Predictive Modeling of CO₂ Adsorption in Metal–Organic Frameworks Using Hybrid Machine Learning Approaches

Carbon dioxide (CO2) emissions from industrial and energy sources lead to global warming and drive climate change. To address the critical challenge of rising atmospheric CO2 levels, strategies such as CO2 capture are essential for limiting emissions and mitigating environmental impact. Among various materials, metal–organic frameworks (MOFs) have emerged as highly promising and effective adsorbents for CO2 capture, offering high selectivity and capacity. This study evaluates the performance of various machine learning (ML) models, including artificial neural network-particle swarm optimization (ANN-PSO), coupled simulated annealing-least squares support vector machine (CSA-LSSVM), and adaptive neuro-fuzzy inference system (ANFIS), in predicting the CO2 capture capacity of MOFs. Additionally, gene expression programming (GEP) is utilized to find a mathematical correlation between CO2 capture capacity and key operating variables such as pressure and surface area. The model development considered the input variables: temperature, pressure, surface area, pore volume, and enthalpy. Performance metrics such as mean square error (MSE) and coefficient of determination (R2) are calculated for the training and testing phases to assess the model accuracy and reliability. Among the evaluated models, CSA-LSSVM exhibits the best performance, achieving R2 values of 0.971 and 0.915 and MSE values of 0.0025 and 0.0034 for the training and testing phases, respectively. Sensitivity analysis conducted on the CSA-LSSVM model reveals that pressure significantly influences CO2 capture capacity, followed by surface area. The results demonstrate the potential of ML models, particularly CSA-LSSVM, as powerful tools for predicting and optimizing the performance of MOFs in CO2 capture, considering cost, energy efficiency, and environmental sustainability.