Towards estimation and mechanism of CO2 adsorption on zeolite adsorbents using molecular simulations and machine learning

Abstract

The earth's average temperature is being elevated owing to burning of fossil fuels, explicitly releasing enormous amounts of CO2 into the atmosphere. Therefore, effective collection strategies are needed to minimize the concentration of CO2. The current work presents use of molecular simulations to examine CO2 adsorption on zeolites from the zeolite database. Among the 218 zeolites, the Linde Type A zeolite (LTA) stands out as the best for CO2 adsorption with a percentage weight of 69.88% at 298 K and 1000 kPa. This calculated value is in agreement with a recent experimental study that reported 63.65% at 298 K and 0.1 MPa, indicating that our model could be reliable in predicting the CO2 loading capacity. The structural features of the zeolites were also calculated and their significance on CO2 adsorption, evaluated using a machine learning algorithm. Gravimetric surface area seemed to be the most significant structural features. Hence, its feature engineering may expedite adsorption of CO2. Quantum chemical calculations were further used to provide insights into the sorption mechanisms, whereby it is thermodynamically favorable for CO2 to adsorb through carbon bonding as opposed to dissociative adsorption or hydrogen bonding. Finally, the effect of doping the zeolites with selected cations on the loading capacity was investigated, whereby doping LTA with lithium significantly improved the adsorption of CO2.