Rational Design of Deep Eutectic Solvents for CO 2 Capture: Integrated Experimental Analysis and Machine Learning-Guided Screening

The current research investigates the CO 2 capture capabilities of eight formulated deep eutectic solvents (DESs). These DESs are based on the ammonium and phosphonium-based hydrogen bond donor (HBDs) and amine and alcohol-based hydrogen bond acceptor (HBAs). CO 2 solubility in the formulated DESs was conducted at varying pressures using a high-pressure volumetric sorption method. The results demonstrated remarkably higher CO 2 uptake capacities in the DES5 (methyltriphenylphosphonium bromide/1,6-hexanediol, 1:4), with the CO 2 uptake (1.43 mol of CO 2 per mol of DES at 2.024 MPa and 30 °C), whereas DES6 (guanidine hydrochloride/2-ethylaminoethanol, 1:2) showed the least CO 2 uptake (0.80 mol of CO 2 per mol of DES at 2.007 MPa and 30 °C). Various physicochemical properties of solvents were computed, and comprehensive quantitative structure–property relationship (QSPR) based thermodynamic framework were developed using seven machine learning models and were thoroughly validated through leave-one-out cross-validation. Additionally, a practical performance index was developed to strategically balance CO 2 uptake capacities of DESs with the limitation of transport property, aiding in the rational selection of the DESs for the CO 2 capture process. The results showed that DES7 and DES1 may be suitable solvents for industrial applications.