In recent years, the absorption of carbon dioxide by water-lean solvents has received special attention. In this study, carbon dioxide absorption was performed in a microfluidic device using a water-lean monoethanolamine solution. The effect of different operating conditions, including inlet solvent flow, solvent concentration, and temperature, on CO2 removal efficiency, overall mass transfer coefficient, and mass transfer flux were investigated. Response surface methodology (RSM) was used to analyze and optimize the responses. The maximum removal efficiency of 92.24% and mass transfer coefficient of 155.12 kmol/m3hrkPa were achieved at the solvent concentration of 30 wt %, 40 °C, and solvent flow of 9 mL/min. For amine-methanol solvent in these conditions, the overall volume mass transfer coefficient was 18.09% higher than the aqueous solvent. The mass transfer coefficient obtained in this study was significantly higher than the values reported for packed towers. Furthermore, the artificial neural network (ANN) method as a branch of machine learning (ML) models was utilized for modeling the CO2 removal efficiency of the water-lean monoethanolamine solvent. The number of neurons and different transfer functions have been optimized in MLP and RBF models to select optimum ANN. The results show that the MLP model with a tangent sigmoid transfer function showed the best performance with an RMSE value of 0.35103, which indicates that the utilized ML method predicts the CO2 removal efficiency of water-lean amine solution to satisfactory levels.
