Polyparameter Linear Free Energy Relationships Research Articles

Concentration division for adsorption coefficient prediction using machine learning with Abraham descriptors: Data-splitting approach comparison and critical factors identification

Machine learning (ML) including Abraham descriptors from polyparameter linear free energy relationships (pp-LFERs) has been a popular method for the adsorption coefficient (Kd) prediction. However, Abraham descriptors from pp-LFERs are concentration-dependent and the significance of these descriptors can change over different adsorbate concentrations. Ignoring concentration effects on the adsorption process and Kd prediction will hinder the understanding of interactions among solutes, solvents and adsorbents at different equilibrium concentration (Ce) ranges. Therefore, our study first systematically investigated the concentration effects on micropollutant adsorption to carbon-based adsorbents using ML with Abraham descriptors. Concentration-selection approach, as a new data-splitting approach, divided the whole dataset according to the different Ce ranges. This concentration-selection approach performed better than the data-splitting approach used in previous studies. After the ML models were built in different Ce subsets, Shapley values were calculated to quantify input descriptor contributions. The results indicated specific surface area (BET) was the only critical factor when Ce was in the highest range. The importance of Abraham descriptors increased gradually when Ce decreased. Total pore volume (Vt) was a far less important feature than BET for Kd prediction. Critical factors identified at different Ce ranges for Kd prediction provide a guidance for novel carbon-based adsorbent design.

Estimating Octanol-Water Partition Coefficients of Novel Brominated Flame Retardants by Reversed-Phase High-Performance Liquid Chromatography and Computational Models.

Legacy brominated flame retardants, including polybrominated diphenyl ethers (PBDEs), have been classified as persistent organic pollutants and replaced with novel brominated flame retardants (NBFRs). The octanol-water partition coefficients (log KOW) of NBFRs have been computationally estimated, but the log KOW values provided by these methods can differ by 1 to 3 orders of magnitude. Given the importance of this parameter in fate and toxicity models, we indirectly measured the log KOW values of eight NBFRs by their capacity factor (k') on a reversed-phase high-performance liquid chromatography (HPLC) C18 column by isocratic elution and compared these measured values with those estimated by nine computational models. Log KOW values were obtained for the NBFRs 1,2-bis(2,4,6-tribromophenoxy) ethane, pentabromobenzene, pentabromoethylbenzene, pentabromotoluene, 2-ethylhexyl 2,3,4,5-tetrabromobenzoate, allyl 2,4,6-tribromophenylether, 2,3-dibromopropyl-2,4,6-tribromophenyl ether, and bis(2-ethylhexyl) tetrabromophthalate. A training set of phthalates, polychlorinated biphenyls, PBDEs, and halogenated benzenes were chosen to obtain the log k'-log KOW calibration for the NBFRs. The computational models KowWIN, XLogP3, EAS-E Suite, COSMOtherm, DirectML, and Abraham polyparameter linear free energy relationships all predicted the log KOW values of the calibration compounds to within 1 order of magnitude without significant bias. The median of these models predicted log KOW values for the calibration compounds that were close to those known in the literature with root mean square error (RMSE) = 0.224 and for the NBFRs that were close to those measured by HPLC (RMSE = 0.334). Environ Toxicol Chem 2024;00:1-10. © 2024 SETAC.

Polyparameter Linear Free Energy Relationships for Partitioning of Neutral Organic Compounds to Storage Lipids.

Storage lipids are an important compartment in the bioaccumulation of neutral organic compounds. Reliable models for predicting storage lipid-water and storage lipid-air partition coefficients (Kislip/w and Kislip/a), as well as their temperature dependence, are considered useful. Polyparameter linear free energy relationships (PP-LFERs) are accurate, general, and mechanistically clear models for predicting partitioning-related physicochemical quantities. About a decade ago, PP-LFERs were calibrated for Kislip/w at the physiological temperature of 37 °C. However, to date, a comprehensive collection and sufficiently reliable PP-LFERs for Kislip/w and Kislip/a at the most common standard temperature of 25 °C are still lacking. In this study, experimentally based Kislip/w and/or Kislip/a values at 25 °C for 278 compounds were extensively collected or converted from the literature. Subsequently, PP-LFERs were calibrated for Kislip/w and Kislip/a at 25 °C, performing well over 10 orders of magnitude with root-mean-square errors of 0.17-0.21 log units for compounds with reliable descriptors. Furthermore, standard internal energy changes of transfer from water or air to storage lipids for 158 compounds were derived and used to calibrate PP-LFERs for estimating the temperature dependence of Kislip/w or Kislip/a. Additionally, using PP-LFERs, low-density polyethylene was confirmed to be a better storage lipid analogue than silicone and polyoxymethylene in the equilibrium passive sampling of nonpolar and H-bond acceptor polar compounds.

Identifying uncertainty in physical–chemical property estimation with IFSQSAR

This study describes the development and evaluation of six new models for predicting physical–chemical (PC) properties that are highly relevant for chemical hazard, exposure, and risk estimation: solubility (in water SW and octanol SO), vapor pressure (VP), and the octanol–water (KOW), octanol–air (KOA), and air–water (KAW) partition ratios. The models are implemented in the Iterative Fragment Selection Quantitative Structure–Activity Relationship (IFSQSAR) python package, Version 1.1.0. These models are implemented as Poly-Parameter Linear Free Energy Relationship (PPLFER) equations which combine experimentally calibrated system parameters and solute descriptors predicted with QSPRs. Two other ancillary models have been developed and implemented, a QSPR for Molar Volume (MV) and a classifier for the physical state of chemicals at room temperature. The IFSQSAR methods for characterizing applicability domain (AD) and calculating uncertainty estimates expressed as 95% prediction intervals (PI) for predicted properties are described and tested on 9,000 measured partition ratios and 4,000 VP and SW values. The measured data are external to IFSQSAR training and validation datasets and are used to assess the predictivity of the models for “novel chemicals” in an unbiased manner. The 95% PI intervals calculated from validation datasets for partition ratios needed to be scaled by a factor of 1.25 to capture 95% of the external data. Predictions for VP and SW are more uncertain, primarily due to the challenges in differentiating their physical state (i.e., liquids or solids) at room temperature. The prediction accuracy of the models for log KOW, log KAW and log KOA of novel, data-poor chemicals is estimated to be in the range of 0.7 to 1.4 root mean squared error of prediction (RMSEP), with RMSEP in the range 1.7–1.8 for log VP and log SW. Scientific contributionNew partitioning models integrate empirical PPLFER equations and QSARs, allowing for seamless integration of experimental data and model predictions. This work tests the real predictivity of the models for novel chemicals which are not in the model training or external validation datasets.Graphical

Insights into the role of VOCs properties on thermal desorption behaviors of two porous polymeric resins

Desorption is a critical process in the recovery or post-treatment of adsorbents saturated with volatile organic compounds (VOCs). In this study, the thermal desorption behaviors for eight VOCs on hypercrosslinked polymeric resin (HPR) and macroporous polymeric resin (MPR) were investigated through isothermal desorption and temperature programmed desorption (TPD). Compared with MPR, HPR with more micropores exhibited a lower desorption rate constant, lower desorption efficiency and higher desorption activation energy due to the strong binding energy generated between VOCs molecules and narrow micropores. As the polarizability of VOCs increased, the desorption rate constants on two porous polymeric resins decreased, while the desorption activation energy showed an incremental trend. Excellent linear correlations were observed between VOC polarizability and desorption rate constants (R2 = 0.957 for HPR and R2 = 0.940 for MPR) as well as between VOC polarizability and desorption activation energy (R2 = 0.981 for HPR and R2 = 0.969 for MPR). Furthermore, a polyparameter linear free energy relationship (PP-LFER) was developed to explore the influences of intermolecular interactions on desorption behaviors of VOCs on porous polymeric resins. The results indicated that the dispersive interaction, which is directly related to polarizability of VOCs, was the primary factor influencing the desorption activation energy of VOCs on porous polymeric resins. The find from this study helps evaluate fleetly and availably the desorption properties of VOCs based on their polarizability.

Modeling sorption of environmental organic chemicals from water to soils

Reliable estimation of chemical sorption from water to solid phases is an essential prerequisite for reasonable assessments of chemical hazards and risks. However, current fate and exposure models mostly rely on algorithms that lack the capability to quantify chemical sorption resulting from interactions with multiple soil constituents, including amorphous organic matter, carbonaceous organic matter, and mineral matter. Here, we introduce a novel, generic approach that explicitly combines the gravimetric composition of various solid constituents and poly-parameter linear free energy relationships to calculate the solid-water sorption coefficient (Kd) for non-ionizable or predominantly neutral organic chemicals with diverse properties in a neutral environment. Our approach demonstrates an overall statistical uncertainty of approximately 0.9 log units associated with predictions for different types of soil. By applying this approach to estimate the sorption of 70 diverse chemicals from water to two types of soils, we uncover that different chemicals predominantly exhibit sorption onto different soil constituents. Moreover, we provide mechanistic insights into the limitation of relying solely on organic carbon normalized sorption coefficient (KOC) in chemical hazard assessment, as the measured KOC can vary significantly across different soil types, and therefore, a universal cut-off threshold may not be appropriate. This research highlights the importance of considering chemical properties and multiple solid constituents in sorption modeling and offers a valuable theoretical approach for improved chemical hazard and exposure assessments.

First insights into the bioaccumulation, biotransformation and trophic transfer of typical tetrabromobisphenol A (TBBPA) analogues along a simulated aquatic food chain

Tetrabromobisphenol A (TBBPA) analogues have been investigated for their prevalent occurrence in environments and potential hazardous effects to humans and wildlife; however, there is still limited knowledge regarding their toxicokinetics and trophic transfer in aquatic food chains. Using a developed toxicokinetic model framework, we quantified the bioaccumulation, biotransformation and trophic transfer of tetrabromobisphenol S (TBBPS) and tetrabromobisphenol A di(allyl ether) (TBBPA-DAE) during trophic transfer from brine shrimp (Artemia salina) to zebrafish (Danio rerio). The results showed that the two TBBPA analogues could be readily accumulated by brine shrimp, and the estimated bioconcentration factor (BCF) value of TBBPS (5.68 L kg-1 ww) was higher than that of TBBPA-DAE (1.04 L kg-1 ww). The assimilation efficiency (AE) of TBBPA-DAE in zebrafish fed brine shrimp was calculated to be 16.3%, resulting in a low whole-body biomagnification factor (BMF) in fish (0.684 g g-1 ww). Based on the transformation products screened using ultra-high-performance liquid chromatograph-high resolution mass spectrometry (UPLC-HRMS), oxidative debromination and hydrolysis were identified as the major transformation pathways of TBBPS, while the biotransformation of TBBPA-DAE mainly took place through ether bond breaking and phase-II metabolism. Lower accumulation of TBBPA as a metabolite than its parent chemical was observed in both brine shrimp and zebrafish, with metabolite parent concentration factors (MPCFs) < 1. The investigated BCFs for shrimp of the two TBBPA analogues were only 3.77 × 10-10 - 5.59 × 10-3 times of the theoretical Kshrimp-water based on the polyparameter linear free energy relationships (pp-LFERs) model, and the BMF of TBBPA-DAE for fish was 0.299 times of the predicted Kshrimp-fish. Overall, these results indicated the potential of the trophic transfer in bioaccumulation of specific TBBPA analogues in higher trophic-level aquatic organisms and pointed out biotransformation as an important mechanism in regulating their bioaccumulation processes. Environmental ImplicationThe internal concentration of a pollutant in the body determines its toxicity to organisms, while bioaccumulation and trophic transfer play important roles in elucidating its risks to ecosystems. Tetrabromobisphenol A (TBBPA) analogues have been extensively investigated for their adverse effects on humans and wildlife; however, there is still limited knowledge regarding their toxicokinetics and trophic transfer in aquatic food chains. This study investigated the bioaccumulation, biotransformation and trophic transfer of TBBPS and TBBPA-DAE in a simulated di-trophic food chain. This state-of-art study will provide a reference for further research on this kind of emerging pollutant in aquatic environments.

Predictive modeling for VOC adsorption on TiO2 filters: Implications for enhanced photocatalytic oxidation in indoor air quality management

Adsorption of volatile organic compounds (VOCs) onto the surface of a TiO2 filter is essential for initiating photocatalytic oxidation (PCO). This study investigated the adsorption coefficients (K) of 14 VOCs under different relative humidities (RH) and explored the influence of different air velocities on the adsorption of Methyl isobutyl ketone (MIBK). Consequently, a polyparameter linear free energy relationship (pp-LFER) model was established to predict lnK at 35 % RH (lnK35%). Moreover, the study found that lnK has a linear relationship with RH, represented as lnK=αRH+β. The parameters ′α′ and ′β′ can be accurately predicted using the Henry’s Law constant (K°OH) and lnK35%, respectively. By combining the pp-LFER model with regression equations for ′α′ and ′β′, it is possible to predict lnK for VOCs across various RH levels. The study also revealed that the adsorption coefficient of increases with escalating air velocities (0.04 – 0.34 m/s), primarily due to mass transfer. The observed low adsorption performance for aldehydes, commonly seen as by-products in the PCO process, suggests cautious usage of the PCO filtration system in practice before the by-products are eliminated downstream. This study offers valuable insights for the practical application of PCO technologies in the comprehensive removal of indoor VOCs.

Acetal Formation of Flavoring Agents with Propylene Glycol in E-Cigarettes: Impacts on Indoor Partitioning and Thirdhand Exposure.

The widespread use of flavored e-cigarettes has led to a significant rise in teenage nicotine use. In e-liquids, the flavor carbonyls can form acetals with unknown chemical and toxicological properties. These acetals can cause adverse health effects on both smokers and nonsmokers through thirdhand exposure. This study aims to explore the impacts of these acetals formed in e-cigarettes on indoor partitioning and thirdhand exposure. Specifically, the acetalization reactions of commonly used flavor carbonyls in laboratory-made e-liquids were monitored using proton nuclear magnetic resonance (1H NMR) spectroscopy. EAS-E Suite and polyparameter linear free energy relationships (PP-LFERs) were employed to estimate the partitioning coefficients for species. Further, a chemical two-dimensional partitioning model was applied to visualize the indoor equilibrium partitioning and estimate the distribution of flavor carbonyls and their acetals in the gas phase, aerosol phase, and surface reservoirs. Our results demonstrate that a substantial fraction of carbonyls were converted into acetals in e-liquids and their chemical partitioning was significantly influenced. This study shows that acetalization is a determinant factor in the exposure and toxicology of harmful carbonyl flavorings, with its impact extending to both direct exposure to smokers and involuntary exposure to nonsmokers.

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.

Improved Predictions of Dissolved Organic Carbon–Water Partitioning and Unveiling the Chemodiversity of Dissolved Organic Matter

Dissolved organic matter (DOM) is a complex mixture of organic compounds with chemical heterogeneity in natural soils, sediments, and waters. Predicting the dissolved organic carbon (DOC)–water partition coefficients (KDOC) of organic pollutants to DOMs from various sources is important for assessing their fate and bioavailability. However, the accuracy of the recently developed polyparameter linear free energy relationship (pp-LFER) KDOC models is generally restricted by the small data set, inclusion of experimental data from unreliable measurement methods, or undefined selection standards. The aim of this study was to establish pp-LFER models for predicting the KDOC of nonionic organic chemicals to a variety of sources of DOMs and to get an understanding of DOM chemodiversity. A reliable and expanded experimental KDOC data set was compiled. Improved pp-LFER models were developed and assessed for All DOM (DOM from all sources), Aldrich humic acid (HA), Roth HA, soil porewater DOM, sediment porewater DOM, natural aquatic DOM, natural terrestrial DOM, natural DOM, and commercial DOM. The models developed in this study were reasonably robust and accurate with a root mean square error (RMSE) of 0.538 for the All DOM model and RMSE from 0.196 to 0.860 for the specific DOMs. Also, the models generally performed better than previously published ones. Moreover, the system parameters of the models well described the chemical variabilities between soil porewater DOM and sediment porewater DOM, between natural aquatic DOM and natural terrestrial DOM, and between natural DOM and commercial DOM regarding polarizability, dipolarity, H-bond-accepting and -donating ability, and cavity formation energy. This study provides effective tools to assess the tendencies of organic chemicals to DOMs and improves our understanding of the chemical heterogeneity of DOMs from different sources.

Recalibrating polyparameter linear free energy relationships and reanalyzing mechanisms for partition of nonionic organic compounds to low-density polyethylene passive sampler

Equilibrium passive sampling techniques based on the low-density polyethylene (LDPE) film are increasingly used for determining the concentration of contaminants in water and air. Reliable models capable of predicting LDPE–water and LDPE–air partition coefficients (KiLDPEw and KiLDPEa) would be very useful. In previous studies, polyparameter linear free energy relationships (PP-LFERs) based on Abraham's solute descriptors were calibrated for LDPE–water and LDPE–air systems. Unfortunately, a portion of unreliable partition coefficients and solute descriptors were included in the calibration sets of these previous studies, leading to unexpected system parameters and predictive performance in the regression results. In this study, more reliable PP-LFERs were recalibrated for LDPE–water and LDPE–air systems (20‒25 °C) using carefully collected reliable partition coefficients and solute descriptors of various polar and nonpolar compounds (over one hundred and with low redundancy) from the literature, as well as the robust regression method. The PP-LFERs performed well with root-mean-square errors of 0.15–0.25 log units and successfully predicted KiLDPEw and KiLDPEa values spanning over 10 orders of magnitude for compounds with reliable descriptors. The partitioning mechanisms of compounds to LDPE were also reanalyzed and compared in detail with n-alkanes (C6–C16). Generally, LDPE is more prone to form dispersion interactions with solutes than n-alkanes, while it is more difficult to form cavities in LDPE. In addition, the crystallinity of LDPE is not the sole reason for the distinct constant terms presenting in PP-LFERs for LDPE–water and n-hexadecane–water systems.

Expanding the Applicability Range of the Generator Column Technique to Measure the Octanol–Air Partition Ratio of Volatile Compounds

Bioaccumulation modeling has identified a lower limit of ∼5 for the log of the octanol–air partition ratio (log KOA) of chemicals that potentially bioaccumulate in air-breathing organisms. Because existing techniques are not well suited to this volatility range (vapor pressure between 1 and 1000 Pa), we adjusted the generator column method to allow for KOA determination of more volatile compounds. We reduced gas flow rates and experimental run times and examined the potential depletion of octanol within the generator column. By recording KOA values for hexachlorobenzene, 1,2,3,4-tetrachlorobenzene, and 1,2,3-trichlorobenzene that agree with earlier measurements, we first confirmed our ability to obtain reliable data with the generator column technique. The log KOA values between 5 and 45 °C for 1,3-dichlorobenzene, trans-decalin, 2,6-dimethyldecane, 1,2,3,4-tetramethylbenzene, hexylcyclohexane, tri-isopropyl phosphate, and tetradecene have been obtained with the technique. The lowest log10 KOA value measured was 4.69 for trans-decalin at 25 °C. Good agreement with predictions made with poly-parameter linear free energy relationships lends confidence to these data. Reducing gas flow rates and experimental run times are key to obtaining reliable KOA data when applying the generator column technique to volatile compounds. Depletion of the octanol concentration in the generator column did not notably affect the KOA obtained from the measurements.

Models Used to Predict Chemical Bioaccumulation in Fish from in Vitro Biotransformation Rates Require Accurate Estimates of Blood-Water Partitioning and Chemical Volume of Distribution.

Methods for extrapolating measured in vitro intrinsic clearance to a whole-body biotransformation rate constant (kB ) have been developed to support modeled bioaccumulation assessments for fish. The inclusion of extrapolated kB values into existing bioaccumulation models improves the prediction of chemical bioconcentration factors (BCFs), but there remains a tendency for these methods to overestimate BCFs relative to measured values. Therefore, a need exists to evaluate the extrapolation procedure to assess potential sources of error in predicted kB values. We examined how three different approaches (empirically based, composition based, and polyparameter linear free energy relationships [ppLFERs]) used to predict chemical partitioning in vitro (liver S9 system; KS9W ), in blood (KBW ), and in whole fish tissues (KFW ) impact the prediction of a chemical's hepatic clearance binding term (fU ) and apparent volume of distribution (VD ), both of which factor into the calculation of kB and the BCF. Each approach yielded different KS9W , KBW , and KFW values, but resulted in fU values that were of similar magnitude and remained relatively constant at log octanol-water partition ratios (KOW ) greater than 4. This is because KBW and KS9W values predicted by any given approach exhibit a similar dependence on log KOW (i.e., regression slope), which results in a cancelation of "errors" when fU is calculated. In contrast, differences in KBW values predicted by the three approaches translate to differences in VD , and by extension kB and the BCF, which become most apparent at log KOW greater than 6. There is a need to collect KBW and VD data for hydrophobic chemicals in fish that can be used to evaluate and improve existing partitioning prediction approaches in extrapolation models for fish. Environ Toxicol Chem 2023;42:33-45. © 2022 SETAC.

Universal machine-learning algorithm for predicting adsorption performance of organic molecules based on limited data set: Importance of feature description

Adsorption of organic molecules from aqueous solution offers a simple and effective method for their removal. Recently, there have been several attempts to apply machine learning (ML) for this problem. To this end, polyparameter linear free energy relationships (pp-LFERs) were employed, and poor prediction results were observed outside model applicability domain of pp-LFERs. In this study, we improved the applicability of ML methods by adopting a chemical-structure (CS) based approach. We used the prediction of adsorption of organic molecules on carbon-based adsorbents as an example. Our results show that this approach can fully differentiate the structural differences between any organic molecules, while providing significant information that is relevant to their interaction with the adsorbents. We compared two CS feature descriptors: 3D-coordination and simplified molecular-input line-entry system (SMILES). We then built CS-ML models based on neural networks (NN) and extreme gradient boosting (XGB). They all outperformed pp-LFERs based models and are capable to accurately predict adsorption isotherm of isomers with similar physiochemical properties such as chiral molecules, even though they are trained with achiral molecules and racemates. We found for predicting adsorption isotherm, XGB shows better performance than NN, and 3D-coordinations allow effective differentiation between organic molecules.

Predicting the Temperature Dependence of the Octanol–Air Partition Ratio: A New Model for Estimating Delta {U^{ circ}_{text{OA}}}

The octanol–air partition ratio (KOA) describes the partitioning of a chemical between air and octanol and is often used to approximate other partitioning phenomena in environmental chemistry (e.g., blood–air, atmospheric particulate matter–air, polyurethane foam-air). Such partitioning processes often occur at environmental temperatures other than 25 °C. Enthalpies Delta {H^{ circ}_{text{OA}}} or internal energies Delta {U^{ circ}_{text{OA}}} of phase transfer are used to express the temperature dependence of the KOA. Existing poly-parameter linear free energy relationships (ppLFERs) for predicting Delta {H^{ circ}_{text{OA}}} were developed using a relatively small dataset. In this work we utilize a recently developed comprehensive KOA database to create and curate a Delta {U^{ circ}_{text{OA}}} dataset containing 195 chemicals and use this dataset in the development of new predictive equations. Using the QSAR development platform QSARINS we evaluate the use of Abraham descriptors, other molecular descriptors, and the log10KOA at 25 °C as variables in different multilinear regression equations for Delta {U^{ circ}_{text{OA}}}. The Delta {U^{ circ}_{text{OA}}} of neutral organic chemicals can be reliably predicted using only the log10KOA (RMSEEXT = 6.86 kJ·mol−1, {text{R}^{2} _{text{adj}}} = 0.94), only the solute’s hydrogen acidity A and the logarithm of the hexadecane–air partition ratio L (RMSEEXT = 7.23 kJ·mol−1, {text{R}^{2} _{text{adj}}} = 0.93), or A and log10KOA (RMSEEXT = 6.76 kJ·mol−1, {text{R}^{2} _{text{adj}}} = 0.95).

QSPRs for Predicting Equilibrium Partitioning in Solvent–Air Systems from the Chemical Structures of Solutes and Solvents

Poly-parameter Linear Free Energy Relationships (PPLFERs) based on the Abraham solvation model are a useful tool for predicting and interpreting equilibrium partitioning of solutes in solvent systems. The focus of this work is neutral organic solutes partitioning in neutral organic liquid solvent-air systems. This is a follow-up to previous work (Brown, 2021) which developed predictive empirical correlations between solute descriptors and system parameters, allowing system parameters to be predicted from the solute descriptors of the solvent. A database of solute descriptors, and a database of system parameters supplemented by empirical predictions, form the basis for the development of new Quantitative Structure Property Relationships (QSPRs). A total of 11 QSPRs have been developed for the E, S, A, B and L solute descriptors, and the s, a, b, v, l, and c system parameters. The QSPRs were developed using a group-contribution method referred to as Iterative Fragment Selection. The method includes robust internal and external model validation and a well-defined Applicability Domain, including estimates of prediction uncertainty. System parameters can also be predicted by combining the solute descriptor QSPRs and the empirical correlations. The predictive power of PPLFERs applied using different combinations of experimental data, empirical correlations, and QSPRs are externally validated by predicting partition ratios between solvents and air. The uncertainty for predicting the log10KSA of diverse solutes in diverse solvents using only the new QSPRs and empirical correlations is estimated to be one log10 unit or less.

Applicability Domain of Polyparameter Linear Free Energy Relationship Models Evaluated by Leverage and Prediction Interval Calculation.

Polyparameter linear free energy relationships (PP-LFERs) are accurate and robust models employed to predict equilibrium partition coefficients (K) of organic chemicals. The accuracy of predictions by a PP-LFER depends on the composition of the respective calibration data set. Generally, extrapolation outside the domain defined by the calibration data is likely to be less accurate than interpolation. In this study, the applicability domain (AD) of PP-LFERs was systematically evaluated by calculating the leverage (h) and prediction interval (PI). Repeated simulations with experimental data showed that the root mean squared error of predictions increased with h. However, the analysis also showed that PP-LFERs calibrated with a large number (e.g., 100) of training data were highly robust against extrapolation error. For such PP-LFERs, the common definition of extrapolation (h > 3 hmean, where hmean is the mean h of all training compounds) may be excessively strict. Alternatively, the PI is proposed as a metric to define the AD of PP-LFERs, as it provides a concrete estimate of the error range that agrees well with the observed errors, even for extreme extrapolations. Additionally, published PP-LFERs were evaluated in terms of their AD using the new concept of AD probes, which indicated the varying predictive performance of PP-LFERs in the existing literature for environmentally relevant compounds.

Simulating and Predicting Adsorption of Organic Pollutants onto Black Phosphorus Nanomaterials.

Layered black phosphorus (BP) has exhibited exciting application prospects in diverse fields. Adsorption of organics onto BP may influence environmental behavior and toxicities of both organic pollutants and BP nanomaterials. However, contributions of various intermolecular interactions to the adsorption remain unclear, and values of adsorption parameters such as adsorption energies (Ead) and adsorption equilibrium constants (K) are lacking. Herein, molecular dynamic (MD) and density functional theory (DFT) was adopted to calculate Ead and K values. The calculated Ead and K values for organics adsorbed onto graphene were compared with experimental ones, so as to confirm the reliability of the calculation methods. Polyparameter linear free energy relationship (pp-LFER) models on Ead and logK were developed to estimate contributions of different intermolecular interactions to the adsorption. The adsorption in the gaseous phase was found to be more favorable than in the aqueous phase, as the adsorbates need to overcome cohesive energies of water molecules onto BP. The affinity of the aromatics to BP was comparable to that of graphene. The pp-LFER models performed well for predicting the Ead and K values, with external explained variance ranging from 0.90 to 0.97, and can serve as effective tools to rank adsorption capacities of organics onto BP.

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.