English

• 1. Introduction

  Antibiotics are prevalent environmental pollutants frequently detected in aquatic environments due to widespread use in medicine and agriculture [1–3]. Antibiotics’ environmental impact extends beyond chemical toxicity; they also facilitate the spread of antibiotic-resistance genes (ARGs) and antibiotic-resistant bacteria (ARBs) [4,5]. The spread of ARGs and ARBs through wastewater disrupts microbial communities, poses public health risks, and accelerates antibiotic-resistant bacteria proliferation, increasing reliance on antibiotics [6]. Antibiotics have been frequently identified in various water sources, including surface water, groundwater, municipal sewage, and even drinking water, due to their extensive use in hospitals, animal husbandry, and poultry farming [7–9]. For example, sulfamethoxazole (SMX) was detected in hog manure digests at microgram levels 1–2 orders of magnitude higher than in the surrounding environment [10–12]. These contaminants are challenging to fully eliminate, highlighting the need for enhanced source-control strategies in farms and pharmaceutical sites [13].

  Advanced oxidation processes (AOPs) are highly effective for water purification, generating reactive oxygen species (ROS) like ^•OH, ¹O₂, and ^•O₂⁻ to degrade persistent pollutants (Fig. 1). Many studies have focused on applying AOPs, especially for removing stubborn organic pollutants [14,15]. Fenton processes, which use iron ions and hydrogen peroxide to generate powerful oxidizers, have been widely applied in treating complex pollutants like antibiotics, dyes, and pesticides, showing high efficiency and cost-effectiveness [16,17]. To overcome limitations in traditional Fenton processes, modified Fenton-like methods, incorporating metal and biochar-doped catalysts, have been developed. These adaptations address issues like iron precipitation, extend the effective pH range, and improve catalyst recyclability [18,19].

  Figure 1

  

  Figure 1.  Classification of AOPs using Fenton methods for antibiotic degradation.

  DownLoad: Full-Size Img PowerPoint

  In the past 15 years, publications and citations on Fenton-like methods for treating antibiotic-contaminated wastewater have steadily increased, with a notable surge from 2016 to 2022 (Fig. 2), reflecting growing academic interest in this field. Concurrently, an analysis of keywords from related articles published between 2000 and 2024 in the Web of Science Core Collection, conducted using the Citespace tool with γ [0,1] = 0.7, is also depicted in Fig. 2 [20,21]. Initially focused on biodegradability, research gradually shifted towards catalytic materials like boron-doped diamond and, later, nanomaterials such as graphene oxide. Research emphasizes antibiotic contamination, nanomaterial-based AOPs, and toxicity assessment of degradation byproducts. Key concerns remain around ARGs and persistent pollutants, including pharmaceuticals and personal care products (PPCPs), which continue to challenge water treatment practices [22].

  Figure 2

  

  Figure 2.  Keywords and publication of Fenton approaches on antibiotic degradation since 2000. Data sources: top 500 sorted by citation high-count in Web of Science Core Collection, from 2000 to 2024. The pie chart depicts the percentage of Fenton’s method in each category.

  DownLoad: Full-Size Img PowerPoint

  Despite advancements, the toxicity of antibiotic-containing wastewater can vary during degradation, with limited assessment of treated water’s environmental acceptability despite antibiotics’ low toxicity and concerns over toxic byproducts [23]. Existing reviews often emphasize optimizing reaction conditions and catalysts and comparing technologies like Fenton and photo-Fenton reactions but lack focus on byproduct toxicity and real-world environmental risks [6,24,25]. While these reviews summarize technological advances, they often overlook degradation byproduct toxicity and real-world environmental risks, with some studies noting increased toxicity during treatment [26,27]. The potential hazards of oxidative degradation are commonly underestimated, posing risks to aquatic organisms and human health. Additionally, current research predominantly focuses on the toxicity of parent compounds, with limited attention to byproducts, leaving water quality toxicity a persistent and unresolved environmental challenge.

  A thorough evaluation of Fenton-like treatments is crucial for remediating antibiotic-containing wastewater, with a focus on the toxicity of real-world samples [28]. Toxicity tests predict the potential impacts of bioavailable fractions of substances, mixtures, or whole environmental samples on living non-target organisms upon their release into the environment. These tests aim to obtain information about environmental contaminants’ toxicity, genotoxicity, or other toxicological outcomes [29]. Toxicity identification evaluation (TIE) and effect-directed analysis (EDA) methods can identify specific toxicants in complex mixtures, which QSAR models can then analyze to predict toxicity for similar compounds [30,31]. Toxicity tests, classified as biotic or abiotic, favor biotic testing for its accuracy and simplicity [32,33].

  Antibiotics often co-exist with substances of uncertain concentrations, proportions, and compositions, limiting data on degradation pathways and intermediates. Computer modeling using abiotic tests can predict the biological activity of unfamiliar chemicals. Integrating experimental methods like EDA with QSAR models offers a comprehensive approach, where EDA validates intermediates and QSAR provides predictive insights [34,35]. Although QSAR models play a crucial role in assessing chemical toxicity, they have limitations when evaluating the toxicity of byproducts generated during antibiotic degradation. In particular, QSAR alone may not fully capture the potential ecological risks associated with antibiotic resistance. Therefore, a comprehensive approach combining experimental validation and more extensive models is essential for accurately assessing antibiotics’ risks and byproducts​.

  This article examines the application of Fenton-like methods for effectively degrading antibiotics in wastewater, highlighting the chemical and ecological risks associated with toxic byproducts. Through the integration of QSAR modeling and biological assessments, it provides a comprehensive framework for evaluating degradation efficiency and toxicity, offering practical strategies to enhance environmental safety and mitigate health risks. Over the past 25 years, Fenton-like research has primarily focused on eliminating antibiotics, often without comprehensive toxicity evaluations. This review systematically assesses the formation and potential toxicity of byproducts in antibiotic degradation pathways, filling a critical gap in the current literature. Summarizing toxicity assessment methods and byproduct patterns, this work offers insights to enhance antibiotic wastewater treatment and reduce secondary pollution risks. Furthermore, integrated computer models are highlighted as tools for accurate toxicity evaluation, offering solutions to overcome current challenges. Finally, the review identifies key challenges in toxicity evaluation and suggests improvements to modeling approaches, emphasizing the critical need for safe and effective application of Fenton-based technologies in real-world wastewater treatment.

  2. Fenton-like methods: Performance in antibiotic degradation

  Advanced oxidation processes can effectively degrade persistent pharmaceuticals; however, improper monitoring and operation may lead to the formation of toxic oxidation intermediates or by-products [36]. If inadequately managed, antibiotic-containing wastewater may be discharged into aquatic ecosystems through various pathways, potentially posing a severe threat to the environment. These residues could lead to the emergence of resistant bacteria and genes, which is a significant threat to human health and ecosystems [37,38]. Tetracyclines (TCs) and sulfonamides (SAs) have attracted extensive research due to their widespread use in wastewater treatment and natural environments [39,40]. AOPs are frequently employed for the efficient removal of antibiotics from wastewater [41–43]. Among the chemical oxidation methods, the Fenton method has been recognized as superior for treating certain refractory organic matter, such as phenol and aniline [44–46]. The Fenton method offers cost efficiency and improved treatment selectivity over electrochemical and ozone oxidation [47–49].

  Researchers have modified the Fenton method to boost its oxidation efficiency. Recently, integrating UV and electrochemical oxidation into the Fenton process has significantly enhanced its capabilities. For example, a UV/chlorination system effectively removes ARGs and increases the removal of antibiotics and antibiotic-resistant bacteria [50]. Concurrently, a heterogeneous Fenton catalytic system has been constructed, which utilizes highly oxidized active radicals produced in situ to achieve higher removal efficiencies of antibiotics and total organic carbon (TOC). Research has demonstrated that introducing l-cysteine into the Fe(Ⅲ)/CaO₂ system can significantly enhance the degradation of sulfadiazine (SDZ) due to improved Fe(Ⅱ)/Fe(Ⅲ) cycling, leading to a marked increase in degradation efficiency [51]. Yu used reduced graphene rGO/SA-TCPP photocatalyst and Fe(Ⅲ) with high H₂O₂ yield to remove sulfamdimethadiazine in environmental water, and the degradation rate is 17.7 times that of the Fenton system [52].

  The Fenton-like method relies on the generation of potent oxidant reactive radicals, such as ^•OH and ^•O₂⁻, to effectively eliminate antibiotics and TOC [53,54]. Reactive oxygen species attack the chemical bonds of antibiotic molecules, leading to their degradation. In both homogeneous and heterogeneous Fenton reactions, the types and concentrations of ROS are influenced by several factors, including the Fe(Ⅱ) source, H₂O₂ concentration, and pH conditions [55]. Among these, excessively high concentrations of ^•OH may lead to non-specific degradation, generating a variety of degradation products, and could even result in the formation of some toxic intermediate products [56]. The diffusion capacity and localization of ^•OH are identified as key factors determining pollutant degradation’s efficacy [48].

  Researchers have proposed numerous theories and experimental findings to investigate the complexities of organic matter oxidation reactions [57,58]. Among these, the solid-liquid boundary layer perspective stands out, suggesting that oxidation primarily occurs at the interface between solid and liquid phases rather than within the bulk solution or solely on the surface. This theory is widely regarded as one of the most broadly applicable, as it encompasses both the oxidation of H₂O₂ at the boundary layer and the diffusion of ^•OH radicals across a defined distance [59–63]. This viewpoint sheds light on the diffusion and reaction dynamics of ^•OH in natural wastewater treatment processes. It also lays a theoretical foundation to boost the effectiveness of Fenton-like methods.

  In applying the Fenton-like method for treating actual organic wastewater, the presence of inorganic ions such as Na⁺, Mg²⁺, Cl⁻, HCO₃⁻, and NO₃⁻ can affect the oxidation process to varying degrees. It has been observed that halogen salt ions may interfere with the system [64]. This emphasizes the importance of salt interference levels and the relationship between hydroxyl radical activity and the type of catalyst used. Enhancing catalyst activity, especially Fe(Ⅱ)/Fe(Ⅲ) cycling efficiency, is crucial for boosting ^•OH production, dispersion, and reactivity within the solid-liquid boundary layer. Such optimizations support broader applications of the Fenton-like method in real-world wastewater treatment.

  Researchers have improved Fe(Ⅱ)/Fe(Ⅲ) cycling and system stability by incorporating metals like Mo, Co, Cr, Cu, and Ti into magnetite or complexing them with carbon materials. The introduction of doping can alter the coordination and electronic environment of iron oxides, thereby improving the catalyst’s performance [65–67]. For example, Fe₃O₄@MoS₂ catalysts, shown in Figs. 3a-d, leverage Mo(Ⅳ) to enhance the Fe(Ⅱ)/Fe(Ⅲ) cycle, achieving up to a fourfold increase in the removal efficiency of sulfonamides and other pollutants compared to conventional systems (Fig. 3e) [18]. This catalyst also facilitates magnetic separation, crucial for practical industrial applications [68].

  Figure 3

  

  Figure 3.  (a) TEM images of Fe₃O₄, (b) Fe₃O₄@MoS₂–3. (c, d) Demonstration of Fe₃O₄ and Fe₃O₄@MoS₂–3. (e) Effect of peroxymonosulfate (PMS) dose on the degradation of SA by Fe₃O₄@MoS₂–3. Copied with permission [18]. Copyright 2021, Chemical Engineering Journal.

  DownLoad: Full-Size Img PowerPoint

  Enhancing the catalytic efficiency of Fenton-like methods involves incorporating highly reactive metals and utilizing synergistic effects with biomaterials like biochar. Combining biochar with metal oxides, such as magnetite (Fe₃O₄), forms biochar-metal composites that enhance ^•OH generation. These composites significantly boost the oxidative degradation of pollutants like TCs and SAs in antibiotic-contaminated wastewater [68,69]. Biochar prevents catalyst particle aggregation and promotes electron transfer, serving both as an adsorbent and a facilitator of catalytic performance [70,71].

  This combined catalytic and adsorptive effect aligns with previous research [72,73]. Presently, two main academic theories are proposed to elucidate the mechanism by which carbon materials participate in the Fe(Ⅲ) reduction process. One theory suggests that reducing groups are directly involved in electron transfer, whereas the other proposes that conductive carbon acts as an electron mediator (Fig. 4) [72,74]. Although these theories differ, both aim to improve catalyst cycling efficiency and overall system stability for enhanced pollutant removal. Functionalizing carbon materials and optimizing their structure can facilitate free radical oxidation and the mineralization of organic pollutants even at low concentrations [75,76].

  Figure 4

  

  Figure 4.  Pathways of electron transfer for Fe(Ⅲ) reduction facilitated by carbon materials in antibiotic degradation.

  DownLoad: Full-Size Img PowerPoint

  3. Risks in degradation: Chemical and ecological considerations

  3.1 Chemical toxicity in degradation by-products

  The Fenton process effectively treats antibiotic-contaminated wastewater, as highlighted in Table 1 [54,77–81], with studies showcasing its capacity to degrade antibiotics and reduce pollution. Recent advances focus on doping reactive metals and integrating diverse materials to enhance its efficacy.

  Table 1

  

  Table 1.  Effectiveness of different classes of Fenton systems in removing antibiotics, toxicity reduction.

  DownLoad: CSV

  Catalyst	Antibiotic	Main ROS	Efficiency	Highly toxic intermediates	Highly toxic intermediate	Overall toxicity
  Pt-FeOX/G [77]	SDZ	•OH	98%, 10 min	B2 (m/z 96) is highly toxic to both daphinds and green algae with higher chronic toxicity than SD itself. However, the majority of its subsequent-generation products are less detrimental than SD, indicating that most generated products were less harmful than SD.		As intermediates continued to mineralize, toxicity towards luminescent bacteria rapidly decreased, with inhibition rates dropping to as low as 4.37%, corresponding to improved TOC removal.
  Fe2O3/MXene-x [78]	SMX	•OH and •O2–	95%, 20 min	The developmental toxicity of compound P285 (m/z 284) exceeded that of SMX. Moreover, the developmental toxicity of the remaining degradation intermediates was notably diminished or completely absent.		A significant reduction in overall toxicity was observed within 30 min for most intermediates, with developmental toxicity and bioaccumulation factors lower than SMX, except for P285.
  MnFe-LDH@BC [79]	TC	•OH	94.2%, 500 min	The ChV content of intermediate products in various degradation pathways exceeded 0.000002 mol/L, indicating potential relevance for organisms such as fish, Daphnia, and green algae due to their weak chronic toxicity.		Toxicity of TC intermediates decreased over time, with acute toxicity remaining low, while chronic toxicity was eventually eliminated after complete mineralization.
  Fe-POM/BMO [80]	TC	•OH	99.76%, 18 min	Among the processed products, TC(Ⅶ) (m/z 417), TC(Ⅷ) (m/z 343), and TC(Ⅹ) (m/z 402) exhibited higher acute toxicity towards the blackhead minnow Daphnia magna compared to TC, along with a greater predicted bioaccumulation factor.		Acute toxicity, bioaccumulation, and mutagenicity significantly decreased after degradation, with luminescent bacteria inhibition dropping from 56% to 21%, indicating reduced environmental risks.
  Ag3PO4/MIL-101 [54]	TC	•OH and 1O2	91%, 25 min	P2 (m/z 431), P4 (m/z 417), and P12 (m/z 277) exhibited higher acute toxicity levels compared to TC in rats, black-headed minnow, and Daphnia magna, whether administered orally or through other means.		In the Ag3PO4 system, some intermediates showed higher toxicity than TC, but toxicity decreased in the AP/MF-0.5 systems over time.
  Ce4O7/Bi4MoO9 [81]	TC	•OH and •O2–	100%, 120 min	P(Ⅶ) (m/z 417) and P(Ⅷ) (m/z 402) exhibited significant acute toxicity towards black-headed minnow (Daphnia magna) and orally administered rats.		Although some intermediates showed higher toxicity during the Photo-Fenton degradation of tetracycline, most intermediates exhibited lower toxicity over time, with bioaccumulation reduced and toxicity towards aquatic organisms mitigated.

  Despite its promising results, the Fenton-like method can produce highly toxic intermediates, posing environmental risks. For example, a study on SDZ degradation using a Pt-FeOX/G system found that inevitable byproducts exhibited more significant chronic toxicity toward daphnids and green algae than SDZ. This highlights the need for further research into intermediate toxicity to assess the environmental impact of Fenton-like processes fully. Studies indicate mineralizing intermediate B2 (Table 1) can yield less toxic substances [26,54,77]. Some researchers highlighted the detoxification and mineralization of highly toxic intermediates by extending the reaction time, leading to the formation of non-toxic byproducts such as carbon dioxide and water [64,77,78]. Upon reviewing Table 1, it becomes clear that common highly toxic intermediates, such as those with m/z 402 and m/z 417, are key areas for further research into the toxicity of antibiotic degradation products [79–81]. Thus, further mineralization processes are critical for reducing the overall toxicity of the antibiotic degradation system in water [42].

  In Fenton-like antibiotic wastewater treatment, oxidizing and reducing agents enhance degradation efficiency but may also increase toxicity risks. Their interaction with pollutants has a notable impact on water quality toxicity [58]. For instance, excess PMS may lead to self-consumption phenomena, where PMS molecules are adsorbed or degraded into sulfate rather than being converted into reactive oxygen species (ROS) [82–84]. Additionally, residual organic reductants and various valence states of iron can present toxicity risks to aquatic organisms, such as luminescent bacteria [85]. The oxidation of micropollutants like antibiotics, using pertechnetate, can significantly increase water toxicity due to the formation of intermediate products like Fe(Ⅴ), as evidenced by a 22% reduction in the luminescence intensity of Vibrio fischeri [42,86]. Furthermore, excess organic reductants can reduce reaction efficiency by quenching free radicals. Although antibiotic concentrations in wastewater are typically low, the high concentrations of oxidants added to achieve the desired degradation effect may result in toxic by-products that pose significant risks to water quality [87–89].

  The initial toxicity increase, often due to toxic intermediates, can be mitigated by extending reaction time, allowing these intermediates to convert into less toxic or non-toxic molecules like carbon dioxide and water [90]. Additionally, although the use of oxidizing and reducing agents poses biological hazards, these risks can be effectively managed through strict safety threshold standards [52]. Current research highlights the importance of controlling reaction time and catalyst concentration to reduce system toxicity. Careful parameter adjustments optimize Fenton-like processes, ensuring efficient antibiotic removal while minimizing toxic by-products and enhancing wastewater treatment safety.

  These intermediates may pose both acute and chronic toxicity risks to aquatic organisms, including microorganisms, phytoplankton, and fish, potentially impacting their long-term health and disrupting ecosystem balance. QSAR models are widely utilized to predict chemical toxicity, particularly in the absence of experimental data. The complexity of antibiotic degradation intermediates limits QSAR models in predicting long-term impacts in diverse aquatic environments, highlighting the need for integrated approaches to address complex water systems.

  3.2 Ecological risks from degradation by-products

  The oxidative degradation of antibiotics, such as through Fenton-like methods, can generate intermediate by-products that may pose ecological risks to aquatic environments [25]. A key concern is whether the conversion products of antibiotics or other pollutants generated during advanced chemical oxidation processes promote antibiotic resistance [91–93].

  The interactions of these by-products within ecosystems, especially with microorganisms, can cause disruptions to microbial community structures and facilitate the spread of ARGs [94]. ARGs can be transferred horizontally between microorganisms, further increasing antibiotic resistance in bacteria, including pathogenic strains [95,96]. This horizontal gene transfer could lead to a more widespread distribution of resistance traits in previously unexposed environments, significantly affecting microbial community dynamics and the overall health of aquatic ecosystem [97]. Lastly, the presence of ARGs in ecosystem samples is identified to be a nuisance to public health [23].

  Studies have confirmed that antibiotics and their degradation products pose acute and chronic toxicity risks to aquatic organisms. Species like fish, algae, and zooplankton are particularly vulnerable to prolonged exposure, experiencing physiological stress, altered reproduction, or mortality, potentially disrupting food chains and ecosystem stability [98,99]. For example, algae, as primary producers, play a critical role in aquatic ecosystems; any reduction in their population due to toxicity could have cascading effects on higher trophic levels, including herbivores and carnivores [100]. The persistence of oxidation by-products from advanced oxidation processes and their potential toxicity exacerbates these ecological risks. This issue is often overlooked, highlighting the need for rigorous monitoring of effluent toxicity to ensure that improvements in degradation efficiency do not inadvertently increase environmental harm.

  However, much of the existing toxicity data focuses solely on the properties of parent antibiotics without considering the toxicity of their transformation products [101]. The limited data on by-product toxicity often leads to an underestimation of the actual environmental impact of antibiotic degradation. Testing water samples with unknown compositions, especially post-degradation, is essential. Such testing helps predict potential ecological risks [102].

  Biological approaches like bioluminescence tests, Daphnia magna assays, and algal tests detect acute and chronic toxicity but overlook longer-term or subtle impacts. They focus on immediate effects rather than broader changes in microbial communities or gene transfer [32,103]. Biological approaches encompass acute and chronic evaluations, incorporating parameters like the no-observed-effect-concentration (NOEC), 10% or 50% effect concentration (EC₁₀ or EC₅₀), lowest-observed-effect-concentration (LOEC), lethal concentration or dose 50% (LC₅₀ or LD₅₀). These data are typically derived from extensive testing with numerous organisms, which can be time-consuming and may not fully capture the toxicity profile for ecosystems and critical species [104,105].

  A more comprehensive ecological risk assessment is necessary to address these gaps, one that goes beyond traditional chemical and biological toxicity tests. This approach should include long-term monitoring of microbial community dynamics and resistance gene propagation within aquatic ecosystems [106,107]. By integrating resistance gene monitoring with existing risk assessment frameworks, researchers can better evaluate the broader ecological impacts of antibiotic degradation and develop more effective strategies to mitigate these risks [108].

  4. Toxicity in computer-prediction

  Toxicity evaluation integrates experimental and computational methods, offering significant insights into water quality. While traditional biotoxicity indicators are valuable, they may not fully capture the complexity of diverse pollutant mixtures. For instance, combining QSAR with EDA has shown promise for comprehensive assessments [35]. However, these methods still have limitations, and new approaches are needed to enhance the accuracy and comprehensiveness of toxicity evaluations. Artificial intelligence demonstrates significant potential in toxicity prediction and data analysis. It enhances prediction accuracy and addresses data imbalance through multitask learning [109]. At the same time, it utilizes AI-driven image and video analysis to efficiently process data and provide precise support for toxicity mechanism research. These technologies accelerate evaluation efficiency and facilitate in-depth studies [110]. Further refining computational approaches can address existing gaps and improve evaluations of environmental contaminants.

  In computational toxicology, various computational methods are employed for toxicity prediction, including QSAR, Read-Across, threshold of toxicological concern, molecular modeling, and quantum chemical calculations. Computational toxicology, especially when evaluating the impacts of pollutants in complex aquatic environments, provides a comprehensive approach. The validated QSAR model has become a vital tool within this field. However, capturing the complexity and long-term ecological impacts of antibiotic degradation by-products remains a challenge. Building on existing advancements, integrating QSAR with biological toxicity assessments could further enhance the evaluation framework, contributing to more comprehensive and reliable environmental impact analyses.

  4.1 Principles of computer-prediction

  QSAR is constructed based on a dataset comprising molecular structures and properties of known active compounds. It aims to forecast the biological activity of a chemical substance by employing quantitative ontological relationship modeling [111]. This approach is characterized by its simplicity and ease of manipulation compared to other intricate network-based computational methods, such as analyzing shared targets and pathways linked to adverse drug reactions [112]. Wang employed QSAR analysis to forecast the high toxicity of 3 intermediates as obtained compared to TC, which were in solid agreement with the empirical findings [113].

  This consistency further supports the reliability and validity of QSAR in toxicity prediction [114]. These models act not only as a reference for experimental testing but can also forecast unknown toxicities when actual measurement data is unavailable [101]. By continually validating these predictions with new data, confidence in QSAR applications for toxicity assessments is strengthened. However, while effective for known structures, QSAR models still depend on empirical validation when applied to unknown byproducts.

  In recent studies, integrating QSAR predictions with EDA has enabled researchers to accurately identify and confirm the toxicity of previously uncharacterized intermediates formed during degradation processes [115]. This iterative process of prediction and validation enhances the reliability of QSAR models, making them essential for assessing the toxicity of known and emerging contaminants. They accurately predict toxicological effects and are valuable in hazard, exposure, and risk assessments. Constructing a canonical QSAR model typically involves several steps (Fig. 5): Dataset preparation, molecular descriptor computation, dataset division, descriptor selection, QSAR model building, model validation, model application domain definition, and model mechanism interpretation [116].

  Figure 5

  

  Figure 5.  Block diagram of QSAR research method.

  DownLoad: Full-Size Img PowerPoint

  The development of QSAR models faces several challenges, mainly due to inherent errors in experimental data. For example, inconsistencies in the QSAR model by Nils led to differing classifications of 16 compounds as specifically or non-specifically acting [117]. From this perspective, variations in experimental conditions and data standards can pose challenges in attaining consistency and inclusiveness in the compiled dataset. Critical evaluation of databases is essential, including bioaccumulation in fish, half-lives, and toxicity data across various species.

  Creating a professional and credible dataset is challenging and requires significant investment. Thorough data assessment is crucial for ensuring QSAR model reliability. While many QSAR models exist, selecting and refining 2D or 3D models is vital for accuracy. 3D-QSAR excels in estimating relative toxicity (Re%) for pollutants like sulfonamide antibiotics, while 2D-QSAR performs well across various contaminants [118]. Pollutants with unique active functional groups can vary significantly in biochemical reactivity, impacting toxicity predictions. Therefore, it is crucial to tailor QSAR models to reflect compound behavior in diverse environmental conditions. This approach enhances the accuracy of toxicity forecasts.

  4.2 Enhancements in computer-prediction

  Since implementing new regulations in 2004 by the Organization for Economic Co-operation and Development (OECD) to standardize QSAR modeling with the establishment of five fundamental principles, the research and application of QSAR technology have seen rapid development [119]. However, the growth of modes of action (MOA)-based local models has been limited by their inherent characteristics, while global modeling approaches remain the standard method for building QSAR models [120].

  Researchers have refined molecular descriptor selection and optimized model parameters to improve prediction accuracy, enhancing QSAR models’ internal and external predictive capabilities. For example, Yu developed 20 RBF submodels with PaDEL descriptors, achieving R² (Eq. 1) values above 0.84 and R²cv10 values over 0.70, showing superior robustness and fitting ability compared to single RBF models, with a dataset applicability of 99.34% [121]. Similarly, Wu et al. applied a QSAR model to predict antibiotic degradation toxicity, achieving a relative error under 3% between predicted and experimental toxicity values, highlighting the model’s reliability [122]. Although high R² values indicate good internal performance, external predictive power is also essential. For instance, a consensus model with lower prediction error outperformed others with similar R² values (Table 2) [118,121,123].

  $ R^2=1-\left[\frac{\sum\limits_{i=1}^N(y(i)-\hat{y}(i))^2}{\sum\limits_{i=1}^N(y(i)-y)^2}\right] $

  (1)

  Table 2

  

  Table 2.  QSAR models for common toxicity prediction.

  DownLoad: CSV

  Compound	Equation	R2
  126 kinds of organic volatile molecules [123]	Model 1: pNOAEC = 6.906–0.605 × nF - 0.044 × H% + 1.082 × Psi_i_0d + 2.065 × X3A-0.2897 × nR	0.743
  	Model 2: pNOAEC = 7.195 - 0.620 × nF - 0.049 × H% + 1.711 × Psi_i_0d + 2.124 × X3A - 0.319 × nR	0.739
  	Model 3: pNOAEC = 6.127 - 0.491 × nF - 0.034 × H% + 0.283 × nHM + 1.574 × Psi_i_0d + 1.800 × X3A + 0.422 × O-060 + 0.859 × B01[C—N] + 0.755 × B01[C—O] + 2.087 × B05[C—N]	0.747
  15 kinds of organophosphates [118]	-logEC50 = 2.979 + 0.246P + 0.220logD + 0.204NAr + 0.186TSA - 0.129LUMO - 0.084NA + 0.080ND	0.90

  Compounds can be subjected to prefiltering using substructure indicators to differentiate between baseline and excess toxicity. M. NENDZA, in an approach targeting 115 chemicals from 9 distinct MOA classes, introduced three classification schemes with the goal of categorizing these chemicals into baseline and overdose toxicants [124]. Following this, a hierarchical categorization process was applied to cluster compounds with similar functionalities. The pre-filtering step was expected to simplify the identification of chemical substances, reducing the overall effort. However, the challenge of classifying chemicals remains due to the limited universal applicability of classification schemes across various endpoints, such as those affecting algae and fish. The variability of MOA for a given compound, which may change with concentration and exposure duration, adds complexity to real-world applications of these classification schemes [124,125]. This variability highlights the constraints of MOA classification in practical scenarios and adds another layer of complexity to modeling applications.

  Relying on a single classification method limits toxicity predictions due to MOA specificity and data inaccuracies. Researchers are exploring advanced approaches, including consensus models that combine validated QSAR single models (SMs) to mitigate individual deficiencies and expand predictive coverage. A consensus model developed with three qualified SMs showed superior statistical performance compared to single models, broadening applicability for environmental toxicity assessment [126]. While classification-based modeling improves toxicity predictions for individual antibiotics, real-world water contamination is far more complex. The combined effects of contaminants in a water system can be either synergistic or antagonistic, yet no effective method currently exists to predict these mixture toxicities [127].

  Researchers have shifted their focus from individual pollutant classification to prioritizing pollutants to address this challenge. The environmental health prioritization index (EHPi) offers a new method for evaluating key pollutants by considering their interactions [128]. Current multicriteria decision-making (MCDM) tools evaluate a limited range of contaminants based on toxicity, persistence, and detection data. To better address exposure risks, the risk assessment, identification, and ranking–indoor and consumer exposure (RAIDAR-ICE) framework has been introduced as an integrated approach [129]. This framework combines screening and prioritization strategies, offering new perspectives on direct and indirect exposure approaches, as well as far-field and near-field considerations.

  In conclusion, enhancing the accuracy of QSAR models for toxicity prediction relies on two main aspects. First, it is essential to ensure that the models are selected with consideration of their domain of applicability [130]. Second, refining the descriptors and related parameters is crucial for model optimization. In practical applications, especially in artificial oxidation treatments for antibiotic-contaminated wastewater, QSAR models have become indispensable. The need for these models is driven by the dynamic nature of intermediate compounds in wastewater, which change due to the evolving structure of the intermediates throughout the treatment process. QSAR models can swiftly predict the toxicity of compounds in the water column, making them an ideal choice for a comprehensive assessment of water quality toxicity changes throughout the treatment of antibiotic-containing wastewater using AOPs, such as Fenton-like approaches.

  5. Perspectives

  The Fenton-like method shows great potential in treating antibiotic-laden wastewater by leveraging metal-doped catalysts and novel adaptations to improve degradation efficiency and stability. A vital advantage of this approach is its ability to break down antibiotics while reducing toxicity, a critical factor in water treatment evaluation. Although early efforts prioritized catalyst efficiency, worries about increased toxicity due to toxic byproducts remain. This toxicity can be effectively managed through combined adsorption and oxidation processes, optimized reaction times, and precise control of oxidizing/reducing agent concentrations to limit toxic byproduct formation.

  A meticulous evaluation of Fenton-like techniques’ effectiveness in mitigating real wastewater toxicity is imperative. This necessity extends to developing and refining computer models, which show significant potential in predicting the complex interactions and outcomes of treatment processes. The toxicity includes assessing not only chemical toxicity but also ecological risks associated with toxic intermediates, which may disrupt microbial communities and propagate antibiotic resistance. Developing and refining computer models, such as QSAR, offers significant potential for predicting the complex interactions and outcomes of treatment processes. QSAR models, when integrated with experimental validation and biological toxicity assessments, can provide valuable insights into the environmental impact of wastewater treatment processes. Refining these models through better descriptor selection, parameter optimization, and the inclusion of more comprehensive degradation pathways can enhance their predictive power, contributing to a more holistic risk assessment framework.

  Future research should focus on refining QSAR models by expanding the database to encompass a broader range of antibiotics and degradation intermediates, thereby improving predictive accuracy and facilitating real-time assessments of environmental risks. Investigations into the long-term impacts of treated wastewater on aquatic ecosystems, alongside the potential for reusing treated water, are also critical areas for study. Furthermore, interdisciplinary approaches that integrate environmental science, chemistry, and computational modeling will be pivotal in advancing the field of wastewater treatment and ensuring both efficiency and safety. This review underscores the importance of robust, multi-faceted toxicity evaluations within antibiotic degradation processes, advocating for approaches that prioritize both treatment efficiency and environmental protection. We hope this analysis catalyzes continued research into practical, ecologically sustainable water treatment technologies.

  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

  Jingyi Yang: Writing – original draft, Formal analysis. Sihan Wang: Writing – review & editing. Xubiao Luo: Writing – review & editing. Zhenyang Yu: Writing – review & editing. Yanbo Zhou: Writing – review & editing, Funding acquisition.

  Acknowledgments

  This work was supported by the National Natural Science Foundation of China (No. 52370168) and the Key Laboratory of Functional Biology and Pollution Control in red soil regions of Jiangxi Province (No. 2023SSY02051).

  
  
• 1. [1]

     Y. Zhou, J. He, J. Lu, et al., Chin. Chem. Lett. 31 (2020) 2623–2626.

  2. [2]

     C. Bhagat, M. Kumar, V.K. Tyagi, P.K. Mohapatra, npj Clean. Water 3 (2020) 42.

  3. [3]

     Y. Zhou, Q. Liu, J. Lu, et al., J. Hazard. Mater. 393 (2020) 122414.

  4. [4]

     E.Y. Klein, M. Milkowska-Shibata, K.K. Tseng, et al., Lancet. Infect. Dis. 21 (2021) 107–115.

  5. [5]

     C. Monahan, S. Harris, D. Morris, E. Cummins, Sci. Total Environ. 826 (2022) 154008.

  6. [6]

     M. Hazra, J.E.M. Watts, J.B. Williams, H. Joshi, Sci. Total Environ. 917 (2024) 170433.

  7. [7]

     N. Jallouli, L.M. Pastrana-Martínez, A.R. Ribeiro, et al., Chem. Eng. J. 334 (2018) 976–984.

  8. [8]

     M.S. Fouad, E.F. Mustafa, M.S. Hellal, M.A. Mwaheb, Sci. Rep. 14 (2024) 18849.

  9. [9]

     J. Guo, X. Jin, Y. Zhou, et al., J. Environ. Manage. 371 (2024) 123125.

  10. [10]

      Z. Huang, L. Hu, J. Yang, et al., Environ. Int. 181 (2023) 108304.

  11. [11]

      S. Zhang, H. Zheng, P.G. Tratnyek, Nat. Water 1 (2023) 666–681. doi: 10.1038/s44221-023-00098-1

  12. [12]

      M.I.R. Baig, P. Kadu, P. Bawane, et al., J. Antibiot. 76 (2023) 629–641. doi: 10.1038/s41429-023-00649-4

  13. [13]

      S. Wang, J. Yang, X. Jin, et al., Sep. Purif. Technol. 356 (2025) 129841.

  14. [14]

      J. Guo, L. Ling, Q. Lu, et al., Chin. Chem. Lett. 36 (2025) 110380.

  15. [15]

      E.M. Abd El-Monaem, H.M. Elshishini, S.S. Bakr, et al., npj Clean Water 6 (2023) 34.

  16. [16]

      S. Xiao, M. Cheng, H. Zhong, et al., Chem. Eng. J. 384 (2020) 123265.

  17. [17]

      H. Zhou, H. Zhang, Y. He, et al., Appl. Catal. B 286 (2021) 119900.

  18. [18]

      J. Lu, Y. Zhou, Y. Zhou, Chem. Eng. J. 422 (2021) 130126.

  19. [19]

      S.A. Walling, W. Um, C.L. Corkhill, N.C. Hyatt, npj Mater. Degrad. 5 (2021) 50.

  20. [20]

      J. Ma, R. Zhou, F. Yu, Desalination 571 (2024) 117107.

  21. [21]

      L. Jiang, A. Raza, A.B. El Ariss, et al., Sci. Rep. 14 (2024) 2419.

  22. [22]

      H. Huang, R. Ma, H. Ren, Front. Environ. Sci. 18 (2024) 72.

  23. [23]

      G. Prasannamedha, P.S. Kumar, J. Clean. Prod. 250 (2020) 119553.

  24. [24]

      V. Vinayagam, K.N. Palani, S. Ganesh, et al., Environ. Res. 240 (2024) 117500.

  25. [25]

      F. Li, K. Liu, Y. Bao, et al., Water Res. 254 (2024) 121373.

  26. [26]

      Y. Huang, K. Zhu, Z. Hu, et al., J. Hazard. Mater. 466 (2024) 133611.

  27. [27]

      Z.Y. Yu, G. Sun, Y.J. Liu, et al., Ecotoxicol. Environ. Saf. 135 (2017) 312–318.

  28. [28]

      X.Y. Du, J.Y. Yang, Sci. Total Environ. 919 (2024) 170745.

  29. [29]

      M. Burkhardt, M. Rohr, I.H. und, S. Gartiser, K.W. Korrespond, Wasserwirtsch 08 (2020) 001.

  30. [30]

      W. Brack, B.I. Escher, E. Müller, et al., Environ. Sci. Eur. 30 (2018) 33.

  31. [31]

      M.A. Aljallal, Q. Chaudhry, N.R. Price, Sci. Rep. 14 (2024) 18330.

  32. [32]

      I.A. Vasiliadou, R. Molina, F. Martinez, et al., Sci. Total Environ. 630 (2018) 809–819.

  33. [33]

      R. Ricky, S. Shanthakumar, Sci. Rep. 13 (2023) 13911.

  34. [34]

      K.E. Manz, A. Feerick, J.M. Braun, et al., J. Expo. Sci. Environ. Epidemiol. 33 (2023) 524–536. doi: 10.1038/s41370-023-00574-6

  35. [35]

      X. Li, G. Li, B. Chen, et al., Environ. Sci. Pollut. 28 (2021) 57530–57542. doi: 10.1007/s11356-021-14672-1

  36. [36]

      H. Ghazal, E. Koumaki, J. Hoslett, et al., J. Clean. Prod. 361 (2022) 132079.

  37. [37]

      Q. Lu, J. Lu, J. Lei, et al., Chin. Chem. Lett. 36 (2025) 110017.

  38. [38]

      W. Yao, Y. Qi, Y. Han, et al., Water Res. 249 (2024) 120881.

  39. [39]

      J. Lu, Y. Zhang, J. Wu, et al., Environ. Pollut. 252 (2019) 450–460.

  40. [40]

      S. Liu, Q. Xu, S. Lou, et al., Ecotoxicol. Environ. Saf. 259 (2023) 115025.

  41. [41]

      S. Babić, L. Ćurković, D. Ljubas, M. Čizmić, Curr. Opin. Green Sustain. Chem. 6 (2017) 34–41.

  42. [42]

      Q. Zhang, Y. Peng, Y. Peng, et al., Water Res. 249 (2024) 120931.

  43. [43]

      O.A. Alsager, M.N. Alnajrani, H.A. Abuelizz, I.A. Aldaghmani, Ecotoxicol. Environ. Saf. 158 (2018) 114–122.

  44. [44]

      J. Fu, N. Liu, L. Mei, et al., Sci. Rep. 9 (2019) 18734.

  45. [45]

      A. Wang, Z. Zheng, H. Wang, et al., Appl. Catal. B 277 (2020) 119171.

  46. [46]

      P. Kumari, N. Bahadur, X.A. Conlan, et al., Water Res. 218 (2022) 118519.

  47. [47]

      A. Huang, M. Yan, J. Lin, et al., Int. J. Environ. Res. Public. Health 18 (2021) 4909. doi: 10.3390/ijerph18094909

  48. [48]

      Y. Zhang, J. Guo, X. Zou, et al., Chem. Eng. J. 504 (2025) 159042.

  49. [49]

      D. Cheng, H.H. Ngo, W. Guo, et al., J. Hazard. Mater. 387 (2020) 121682.

  50. [50]

      T. Zhang, Y. Hu, L. Jiang, et al., Chem. Eng. J. 358 (2019) 589–597.

  51. [51]

      J. Lu, T. Wang, Y. Zhou, et al., J. Hazard. Mater. 383 (2020) 121133.

  52. [52]

      Y. Liu, Y. Lin, W. Wang, et al., Environ. Sci. Technol. 58 (2024) 8009–8019. doi: 10.1021/acs.est.3c09086

  53. [53]

      B. Gokulakrishnan, G. Satishkumar, Sep. Purif. Technol. 333 (2024) 125690.

  54. [54]

      X. Chen, L. Yao, J. He, et al., J. Hazard. Mater. 449 (2023) 131024.

  55. [55]

      J. Cao, J. Li, B. Yang, et al., Cell. Rep. Phys. Sci. 5 (2024) 101966.

  56. [56]

      S. Pei, S. Wang, Y. Lu, et al., Nano Res. 17 (2024) 9446–9471. doi: 10.1007/s12274-024-6973-y

  57. [57]

      L. Di, X. Chen, J. Lu, et al., J. Water Process. Eng. 53 (2023) 103913.

  58. [58]

      X. Chen, C. Duan, Y. Zhou, et al., J. Clean. Prod. 395 (2023) 136323.

  59. [59]

      W. Gu, X. Huang, Y. Tian, et al., Appl. Surf. Sci. 538 (2021) 147813.

  60. [60]

      R. Gonzalez-Olmos, F.D. Kopinke, K. Mackenzie, A. Georgi, Environ. Sci. Technol. 47 (2013) 2353–2360. doi: 10.1021/es303885y

  61. [61]

      J. He, C.J. Miller, R. Collins, et al., Environ. Sci. Technol. 54 (2020) 1167–1176. doi: 10.1021/acs.est.9b03975

  62. [62]

      J. He, X. Yang, B. Men, et al., Chem. Eng. J. 258 (2014) 433–441.

  63. [63]

      Y. Chen, C.J. Miller, T.D. Waite, Environ. Sci. Technol. 55 (2021) 14414–14425. doi: 10.1021/acs.est.1c00284

  64. [64]

      X. Nie, G. Li, S. Li, et al., Appl. Catal. B 300 (2022) 120734.

  65. [65]

      R.C.C. Costa, M.F.F. Lelis, L.C.A. Oliveira, et al., J. Hazard. Mater. 129 (2006) 171–178.

  66. [66]

      P. Baldrian, V. Merhautová, J. Gabriel, et al., Appl. Catal. B 66 (2006) 258–264.

  67. [67]

      Y. Zhou, Y. He, M. Gao, et al., Chin. Chem. Lett. 35 (2024) 108690.

  68. [68]

      C. Lai, F. Huang, G. Zeng, et al., Chemosphere 224 (2019) 910–921.

  69. [69]

      P. Manjuri Bhuyan, S. Borah, B. Kumar Bhuyan, et al., Sep. Purif. Technol. 312 (2023) 123387.

  70. [70]

      J. Scaria, K.V. Anupama, P.V. Nidheesh, Sci. Total Environ. 771 (2021) 145291.

  71. [71]

      T. Wang, J. He, J. Lu, et al., Chin. Chem. Lett. 33 (2022) 3585–3593.

  72. [72]

      J. Wu, J. Wang, C. Liu, et al., Environ. Sci. Technol. 56 (2022) 13996–14007. doi: 10.1021/acs.est.2c03590

  73. [73]

      V. Jurado-Davila, K.G.P. Nunes, G.P. Oshiro, L.A. Féris, J. Environ. Chem. Eng. 11 (2023) 111423.

  74. [74]

      W. Zhao, B. Zhou, Sep. Purif. Technol. 259 (2021) 118218.

  75. [75]

      K. Lv, L. Ling, Q. Lu, et al., Sep. Purif. Technol. 344 (2024) 127207.

  76. [76]

      Q. Chen, F. Lv, H. Zhang, P. He, Water Res. 229 (2023) 119479.

  77. [77]

      X. Song, M. Zhang, X. Xiu, et al., J. Environ. Manage. 349 (2024) 119541.

  78. [78]

      H. Liu, M. Cui, Y. Liu, et al., J. Ind. Eng. Chem. 125 (2023) 360–369.

  79. [79]

      M. Xu, J. Wei, X. Chen, et al., J. Environ. Chem. Eng. 10 (2022) 108409.

  80. [80]

      G. Yang, Y. Liang, H. Zheng, et al., Sep. Purif. Technol. 310 (2023) 123167.

  81. [81]

      G. Yang, Y. Liang, Z. Xiong, et al., Chem. Eng. J. 425 (2021) 130689.

  82. [82]

      L. Di, T. Wang, Q. Lu, et al., Sep. Purif. Technol. 339 (2024) 126740.

  83. [83]

      J. Lu, Y. Zhou, L. Ling, Y. Zhou, Chem. Eng. J. 446 (2022) 137067.

  84. [84]

      Q. Lu, J. Lu, Y. Zhou, Y. Zhou, Sep. Purif. Technol. 358 (2025) 130458.

  85. [85]

      X. Ma, X. Liu, X. Shang, et al., J. Hazard. Mater. 469 (2024) 134087.

  86. [86]

      Z. Wang, Y. Du, T. Liu, et al., Environ. Sci. Technol. 58 (2024) 4812–4823. doi: 10.1021/acs.est.3c10800

  87. [87]

      J. Yao, B. Yu, H. Li, et al., Chem. Eng. Sci. 276 (2023) 118779.

  88. [88]

      Y. Zhou, X. Fang, Z. Zhang, et al., Environ. Technol. 38 (2017) 2793–2799.

  89. [89]

      X.Y. Zhang, T.S. Liu, J.Y. Hu, Water Res. 261 (2024) 122069.

  90. [90]

      A. Tufail, J. Al-Rifai, W.E. Price, et al., Chemosphere 350 (2024) 140978.

  91. [91]

      N.J. Zhu, S. Ghosh, M.A. Edwards, A. Pruden, Environ. Sci. Technol. 55 (2021) 8329–8340. doi: 10.1021/acs.est.1c01199

  92. [92]

      M. Ahmed, M.O. Mavukkandy, A. Giwa, et al., npj Clean Water 5 (2022) 12. doi: 10.22271/plants.2022.v10.i1a.1355

  93. [93]

      Y. Yu, Y. Zhou, Z. Wang, et al., Sci. Rep. 7 (2017) 4168.

  94. [94]

      V.M. D’Costa, C.E. King, L. Kalan, et al., Nature 477 (2011) 457–461. doi: 10.1038/nature10388

  95. [95]

      Z.J. Lin, Z.C. Zhou, X.-Y. Shuai, et al., Environ. Sci. Technol. 58 (2024) 6793–6803. doi: 10.1021/acs.est.3c10895

  96. [96]

      Q.H. Liu, L. Yuan, Z.H. Li, et al., Environ. Sci. Technol. 58 (2024) 17990–17998. doi: 10.1021/acs.est.4c08211

  97. [97]

      S. Li, Y. Zhu, G. Zhong, et al., Environ. Sci. Technol. 58 (2024) 5534–5547. doi: 10.1021/acs.est.4c00993

  98. [98]

      X. Deng, J. Chen, L.-A. Hansson, et al., Natl. Sci. Rev. 8 (2021) nwaa140.

  99. [99]

      Y. Yang, Z.H. Xie, H. Wang, et al., Sci. Total Environ. 950 (2024) 175401.

  100. [100]

       X. Liu, X. Zheng, L. Zhang, et al., J. Hazard. Mater. 437 (2022) 129355.

  101. [101]

       S. Zhang, Y. Han, J. Peng, et al., Environ. Int. 171 (2023) 107700.

  102. [102]

       M. Gustavsson, S. Käll, P. Svedberg, et al., Sci. Adv. 10 (2024) eadk6669.

  103. [103]

       S. Li, Y. Wu, H. Zheng, et al., Chemosphere 311 (2023) 136977.

  104. [104]

       S.J. Markich, J.P. Hall, J.M. Dorsman, P.L. Brown, Mar. Pollut. Bull. 192 (2023) 114964.

  105. [105]

       D. Giesen, C.A.M. van Gestel, Chemosphere 90 (2013) 2667–2673.

  106. [106]

       D. Dai, C. Brown, H. Bürgmann, et al., Microbiome 10 (2022) 20.

  107. [107]

       T.U. Berendonk, C.M. Manaia, C. Merlin, et al., Nat. Rev. Microbiol. 13 (2015) 310–317. doi: 10.1038/nrmicro3439

  108. [108]

       W. Li, C. Liu, H.C. Ho, et al., Lancet. Reg. Health. West. Pac. 30 (2023) 100628.

  109. [109]

       R. Li, J. Lu, Z. Liu, et al., Nat. Mach. Intell. 6 (2024) 1457–1466. doi: 10.1038/s42256-024-00923-6

  110. [110]

       N. Wang, G. Dong, R. Qiao, et al., Environ. Sci. Technol. 58 (2024) 9487–9499. doi: 10.1021/acs.est.4c00480

  111. [111]

       P. Zhang, M. Sun, C. Zhou, et al., Environ. Sci. Technol. 58 (2024) 4781–4791. doi: 10.1021/acs.est.3c06252

  112. [112]

       Z. Jinmiao, Z. Ziting, Y. Saihong, Acta Sci. Circumst. 42 (2022) 1–11.

  113. [113]

       J. Wang, D. Zhi, H. Zhou, et al., Water Res. 137 (2018) 324–334.

  114. [114]

       L. Zhilei, F. Yongjie, W. Chunshan, et al., Environ. Chem. 43 (2024) 82–91.

  115. [115]

       J. Neuwoehner, T. Zilberman, K. Fenner, B.I. Escher, Aquat. Toxicol. 97 (2010) 58–67.

  116. [116]

       D.E. Dawson, B.L. Ingle, K.A. Phillips, et al., Environ. Sci. Technol. 55 (2021) 6505–6517. doi: 10.1021/acs.est.0c06117

  117. [117]

       N. Klüver, C. Vogs, R. Altenburger, et al., Chemosphere 164 (2016) 164–173.

  118. [118]

       J. Dan, Z. Jianguo, L. Na, et al., Asian J. Ecotoxicol. Chemosphere 9 (2014) 71–80.

  119. [119]

       C. Hansch, A. Kurup, R. Garg, H. Gao, Chem. Rev. 101 (2001) 619–672.

  120. [120]

       H. Liu, E. Papa, P. Gramatica, Chem. Res. Toxicol. 19 (2006) 1540–1548. doi: 10.1021/tx0601509

  121. [121]

       Y.K Wang, QSAR Integrated Modeling Technology in ADMET Evaluation and Acute Toxicity Prediction, University of Science and Technology Liaoning, Liaoning, China, 2022.

  122. [122]

       X. Wu, Q. Zhang, J. Hu, SAR QSAR Environ. Res. 27 (2016) 147–164.

  123. [123]

       A. Nath, P. De, K. Roy, Chemosphere. 287 (2022) 131954.

  124. [124]

       M. Nendza, M. Müller, SAR QSAR Environ. Res. 18 (2007) 155–168. doi: 10.1080/10629360601054354

  125. [125]

       M. Nendza, A. Wenzel, Environ. Sci. Pollut. Res. 13 (2006) 192–203. doi: 10.1065/espr2006.01.013

  126. [126]

       S. Chen, G. Sun, T. Fan, et al., Sci. Total Environ. 876 (2023) 162736.

  127. [127]

       L.T. Qin, Y.H. Chen, X. Zhang, et al., Chemosphere 198 (2018) 122–129.

  128. [128]

       R. He, X. Wu, H. Mu, et al., Water Res. 243 (2023) 120338.

  129. [129]

       L. Li, J.N. Westgate, L. Hughes, et al., Environ. Sci. Technol. 52 (2018) 14235–14244. doi: 10.1021/acs.est.8b04059

  130. [130]

       Z. Cheng, Q. Chen, S. Liu, et al., J. Hazard. Mater. 430 (2022) 128269.

• 

  Figure 1  Classification of AOPs using Fenton methods for antibiotic degradation.

  下载: 全尺寸图片 幻灯片
  

  Figure 2  Keywords and publication of Fenton approaches on antibiotic degradation since 2000. Data sources: top 500 sorted by citation high-count in Web of Science Core Collection, from 2000 to 2024. The pie chart depicts the percentage of Fenton’s method in each category.

  下载: 全尺寸图片 幻灯片
  

  Figure 3  (a) TEM images of Fe₃O₄, (b) Fe₃O₄@MoS₂–3. (c, d) Demonstration of Fe₃O₄ and Fe₃O₄@MoS₂–3. (e) Effect of peroxymonosulfate (PMS) dose on the degradation of SA by Fe₃O₄@MoS₂–3. Copied with permission [18]. Copyright 2021, Chemical Engineering Journal.

  下载: 全尺寸图片 幻灯片
  

  Figure 4  Pathways of electron transfer for Fe(Ⅲ) reduction facilitated by carbon materials in antibiotic degradation.

  下载: 全尺寸图片 幻灯片
  

  Figure 5  Block diagram of QSAR research method.

  下载: 全尺寸图片 幻灯片

  Table 1.  Effectiveness of different classes of Fenton systems in removing antibiotics, toxicity reduction.

  Catalyst	Antibiotic	Main ROS	Efficiency	Highly toxic intermediates	Highly toxic intermediate	Overall toxicity
  Pt-FeOX/G [77]	SDZ	•OH	98%, 10 min	B2 (m/z 96) is highly toxic to both daphinds and green algae with higher chronic toxicity than SD itself. However, the majority of its subsequent-generation products are less detrimental than SD, indicating that most generated products were less harmful than SD.		As intermediates continued to mineralize, toxicity towards luminescent bacteria rapidly decreased, with inhibition rates dropping to as low as 4.37%, corresponding to improved TOC removal.
  Fe2O3/MXene-x [78]	SMX	•OH and •O2–	95%, 20 min	The developmental toxicity of compound P285 (m/z 284) exceeded that of SMX. Moreover, the developmental toxicity of the remaining degradation intermediates was notably diminished or completely absent.		A significant reduction in overall toxicity was observed within 30 min for most intermediates, with developmental toxicity and bioaccumulation factors lower than SMX, except for P285.
  MnFe-LDH@BC [79]	TC	•OH	94.2%, 500 min	The ChV content of intermediate products in various degradation pathways exceeded 0.000002 mol/L, indicating potential relevance for organisms such as fish, Daphnia, and green algae due to their weak chronic toxicity.		Toxicity of TC intermediates decreased over time, with acute toxicity remaining low, while chronic toxicity was eventually eliminated after complete mineralization.
  Fe-POM/BMO [80]	TC	•OH	99.76%, 18 min	Among the processed products, TC(Ⅶ) (m/z 417), TC(Ⅷ) (m/z 343), and TC(Ⅹ) (m/z 402) exhibited higher acute toxicity towards the blackhead minnow Daphnia magna compared to TC, along with a greater predicted bioaccumulation factor.		Acute toxicity, bioaccumulation, and mutagenicity significantly decreased after degradation, with luminescent bacteria inhibition dropping from 56% to 21%, indicating reduced environmental risks.
  Ag3PO4/MIL-101 [54]	TC	•OH and 1O2	91%, 25 min	P2 (m/z 431), P4 (m/z 417), and P12 (m/z 277) exhibited higher acute toxicity levels compared to TC in rats, black-headed minnow, and Daphnia magna, whether administered orally or through other means.		In the Ag3PO4 system, some intermediates showed higher toxicity than TC, but toxicity decreased in the AP/MF-0.5 systems over time.
  Ce4O7/Bi4MoO9 [81]	TC	•OH and •O2–	100%, 120 min	P(Ⅶ) (m/z 417) and P(Ⅷ) (m/z 402) exhibited significant acute toxicity towards black-headed minnow (Daphnia magna) and orally administered rats.		Although some intermediates showed higher toxicity during the Photo-Fenton degradation of tetracycline, most intermediates exhibited lower toxicity over time, with bioaccumulation reduced and toxicity towards aquatic organisms mitigated.

  下载: 导出CSV

  Table 2.  QSAR models for common toxicity prediction.

  Compound	Equation	R2
  126 kinds of organic volatile molecules [123]	Model 1: pNOAEC = 6.906–0.605 × nF - 0.044 × H% + 1.082 × Psi_i_0d + 2.065 × X3A-0.2897 × nR	0.743
  	Model 2: pNOAEC = 7.195 - 0.620 × nF - 0.049 × H% + 1.711 × Psi_i_0d + 2.124 × X3A - 0.319 × nR	0.739
  	Model 3: pNOAEC = 6.127 - 0.491 × nF - 0.034 × H% + 0.283 × nHM + 1.574 × Psi_i_0d + 1.800 × X3A + 0.422 × O-060 + 0.859 × B01[C—N] + 0.755 × B01[C—O] + 2.087 × B05[C—N]	0.747
  15 kinds of organophosphates [118]	-logEC50 = 2.979 + 0.246P + 0.220logD + 0.204NAr + 0.186TSA - 0.129LUMO - 0.084NA + 0.080ND	0.90

  下载: 导出CSV
•