چكيده به لاتين
Climate change caused by greenhouse gas emissions, particularly carbon dioxide (CO2), is one of the most critical challenges facing humanity. Fossil fuel combustion is the primary source of CO2 emissions, and Carbon Capture and Storage (CCS) has been proposed as a key strategy to mitigate global warming. In this approach, CO2 is separated from gas streams, compressed, and stored in underground reservoirs such as saline aquifers or depleted oil and gas fields. Additionally, CO2 can be utilized in various industrial processes to produce value-added materials. Among the available CO2 capture technologies, post-combustion absorption using chemical solvents is one of the most promising options. However, conventional amine-based solvents, such as monoethanolamine (MEA), suffer from major drawbacks, including high energy consumption during solvent regeneration and corrosion issues. As a result, alternative absorbents are needed. Ionic liquids (ILs) have emerged as promising candidates due to their unique physicochemical properties, such as negligible vapor pressure, high CO2 absorption capacity, and excellent thermal stability. The vast structural diversity of ILs, derived from different combinations of cations and anions, allows for the design of numerous ILs with different CO2 adsorption capacities. Given the time and cost limitations associated with experimentally evaluating all potential ILs, in this study, new machine learning-based models have been developed for the prediction of CO2 solubility in different ILs. An extensive dataset comprising 16,480 experimental data points of CO2 solubility in 296 ILs, consisting of 103 different cation and 78 different anion structures, was utilized for this purpose. Quantitative Structure-Property Relationship (QSPR) models were developed using linear and nonlinear methods based on this large dataset. To consider the effect of cation and anion structures on CO2 solubility, basic descriptors were calculated, including zero-dimensional, one-dimensional, and fingerprint descriptors (a category of two-dimensional descriptors). Subsequently, the most relevant variables were identified through the StepWise Regression (SWR), resulting in the selection of 18 categories of cationic and anionic descriptors, in addition to temperature and pressure, as inputs for the nonlinear machine learning (ML) models such as MultiLayer Perceptron (MLP), Radial Basis Function (RBF), Random Forest (RF), and Least-Squares Boosting (LSBoost). Internal and external validation of the models indicated that the LSBoost model displayed the highest accuracy in predicting CO2 solubility and demonstrated superior capability in modeling complex data. R2 and MSE values for this model were 0.9962 and 0.0070 for the training set and 0.9243 and 0.1277 for the test set, respectively. Furthermore, comparisons between the LSBoost model and the available models in the literature demonstrated that the LSBoost model surpasses the other models in performance, proving to be reliable for predicting CO2 solubility in new ILs, thereby aiding in the design and selection of ILs for CO2 capture.
