Development of the CO<sub>2</sub> Adsorption Model on Porous Adsorbent Materials Using Machine Learning Algorithms
Hossein Mashhadimoslem, Mohammad Ali Abdol, Kourosh Zanganeh, Ahmed Shafeen, Ali A. AlHammadi, Milad Kamkar, Ali Elkamel
Abstract
Porous adsorbents have common characteristics, such as high porosity and a large specific surface area. These characteristics, attributed to the internal structure of the material, significantly affect their adsorption performance. In this research study, we created a data set and collected data points from porous adsorbents (2789) from 21 published papers, including carbon-based, porous polymers, metal–organic frameworks (MOFs), and zeolites, to understand their characteristics for CO 2 adsorption. Different machine learning (ML) algorithms, such as NN, MLP-GWO, XGBoost, RF, DT, and SVM, have been applied to display the CO 2 adsorption performance as a function of characteristics and adsorption isotherm parameters. XGBoost was selected as the best ML algorithm due to its highest accuracy ( R 2 = 0.9980; MSE = 0.0001). The predicted results revealed that the adsorption pressure parameter is the most effective in all of the mentioned porous adsorbents. With regard to materials type, while carbon-based materials require higher pressures for a more effective CO 2 adsorption, MOFs exhibit a higher potential for adsorbing CO 2 under lower pressure conditions. The study also revealed that carbon-based adsorbents, zeolites, and porous polymers with smaller pore diameters demonstrate a high level of CO 2 uptake. In contrast, the adsorption performance of MOFs does not show a consistent trend with respect to pore sizes. Also, in all adsorbents, the effect of a pore size smaller than 1 nm on more CO 2 adsorption was evident.