Fresh biomass-derived adsorbents represent a sustainable alternative for wastewater treatment; however, their practical application is often hindered by limited recoverability and insufficient process optimization. In this study, a magnetically recoverable adsorbent (M-DSWP) was prepared from fresh durian shell waste through a simple Fe₃O₄ impregnation route without carbonization or chemical activation and applied for methylene blue (MB) removal from water. The adsorption process was comprehensively evaluated through conventional optimization, multi-response response surface methodology (RSM), adsorption modeling, regeneration studies, real-water treatment, and fixed-bed column experiments. Multi-response RSM was employed to simultaneously optimize MB removal efficiency (%RE) and adsorption capacity (qₑ), yielding optimum conditions of 25 °C, 90min, pH 9.0, and 0.15g adsorbent dosage. Under these conditions, experimental %RE and qₑ values of 88.36 ± 0.30% and 14.86 ± 0.05mgg⁻¹ were achieved, respectively. Adsorption was governed primarily by electrostatic interactions, with additional contributions from hydrogen bonding and π–π interactions, as supported by FTIR, XPS, pHpzc, and interference studies. M-DSWP exhibited excellent reusability, maintaining removal efficiencies above 83% after ten adsorption–desorption cycles. The adsorbent remained effective in diverse water matrices, achieving MB removal efficiencies of 52.31–98.78% in real water samples. In continuous-flow operation, overall removal efficiencies of 95.31% and 98.95% were obtained for standard MB solution and MB-spiked wastewater, respectively. The combination of resource-efficient preparation, magnetic recoverability, multi-response optimization, regeneration stability, and successful application under batch and continuous-flow conditions highlights the potential of M-DSWP as a sustainable adsorbent for practical wastewater treatment.
The present work focuses on the synthesis of sustainable Fe-based biochar using bagasse and red mud and examines its efficacy in a column-based system for Se (IV) removal. The adsorption performance of a fixed-bed column is investigated under varying bed heights of 1, 3, and 5cm, flow rates of 1, 2, and 3ml/min, and initial concentrations of 1 and 5mg/l. Experimental breakthrough curves are fitted with three different kinetic models, such as Thomas, Adams Bohart, and Yoon Nelson model. The models described the breakthrough behavior, predicting a maximum adsorption capacity (qo) of 3.98mg/g, bed saturation concentration (No) of 623.7mg/l, and 50% breakthrough time (τ) of 990min at optimized parameters. To further understand the adsorption mechanism, density functional theory (DFT) analysis is performed on the aqueous Se (IV) species. The results revealed that stable complexes are formed between biochar and Se (IV) forms i.e., SeO32- and HSeO3- with the adsorption energies of -29.56kcal/mol and -21.05kcal/mol, respectively. SeO32- adsorption was thermodynamically more favorable than HSeO3- adsorption, indicating a stronger affinity of the biochar toward the Se (IV) species. The combined experimental and theoretical findings demonstrate that Fe-based biochar is a promising adsorbent for Se (IV) removal and provides a sustainable pathway for the valorization of agricultural and industrial residues for water treatment applications.
This study presents a lab-scale integrated biomass valorization approach for the sequential recovery of value-added biomolecules from Schizochytrium aggregatum biomass, followed by utilization of the residual biomass for heavy metal remediation. The approach was designed to recover lipids, proteins, and carbohydrates sequentially while exploring the potential of the residual biomass for biosorption applications. Lipid extraction achieved a maximum yield of 48 ± 1.1% using ultrasonic probe-assisted extraction, providing higher recovery than the other extraction methods evaluated. Fatty acid profiling revealed docosahexaenoic acid (DHA) and palmitic acid as dominant components, highlighting the biomass potential for nutraceutical and pharmaceutical applications. Sequential extraction of proteins and carbohydrates further enabled comprehensive biomass valorization. Proximate composition and CHNS analysis confirmed efficient biomolecule recovery across fractions. The residual biomass demonstrated effective adsorption of chromium and cadmium ions, supported by surface and structural characterization. Unlike conventional biorefinery strategies focused solely on biomolecule recovery, this integrated waste-minimizing biorefinery approach enhances biomass utilization efficiency and promotes circular bioeconomy principles, aligning with sustainable waste valorization and water remediation strategies. This approach aligns with the United Nations Sustainable Development Goals, especially SDG 12 (Responsible Consumption and Production) and SDG 6 (Clean Water and Sanitation), providing a lab scale framework for industrial and environmental applications.
Despite containing high levels of functional polyphenols, fallen leaves are regarded as low-value waste in urban environments. We investigated ultrasound-assisted extraction as an approach for recovering functional polyphenolics from urban leaf biomass, with particular emphasis on biomass type. Hardwood fallen-leaf extract (HWLE), softwood fallen-leaf extract (SWLE), and heterogeneous fallen-leaf extract (HTLE) were evaluated in terms of polyphenolic recovery and functional performance. The effects of ethanol concentration (40–80%), ultrasonic amplitude (40–60%), and extraction time were initially screened based on extraction yield, followed by comparative evaluation of total polyphenol, tannin, and condensed tannin contents under selected representative conditions. The highest polyphenolic recovery among the evaluated conditions was achieved using 60% ethanol with 60% ultrasonic amplitude for 20min. The HWLE extract exhibited the highest polyphenolic content, followed by the HTLE and SWLE extracts. However, HTLE retained substantial antioxidant and antibacterial activities despite its lower polyphenolic content relative to HWLE. Targeted LC–MS/MS analysis of HTLE identified ellagic and gallic acids as the most abundant among the quantified phenolic compounds, along with flavonols such as quercetin and kaempferol. Structural and thermal analyses further characterized the chemical heterogeneity and thermal behavior of the leaf-derived extracts. Importantly, these findings indicate that, under the conditions evaluated in this study, heterogeneous fallen-leaf biomass can retain substantial polyphenolic recovery and functional activity without prior separation into hardwood and softwood fractions. Overall, this study provides a comparative experimental basis for the potential valorization of heterogeneous urban fallen-leaf biomass as a source of functional polyphenolic compounds.
Accurately characterizing the 3D spatial distribution of soil heavy metal contamination is fundamental for remediation-boundary delineation and risk management at industrial legacy sites. However, intensive drilling and sampling campaigns are costly, whereas excessive sample reduction may miss contamination hotspots and distort the spatial distribution and morphology of contamination bodies. To address this issue, this study developed an integrated framework coupling quantum-inspired annealing-based sampling optimization with high-precision 3D contamination characterization. A multi-objective evaluation system integrating spatial distribution similarity, 3D volumetric overlap, pollution-risk consistency, and heavy-metal migration consistency was established, and a quantum-inspired annealing algorithm (QIAA) was employed to identify representative sampling points under 30%–80% retention ratios. The framework was applied to a decommissioned chemical industrial park in southern China, where 1,575 soil samples were analyzed for eight heavy metals. Results showed that retaining 50% of the original sampling locations achieved the optimal balance between cost reduction and information preservation. Under this retention level, the sampling network was reduced to 787 points while preserving essential spatial coverage and vertical distribution characteristics. Compared with genetic algorithm, particle swarm optimization, and simulated annealing, the QIAA-based approach achieved higher reconstruction accuracy and lower prediction error under the same sampling density. The 3D reconstruction and hotspot analysis further confirmed that the optimized sampling design preserved the morphology, spatial extent, and core high-risk clusters of major contamination bodies. Overall, the proposed framework provides a robust, reproducible, and cost-effective approach for contaminated-site investigation, 3D pollution characterization, hotspot identification, and remediation-boundary delineation.
Innovations in bioenergy are a key tool for reducing carbon dioxide (CO₂) emissions and are critical to achieving Nationally Determined Contributions (NDCs) under the Paris Agreement. However, existing research largely relies on single-equation specifications in which innovation is treated as exogenous to emissions, with little testing of their joint determination. This study addresses this by simultaneously estimating the relationship using established econometric methods. The balanced panel dataset covers 49 countries (26 developed, 23 developing) from 2000 to 2023 and examines CO₂ emissions per capita and bioenergy patents together. Results indicate statistically significant effects in both directions: reducing CO₂ emissions reduces bioenergy innovation, and larger increases in CO₂ emissions reduce bioenergy patenting. NDC commitments trigger bioenergy patenting in developing economies but have a negative net effect in developed economies, consistent with reduced policy effectiveness in countries with more mature innovation systems. Contrary to the Environmental Kuznets Curve (EKC), the relationship between income and emissions suggests that emissions may rise again as income continues to increase. Many developing economies exhibit high biomass potential but low institutional capacity. Policy approaches should be development–sensitive, highlighting frontier research and development (R&D) and innovation diversification in developed economies, while stimulating targeted technology transfer, NDC-linked finance, and R&D capacity building in developing economies.
The disinfection of drinking water lowers pathogenic infection risks, but it may introduce chemical hazards from the formation of disinfection byproducts (DBPs), particularly via the reaction of the disinfectant with natural organic matter in the aqueous phase. Increased exposure to DBPs poses significant risks to human health, owing to their pronounced cytotoxic, mutagenic, and carcinogenic properties. Therefore, efficient real-time detection of these micropollutants remains a challenge.In this study, a portable sensing system for the simultaneous detection of multiple environmentally relevant DBPs in water was developed. Advanced mid-infrared fiber optic evanescent wave spectroscopy (MIR-FEWS) system facilitates the in situ simultaneous detection and quantification of multiple DBPs, including trihalomethanes, haloacetic acids, haloacetonitriles, oxyhalide compounds, and haloketones. The active optical transducer (an MIR-transparent silver halide fiber sensing element) was coated with a polymeric molecular recognition membrane (namely poly(styrene-co-butadiene)) mounted inside a micro-flow cell and coupled to a compact broadband Fourier transform infrared spectrometer. The hydrophobic polymer coating enriches non-polar DBPs within the penetration depth of the evanescent wave and improves the limit of detection of the fifteen tested DBPs. Linear calibration functions of simultaneously detected DBPs revealed limits of detection at low ppb concentration levels close to the EU regulatory limits of drinking water. Such an innovative label-free analytical platform with improved detection sensitivity is an essential tool for the safety of water sources. The developed MIR-FEWS sensing prototype was successfully tested in the first-field deployment study, demonstrating strong potential for long-term, real-time, low-ppb in situ detection of multicomponent DBPs.
Plant diseases pose a major threat to global food security by reducing crop productivity and increasing reliance on chemical pesticides. However, conventional agrochemicals often suffer from poor stability, low target specificity, rapid degradation, and environmental toxicity, highlighting the need for sustainable crop protection strategies. Natural polymeric nanotherapeutics have emerged as promising alternatives due to their biodegradability, biocompatibility, and structural versatility. This review summarizes recent advances in natural polymer-based engineered nanomaterials (ENMs) for plant disease management. Major classes of natural polymers, including polysaccharides (e.g., chitosan, alginate, cellulose, starch, pectin, and dextran), proteins, lipids, and lignin-derived materials, are discussed as versatile nanoplatforms for agrochemical delivery. These materials enable efficient encapsulation, improved stability, and controlled release of bioactive compounds. The review further highlights the antimicrobial mechanisms of polymeric nanotherapeutics, including reactive oxygen species generation, membrane disruption, and metabolic interference, as well as indirect effects through activation of plant immune responses such as induced systemic resistance and systemic acquired resistance. In addition, the uptake, translocation, and transformation of polymeric nanoparticles within plant systems are discussed to better understand their bioavailability and protective efficacy. Emerging innovations, including stimuli-responsive nanoformulations and artificial intelligence-assisted design, are also highlighted, offering promising opportunities for developing sustainable and precise crop protection technologies.
Per- and polyfluoroalkyl substances (PFAS) are persistent contaminants that are challenging to remove using conventional water-treatment processes. In this study, coal fly ash (FA) was functionalized with cetyltrimethylammonium bromide (CTAB) and subsequently loaded with Fe3O4 to produce a waste-derived, magnetically responsive adsorbent for perfluorooctanoic acid (PFOA) removal. FTIR, XRD, SEM–EDS, XPS, and N2 adsorption–desorption analyses confirmed CTAB functionalization, Fe3O4 incorporation, and the development of a mesoporous composite structure. Mag–CTAB–FA exhibited a BET surface area of 82.35 m2 g−1, a total pore volume of 0.224 cm3 g⁻1, and an estimated mean pore diameter of 10.88nm. A CTAB:FA ratio of 1:10 provided the highest performance, while the final Mag–CTAB–FA composite achieved 88.3% PFOA removal under the selected screening conditions. Adsorption increased markedly with adsorbent dosage, remained effective over a broad pH range, and was only slightly affected by temperature. Equilibrium uptake increased to a maximum experimentally measured capacity of 32.03mgg⁻1 at an initial PFOA concentration of 100mgL⁻1. The Langmuir model provided a suitable representation of the equilibrium data and estimated a maximum adsorption capacity of 38.10mgg⁻1. Kinetic analysis showed rapid initial uptake, with the fractal-like Vermeulen model providing the best statistical description of the experimental data. The multilinear Weber–Morris plot revealed three uptake stages and indicated concurrent contributions from boundary-layer transport and intraparticle diffusion. Fe release remained below 28.39μgL⁻1 and 0.00011% over 192h, supporting the aqueous stability of the magnetic phase. The adsorbent retained 89.7% of its initial capacity after five adsorption–desorption cycles and achieved up to 99.9% PFOA removal from spiked municipal wastewater at sufficient dosage. Overall, Mag–CTAB–FA combines industrial-waste valorization, high PFOA removal, magnetic responsiveness, limited Fe release, and repeated-use capability, demonstrating its potential for further development in water-treatment applications.
This multi-year study evaluated airborne shortwave infrared (SWIR) imaging spectroscopy and repeated flux-chamber measurements for characterizing methane emissions at a municipal solid waste landfill during the transition from active operation to post-closure conditions. The objectives were to assess the field performance and repeatability of the general-purpose SASI-600 airborne SWIR imaging spectrometer, characterize the spatial and temporal variability of chamber-derived methane surface fluxes across selected landfill locations, and assess the observational capabilities and limitations of the two approaches. Across five airborne acquisitions and six ground-based campaigns, methane column enhancements reached approximately 250 ppm·m near the source and decreased to about 50 ppm·m within detected plumes. Integrated mass enhancement estimates indicated emission rates of 7.8–15.1kg CH₄ h⁻¹. Flux-chamber measurements showed strong spatial variability, with fluxes reaching approximately 15,000mg CH₄ m⁻² d⁻¹ in newly deposited waste mixed with soil and in waste deposited 1–3 years previously, and up to 38,000mg CH₄ m⁻² d⁻¹ in mixed-age waste comprising material up to 10 years old. High chamber-derived fluxes did not consistently correspond to detectable airborne enhancements, indicating that spatially confined, intermittent, or rapidly dispersed emissions may remain undetected during an airborne overpass. Conversely, airborne imaging enabled site-wide screening and detected localized plume-forming sources outside the chamber sampling locations. These results provide site-specific evidence that repeated airborne and surface measurements, can support landfill methane-emission characterization and monitoring during the transition to post-closure conditions and highlight the potential for integrating airborne enhancement detection with field-based flux measurements in landfill environments.
The accumulation of copper tailings has caused significant environmental concerns due to land occupation and potential heavy metal risks, while efficient control of elemental mercury (Hg0) emissions from industrial flue gas remains challenging. This study presents a waste-to-resource strategy by converting thermally activated copper tailings into multifunctional catalytic adsorptive materials for Hg0 removal through a wet H₂O₂/K₂S₂O₈ oxidation system. The abundant mineral components and transition metal species (Fe and Mn) in copper tailings provided catalytic active sites for oxidant activation, promoting the transformation of Hg0 into stable oxidized mercury species. Under optimized conditions (solid–liquid ratio of 10%, reaction temperature of 45 °C, SO₂ concentration of 2000 mg/m³, NOₓ concentration of 600mg/m³, and O₂ content of 10%), the catalytic system achieved a Hg0 removal efficiency of 91.99%. The enhanced performance was attributed to the synergistic interaction between activated copper tailings and composite oxidants, as well as the improved surface reactivity of the catalyst. Furthermore, the catalyst exhibited good resistance to flue gas components and operational stability, demonstrating the feasibility of copper tailings as low-cost catalysts for mercury emission control. This work provides a sustainable pathway for simultaneous copper tailings valorization and atmospheric mercury pollution mitigation.
Microplastics can modify the behavior and ecological effects of coexisting heavy metals in soil, but the response of microplastic–heavy-metal systems to [S,S]-ethylenediamine disuccinic acid (EDDS) remains poorly understood. A pot experiment was conducted using sorghum in soil co-contaminated with cadmium (Cd), lead (Pb), and polybutylene succinate (PBS). Different cumulative EDDS doses were applied to evaluate treatment-associated changes in plant growth, metal accumulation, soil properties, and rhizosphere bacterial and fungal communities. Pb-Cd co-contamination significantly inhibited sorghum growth, and, under the tested heavy-metal condition, PBS addition was associated with greater growth inhibition. EDDS slightly increased shoot length in some treatments but did not consistently improve biomass. EDDS treatments generally increased DTPA-extractable Cd and Pb, but these increases did not consistently translate into greater plant metal accumulation because biomass was not simultaneously improved. Increasing PBS loading under the fixed EDDS condition was associated with greater metal extractability but progressively lower sorghum biomass, indicating a trade-off between metal mobilization and plant performance. Treatment groups differing in PBS and EDDS levels also exhibited distinct rhizosphere soil properties and microbial community structures. Bacterial community differentiation was more pronounced among treatments, whereas changes in fungal richness indices were relatively prominent. DTPA-extractable Pb and Cd, available potassium, and available phosphorus were significantly associated with rhizosphere microbial community variation. Variations in DTPA-extractable metals were also associated with plant metal accumulation patterns. These findings characterize treatment-associated plant–soil–microbial responses to EDDS under the tested PBS–Pb–Cd co-contamination conditions and highlight a trade-off between metal mobilization and maintenance of plant performance during chelator-assisted phytoextraction.
Antibiotic residues persist in water because continuous discharge, molecular stability and the formation of bioactive transformation products cannot be eliminated by adsorption alone. Nevertheless, integrating pollutant enrichment with in situ photocatalytic conversion in a single material remains difficult because suspension photocatalysis is hindered by weak interfacial affinity, short radical lifetimes, and the spatial mismatch between adsorbed pollutants and photoactive domains. Herein, we knit triazine units into waste-plastic-derived hyper-crosslinked polymers (HCP-Ns) to couple microporous enrichment with visible-light-driven tetracycline (TC) oxidation, in which the HCP skeleton supplies the enrichment microenvironment while triazine domains concurrently tune the porous architecture, narrow the optical band gap, and accelerate interfacial charge transfer to drive oxygen activation. The optimized HCP-N-4 combines a BET surface area of 1164.71 m2 g-1, a hierarchical micro/mesoporous network, and a 2.29eV band gap, and removes 92.62% of TC at pH 6.0 with 0.15gL-1 catalyst and 80mgL-1 TC. LC-MS, EPR, and scavenger experiments reveal that TC transformation proceeds through adsorption-assisted demethylation, deamidation, dehydroxylation, and ring-opening routes, with ·O2- showing the strongest measured contribution among the reactive species probed, while ·OH and h⁺ also. This work establishes a structure-activity relationship in which triazine-regulated waste-plastic polymers unify pore enrichment, charge separation and oxygen activation, linking plastic-waste valorization to the adsorption-photocatalysis synergistic removal of antibiotics.
Bauxite residue, a highly alkaline by-product of alumina production, poses significant environmental challenges worldwide. This study presents an integrated process for the comprehensive treatment of bauxite residue, coupling the recovery of valuable metals (Al and Fe) with CO2 sequestration. The process involves low-temperature roasting of bauxite residue with ammonium bisulfate, followed by water leaching, sequential recovery of Al and Fe from the leaching solution, and carbonation of the leaching residue. Under optimized roasting conditions (ammonium bisulfate dosage of 1.8 times the theoretical amount, 400 °C, 90min) and leaching conditions (95 °C, liquid−solid ratio of 3mL/g, 60min), extraction efficiencies of 95.9% for Al and 98.6% for Fe were achieved. Aluminum was selectively recovered from the leaching solution as ammonium alum via crystallization. After recrystallization under optimized conditions, the recrystallization efficiency reached 89.3% with an Fe impurity content as low as 0.068wt.%. Iron was subsequently precipitated as an Fe-rich product (42.06wt.% Fe) with >99.99% efficiency. The leaching residue was carbonated using (NH4)2CO3, achieving a carbonation efficiency of 96.28% and a CO2 sequestration capacity of 100.61kg per 1000kg of bauxite residue. This process also effectively removed sodium (from 6.31wt.% in the original bauxite residue to 0.09wt.% in the carbonated slag), rendering the slag suitable for construction materials or soil improvement. The recovered sulfates can be recycled as a roasting agent for the subsequent bauxite residue treatment. This integrated approach offers a promising circular economy strategy for the sustainable management of bauxite residue, combining valuable metal recovery with effective CO2 sequestration.
Cadmium (Cd) contamination in paddy soil threatens rice safety, but traditional rhizosphere regulation fails to effectively suppress the internal remobilization of Cd from vegetative tissues to grains during grain filling. Here, we investigated the efficacy and mechanisms of foliar-applied glycerol (GL, 0.8mM) in reducing Cd accumulation in hydroponically grown indica rice, using nano-silicon (Si, 5mM) as a comparative inhibitor. We hypothesized that GL restricts Cd translocation to grains by enhancing vacuolar sequestration and downregulating xylem loading transporters. GL significantly reduced grain Cd concentration by 39%–43%, comparable to the high-concentration Si. Under a pre-heading Cd exposure regime (HCd), where exogenous Cd was terminated at full heading stage GL still markedly reduced grain Cd concentration by 42.9%, demonstrating that its primary mode of action is suppressing internal Cd redistribution rather than root uptake. GL decreased the husk-to-rice translocation factor by 36.2% and increased Cd retention in flag leaves. Subcellular and chemical speciation analyses showed that unlike Si which primarily immobilizes Cd via cell wall adsorption, GL promotes intracellular Cd-chelate formation and vacuolar sequestration. At the molecular level, GL downregulated the xylem-loading transporter gene OsCCX2 by 82%, while upregulating phytochelatin synthase genes OsPCS1/2 and vacuolar transporters OsHMA3 and OsABCC1. Transcriptome analysis further showed that GL activated glutathione metabolism, nitrogen assimilation, and photosynthetic pathways, providing metabolic support for Cd detoxification. These findings demonstrate that foliar-applied glycerol establishes a multi-level regulatory network, restricting Cd export while enhancing vacuolar sequestration, offering a green, low-cost strategy for mitigating Cd contamination in rice.
Cooking aerosols (CAs) are a source of indoor particulate matter, but the effects of food material composition on their physicochemical properties and cellular responses remain insufficiently characterized. This study compared CAs generated from four food powder and soybean oil mixtures, including soy, beef, soy protein and beef protein, under controlled heating at 300 and 400 °C. Aerosols were characterized by mass based particle size distribution, morphology, elemental composition, selected carbonyl and polycyclic aromatic hydrocarbon (PAH) contents, and in vitro responses of A549 lung epithelial cells. CA emission rates increased substantially at 400 °C relative to 300 °C. At 300 °C, the collected aerosols were predominantly liquid-like, whereas 400 °C aerosols contained abundant carbonaceous solid particles. Beef-related mixtures generally showed higher relative abundances of carbonaceous particles and higher concentrations of selected PAHs than soy-related mixtures, particularly at 400 °C. Exposure of A549 cells to 400 °C CA suspensions was associated with reduced cell viability and altered reactive oxygen species and IL-6 responses, with stronger effects generally observed for beef-related samples. These cellular responses should be interpreted as combined effects of particle associated physical interactions and measured and unmeasured chemical constituents. Overall, the results demonstrate that the four tested food powder and soybean oil mixtures generated aerosols with distinct physicochemical characteristics and cellular effects under controlled high-temperature conditions. These findings provide a comparative basis for future studies that more rigorously separate the roles of food composition, oil decomposition, particle properties, and chemical constituents in cooking-related aerosol toxicity.
Microplastics (MPs) in indoor air pose emerging health risks, yet their accumulation dynamics on filtration systems remain poorly quantified. We employed laser direct infrared imaging to characterize MPs (abundance, polymer composition, size) across residential air filters (≥3 years’ service). Spectral deconvolution distinguished material-derived MPs from airborne particles. Bedroom screens showed peak MPs loads (3.71 × 10⁶ particles/g; 34.5% acrylonitrile-butadiene), whereas fan blades were enriched with polyacrylamide (87.1%, 2.63 × 10⁶ particles/g). The air conditioner retained 45.0% polyethylene but exhibited a 45.7% MPs reduction at the outlets, with spectral evidence of filter-derived polyamide (PA). Principal component analysis confirmed dual MPs origins: airborne (heterogeneous spectra) and material degradation, evidenced by clusters of PA and polytetrafluoroethylene (PTFE). This study redefines indoor filters not merely as sinks but as active MPs sources, highlighting an urgent need for innovative designs to mitigate secondary pollution and associated health risks.
Riverine microbial biogeography is increasingly reshaped by anthropogenic pollution, yet the mechanisms governing microbial exchange between the water and sediments in high-turbidity rivers remain unclear. Here, we developed DiffusionSedModel, a fugacity-inspired deep learning framework that couples mass transfer theory with Long Short-Term Memory (LSTM) networks, to disentangle and quantify the relative contributions of sedimentation and diffusion in structuring microbial communities across pollution gradients. The model was trained using paired 16S rRNA gene datasets from 78 synchronous water–sediment samples. By integrating cross-sectional abundance data with hydrodynamic constraints, the framework infers position-specific sedimentation and diffusion coefficients without relying on direct flux measurements. Model performance exceeded 80% predictive accuracy for both compartments. Model simulations reveal that vertical sedimentation overwhelms horizontal diffusion by an order of magnitude, with mean rate coefficients of 0.1193 and 0.0108, respectively, identifying particle-associated settling as the dominant driver of microbial spatial organization. Sediment microbial assemblages remain structurally stable across diverse pollution sources, whereas waterborne communities exhibit pronounced pollution-driven differentiation. Taxon-level analyses further link transport pathways to functional roles, highlighting diffusion-associated planktonic taxa such as putative Limnohabitans parvus and sedimentation-associated sulfur oxidizers such as putative Sulfuricaulis limicola. These results indicate that microbial dispersal in turbid rivers is governed less by advective flow than by particle-mediated vertical fluxes. By drawing conceptual analogy from fugacity theory and integrating it with data-driven learning, this study proposes a mechanistic framework for predicting microbial fate and function in sediment-dominated river systems under increasing anthropogenic pressure, whose transferability to other systems awaits cross-system validation.
Tetracycline (TC), a persistent antibiotic micropollutant ubiquitously detected in surface water bodies at ecologically concerning concentrations, poses escalating risks to aquatic ecosystems and human health, making its effective removal a critical priority in advanced water treatment. Conventional experimental approaches in forward osmosis (FO) membrane systems, however, often face limitations in precisely predicting and optimizing the dual objectives of high TC Rejection and Water Flux. To bridge this gap, we developed an interpretable, data-driven framework leveraging advanced machine learning (ML) to accurately forecast FO membrane performance and uncover the governing mechanisms. Among various ML models compared, CatBoost achieves the best performance with Test R²values exceeding 0.91 for both TC Rejection and Water Flux. Subsequently, CatBoost results are interpreted by SHapley Additive exPlanations (SHAP) and Partial Dependence Plot (PDP) analyses, identifying Membrane Pore Size and Pure Water Permeability Coefficient (A) having the most synergistic effects on the trade-off between the predicted TC Rejection and Water Flux. Building on these insights, particle swarm optimization (PSO) is applied to determine optimal conditions that effectively balance TC removal and permeate flux. The proposed configuration suggests an optimal operating region with TC Rejection of 99.45% alongside Water Flux of 48.12 LMH. This paper highlights the considerable promise of interpretable ML in guiding the rational design and multi-objective optimization of FO membrane systems, providing a data-driven framework applicable to micropollutant removal in diverse water treatment scenarios, including surface water systems.
Water quality prediction in freshwater systems and engineered water infrastructure is increasingly supported by machine learning (ML), yet limited model interpretability constrains its scientific and operational utility. Explainable artificial intelligence (XAI) has emerged as a critical framework for addressing this limitation by providing insights into model behavior, dominant predictors, nonlinear responses, and context-dependent interactions. However, existing reviews have largely treated XAI as a complementary component, offering descriptive summaries without critically examining methodological assumptions, interpretive boundaries, or cross-system applicability. In contrast, this review provides a comprehensive and critical synthesis that positions explainability as a central analytical framework and integrates applications across both freshwater systems and engineered water infrastructure. More than 70 studies published since 2018 were analyzed to identify recurring explanatory patterns across key water quality parameters, including nutrients, algal indicators, oxygen- and organic matter–related variables, composite indices, and operational parameters. Across these domains, XAI reveals that model-learned water quality patterns are often structured by nonlinear, threshold-dependent, and context-specific interactions among temperature, nutrient availability, hydrological conditions, light availability, and operational variables. The synthesis highlights the complementary roles of global and local explanations in capturing system heterogeneity and supporting both process understanding and operational decision-making. Key limitations include the non-causal nature of feature attribution and sensitivity to data structure, predictor collinearity, model configuration, and explanation stability. Future work should prioritize integration with process-based and causal inference frameworks, uncertainty-aware explainability, standardized evaluation protocols, and applications across upstream–downstream linkages. Overall, this review reframes XAI as an integrative framework linking prediction, interpretation, and decision support across the water cycle.
