Machine Learning for Ionic Liquid Toxicity Prediction

Hydrocarbon Extraction with Ionic Liquids

Separation and reaction processes are key components employed in the modern chemical industry, and the former accounts for the majority of the energy consumption therein. In particular, hydrocarbon separation and purification processes, such as aromatics extraction, desulfurization, and denitrification, are challenging in petroleum refinement, an industrial cornerstone that provides raw materials for products used in human activities. The major technical shortcomings in solvent extraction are volatile solvent loss, product entrainment leading to secondary pollution, low separation efficiency, and high regeneration energy consumption due to the use of traditional organic solvents with high boiling points as extraction agents. Ionic liquids (ILs), a class of designable functional solvents or materials, have been widely used in chemical separation processes to replace conventional organic solvents after nearly 30 years of rapid development. Herein, we provide a systematic and comprehensive review of the state-of-the-art progress in ILs in the field of extractive hydrocarbon separation (i.e., aromatics extraction, desulfurization, and denitrification) including (i) molecular thermodynamic models of IL systems that enable rapid large-scale screening of IL candidates and phase equilibrium prediction of extraction processes; (ii) structure-property relationships between anionic and cationic structures of ILs and their separation performance (i.e., selectivity and distribution coefficients); (iii) IL-related extractive separation mechanisms (e.g., the magnitude, strength, and sites of intermolecular interactions depending on the separation system and IL structure); and (iv) process simulation and design of IL-related extraction at the industrial scale based on validated thermodynamic models. In short, this Review provides an easy-to-read exhaustive reference on IL-related extractive separation of hydrocarbon mixtures from the multiscale perspective of molecules, thermodynamics, and processes. It also extends to progress in IL analogs, deep eutectic solvents (DESs) in this research area, and discusses the current challenges faced by ILs in related separation fields as well as future directions and opportunities.

Energy Applications of Ionic Liquids: Recent Developments and Future Prospects

Ionic liquids (ILs) consisting entirely of ions exhibit many fascinating and tunable properties, making them promising functional materials for a large number of energy-related applications. For example, ILs have been employed as electrolytes for electrochemical energy storage and conversion, as heat transfer fluids and phase-change materials for thermal energy transfer and storage, as solvents and/or catalysts for CO2 capture, CO2 conversion, biomass treatment and biofuel extraction, and as high-energy propellants for aerospace applications. This paper provides an extensive overview on the various energy applications of ILs and offers some thinking and viewpoints on the current challenges and emerging opportunities in each area. The basic fundamentals (structures and properties) of ILs are first introduced. Then, motivations and successful applications of ILs in the energy field are concisely outlined. Later, a detailed review of recent representative works in each area is provided. For each application, the role of ILs and their associated benefits are elaborated. Research trends and insights into the selection of ILs to achieve improved performance are analyzed as well. Challenges and future opportunities are pointed out before the paper is concluded.

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.

A Review of Process Systems Engineering (PSE) Tools for the Design of Ionic Liquids and Integrated Biorefineries

Over the years, the global process industry is continually improving in product development and process performances [...]

Deep Probabilistic Learning Model for Prediction of Ionic Liquids Toxicity

Identification of ionic liquids with low toxicity is paramount for applications in various domains. Traditional approaches used for determining the toxicity of ionic liquids are often expensive, and can be labor intensive and time consuming. In order to mitigate these limitations, researchers have resorted to using computational models. This work presents a probabilistic model built from deep kernel learning with the aim of predicting the toxicity of ionic liquids in the leukemia rat cell line (IPC-81). Only open source tools, namely, RDKit and Mol2vec, are required to generate predictors for this model; as such, its predictions are solely based on chemical structure of the ionic liquids and no manual extraction of features is needed. The model recorded an RMSE of 0.228 and R2 of 0.943. These results indicate that the model is both reliable and accurate. Furthermore, this model provides an accompanying uncertainty level for every prediction it makes. This is important because discrepancies in experimental measurements that generated the dataset used herein are inevitable, and ought to be modeled. A user-friendly web server was developed as well, enabling researchers and practitioners ti make predictions using this model.

High-Throughput Computational Screening of Ionic Liquids for Butadiene and Butene Separation

The separation of 1,3-butadiene (1,3-C4H6) and 1-butene (n-C4H8) is quite challenging due to their close boiling points and similar molecular structures. Extractive distillation (ED) is widely regarded as a promising approach for such a separation task. For ED processes, the selection of suitable entrainer is of central importance. Traditional ED processes using organic solvents suffer from high energy consumption. To tackle this issue, the utilization of ionic liquids (ILs) can serve as a potential alternative. In this work, a high-throughput computational screening of ILs is performed to find proper entrainers, where 36,260 IL candidates comprising of 370 cations and 98 anions are involved. COSMO-RS is employed to calculate the infinite dilution extractive capacity and selectivity of the 36,260 ILs. In doing so, the ILs that satisfy the prespecified thermodynamic criteria and physical property constraints are identified. After the screening, the resulting IL candidates are sent for rigorous process simulation and design. 1,2,3,4,5-pentamethylimidazolium methylcarbonate is found to be the optimal IL solvent. Compared with the benchmark ED process where the organic solvent N-methyl-2-pyrrolidone is adopted, the energy consumption is reduced by 26%. As a result, this work offers a new IL-based ED process for efficient 1,3-C4H6 production.

Atom surface fragment contribution method for predicting the toxicity of ionic liquids

In this study, a novel method-atom surface fragment contribution (ASFC)-was proposed for assessing the properties of compounds. We developed a predictive model using the ASFC method based on the sigma surface areas (S_σ-surface) of fragments/groups for estimating the toxicity of ILs. A toxicity dataset of 140 ILs towards leukemia rat cell line (ICP-81) was gathered and employed to train and validate models. The S_σ-surface values of atoms in each group were firstly calculated from the COSMO profiles of cations and anions for ILs. Then the S_σ-surface values of 26 groups were obtained and used as input descriptors for modelling. The R² and MSE of the built ASFC model were 0.924 and 0.071, respectively. Results indicate that the ASFC model developed by the new approach possesses great accuracy and reliability. In total, the ASFC method has extensive potential for the application of estimating diverse properties of ILs and other compounds due to its remarkable advantages.