Experimental optimization and machine learning modeling of CO₂ capture in choline chloride-ethylene glycol

Deep eutectic solvents are emerging as sustainable and environmentally friendly alternatives for carbon dioxide absorption. This experimental study investigated the efficient capture of CO₂ using deep eutectic solvents. The pressure drop method was employed to evaluate the absorption capabilities of various deep eutectic solvent formulations. The response surface methodology was used to identify and optimize the experimental parameters and the best operational conditions for CO₂ absorption. Solubility studies were conducted under controlled conditions, such as deep eutectic solvent mole ratios (ranging from 1:1 to 1:6), contact times (from 63 to 693 min), and water percentages (from 17.5 % to 30 %). The results indicated that the optimal CO₂ absorption occurred at a deep eutectic solvent molar ratio of 1:4.75, a contact time of 693 min, and a starting pressure of 15 bar. To further analyze the physical property data obtained from our experiments, ANOVA was used to assess the significance of various influencing factors. Fourier transform infrared spectroscopy confirmed the presence of CO₂ in the DES after absorption. The XGBOOST machine learning algorithm, which was employed to build the prediction model for solubility, successfully predicted the solubility with a maximum error of 8.1 %. This study highlights the potential of deep eutectic solvents for effective CO₂ capture and provides a framework for further optimisation and application.