Adsorption of CO₂ and SO₂ on multi-walled carbon nanotubes: experimental data and modeling using artificial neural network

Abstract

Multi-walled carbon nanotubes (MWCNTs) containing hydroxylgroups (OH-MWCNT) were modified by functionalization with 3-[2-(2-aminoethylamino)ethylamino]propyl trimethoxysilane (TRI). Adsorption isotherms of pure CO₂ and SO₂ on the pristine MWCNT, OH-MWCNT, and amine functionalized MWCNT (amine-MWCNT) were measured at two temperatures of 313.2 K and 323.2 K and pressures up to 2.1 bar by a static volumetric method. Capacities of all three types of adsorbents for CO₂ adsorption are greater than those of CO₂. The performance of amine-MWCNT in adsorpting CO₂ is higher than the other two adsorbents. The average saturated capacity of amine-MWCNT for adsorption of pure CO₂ at 313.2 K are about 38.6% and 20.8% higher than OH-MWCNT and pristine-MWCNT, respectively. Corresponding values for adsorption of pure CO₂ are about 51.3% and 89.65%. Also, the equilibrium adsorption capacity of pristine MWCNT and amine-MWCNT for mixtures for CO₂, nitrogen, and water vapor at 299.2 K was obtained. The equilibrium adsorption of CO₂ increases as the water content increases in the presence of diluting gas (nitrogen). Freundlich and Langmuir equations were fitted on experimental adsorption isotherms. The Freundlich equation predicts experimental data better than the Langmuir equation. A multi-layer perceptron artificial neural network (ANN) model has been also proposed for predicting adsorption experimental data. The average and maximum difference between experimental data and values predicted by ANN model are about 3% and 24%, respectively.