Quantitative structure activity relationships (QSARs) and machine learning models for abiotic reduction of organic compounds by an aqueous Fe(II) complex

Origins of Selective Oxidation in Carbon-Based Nonradical Oxidation Processes toward Organic Pollutants: Quantitative Structure-Activity Relationships (QSARs)

Metal-free carbon material-mediated nonradical oxidation processes (C-NOPs) have emerged as a research hotspot due to their excellent performance in selectively eliminating organic pollutants in aqueous environments. However, the selective oxidation mechanisms of C-NOPs remain obscure due to the diversity of organic pollutants and nonradical active species. Herein, quantitative structure-activity relationship (QSAR) models were employed to unveil the origins of C-NOP selectivity toward organic pollutants in different oxidant systems. QSAR analysis based on adsorption and oxidation descriptors revealed that C-NOP selectivity depends on the oxidation potentials of organic pollutants rather than on adsorption interactions. However, the dominance of electronic effects in selective oxidation decreases with increasing structural complexity of organic pollutants. Moreover, the oxidation threshold solely depends on the inherent electronic nature of organic pollutants and not on the reactivity of nonradical active species. Notably, the accuracy of substituent descriptors (Hammett constants) and theoretical descriptors (e.g., highest occupied molecular orbital energy, ionization potential, and single-electron oxidation potential) is significantly influenced by the complexity and molecular state of organic pollutants. Overall, the study findings reveal the origins of organic pollutant-oriented selective oxidation and provide insight into the application of descriptors in QSAR analysis.

Machine learning-assisted assessment of key meteorological and crop factors affecting historical mulch pollution in China

Degraded mulch pollution is of a great concern for agricultural soils. Although numerous studies have examined this issue from an environmental perspective, there is a lack of research focusing on crop-specific factors such as crop type. This study aimed to explore the correlation between meteorological and crop factors and mulch contamination. The first step was to estimate the amounts of mulch-derived microplastics (MPs) and phthalic acid esters (PAEs) during the rapid expansion period (1993-2012) of mulch usage in China. Subsequently, the Elastic Net (EN) and Random Forest (RF) models were employed to process a dataset that included meteorological, crop, and estimation data. At the national level, the RF model suggested that coldness in fall was crucial for MPs generation, while vegetables acted as a key factor for PAEs release. On a regional scale, the EN results showed that crops like vegetables, cotton, and peanuts remained significantly involved in PAEs contamination. As for MPs generation, coldness prevailed over all regions. Aridity became more critical for southern regions compared to northern regions due to solar radiation. Lastly, each region possessed specific crop types that could potentially influence its MPs contamination levels and provide guidance for developing sustainable ways to manage mulch contamination.

Predicting reactivity dynamics of halogen species and trace organic contaminants using machine learning models

Reactions of reactive halogen species (Cl•, Br•, and Cl2•-) with trace organic contaminants (TrOCs) have received much attention in recent years, and their k values are fundamental parameters for understanding their reaction mechanisms. However, k values are usually unknown. In this study, we developed machine learning (ML)-based quantitative structure-activity relationship (QSAR) models to predict k values. We tested five algorithms, namely, random forest, neural network, XGBoost, support vector machine (SVM), and multilinear regression, using molecular descriptors (MDs) and molecular fingerprints (MFs) as inputs. The optimal algorithms were MD-XGBoost for Cl• and Br•, and MF-SVM for Cl2•-, respectively, with R2test values of 0.876, 0.743, and 0.853. We found that electron-withdrawing/donating groups tended to interfere with the reactivity of Cl2•- more than Cl• and Br•. This explains why MFs are better inputs for predictive models of Cl2•-, whereas MDs are more suitable for Cl• and Br•. Furthermore, we interpreted the models using SHAP analysis, and the results indicated that our models accurately predicted k values both statistically and mechanistically. Our models provide useful tools for obtaining unknown k values and help researchers understand the inherent relationships between the models.

Machine Learning Models for Inverse Design of the Electrochemical Oxidation Process for Water Purification

In this study, a machine learning (ML) framework is developed toward target-oriented inverse design of the electrochemical oxidation (EO) process for water purification. The XGBoost model exhibited the best performances for prediction of reaction rate (k) based on training the data set relevant to pollutant characteristics and reaction conditions, indicated by Rext2 of 0.84 and RMSEext of 0.79. Based on 315 data points collected from the literature, the current density, pollutant concentration, and gap energy (Egap) were identified to be the most impactful parameters available for the inverse design of the EO process. In particular, adding reaction conditions as model input features allowed provision of more available information and an increase in the sample size of the data set to improve the model accuracy. The feature importance analysis was performed for revealing the data pattern and feature interpretation by using Shapley additive explanations (SHAP). The ML-based inverse design for the EO process was generalized to a random case for tailoring the optimum conditions with phenol and 2,4-dichlorophenol (2,4-DCP) serving as model pollutants. The resulting predicted k values were close to the experimental k values by experimental verification, accounting for the relative error lower than 5%. This study provides a paradigm shift from conventional trial-and-error mode to data-driven mode for advancing research and development of the EO process by a time-saving, labor-effective, and environmentally friendly target-oriented strategy, which makes electrochemical water purification more efficient, more economic, and more sustainable in the context of global carbon peaking and carbon neutrality.

Count-Based Morgan Fingerprint: A More Efficient and Interpretable Molecular Representation in Developing Machine Learning-Based Predictive Regression Models for Water Contaminants' Activities and Properties

In this study, we introduce the count-based Morgan fingerprint (C-MF) to represent chemical structures of contaminants and develop machine learning (ML)-based predictive models for their activities and properties. Compared with the binary Morgan fingerprint (B-MF), C-MF not only qualifies the presence or absence of an atom group but also quantifies its counts in a molecule. We employ six different ML algorithms (ridge regression, SVM, KNN, RF, XGBoost, and CatBoost) to develop models on 10 contaminant-related data sets based on C-MF and B-MF to compare them in terms of the model's predictive performance, interpretation, and applicability domain (AD). Our results show that C-MF outperforms B-MF in nine of 10 data sets in terms of model predictive performance. The advantage of C-MF over B-MF is dependent on the ML algorithm, and the performance enhancements are proportional to the difference in the chemical diversity of data sets calculated by B-MF and C-MF. Model interpretation results show that the C-MF-based model can elucidate the effect of atom group counts on the target and have a wider range of SHAP values. AD analysis shows that C-MF-based models have an AD similar to that of B-MF-based ones. Finally, we developed a "ContaminaNET" platform to deploy these C-MF-based models for free use.

Abiotic Reduction of Organic and Inorganic Compounds by Fe(II)-Associated Reductants: Comprehensive Data Sets and Machine Learning Modeling

Iron-associated reductants play a crucial role in providing electrons for various reductive transformations. However, developing reliable predictive tools for estimating abiotic reduction rate constants (logk) in such systems has been impeded by the intricate nature of these systems. Our recent study developed a machine learning (ML) model based on 60 organic compounds toward one soluble Fe(II)-reductant. In this study, we built a comprehensive kinetic data set covering the reactivity of 117 organic and 10 inorganic compounds toward four major types of Fe(II)-associated reductants. Separate ML models were developed for organic and inorganic compounds, and the feature importance analysis demonstrated the significance of resonance structures, reducible functional groups, reductant descriptors, and pH in logk prediction. Mechanistic interpretation validated that the models accurately learned the impact of various factors such as aromatic substituents, complexation, bond dissociation energy, reduction potential, LUMO energy, and dominant reductant species. Finally, we found that 38% of the 850,000 compounds in the Distributed Structure-Searchable Toxicity (DSSTox) database contain at least one reducible functional group, and the logk of 285,184 compounds could be reasonably predicted using our model. Overall, the study is a significant step toward reliable predictive tools for anticipating abiotic reduction rate constants in iron-associated reductant systems.

Predicting aqueous sorption of organic pollutants on microplastics with machine learning

Microplastics (MPs) are ubiquitously distributed in freshwater systems and they can determine the environmental fate of organic pollutants (OPs) via sorption interaction. However, the diverse physicochemical properties of MPs and the wide range of OP species make a deeper understanding of sorption mechanisms challenging. Traditional isotherm-based sorption models are limited in their universality since they normally only consider the nature and characteristics of either sorbents or sorbates individually. Therefore, only specific equilibrium concentrations or specific sorption isotherms can be used to predict sorption. To systematically evaluate and predict OP sorption under the influence of both MPs and OPs properties, we collected 475 sorption data from peer-reviewed publications and developed a poly-parameter-linear-free-energy-relationship-embedded machine learning method to analyze the collected sorption datasets. Models of different algorithms were compared, and the genetic algorithm and support vector machine hybrid model displayed the best prediction performance (R2 of 0.93 and root-mean-square-error of 0.07). Finally, comparison results of three feature importance analysis tools (forward step wise method, Shapley method, and global sensitivity analysis) showed that chemical properties of MPs, excess molar refraction, and hydrogen-bonding interaction of OPs contribute the most to sorption, reflecting the dominant sorption mechanisms of hydrophobic partitioning, hydrogen bond formation, and π-π interaction, respectively. This study presents a novel sorbate-sorbent-based ML model with a wide applicability to expand our capacity in understanding the complicated process and mechanism of OP sorption on MPs.

Quantitative structure-activity relationship models for predicting apparent rate constants of organic compounds with ferrate (VI)

Ferrate (VI) (Fe (VI)) is a promising, environmentally friendly multifunctional oxidant widely applied in organic compound degradation. Oxidative kinetics of the apparent second-order rate constants (k_app) of Fe (VI) with organic compounds are critical for modeling oxidation processes. Herein, a quantitative structure-activity relationship (QSAR) model was developed using particle swarm optimization and an extreme learning machine to better understand the laws of the k_app values of organic compounds, including 33 aliphatic and aromatic hydrocarbon derivatives, during degradation by Fe (VI). Seven components-electronic hardness (H), electronic softness (S), ratio of oxygen to carbon atoms (O_n/C_n), energy of the highest occupied molecular orbital (E_HOMO), vertical ionization potential (VIP), maximum nucleophilic reaction index (f(+)_x), and minimum relative electrophilicity index (RE_n) constitute the critical molecular parameters. The developed QSAR model was verified on the basis of the coefficient of determination (R²) and the root mean square error (RMSE): for the training set, R² = 0.924 and RMSE = 1.186, whereas for the test set, R² = 0.996, and RMSE = 0.352. The applicability, reliability, and predictability of the model were verified by estimating the applicability domain (AD) of the model. Furthermore, QSAR models constructed using different methods were compared, and the main impact descriptors and conclusions obtained from previous studies were theoretically analyzed. Results indicate that constructing the QSAR model facilitates k_app prediction for Fe (VI) in the degradation of various organic compounds, improves the understanding of the degradation mechanism, and reduces the pressure on human and material resources caused by experiments.

Predictive models for the aqueous phase reactivity of inorganic radicals with organic micropollutants

Single-electron transfer (SET) is one of the most common reaction mechanisms for degrading organic micropollutants (OMPs) in advanced oxidation processes. We collected 300 SET reactions (CO₃^•-, SO₄^•-, Cl₂^•-, and Br₂^•--mediated) and calculated three key parameters for understanding the SET mechanism: aqueous phase free energies of activation (ΔG^‡), free energies of reactions (ΔG), and orbital energy gaps of reactants (E_OMPsHOMO-E_RadiLUMO). Then, we classified the OMPs according to their structure, developed and evaluated linear energy relationships of the second-order rate constants (k) with ΔG^‡, ΔG, or E_OMPsHOMO-E_RadiLUMO in each class. Considering that a single descriptor cannot capture all the chemical diversity, we combined ΔG^‡, ΔG, and E_OMPsHOMO-E_RadiLUMO as inputs to develop multiple linear regression (MLR) models. Chemical classification is critical to the linear model described above. However, OMPs usually have multiple functional groups, making the classification challenging and uncertain. Therefore, we tried machine learning algorithms to predict k values without chemical classification. We found that decision trees (R² = 0.88-0.95) and random forest (R² = 0.90-0.94) algorithms show better performance on the prediction of the k values, whereas boosted tree algorithm cannot make an accurate prediction (R² = 0.19-0.36). Overall, our study provides a powerful tool to predict the aqueous phase reactivity of OMP to certain radicals without the need for chemical classification.

Electrochemical Reduction of Halogenated Alkanes and Alkenes Using Activated Carbon-Based Cathodes

Granular activated carbon (GAC) is used to sorb a broad range of halogenated contaminant classes, but spent GAC disposal is costly. Taking advantage of GAC's conductivity, this study evaluated the conversion of the GAC to cathodes for electrochemical reductive dehalogenation of 15 halogenated alkanes and alkenes exhibiting a diversity of structures (type of halogen, number of halogens, functional groups) and including contaminants of practical importance (e.g., trichloroethylene). Alkane degradation rates increased with the number of halogens and in the order: chlorine < bromine < iodine. Quantitative structure-activity relationships (QSARs) correlating experimental first-order degradation rate constants for alkanes with molecular descriptors associated with an outer-sphere one-electron transfer calculated using density functional theory indicated that correlations with molecular descriptors improved in the order: aqueous phase reduction potentials (E^0,aq) < energy of the substrate's lowest unoccupied molecular orbital (E_LUMO) < Marcus theory activation free energies (ΔG^‡) ∼ gas-phase standard reduction free energies (ΔG^0,gas). Chlorinated alkene degradation rates increased with decreasing number of chlorines, and QSAR correlations were opposite those of alkanes, indicating a different reaction mechanism. Degradation timescales ranged from 1 min to 3 h with halides as predominant products. These results suggest that the electrochemical reduction of halogenated alkanes and alkenes can be used to regenerate spent GAC.

Computational Predictions of the Hydrolysis of 2,4,6-Trinitrotoluene (TNT) and 2,4-Dinitroanisole (DNAN)

Hydrolysis is a common transformation reaction that can affect the environmental fate of many organic compounds. In this study, three proposed mechanisms of alkaline hydrolysis of 2,4,6-trinitrotoluene (TNT) and 2,4-dinitroaniline (DNAN) were investigated with plane-wave density functional theory (DFT) combined with ab initio and classical molecular dynamics (AIMD/MM) free energy simulations, Gaussian basis set DFT calculations, and correlated molecular orbital theory calculations. Most of the computations in this study were carried out using the Arrows web-based tools. For each mechanism, Meisenheimer complex formation, nucleophilic aromatic substitution, and proton abstraction reaction energies and activation barriers were calculated for the reaction at each relevant site. For TNT, it was found that the most kinetically favorable first hydrolysis steps involve Meisenheimer complex formation by attachment of OH^- at the C1 and C3 arene carbons and proton abstraction from the methyl group. The nucleophilic aromatic substitution reactions at the C2 and C4 arene carbons were found to be thermodynamically favorable. However, the calculated activation barriers were slightly lower than in previous studies, but still found to be ΔG^‡ ≈ 18 kcal/mol using PBE0 AIMD/MM free energy simulations, suggesting that the reactions are not kinetically significant. For DNAN, the barriers of nucleophilic aromatic substitution were even greater (ΔG^‡ > 29 kcal/mol PBE0 AIMD/MM). The most favorable hydrolysis reaction for DNAN was found to be a two-step process in which the hydroxyl first attacks the C1 carbon to form a Meisenheimer complex at the C1 arene carbon C1-(OCH₃)OH^-, and subsequently, the methoxy anion (-OCH₃) at the C1 arene carbon dissociates and the proton shuttles from the C1-OH to the dissociated methoxy group, resulting in methanol and an aryloxy anion.

Versatile in silico modelling of microplastics adsorption capacity in aqueous environment based on molecular descriptor and machine learning

To comprehensively evaluate the hazards of microplastics and their coexisting organic pollutants, the sorption capacity of microplastics is a major issue that is quantified through the microplastic-aqueous sorption coefficient (K_d). Almost all quantitative structure-property relationship (QSPR) models that describe K_d apply only to narrow, relatively homogeneous groups of reactants. Herein, non-hybrid QSPR-based models were developed to predict PE-water (K_PE-w), PE-seawater (K_PE-sw), PVC-water (K_PVC-w) and PP-seawater (K_PP-sw) sorption coefficients at different temperatures, with eight machine learning algorithms. Moreover, novel hybrid intelligent models for predicting K_d more accurately were innovatively developed by applying GA, PSO and AdaBoost algorithms to optimize MLP and ELM models. The results indicated that all three optimization algorithms could improve the robustness and predictability of the standalone MLP and ELM models. In all models trained with K_PE-w, K_PE-sw, K_PVC-w and K_PP-sw data sets, GBDT-1 and XGBoost-1 models, MLP-GA-2 and MLP-PSO-2 models, MLR-3 and MLR-4 models performed better in terms of goodness of fit (R_adj²: 0.907-0.999), robustness (Q_BOOT²: 0.900-0.937) and predictability (R_ext²: 0.889-0.970), respectively. Analyzing the descriptors revealed that temperature, lipophilicity, ionization potential and molecular size were correlated closely with the adsorption capacity of microplastics to organic pollutants. The proposed QSPR models may assist in initial environmental exposure assessments without imposing heavy costs in the early experimental phase.

Predicting Solute Descriptors for Organic Chemicals by a Deep Neural Network (DNN) Using Basic Chemical Structures and a Surrogate Metric

Solute descriptors have been widely used to model chemical transfer processes through poly-parameter linear free energy relationships (pp-LFERs); however, there are still substantial difficulties in obtaining these descriptors accurately and quickly for new organic chemicals. In this research, models (PaDEL-DNN) that require only SMILES of chemicals were built to satisfactorily estimate pp-LFER descriptors using deep neural networks (DNN) and the PaDEL chemical representation. The PaDEL-DNN-estimated pp-LFER descriptors demonstrated good performance in modeling storage-lipid/water partitioning coefficient (log K_{storage-lipid/water}), bioconcentration factor (BCF), aqueous solubility (ESOL), and hydration free energy (freesolve). Then, assuming that the accuracy in the estimated values of widely available properties, e.g., logP (octanol-water partition coefficient), can calibrate estimates for less available but related properties, we proposed logP as a surrogate metric for evaluating the overall accuracy of the estimated pp-LFER descriptors. When using the pp-LFER descriptors to model log K_{storage-lipid/water}, BCF, ESOL, and freesolve, we achieved around 0.1 log unit lower errors for chemicals whose estimated pp-LFER descriptors were deemed "accurate" by the surrogate metric. The interpretation of the PaDEL-DNN models revealed that, for a given test chemical, having several (around 5) "similar" chemicals in the training data set was crucial for accurate estimation while the remaining less similar training chemicals provided reasonable baseline estimates. Lastly, pp-LFER descriptors for over 2800 persistent, bioaccumulative, and toxic chemicals were reasonably estimated by combining PaDEL-DNN with the surrogate metric. Overall, the PaDEL-DNN/surrogate metric and newly estimated descriptors will greatly benefit chemical transfer modeling.

Predicting Heavy Metal Adsorption on Soil with Machine Learning and Mapping Global Distribution of Soil Adsorption Capacities

Studying heavy metal adsorption on soil is important for understanding the fate of heavy metals and properly assessing the related environmental risks. Existing experimental methods and traditional models for quantifying adsorption, however, are time-consuming and ineffective. In this study, we developed machine learning models for the soil adsorption of six heavy metals (Cd(II), Cr(VI), Cu(II), Pb(II), Ni(II), and Zn(II)) using 4420 data points (1105 soils) extracted from 150 journal articles. After a comprehensive comparison, our results showed that the gradient boosting decision tree had the best performance for a combined model based on all the data. The Shapley additive explanation method was used to identify the feature importance and the effects of these features on the adsorption, based on which six independent models were developed for the six metals to achieve better model performance than the combined model. Using these independent models, the global distribution of heavy metal adsorption capacities on soils was predicted with known soil properties. Reversed models, including one combined model for all the six metals and six independent models, were also built using the same data sets to predict the heavy metal concentration in water when the adsorbed amount is known for a soil/sediment.

Machine Learning: New Ideas and Tools in Environmental Science and Engineering

The rapid increase in both the quantity and complexity of data that are being generated daily in the field of environmental science and engineering (ESE) demands accompanied advancement in data analytics. Advanced data analysis approaches, such as machine learning (ML), have become indispensable tools for revealing hidden patterns or deducing correlations for which conventional analytical methods face limitations or challenges. However, ML concepts and practices have not been widely utilized by researchers in ESE. This feature explores the potential of ML to revolutionize data analysis and modeling in the ESE field, and covers the essential knowledge needed for such applications. First, we use five examples to illustrate how ML addresses complex ESE problems. We then summarize four major types of applications of ML in ESE: making predictions; extracting feature importance; detecting anomalies; and discovering new materials or chemicals. Next, we introduce the essential knowledge required and current shortcomings in ML applications in ESE, with a focus on three important but often overlooked components when applying ML: correct model development, proper model interpretation, and sound applicability analysis. Finally, we discuss challenges and future opportunities in the application of ML tools in ESE to highlight the potential of ML in this field.

Quantification of photocatalytically-generated hydrogen peroxide in the presence of organic electron donors: Interference and reliability considerations

Photocatalytic H₂O₂ production is an innovative on-site H₂O₂ synthesis method to treat organic pollutants through Fenton-like reactions, avoiding the need and potential liability of H₂O₂ storage and transportation. Accurate quantification of H₂O₂ is crucial to explore the mechanism of photocatalytic H₂O₂ production and optimize reaction parameters. In this work, three common H₂O₂ quantification methods (i.e., titration with potassium permanganate (KMnO₄), and colorimetry with ammonium metavanadate (NH₄VO₃) or N,N-diethylp-phenylenediamine-horseradish peroxidase (DPD-POD)) were compared and their susceptibility to interference by seven types of representative organics were considered. Interference mechanisms were explored based on the electron-donating (E_gap) and electron-accepting (E_LUMO) ability of the present organics. The accuracy of the KMnO₄ titration method is greatly compromised by aromatic compounds even at 0.1 mM due to the increased KMnO₄ consumption by direct oxidation. The presence of p-benzoquinone that directly reacts with NH₄VO₃ and DPD compromises these colorimetric methods, especially DPD-POD colorimetry at concentrations as low as 0.1 mM. The DPD-POD method should also be scrutinized in the presence of phenols due to significant disturbance by oxidation byproducts (e.g. hydroquinone inducing immediate color disappearance). A flowchart was generated to provide guidelines for selecting an appropriate H₂O₂ quantification method for different water matrices treated by Fenton-like reactions.

A modified binary particle swarm optimization with a machine learning algorithm and molecular docking for QSAR modelling of cholinesterase inhibitors

The acetylcholinesterase (AChE) and butyrylcholinesterase (BuChE) inhibitors play a key role in treating Alzheimer's disease. This study proposes an approach that integrates a modified binary particle swarm optimization (PSO) with a machine learning algorithm for building QSAR models to predict the activity of inhibitors for AChE and BuChE enzymes. More precisely, it uses a transfer function to convert the continuous search space of PSO to binary. Furthermore, it utilizes the concepts of catfish effect and chaotic map to improve exploration ability in searching for an optimum subset of descriptors for QSAR model constructions. Then, through a statistical method, it employs a machine learning algorithm to evaluate the fitness value of each candidate subset of features. Different combinations of four transfer functions with four machine learning algorithms, including K-nearest neighbour, multiple linear regression, support vector machine, and regression tree, were used to build several variants of the proposed algorithm. QSAR models constructed by each version were verified by internal and external validations. The best variants were selected based on a method called sum of ranking differences.

Fe(II) Redox Chemistry in the Environment

Iron (Fe) is the fourth most abundant element in the earth's crust and plays important roles in both biological and chemical processes. The redox reactivity of various Fe(II) forms has gained increasing attention over recent decades in the areas of (bio) geochemistry, environmental chemistry and engineering, and material sciences. The goal of this paper is to review these recent advances and the current state of knowledge of Fe(II) redox chemistry in the environment. Specifically, this comprehensive review focuses on the redox reactivity of four types of Fe(II) species including aqueous Fe(II), Fe(II) complexed with ligands, minerals bearing structural Fe(II), and sorbed Fe(II) on mineral oxide surfaces. The formation pathways, factors governing the reactivity, insights into potential mechanisms, reactivity comparison, and characterization techniques are discussed with reference to the most recent breakthroughs in this field where possible. We also cover the roles of these Fe(II) species in environmental applications of zerovalent iron, microbial processes, biogeochemical cycling of carbon and nutrients, and their abiotic oxidation related processes in natural and engineered systems.