Prediction of nitrogen solubility in ionic liquids by machine learning methods based on COSMO-derived descriptors

Tian, Yuan; Wang, Xinxin; Liu, Yanrong; Hu, Wenping

doi:10.1016/j.ces.2023.119482

Cited by 10 publications

(2 citation statements)

References 49 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…All ML algorithms were conducted by the open-source code Scikit-learn and PyTorch package in the Python3 environment. A spectrum of supervised ML regression algorithms has proven efficacious, including gradient boosting regression (GBR), artificial neural networks (ANN), kernel ridge regression (KRR), and extreme gradient boosting (XGBoost) . Notably, these algorithms afford the dual benefit of accurately predicting material properties akin to density functional theory (DFT) while also furnishing atomic-level chemical insights.…”

Section: Computational Detailsmentioning

confidence: 99%

Combined Machine Learning and High-Throughput Calculations Predict Heyd–Scuseria–Ernzerhof Band Gap of 2D Materials and Potential MoSi₂N₄ Heterostructures

Zhang,

Guo,

et al. 2024

J. Phys. Chem. Lett.

View full text Add to dashboard Cite

We present a novel target-driven methodology devised to predict the Heyd–Scuseria–Ernzerhof (HSE) band gap of two-dimensional (2D) materials leveraging the comprehensive C2DB database. This innovative approach integrates machine learning and density functional theory (DFT) calculations to predict the HSE band gap, conduction band minimum (CBM), and valence band maximum (VBM) of 2176 types of 2D materials. Subsequently, we collected a comprehensive data set comprising 3539 types of 2D materials, each characterized by its HSE band gaps, CBM, and VBM. Considering the lattice disparities between MoSi2N4 (MSN) and 2D materials, our analysis predicted 766 potential MSN/2D heterostructures. These heterostructures are further categorized into four distinct types based on the relative positions of their CBM and VBM: Type I encompasses 230 variants, Type II comprises 244 configurations, Type III consists of 284 permutations, and 0 band gap comprises 8 types.

show abstract

Section: Computational Detailsmentioning

confidence: 99%

Combined Machine Learning and High-Throughput Calculations Predict Heyd–Scuseria–Ernzerhof Band Gap of 2D Materials and Potential MoSi₂N₄ Heterostructures

Zhang,

Guo,

et al. 2024

J. Phys. Chem. Lett.

View full text Add to dashboard Cite

show abstract

“…Hence, the solubility of N 2 in ILs is vital for the design and development of high-efficiency NRR. Recently, Tian et al used COSMO and machine learning to predict N 2 solubility data. However, the mechanisms underlying the differences in N 2 solubility among various ILs are still not well understood.…”

Section: Introductionmentioning

confidence: 99%

Fluorine Domains Induced Ultrahigh Nitrogen Solubility in Ionic Liquids

Li,

Wang,

Wang

et al. 2024

J. Am. Chem. Soc.

View full text Add to dashboard Cite

Fluorinated ionic liquids (ILs) are well-known as electrolytes in the nitrogen (N 2 ) electroreduction reaction due to their exceptional gas solubility. However, the influence of fluorinated functional group on N 2 solvation and solubility enhancement remains unclear. Massive molecular dynamics simulations and free energy perturbation methods are conducted to investigate the N 2 solubility in 11 traditional and 9 fluorinated ILs. It shows that the fluorinated IL of 1-Ethyl-3-methylimidazolium tris(pentafluoroethyl) trifluorophosphate ([Emim]FAP) exhibits ultrahigh solubility, 4.844 × 10 −3 , approximately 118 times higher than that of traditional IL 1-Ethyl-3-methylimidazolium nitrate ([Emim]NO 3 ). Moreover, fluorinated ILs with more than 10 C−F bonds possess higher N 2 solubility than others and show an exothermic nature during solvation. As the C−F bonds number in ILs decreases, the N 2 solubility decreases significantly and displays the opposite endothermic behavior. To understand the ultrahigh N 2 solubility in fluorinated ILs, we propose a concept of fluorine densification energy (FDE), referring to the average strength of interaction between atoms per unit volume in ILs with fluorine domains, demonstrating a linear relationship with C−F bonds. Physically, lower FDE results in lower N 2 −anion pair dissociation energy and higher free volume, finally enhancing the N 2 solubility. Consequently, medium to long alkyl fluorine tails within a polar environment defines a distinct fluorine domain, emphasizing FDE's role in enhancing N 2 solubility. Overall, these quantitative results will not only deepen the understanding of N 2 solvation in ILs but may also shed light on the rational design of IL-based high-performance N 2 capture and conversion technologies.

show abstract

Insights into the pore structure effect on the mass transfer of fuel cell catalyst layer via combining machine learning and multiphysics simulation

Luo,

Liu,

Zhang

et al. 2025

Chemical Engineering Science

View full text Add to dashboard Cite

Prediction of nitrogen solubility in ionic liquids by machine learning methods based on COSMO-derived descriptors

Cited by 10 publications

References 49 publications

Combined Machine Learning and High-Throughput Calculations Predict Heyd–Scuseria–Ernzerhof Band Gap of 2D Materials and Potential MoSi₂N₄ Heterostructures

Combined Machine Learning and High-Throughput Calculations Predict Heyd–Scuseria–Ernzerhof Band Gap of 2D Materials and Potential MoSi₂N₄ Heterostructures

Fluorine Domains Induced Ultrahigh Nitrogen Solubility in Ionic Liquids

Insights into the pore structure effect on the mass transfer of fuel cell catalyst layer via combining machine learning and multiphysics simulation

Contact Info

Product

Resources

About

Prediction of nitrogen solubility in ionic liquids by machine learning methods based on COSMO-derived descriptors

Cited by 10 publications

References 49 publications

Combined Machine Learning and High-Throughput Calculations Predict Heyd–Scuseria–Ernzerhof Band Gap of 2D Materials and Potential MoSi2N4 Heterostructures

Combined Machine Learning and High-Throughput Calculations Predict Heyd–Scuseria–Ernzerhof Band Gap of 2D Materials and Potential MoSi2N4 Heterostructures

Fluorine Domains Induced Ultrahigh Nitrogen Solubility in Ionic Liquids

Insights into the pore structure effect on the mass transfer of fuel cell catalyst layer via combining machine learning and multiphysics simulation

Contact Info

Product

Resources

About

Combined Machine Learning and High-Throughput Calculations Predict Heyd–Scuseria–Ernzerhof Band Gap of 2D Materials and Potential MoSi₂N₄ Heterostructures

Combined Machine Learning and High-Throughput Calculations Predict Heyd–Scuseria–Ernzerhof Band Gap of 2D Materials and Potential MoSi₂N₄ Heterostructures