 × 10⁷ M⁻¹ s⁻¹ for OH˙ and SO₄˙⁻, respectively) but minimal interaction with ¹O₂.³⁸ As shown in Fig. 4(a), 4 mM MeOH and IPA reduced AZT degradation by 25.37% and 29.90%, respectively, ruling out significant ¹O₂ involvement. The stronger inhibition by IPA (29.90%) compared to MeOH (25.37%) correlates with its higher SO₄˙⁻ quenching rate (8.2 × 10⁷vs. 1.1 × 10⁷ M⁻¹ s⁻¹) and MeOH's residual radical activity. These results establish OH˙ and SO₄˙⁻ as primary oxidative agents, with negligible input from other species.

Nitrobenzene (NB) and ATZ were employed as probe compounds to quantify AZT's second-order rate constants with OH˙ and SO₄˙⁻. NB, stable under UV irradiation,³⁹ served as an OH˙-specific probe due to its preferential reaction with OH˙ (k = 3.9 × 10⁹ M⁻¹ s⁻¹) over SO₄˙⁻ (k < 1.0 × 10⁶ M⁻¹ s⁻¹). ATZ, exhibiting comparable reactivity with both radicals (k = 2.6 × 10⁹ M⁻¹ s⁻¹),⁴⁰ enabled simultaneous evaluation of OH˙/SO₄˙⁻ contributions. Comparative degradation rates of AZT, NB, and ATZ provided mechanistic resolution of radical-specific pathways.

During the experiments, 0.75 μM of AZT and NB were added to the UV/H₂O₂ and UV/PMS systems to determine the degradation rates, and the results are shown in Fig. 4(b). The degradation patterns were in accordance with the proposed first-order kinetic model. Second-order rate constants for AZT with OH˙ and SO₄˙⁻ were derived from eqn (21)–(24). In eqn (21)–(24), the value of k_NB(UV) can be ignored because NB is difficult to be directly degraded by UV. Here, and refer to the secondary reaction rate constants of compound A with OH˙ and SO₄˙⁻, respectively; k_A refers to the proposed first-order rate constant of A in each system; [OH˙] and [SO₄˙⁻] denote the concentrations of OH˙ and SO₄˙⁻ at the steady state, respectively.

As demonstrated in section 3.1, the PMS system alone exhibited limited AZT degradation, underscoring the critical roles of UV irradiation and radical-mediated oxidation. Quenching experiments identified HO˙ and SO₄˙⁻ as critical reactive species, with their relative contributions quantified through second-order rate constants derived in section 3.2. Probe compounds ATZ and NB (20% of target concentration) were introduced into the UV/PMS system under optimized oxidant conditions (6 mg L⁻¹, determined via orthogonal design). Time-resolved concentration profiles of AZT, ATZ, and NB enabled calculation of primary rate constants, which were applied to eqn (23) and (24) to determine steady-state radical concentrations ([OH˙] = 1.2 × 10⁻¹³ M; [SO₄˙⁻] = 8.7 × 10⁻¹⁴ M). Subsequent application of eqn (22) (Fig. S6) revealed AZT degradation contributions as follows: UV photolysis (51.03%), SO₄˙⁻ (30.85%), and OH˙ (18.11%). The dominance of UV photolysis aligns with AZT's pronounced photosensitivity, while SO₄˙⁻, a selective oxidant with high redox potential, emerged as the predominant radical contributor.

3.3 Speculation on the partial degradation pathway of AZT

As demonstrated in previous studies,⁴¹ the highest occupied molecular orbital (HOMO) energy determined a molecule's propensity to donate electrons during electrophilic reactions, while the lowest unoccupied molecular orbital (LUMO) energy reflected its capacity to accept electrons in nucleophilic processes. These frontier molecular orbitals elucidated the electronic reactivity of molecules. Electrostatic potential analysis (Fig. 5) indicated the electron-rich regions in AZT's thymine moiety, making it vulnerable to electrophilic attack by OH˙ and SO₄˙⁻. AZT's preferential reactivity toward radicals stems from its molecular structure and frontier orbital properties: the azide group's N8 (28.584%) and N9 (42.467%) exhibit high LUMO contributions (Fig. 5) with a low LUMO energy (−0.18 eV), making their N–N bonds susceptible to nucleophilic attack by SO₄˙⁻. The C17 C=C bond, with the highest HOMO contribution (24.032%) and a relatively high HOMO energy (−0.35 eV), is prone to electrophilic attack by OH˙. Oxygen atoms in the sugar moiety (O₂, O₃) contribute minimally to the HOMO (0.010–0.129%) but stabilize electron distribution via conjugation, lowering reaction barriers. Compared to probes (NBT: HOMO = −0.42 eV; ATZ: HOMO = −0.40 eV), AZT's higher HOMO energy enhances radical reactivity.⁴²

Fig. 5 Atomic contributions to the frontier molecular orbitals of AZT, along with the net charge distribution and LUMO and HOMO plots of AZT.

The detected parent substances, fragment ions, and degradation products are presented in Table S7. Combined with the results of semi-quantitative analysis based on the signal intensity (abund, counts) of ultra-performance liquid chromatography/quadrupole time-of-flight mass spectrometry (UPLC/Q-TOF MS)—where signal intensity is positively correlated with relative content under the same experimental conditions—which are provided in Table S8, the following observations were made: AZT-1 (hydrogenated derivative), AZT-2 (sodium-bound derivative), and AZT-3 (potassium-bound derivative) all eluted at 5.119 min, with signal intensities in the order of AZT-1 (247 961 counts at 0 min → 34 438 counts at 60 min) > AZT-2 (28 658 counts at 0 min → 4073 counts at 60 min) > AZT-3 (3395 counts at 0 min → 529 counts at 60 min). This is consistent with previous reports that zidovudine (AZT) primarily exists in protonated^43–45 and sodiated forms,^46,47 and the parent compound continuously degraded as the reaction proceeded. The cleavage fragment DP127 (thymine) was not detected at 0 min but increased to 1356 counts at 60 min, serving as an early characteristic product of AZT structural cleavage.^43,45,48

Among the four major degradation products: DP239 (generated by the loss of N₂ from AZT under free radical attack) reached a peak at 30 min (1239 counts) and then decreased to 979 counts; DP223 (formed via azide group elimination) peaked at 30 min (19 188 counts) followed by a sharp decline; DP185 (produced through the cleavage of the C₄H₃ON group) doubled at 60 min (1338 counts) compared to 30 min (539 counts); DP125 (a terminal thymine product) also doubled at 60 min (2676 counts) relative to 30 min (1338 counts). These degradation products exhibited a “mutual growth and decline” trend relative to the signal of the parent compound, confirming the continuity of the AZT degradation pathway.

The proposed degradation pathways (Fig. 6) integrated reactive site analysis and product identification. Pathway P1 involved C11–N5 bond cleavage via radical/UV-induced electron transfer, yielding thymine (DP125). This aligns with the HOMO contributions of C11 (2.044%) and N5 (18.638%) (Fig. 5), indicating their susceptibility to electrophilic attack. These pathways constituted key routes in the UV/PMS-mediated AZT degradation network.

Fig. 6 Possible degradation pathways of AZT.

Pathway P2 initiated with N₂ elimination from the azide group, forming an intermediate that undergoed intramolecular C–H insertion to yield DP239, a hallmark of azide degradation. The N=N bond cleavage arises from both the intrinsic photosensitivity of azide groups and the electronic structure revealed by frontier orbital analysis. As shown in Fig. 5, the LUMO contributions of N6, N8, and N9 atoms were calculated as 11.100%, 28.584%, and 42.467%, respectively. The high LUMO contributions of N8 (28.584%) and N9 (42.467%) indicate electron-deficient character, making them susceptible to nucleophilic attack. This mechanistic interpretation aligns with previous observations of DP239 formation during photolytic degradation of AZT and related antiviral compounds.⁴⁶ The metastable DP239 undergoes subsequent radical-mediated transformations, including ring-opening or double bond cleavage.

Further − NH elimination from DP239 yields DP223, consistent with degradation products identified under hydrolytic conditions.⁴⁵ DP185 forms via a four-step mechanism: C17 demethylation by electrophilic radicals, C15=C17 hydroxylation, N5–C16/C15–C17 bond cleavage, and oxidation of the C15 hydroxyl to an aldehyde.

Notably, the degradation pathway yields several transformation products with potential environmental persistence. Chief among them is DP125 (thymine), a stable aromatic nucleobase exhibiting recalcitrance toward further oxidative attack by SO₄˙⁻ and HO˙ radicals due to its electron-deficient ring structure.⁴⁷ To mitigate this risk and achieve complete detoxification, a sequential treatment process is recommended, wherein UV/PMS serves as a primary step to dismantle the parent compound, followed by a biological polishing stage to mineralize these amenable yet persistent oxidation by-products.⁴⁸

3.4 Toxicity changes during degradation of AZT by UV/PMS technology

3.4.1 Acute toxicity analysis based on Vibrio fischeri. The Vibrio fischeri bioluminescence inhibition assay (VFBIA) is widely employed for toxicity assessment in diverse water bodies due to its advantages of short test duration, high sensitivity, cost-effectiveness, and ease of operation.^49–51 This section investigated the overall toxicity evolution during the degradation of AZT by the UV/PMS process, utilizing luminescent bacteria for comprehensive acute toxicity testing. Acute toxicity tests with luminescent bacteria were conducted at 0, 15, 30, 45, and 60 min (Fig. 7). Initial luminescence intensity (2.5 × 10⁶ at time 0) served as the control. During UV/PMS treatment (0–15 min), luminescence sharply decreased, indicating high toxicity from reaction intermediates. Subsequent intensity recovery (15–60 min) correlated with the degradation of toxic intermediates, demonstrating UV/PMS efficacy in both pollutant removal and risk mitigation. These findings provided critical insights for scaling UV/PMS technology in real wastewater treatment.

Fig. 7 Luminous intensity of the system.

3.4.2 Toxicity prediction based on ECOSAR software. The ECOSAR software calculates LC₅₀, EC₅₀, and ChV values via functional group-based algorithms, selecting the lowest predictions for conservative toxicity assessment. Toxicity categories (Table S9) are defined as: very toxic (LC₅₀/EC₅₀/ChV ≤ 1 mg L⁻¹), toxic (1–10 mg L⁻¹), harmful (10–100 mg L⁻¹), and non-harmful (>100 mg L⁻¹).⁵²

The log K_ow values, acute toxicity, and chronic toxicity of AZT and its degradation products, as predicted by the software, are summarized in Table S10. The log K_ow value, representing the n-octanol–water partition coefficient, is a key parameter for assessing compound hydrophobicity and risk,⁵³ calculated viaeqn (25). Higher values suggest greater organic phase affinity and bioaccumulation potential. Symbols in Table S10: (*) indicates limited solubility for effect measurement; (!) denotes acute-to-chronic ratio methodology (ECOSAR manual).

Table S10 reveals that AZT and its degradation products exhibit log K_ow values < 0, indicating low bioaccumulation potential. Acute toxicity prioritizes green algae as the most sensitive species: AZT and most products (excluding DP185) are very toxic (LC₅₀/EC₅₀ ≤ 1 mg L⁻¹). In contrast, AZT shows no acute toxicity to fish or daphnia, whereas DP223 is harmful to fish. Chronic toxicity confirms green algae sensitivity, with AZT and DP185 posing minimal risks to fish/daphnia. DP239(a)/(b) and DP223 are harmful and toxic to fish/daphnia, respectively, while DP125 shows chronic toxicity to both. Chronic toxicity consistently exceeds acute levels.

Although ECOSAR predictions indicate that AZT and its degradation products exhibit high acute toxicity to green algae (LC₅₀ ≤ 1 mg L⁻¹), experimental toxicity tests using Vibrio fischeri showed an initial increase in toxicity during the early degradation stage, followed by a gradual decrease. Such discrepancies may arise from the fact that ECOSAR, based on QSAR models, fails to fully account for the dynamic changes in degradation intermediates and their bioavailability. In this study, the log K_ow values of AZT and its degradation products are all less than 0, indicating low bioaccumulation potential. However, ECOSAR predictions still suggest high toxicity to green algae, suggesting that beyond hydrophobicity, other molecular properties (e.g., reactivity, molecular size, functional groups) may also influence their ecotoxicity. A study by Wang et al. demonstrated that the cytotoxicity of BPA analogs is significantly negatively correlated with log K_ow (R² = 0.69), indicating that stronger hydrophobicity facilitates membrane penetration, thereby resulting in greater toxicity.⁵⁴ Despite the low log K_ow values of AZT and its products, the reactive functional groups in their structures, such as nitrogen-containing heterocycles and aldehyde groups, may induce toxicity through alternative mechanisms (e.g., oxidative stress, DNA damage).

Despite low bioaccumulation, AZT and intermediates retain ecotoxicity. Pathway P1/P2 products (e.g., DP125, DP223) are more toxic than AZT, whereas pathway P3 (DP185) reduces toxicity. Further studies are needed to identify terminal products and long-term ecological effects.

4. Conclusions

This study systematically investigated the degradation of AZT via PMS advanced oxidation. Three principal findings emerged:

(1) UV/PMS demonstrated superior AZT removal efficiency (89.03% at 59.23 min with 6 mg L⁻¹ PMS) compared to sole UV irradiation (48.12%) or PMS alone. Reaction kinetics followed pseudo-first-order behavior, with degradation enhanced by acidic conditions, elevated PMS concentrations, and nitrate ions. Conversely, common water constituents (Cl⁻, HCO₃⁻, NO₂⁻, humic acid) exhibited concentration-dependent inhibitory effects. In real water matrices such as Yellow River-derived reservoir water, the removal efficiency of AZT decreased by 26.85% under 6 mg L⁻¹ PMS compared to that in ultrapure water, reflecting the inhibitory effect of complex constituents in real water matrices.

(2) Radical quenching experiments and probe-based competition kinetics revealed dual oxidative contributions from sulfate (SO₄˙⁻) and hydroxyl (OH˙) radicals, with calculated second-order rate constants of 5.47 × 10⁹ M⁻¹ s⁻¹ and 6.64 × 10⁹ M⁻¹ s⁻¹, respectively. Quantitative analysis attributed 51.03% of AZT degradation to direct photolysis, with radical-mediated oxidation accounting for 48.97% (30.85% SO₄˙⁻, 18.11% OH˙).

(3) Four degradation products were identified through UPLC-QTOF-MS analysis, with proposed pathways involving azide group de-N₂, thymine formation, demethylation, and C=C bond addition. Vibrio fischeri bioluminescence assays and ECOSAR predictions indicated transient toxicity elevation during early degradation phases, followed by gradual detoxification. Notably, most intermediates exhibited higher acute/chronic toxicity than the parent AZT compound, emphasizing the necessity for process optimization to minimize ecological risks.

These findings establish UV/PMS as an effective strategy for antiviral drug remediation while highlighting critical considerations for practical implementation in complex aqueous matrices.

Author contributions

Yiran Jia: methodology and original draft preparation; Zhenqi Du: validation, supervision, reviewing, and funding acquisition; Xiaohong Wang: investigation; Zhangbin Pan: reviewing; Baozhen Liu: data curation and editing; Guifang Li: investigation; Yonglei Wang: funding acquisition; Ruibao Jia: funding acquisition.

Conflicts of interest

There are no conflicts to declare.

Data availability

The data supporting this article have been included as part of the supplementary information (SI).

Supplementary information: Supplementary data related to this article is available in this appendix. See DOI: https://doi.org/10.1039/d5ew00648a.

Acknowledgements

This study was supported by the National Natural Science Foundation of China (Grant No. 52500007), the National Key Research and Development Program of China (2022YFC3203705, 2022YFC3203704), the Shandong Provincial Natural Science Foundation (Grant No. ZR2025QC1147), the Key R&D Program of Shandong Province (Grant No. 2023TZXD019), and the Postdoctoral Innovation Project of Shandong Province (SDCX-ZG-202303055).

References

1. D. R. Tompa, A. Immanuel, S. Srikanth and S. Kadhirvel, Trends and strategies to combat viral infections: A review on FDA approved antiviral drugs, Int. J. Biol. Macromol., 2021, 172, 524–541.
2. S. Ncube, L. M. Madikizela, L. Chimuka and M. M. Nindi, Environmental fate and ecotoxicological effects of antiretrovirals: a current global status and future perspectives, Water Res., 2018, 145, 231–247.
3. R. Wang, J. Luo, C. Li, J. Chen and N. J. Zhu, Antiviral drugs in wastewater are on the rise as emerging contaminants: A comprehensive review of spatiotemporal characteristics, removal technologies and environmental risks, J. Hazard. Mater., 2023, 457, 131694.
4. N. Morin-Crini, E. Lichtfouse, M. Fourmentin, A. R. L. Ribeiro, C. Noutsopoulos and F. Mapelli, et al., Removal of emerging contaminants from wastewater using advanced treatments. A review, Environ. Chem. Lett., 2022, 20(2), 1333–1375.
5. C. Prasse, M. P. Schlüsener, R. Schulz and T. A. Ternes, Antiviral drugs in wastewater and surface waters: a new pharmaceutical class of environmental relevance?, Environ. Sci. Technol., 2010, 44(5), 1728–1735.
6. L. Yao, Z.-Y. Chen, W.-Y. Dou, Z.-K. Yao, X.-C. Duan and Z.-F. Chen, et al., Occurrence, removal and mass loads of antiviral drugs in seven wastewater treatment plants with various treatment processes, Water Res., 2021, 207, 117803.
7. X. Chen, J. Cui, X. Yang, T. Yuan, J. Ge and M. Ma, et al., Enhanced removal of phenolic contaminants with permanganate oxidation mediated by SiO₂@ TEMPO: Performance, transformation pathway, and selectivity, Chem. Eng. J., 2025, 503, 158210.
8. T. Simetić, J. Nikić, M. Kuč, D. Tamindžija, A. Tubić and J. Agbaba, et al., New insight into the degradation of sunscreen agents in water treatment using UV-Driven advanced oxidation processes, Processes, 2024, 12(6), 1156.
9. X. Wang, L. Zhou, Y. Liu, K. Zhang and G. J. A. E. Xiu, Investigating the photolysis of NO₂ and influencing factors by using a DFT/TD-DFT method, Atmos. Environ., 2020, 230, 117559.
10. W. Ren, Z. Zhou, Y. Zhu, L.-M. Jiang, H. Wei and T. Niu, et al., Effect of sulfate radical oxidation on disintegration of waste activated sludge, Int. Biodeterior. Biodegrad., 2015, 104, 384–390.
11. K. Guo, Z. Wu, C. Chen and J. J. Fang, UV/chlorine process: an efficient advanced oxidation process with multiple radicals and functions in water treatment, Acc. Chem. Res., 2022, 55(3), 286–297.
12. K. H. Lee, J. Y. Kim, I. A. Khan, J. W. Lee and J.-O. Kim, Investigation of degradation and transformation product formation during the UV/H₂O₂-based oxidation process for chemicals of emerging concern, J. Environ. Chem. Eng., 2023, 11(5), 110340.
13. X. Li, X. Wen, J. Lang, Y. Wei, J. Miao and X. Zhang, et al., CoN1O2 single-atom catalyst for efficient peroxymonosulfate activation and selective Cobalt (IV)= O generation, Angew. Chem., 2023, 135(27), e202303267.
14. H. Herrmann, A. Reese and R. Zellner, Time-resolved UV/VIS diode array absorption spectroscopy of SOx−(x= 3, 4, 5) radical anions in aqueous solution, J. Mol. Struct., 1995, 348, 183–186.
15. Y.-J. Shih, W. N. Putra, Y.-H. Huang and J.-C. Tsai, Mineralization and deflourization of 2, 2, 3, 3-tetrafluoro-1-propanol (TFP) by UV/persulfate oxidation and sequential adsorption, Chemosphere, 2012, 89(10), 1262–1266.
16. C. Liang and H.-W. Su, Identification of sulfate and hydroxyl radicals in thermally activated persulfate, Ind. Eng. Chem. Res., 2009, 48(11), 5558–5562.
17. P. Shukla, I. Fatimah, S. Wang, H. Ang and M. O. Tadé, Photocatalytic generation of sulphate and hydroxyl radicals using zinc oxide under low-power UV to oxidise phenolic contaminants in wastewater, Catal. Today, 2010, 157(1–4), 410–414.
18. Y. Qi, M. Zou, J. S. Ajarem, A. A. Allam, Z. Wang and R. Qu, et al., Catalytic degradation of pharmaceutical and personal care products in aqueous solution by persulfate activated with nanoscale FeCoNi-ternary mixed metal oxides, Sep. Purif. Technol., 2023, 314, 123585.
19. X. Ao and W. Liu, Degradation of sulfamethoxazole by medium pressure UV and oxidants: peroxymonosulfate, persulfate, and hydrogen peroxide, Chem. Eng. J., 2017, 313, 629–637.
20. X. Duan, X. Niu, J. Gao, S. Wacławek, L. Tang and D. D. Dionysiou, Comparison of sulfate radical with other reactive species, Curr. Opin. Chem. Eng., 2022, 38, 100867.
21. J. M. Jazić, T. Đurkić, B. Bašić, M. Watson, T. Apostolović and A. Tubić, et al., Degradation of a chloroacetanilide herbicide in natural waters using UV activated hydrogen peroxide, persulfate and peroxymonosulfate processes, Environ. Sci.:Water Res. Technol., 2020, 6(10), 2800–2815.
22. T. Olmez-Hanci, C. Imren, I. Kabdaşlı, O. Tünay, I. J. P. Arslan-Alaton and P. Sciences, Application of the UV-C photo-assisted peroxymonosulfate oxidation for the mineralization of dimethyl phthalate in aqueous solutions, Photochem. Photobiol. Sci., 2011, 10(3), 408–413.
23. X. Wang and L. J. R. a. Zhang, Kinetic study of hydroxyl radical formation in a continuous hydroxyl generation system, RSC Adv., 2018, 8(71), 40632–40638.
24. A. Wang, Z. Hua, Z. Wu, C. Chen, S. Hou and B. Huang, et al., Insights into the effects of bromide at fresh water levels on the radical chemistry in the UV/peroxydisulfate process, Water Res., 2021, 197, 117042.
25. G. Dong, B. Chen, B. Liu, Y. Cao, B. de Jourdan and S. R. Stoyanov, et al., Comparison of O3, UV/O3, and UV/O₃/PS processes for marine oily wastewater treatment: Degradation performance, toxicity evaluation, and flocs analysis, Water Res., 2022, 226, 119234.
26. A. Shad, J. Chen, R. Qu, A. A. Dar, M. Bin-Jumah and A. A. Allam, et al., Degradation of sulfadimethoxine in phosphate buffer solution by UV alone, UV/PMS and UV/H₂O₂: kinetics, degradation products, and reaction pathways, Chem. Eng. J., 2020, 398, 125357.
27. J. Mack and J. R. Bolton, Photochemistry of nitrite and nitrate in aqueous solution: a review, J. Photochem. Photobiol., A, 1999, 128(1–3), 1–13.
28. S. Verma, S. Nakamura and M. J. Sillanpää, Application of UV-C LED activated PMS for the degradation of anatoxin-a, Chem. Eng. J., 2016, 284, 122–129.
29. Y. Lu, Y. Liu, C. Tang, J. Chen and G. Liu, Heat/PMS degradation of atrazine: theory and kinetic studies, Processes, 2022, 10(5), 941.
30. M. Exner, H. Herrmann and R. Zellner, Laser-based studies of reactions of the nitrate radical in aqueous solution, Berichte der Bunsengesellschaft für physikalische Chemie, 1992, 96(3), 470–477.
31. Y. Xu, Z. Lin, Y. Wang and H. Zhang, The UV/peroxymonosulfate process for the mineralization of artificial sweetener sucralose, Chem. Eng. J., 2017, 317, 561–569.
32. D.-h. Kim, J. Lee, J. Ryu, K. Kim and W. Choi, Arsenite oxidation initiated by the UV photolysis of nitrite and nitrate, Environ. Sci. Technol., 2014, 48(7), 4030–4037.
33. C. Chen, Z. Wu, S. Zheng, L. Wang, X. Niu and J. J. Fang, et al., Comparative study for interactions of sulfate radical and hydroxyl radical with phenol in the presence of nitrite, Environ. Sci. Technol., 2020, 54(13), 8455–8463.
34. Q. Shen, X. Song, J. Fan, C. Chen and Z. Guo, Degradation of humic acid by UV/PMS: process comparison, influencing factors, and degradation mechanism, RSC Adv., 2024, 14(32), 22988–23003.
35. J. M. Jazić, A. Gross, B. Glaser, J. Agbaba, T. Simetić and J. Nikić, et al., Boosting advanced oxidation processes by biochar-based catalysts to mitigate pesticides and their metabolites in water treatment: A meta-analysis, J. Environ. Chem. Eng., 2024, 12(6), 114260.
36. L. Wang, X. Lan, W. Peng and Z. Wang, Uncertainty and misinterpretation over identification, quantification and transformation of reactive species generated in catalytic oxidation processes: A review, J. Hazard. Mater., 2021, 408, 124436.
37. D. Wen, W. Li, J. Lv, Z. Qiang and M. Li, Methylene blue degradation by the VUV/UV/persulfate process: Effect of pH on the roles of photolysis and oxidation, J. Hazard. Mater., 2020, 391, 121855.
38. L. Gao, Y. Guo, J. Zhan, G. Yu and Y. Wang, Assessment of the validity of the quenching method for evaluating the role of reactive species in pollutant abatement during the persulfate-based process, Water Res., 2022, 221, 118730.
39. Y.-M. Lee, G. Lee and K.-D. Zoh, Benzophenone-3 degradation via UV/H2O2 and UV/persulfate reactions, J. Hazard. Mater., 2021, 403, 123591.
40. Y. Yang, Y. Cao, J. Jiang, X. Lu, J. Ma and S. Pang, et al., Comparative study on degradation of propranolol and formation of oxidation products by UV/H₂O₂ and UV/persulfate (PDS), Water Res., 2019, 149, 543–552.
41. C. Nannou, A. Ofrydopoulou, E. Evgenidou, D. Heath, E. Heath and D. Lambropoulou, Antiviral drugs in aquatic environment and wastewater treatment plants: a review on occurrence, fate, removal and ecotoxicity, Sci. Total Environ., 2020, 699, 134322.
42. W. Cao, Y. Qi, Y. Zhang, R. Guo, R. Qu and Z. Wang, Insight into the photodegradation of o, p'-dichlorodiphenyltrichloroethane in a broad spectrum of irradiation light, Chem. Eng. J., 2024, 488, 150855.
43. P. S. Devrukhakar, M. S. Shankar, G. Shankar and R. J. Srinivas, A stability-indicating LC–MS/MS method for zidovudine: Identification, characterization and toxicity prediction of two major acid degradation products, J. Pharm. Anal., 2017, 7(4), 231–236.
44. J. Funke, C. Prasse, C. Dietrich and T. A. Ternes, Ozonation products of zidovudine and thymidine in oxidative water treatment, Water Res.: X, 2021, 11, 100090.
45. M. Kurmi, A. Sahu, S. K. Tiwari and S. J. Singh, Stability behaviour of antiretroviral drugs and their combinations. 6: evidence of formation of potentially toxic degradation products of zidovudine under hydrolytic and photolytic conditions, RSC Adv., 2017, 7(30), 18803–18814.
46. C. Prasse, J. Wenk, J. T. Jasper, T. A. Ternes and D. L. Sedlak, Co-occurrence of photochemical and microbiological transformation processes in open-water unit process wetlands, Environ. Sci. Technol., 2015, 49(24), 14136–14145.
47. M. S. Swaraga and M. J. Adinarayana, Kinetics and mechanism of protection of thymine from sulphate radical anion under anoxic conditions, J. Chem. Sci., 2003, 115(2), 123–128.
48. A. Faggiano, M. De Carluccio, A. Fiorentino, M. Ricciardi, R. Cucciniello and A. Proto, et al., Photo-Fenton like process as polishing step of biologically co-treated olive mill wastewater for phenols removal, Sep. Purif. Technol., 2023, 305, 122525.
49. J. Drzymała and J. J. Kalka, Elimination of the hormesis phenomenon by the use of synthetic sea water in a toxicity test towards Aliivibrio fischeri, Chemosphere, 2020, 248, 126085.
50. S. Tongur and S. J. ildiz, Toxicological evaluation of carbamazepine active pharmaceutical ingredient with Lepidium sativum, Daphnia magna and Vibrio fischeri toxicity test methods, Desalin. Water Treat., 2020, 201, 438–442.
51. T. S. Viana, T. C. R. Rialto, J. Brito, A. F. D. Micas, F. R. Abe and E. A. Savazzi, et al., Effects of water produced by oil segment on aquatic organisms after treatment using advanced oxidative processes, J. Toxicol. Environ. Health, 2021, 84(22), 901–913.
52. P. Reuschenbach, M. Silvani, M. Dammann, D. Warnecke and T. J. Knacker, ECOSAR model performance with a large test set of industrial chemicals, Chemosphere, 2008, 71(10), 1986–1995.
53. P. R. Duchowicz, QSPR studies on water solubility, octanol-water partition coefficient and vapour pressure of pesticides, SAR QSAR Environ. Res., 2020, 31(2), 135–148.
54. J. Wang, X. Hong, W. Liu, L. Zhang, S. Yan and Z. Li, et al., Comprehensive assessment of the safety of bisphenol A and its analogs based on multi-toxicity tests in vitro, J. Hazard. Mater., 2025, 486, 136983.

---

Footnote

† These authors contributed to the work equally and should be regarded as co-first authors.

---