Prediction of CO₂ solubility in Ionic liquids for CO₂ capture using deep learning models

Area of Science:

• Materials Science
• Chemical Engineering
• Computational Chemistry

Background:

• Ionic liquids (ILs) are effective solvents for carbon dioxide (CO2) capture.
• Accurate prediction of CO2 solubility in ILs is essential for process optimization.
• Existing prediction methods may lack efficiency or accuracy for diverse ILs.

Purpose of the Study:

• To develop and evaluate deep learning models for predicting CO2 solubility in ILs.
• To compare the performance and efficiency of Artificial Neural Network (ANN) and Long Short-Term Memory (LSTM) models.
• To identify key parameters influencing CO2 solubility through sensitivity analysis.

Main Methods:

• Utilized a comprehensive dataset of 10,116 CO2 solubility data points across 164 ILs.
• Developed and trained ANN and LSTM deep neural network models.
• Performed global sensitivity analysis (GSA) to understand parameter influence.

Main Results:

• Both ANN and LSTM models achieved high prediction accuracy (R² values of 0.986 and 0.985, respectively).
• ANN models demonstrated superior computational efficiency, being approximately 30 times faster than LSTM.
• Sensitivity analysis revealed the impact of process parameters and IL functional groups on CO2 solubility.

Conclusions:

• Deep learning models show significant potential for predicting CO2 solubility in ILs.
• ANN models provide a computationally efficient and accurate approach for IL screening in CO2 capture.
• The study offers valuable insights for designing and selecting ILs for enhanced CO2 capture.