Modeling adsorption of organic pollutants onto single-walled carbon nanotubes with theoretical molecular descriptors using MLR and SVM algorithms

Experimental analysis and prediction of radionuclide solubility using machine learning models: Effects of organic complexing agents

Radioactive wastes contain organic complexing agents that can form complexes with radionuclides and enhance the solubility of these radionuclides, increasing the mobility of radionuclides over great distances from a radioactive waste repository. In this study, four radionuclides (cobalt, strontium, iodine, and uranium) and three organic complexing agents (ethylenediaminetetraacetic acid, nitrilotriacetic acid, and iso-saccharic acid) were selected, and the solubility of these radionuclides was assessed under realistic environmental conditions such as different pHs (7, 9, 11, and 13), temperatures (10 °C, 20 °C, and 40 °C), and organic complexing agent concentrations (10-5-10-2 M). A total of 720 datasets were generated from solubility batch experiments. Four supervised machine learning models such as the Gaussian process regression (GPR), ensemble-boosted trees, artificial neural networks, and support vector machine were developed for predicting the radionuclide solubility. Each ML model was optimized using Bayesian optimization algorithm. The GPR evolved as a robust model that provided accurate predictions within the underlying solubility patterns by capturing the intricate relationships of the independent parameters of the dataset. At an uncertainty level of 95%, both the experimental results and GPR simulated estimations were closely correlated, confirming the suitability of the GPR model for future explorations.

Predicting adsorption of organic compounds onto graphene and black phosphorus by molecular dynamics and machine learning

With an increase in production and application of various engineering nanomaterials (ENMs), they will inevitably be released into the environment. Adsorption of various organic chemicals onto ENMs will impact on their environmental behavior and toxicology. It is unrealistic to experimentally determine adsorption equilibrium constants (K) for the vast number of organics and ENMs due to high cost in expenditure and time. Herein, appropriate molecular dynamics (MD) methods were evaluated and selected by comparing experimental K values of seven organics adsorbed onto graphene with the MD-calculated ones. Machine learning (ML) models on K of organics adsorption onto graphene and black phosphorus nanomaterials were constructed based on a benchmark data set from the MD simulations. Lasso models based on Mordred descriptors outperformed ML models built by support vector machine, random forest, k-nearest neighbor, and gradient boosting decision tree, in terms of cross-validation coefficients (Q2 > 0.90). The Lasso models also outperformed conventional poly-parameter linear free energy relationship models for predicting logK. Compared with previous models, the Lasso models considered more compounds with different functional groups and thus have broader applicability domains. This study provides a promising way to fill the data gap in logK for chemicals adsorbed onto the ENMs.

Modelling enzyme inhibition toxicity of ionic liquid from molecular structure via convolutional neural network model

Deep learning (DL) methods further promote the development of quantitative structure-activity/property relationship (QSAR/QSPR) models by dealing with complex relationships between data. An acetylcholinesterase inhibitory toxicity model of ionic liquids (ILs) was established using a convolution neural network (CNN) combined with support vector machine (SVM), random forest (RF) and multilayer perceptron (MLP). A CNN model was proposed for feature self-learning and extraction of ILs. By comparing with the model results through feature engineering (FE), the model regression results based on the CNN model for feature extraction have been substantially improved. The results showed that all six models (FE-SVM, FE-RF, FE-MLP, CNN-SVM, CNN-RF, and CNN-MLP) had good prediction accuracy, but the results based on the CNN model were better. The hyperparameters of six models were optimized by grid search and the 10-fold cross validation. Compared with the existing models in the literature, the model performance has been further improved. The model could be used as an intelligent tool to guide the design or screening of low-toxicity ILs.

Converting Nanotoxicity Data to Information Using Artificial Intelligence and Simulation

Decades of nanotoxicology research have generated extensive and diverse data sets. However, data is not equal to information. The question is how to extract critical information buried in vast data streams. Here we show that artificial intelligence (AI) and molecular simulation play key roles in transforming nanotoxicity data into critical information, i.e., constructing the quantitative nanostructure (physicochemical properties)-toxicity relationships, and elucidating the toxicity-related molecular mechanisms. For AI and molecular simulation to realize their full impacts in this mission, several obstacles must be overcome. These include the paucity of high-quality nanomaterials (NMs) and standardized nanotoxicity data, the lack of model-friendly databases, the scarcity of specific and universal nanodescriptors, and the inability to simulate NMs at realistic spatial and temporal scales. This review provides a comprehensive and representative, but not exhaustive, summary of the current capability gaps and tools required to fill these formidable gaps. Specifically, we discuss the applications of AI and molecular simulation, which can address the large-scale data challenge for nanotoxicology research. The need for model-friendly nanotoxicity databases, powerful nanodescriptors, new modeling approaches, molecular mechanism analysis, and design of the next-generation NMs are also critically discussed. Finally, we provide a perspective on future trends and challenges.

Novel quantitative structure activity relationship models for predicting hexadecane/air partition coefficients of organic compounds

Predicting the logarithm of hexadecane/air partition coefficient (L) for organic compounds is crucial for understanding the environmental behavior and fate of organic compounds and developing prediction models with polyparameter linear free energy relationships. Herein, two quantitative structure activity relationship (QSAR) models were developed with 1272 L values for the organic compounds by using multiple linear regression (MLR) and support vector machine (SVM) algorithms. On the basis of the OECD principles, the goodness of fit, robustness and predictive ability for the developed models were evaluated. The SVM model was first developed, and the predictive capability for the SVM model is slightly better than that for the MLR model. The applicability domain (AD) of these two models has been extended to include more kinds of emerging pollutants, i.e., oraganosilicon compounds. The developed QSAR models can be used for predicting L values of various organic compounds. The van der Waals interactions between the organic compound and the hexadecane have a significant effect on the L value of the compound. These in silico models developed in current study can provide an alternative to experimental method for high-throughput obtaining L values of organic compounds.

Adsorption of benzene on soils under different influential factors: an experimental investigation, importance order and prediction using artificial neural network

The adsorption of benzene on soils is specifically associated with its migration and transformation. Although previous studies have proved that the adsorption of benzene is affected by various factors, studies simultaneously considering the effects of multiple factors are rare. This study aimed to identify the qualitative and quantitative relationships between multiple influential factors and the adsorption capacity of benzene (BC). Batch adsorption experiments considering different influential factors, including initial concentration (IC), pH, temperature (T), ion strength (IS) and organic matter content (OMC), were conducted in three kinds of soils collected in a chemical industry park. The correlation analysis between different influential factors and BC was carried out based on the experimental data. The artificial neural network (ANN) was applied to predict BC. The results showed that BC increased with the increase of T. As the pH increased, BCs on silty loam and loam increased, while that on sandy loam decreased. Besides, BCs on silty loam and loam raised with increasing OMC, while that on sandy loam remained unchanged. BCs on all three kinds of soils attained their peaks when IS was small and then become stable with an increase in IS. The sequence of correlation between BC and influential factors is listed as IC > OMC > T > IS > pH for silty loam, OMC > IC > T > IS > pH for loam and IC > T > IS > pH > OMC for sandy loam. ANN analysis showed satisfactory accuracy in predicting BC under different influential factors. These results help us understand the important factors affecting benzene adsorption and provide a tool to get the adsorption information easily in complex site conditions.

The state of art on the prediction of efficiency and modeling of the processes of pollutants removal based on machine learning

During the last few years, important advances have been made in big data exploration, complex pattern recognition and prediction of complex variables. Machine learning (ML) algorithms can efficiently analyze voluminous data, identify complex patterns and extract conclusions. In chemical engineering, the application of machine learning approaches has become highly attractive due to the growing complexity of this field. Machine learning allows computers to solve problems by learning from large data sets and provides researchers with an excellent opportunity to enhance the quality of predictions for the output variables of a chemical process. Its performance has been increasingly exploited to overcome a wide range of challenges in chemistry and chemical engineering, including improving computational chemistry, planning materials synthesis and modeling pollutant removal processes. In this review, we introduce this discipline in terms of its accessible to chemistry and highlight studies that illustrate in-depth the exploitation of machine learning. The main aim of the review paper is to answer these questions by analyzing physicochemical processes that exploit machine learning in organic and inorganic pollutants removal. In general, the purpose of this review is both to provide a summary of research related to the removal of various contaminants performed by ML models and to present future research needs in ML for contaminant removal.

Modeling defluoridation of real-life groundwater by a green adsorbent aluminum/olivine composite: Isotherm, kinetics, thermodynamics and novel framework based on artificial neural network and support vector machine

The kinetic, isotherm, and thermodynamics of adsorptive removal of fluoride from the real-life groundwater was evaluated to assess the applicability of a green adsorbent, aluminum/olivine composite (AOC). The isotherm and kinetics were demonstrated by the Freundlich and Elovich model indicating significant surface heterogeneity of AOC in favouring the fluoride sorption. The fluoride removal efficiency of AOC was achieved as 87.5% after 240 min of contact time. The diffusion kinetic model exhibited that both the intra-particle and film diffusion together control the rate-limiting step of fluoride adsorption. A negative value of ΔG⁰ (-19.919 kJ/mol) at 303 K confirmed the spontaneous adsorption reaction of fluoride, and its endothermic nature was supported by the negative value of ΔH⁰ (39.504 kJ/mol). A novel framework for a predictive model by artificial neural network (ANN), and support vector machine (SVM) considering the real and synthetic fluoride-containing water was developed to assess the efficiency of adsorbent under different scenarios. ANN model was observed to be statistically significant (RMSE: 1.0955 and R²: 0.9982) and the proposed method may be instrumental in a similar area for benchmarking the synthetic and real-life samples. The low desorption potential of the spent adsorbent exhibited safe disposal of sludge and the secondary-pollutant-free treated water by the efficient and green adsorbent AOC enhanced the field-scale applicability of the green technology.

Adsorption modeling of organic compounds (OCs) by carbon nanotubes (CNTs): role of OC and CNT properties on the linear solvation energy relationship

This study evaluated a comprehensive database for the adsorption of polar and nonpolar organic compounds (OCs) by carbon nanotubes (CNTs) and to use the linear solvation energy relationship (LSER) technique for developing predictive adsorption models of OCs by multi-walled carbon nanotubes (MWCNTs) and single-walled carbon nanotubes (SWCNTs). The results showed that coefficient of determinations (R²) values for all compounds are higher variability in the 200 g/mol molecular weight cutoff (74-99%). When the molecular weight cutoff of all OCs is higher than 200 g/mol, the trend of their R² values is decreased (less than 70%). Among all adsorbate descriptor coefficients, V and B terms are the most significant descriptors (p-values ≤ 0.05) in LSER equations for adsorption of low molecular weight polar and nonpolar OCs by both CNTs. Besides, K_OW normalization of all K_d values did not have significant impact on the regression of the LSER model, indicating that hydrophobic interactions are not sole mechanism for the adsorption of OCs on CNTs. Lastly, SWCNTs exhibited higher polar OCs uptake than MWCNTs, which was attributed to more polar surface of SWCNTs as suggested by its high oxygen content (%10).

Polystyrene-Fe3O4-MWCNTs Nanocomposites for Toluene Removal from Water

In this research, multi-walled carbon nanotubes (MWCNTs) were functionalized by oxidation with strong acids HNO₃, H₂SO₄, and H₂O₂. Then, magnetite/MWCNTs nanocomposites were prepared and polystyrene was added to prepare polystyrene/MWCNTs/magnetite (PS:MWCNTs:Fe) nanocomposites. The magnetic property of the prepared nano-adsorbent PS:MWCNTs:Fe was successfully checked. For characterization, X-ray diffraction (XRD), scanning electron microscopy (SEM), transmission electron microscopy (TEM), Fourier transform infrared spectroscopy (FTIR), Raman spectroscopy, and BET surface area were used to determine the structure, morphology, chemical nature, functional groups, and surface area with pore volume of the prepared nano-adsorbents. The adsorption procedures were carried out for fresh MWCNTs, oxidized MWCNTs, MWCNTs-Fe, and PS:MWCNTs:Fe nanocomposites in batch experiments. Toluene standard was used to develop the calibration curve. The results of toluene adsorption experiments exhibited that the PS:MWCNTs:Fe nonabsorbent achieved the highest removal efficiency and adsorption capacity of toluene removal. The optimum parameters for toluene removal from water were found to be 60 min, 2 mg nano-sorbent dose, pH of 5, solution temperature of 35 °C at 50 mL volume, toluene concentration of 50 mg/L, and shaking speed of 240 rpm. The adsorption kinetic study of toluene followed the pseudo-second-order kinetics, with the best correlation (R²) value of 0.998, while the equilibrium adsorption study showed that the Langmuir isotherm was obeyed, which suggested that the adsorption is a monolayer and homogenous.

In silico prediction of polyethylene-aqueous and air partition coefficients of organic contaminants using linear and nonlinear approaches

Low-density polyethylene (LDPE) passive sampling is very attractive for use in determining chemicals concentrations. Crucial to the measurement is the coefficient (K_PE) describing partitioning between LDPE and environmental matrices. 255, 117 and 190 compounds were collected for the development of datasets in three different matrices, i.e., water, air and seawater, respectively. Further, 3 pp-LFER models and 9 QSPR models based on classical multiple linear regression (MLR) coupled with prevalent nonlinear algorithms (artificial neural network, ANN and support vector machine, SVM) were performed to predict LDPE-water (K_PE-W), LDPE-air (K_PE-A) and LDPE-seawater (K_PE-SW) partition coefficients. These developed models have satisfying predictability (R²_adj: 0.805-0.966, 0.963-0.991 and 0.817-0.941; RMSE_tra: 0.233-0.565, 0.200-0.406 and 0.260-0.459) and robustness (Q²_ext: 0.840-0.943, 0.968-0.984 and 0.797-0.842; RMSE_ext: 0.308-0.514, 0.299-0.426 and 0.407-0.462) in three datasets (water, air and seawater), respectively. In particular, the reasonable mechanism interpretations revealed that the molecular size, hydrophobicity, polarizability, ionization potential, and molecular stability were the most relevant properties, for governing chemicals partitioning between LDPE and environmental matrices. The application domains (ADs) assessed here exhibited the satisfactory applicability. As such, the derived models can act as intelligent tools to predict unknown K_PE values and fill the experimental gaps, which was further beneficial for the construction of enormous and reliable database to facilitate a distinct understanding of the distribution for organic contaminants in total environment.

Adsorption of low-concentration VOCs on various adsorbents: Correlating partition coefficient with surface energy of adsorbent

Accurately evaluating the adsorption properties of various adsorbents by some parameter is of great significance to select an appropriate adsorbent and remove volatile organic compounds (VOCs) efficiently. In this study, we successfully found a new parameter as a common standard in selecting adsorbents. Six classical adsorbents containing three carbon materials and three porous polymeric resins were used, and their surface energy (γ_s^t) and corresponding gas-solid partition coefficients (K) of eleven VOCs were measured by inverse gas chromatography (IGC) at three different column temperatures of 343 K(or 353 K), 373 K and 403 K. Then, these values at 303 K were calculated according to the linear relationship between lnK and 1/T. It was found that surface energy was significantly correlated with K values for a specific VOC, and could be used as a common standard to well evaluate the adsorption properties of various adsorbents. Furthermore, we employed it to develop a model for predicting the adsorption properties of low-concentration VOCs on various adsorbents at 303 K. The developed model exhibited an excellent predictive ability by external validation. Moreover, the model showed wide applicability and predicted the lnK values of VOCs at 373 K and 403 K in R² of 0.910 and 0.889.

Development of models predicting biodegradation rate rating with multiple linear regression and support vector machine algorithms

Biodegradation is a significant process for removing organic chemicals from water, soil and sediment environments, and therefore biodegradability is critical to evaluate the environmental persistence of organic chemicals. In this study, based on a dataset with 171 compounds, four quantitative structure-activity relationship (QSAR) models were developed for predicting primary and ultimate biodegradation rate rating with multiple linear regression (MLR) and support vector machine (SVM) algorithms. Two MLR models were built with a dataset with carbon atom number ≤9, and two SVM models were built with a dataset with carbon atom number >9. In the MLR models, n_ArX (number of X on aromatic ring) is the most important descriptor governing primary and ultimate biodegradation of organic chemicals. For the SVM models, determination coefficient (R²) values, cross-validated coefficients (Q²_LOO) and external validation coefficient (Q²_ext) values are over 0.9, indicating the SVM models have satisfactory goodness-of-fit, robustness and external predictive abilities. The applicability domains of these models were visualized by the Williams plot. The developed models can be used as effective tools to predict biodegradability of organic chemicals.

Development of a quantitative structure-activity relationship model for mechanistic interpretation and quantum yield prediction of singlet oxygen generation from dissolved organic matter

Singlet oxygen (¹O₂) is capable of degrading organic contaminants and inducing cell damage and inactivation of viruses. It is mainly generated through the interaction of dissolved oxygen with excited triplet states of dissolved organic matter (DOM) in natural waters. The present study aims at revealing the underlying mechanism of ¹O₂ generation and providing a potential tool for predicting the quantum yield of ¹O₂ (Φ_1O2) generation from DOM by constructing a quantitative structure-activity relationship (QSAR) model. The determined Φ_1O2 values for the selected DOM-analogs range from (0.54 ± 0.23) × 10^-2 to (62.03 ± 2.97) × 10^-2. A QSAR model was constructed and was proved to have satisfactory goodness-of-fit and robustness. The QSAR model was successfully used to predict the Φ_1O2 of Suwannee River fulvic acid. Mechanistic interpretation of the descriptors in the model showed that hydrophobicity, molecular complexity and the presence of carbonyl groups in DOM play crucial roles in the generation of ¹O₂ from DOM. The presence of other heteroatoms besides O, such as N and S, also affects the generation of ¹O₂. The results of this study provide valuable insights into the generation of ¹O₂ from DOM in sunlit natural waters.

Chemometric modeling to predict air half-life of persistent organic pollutants (POPs)

We have reported here a quantitative structure-property relationship (QSPR) model for prediction of air half-life of organic chemicals using a dataset of 302 diverse organic chemicals employing only two-dimensional descriptors with definite physicochemical meaning in order to avoid the computational complexity for higher dimensional molecular descriptors. The developed model was rigorously validated using the internationally accepted internal and external validation metrics. The final partial least squares (PLS) regression model obtained at three latent variables comprises six simple and interpretable 2D descriptors. The simple and highly robust model with good quality of predictions explains 66% for the variance of the training set (R²) (64% in terms of LOO variance (Q²)) and 76% for test set variance (R²_pred) (prediction quality). This model might be applicable for data gap filling for determination of POPs in the environment, in case of new or untested chemicals falling within the applicability domain of the model. In general, the model indicates that the air half-life of organic chemicals increases with presence of H-bond acceptor atoms, number of halogen atoms and presence of the R-CH-X fragment and lipophilicity, and decreases with presence of a number of halogens on ring C(sp3) (substitution of halogen atoms on a ring).

Predicting adsorption coefficients of VOCs using polyparameter linear free energy relationship based on the evaluation of dispersive and specific interactions

Predicting adsorption of volatile organic compounds (VOCs) on activated carbons is of major importance to understand activated carbons' adsorption properties and explore their potential applications. In this study, adsorption of 38 VOCs on a commercial granular activated carbon (GAC) was examined using inverse gas chromatography (IGC) at infinite dilution, and the adsorption coefficients (K), dispersive and specific components of adsorption free energy were calculated. We found that the dispersive interaction was well described by adsorbate's molar polarizability (P), and the specific interactions well by dipolarity/polarizability (S), hydrogen-bond acidity (A) and hydrogen-bond basicity (B). Based on the result, a polyparameter linear free energy relationship (PP-LFER) was established: logK = (0.96 ± 0.23) S + (2.23 ± 0.34) A + (0.84 ± 0.25) B + (0.69 ± 0.050) P + (0.13 ± 0.35); (n = 38, R² = 0.859, root mean square error (RMSE) = 0.25), which exhibited a more accurate prediction compared to the classical PP-LFER (E, S, A, B and L as descriptors, R² = 0.765, RMSE = 0.33). Moreover, it overcame the drawbacks of indistinguishable dispersive interaction and unavailable relative contribution of each interaction for classical PP-LFER in explaining adsorption mechanism. As suggested by the developed model, the dispersive interaction was the dominant contribution to the adsorption of VOCs on GAC (42-100%), following by dipole-type interactions (0-30%) and hydrogen bonding (hydrogen-bond acidity 0-32%, hydrogen-bond basicity 0-11%). Additionally, it also accurately predicted the K values of VOCs on other three activated carbons.

Development of Prediction Models on Base-Catalyzed Hydrolysis Kinetics of Phthalate Esters with Density Functional Theory Calculation

Many phthalate esters (PAEs) are chemicals of high production volume and of toxicological concern. The second-order rate constant for base-catalyzed hydrolysis ( k_B) is a key parameter for assessing environmental persistence of PAEs. However, the k_B values for most PAEs are lacking, and the experimental determination of k_B encounters various difficulties. Herein, density functional theory (DFT) methods were selected by comparing empirical k_B values of five PAEs and five carboxylic acid esters with the DFT-calculated ones. Results indicate that PAEs with cyclic side chains are more vulnerable to base-catalyzed hydrolysis than PAEs with linear alkyl side chains, followed by PAEs with branched alkyl side chains. By combining experimental and DFT-calculated second-order rate constants for base-catalyzed hydrolysis of one side chain in PAEs ( k_{B_side chain}), quantitative structure-activity relationship models were developed. The models can differentiate PAEs with the departure of the leaving group (or the nucleophilic attack of OH^-) as the rate-determining step in the hydrolysis and estimate k_B values, which provides a promising way to predict hydrolysis kinetics of PAEs. The half-lives of the investigated PAEs were calculated and vary from 0.001 h to 558 years (pH = 7∼9), further illustrating the necessity of prediction models for hydrolysis kinetics in assessing the environmental persistence of chemicals.

Consensus QSPR modelling for the prediction of cellular response and fibrinogen adsorption to the surface of polymeric biomaterials

In the current study, we have developed predictive quantitative structure-activity relationship (QSAR) models for cellular response (foetal rate lung fibroblast proliferation) and protein adsorption (fibrinogen adsorption (FA)) on the surface of tyrosine-derived biodegradable polymers designed for tissue engineering purpose using a dataset of 66 and 40 biodegradable polymers, respectively, employing two-dimensional molecular descriptors. Best four individual models have been selected for each of the endpoints. These models are developed using partial least squares regression with a unique combination of six and four descriptors for cellular response and protein adsorption, respectively. The generated models were strictly validated using internal and external metrics to determine the predictive ability and robustness of proposed models. Subsequently, the validated individual models for each response endpoints were used for the generation of 'intelligent' consensus models ( http://teqip.jdvu.ac.in/QSAR_Tools/DTCLab/ ) to improve the quality of predictions for the external data set. These models may help in prediction of virtual polymer libraries for rational design/optimization for properties relevant to biomedical applications prior to their synthesis.

Is clay-polycation adsorbent future of the greener society? In silico modeling approach with comprehensive virtual screening

Presence of organic pollutants in the wastewater and aquatic environment is one of the serious concerns worldwide. Superior adsorption of organic pollutants on modified clays with organocations is well approved nowadays. Among hybrid materials, clay-polyelectrolyte nanocomposites (CPN) are one of the specifically designed materials for the efficient adsorption of diverse organic pollutants. Due to higher surface area of the clay mineral coupled with a polymer coating, they have an explicit affinity for the organic pollutants. In this background, we have developed statistically significant and mechanistically interpretable quantitative structure-property relationship (QSPR) model for adsorption coefficient of diverse organic pollutants to the protonated montmorillonite-poly-4-vinylpyridine-co-styrene (Mt-HPVPcoS), a hybrid CPN. Further, the model was employed to predict the logk_d value of ∼0.9 million chemicals from five diverse databases spanning from existing and experimental pharmaceuticals, natural and synthetic chemicals and dyes with unknown logk_d value for the mentioned CPN. The reliability of predicted data is checked with two layers confidence screening i.e. the applicability domain study followed by prediction quality check by 'Prediction Reliability Indicator'. Thus, prediction of each compound can be used for data gap filling by environmental regulatory authorities as well as industries. Followed by, maximum common substructure-based (MCS) algorithm is employed for individual database to extract the important structural scaffold for higher logk_d to the mentioned CPN.