A thermodynamic study of nitrogen and carbon dioxide adsorption on alkali-exchanged chabazite zeolites

Abstract

Zeolites have long been considered as excellent candidate materials for gas separation and purification. Most of the worldwide industrial adsorption applications of molecular sieve zeolites utilize zeolites X, A and Y. Chabazite zeolite however, has pore dimensions between those for zeolites X, A and Y and hence has promising separation features for specific application but has not been utilized in any major industrial separation or purification applications to date. Some early work on alkali-exchanged chabazites revealed a diminished porosity and surface area due to pore blockage. This initial work was based on limited experimental data and preliminary analytical techniques and analysis. As a result, this initial pessimistic assessment of the merits of chabazite zeolite has led to a loss of interest in this material for adsorption applications. In this study, a pure phase chabazite was synthesized and ion-exchanged to produce potassium chabazite (KCHA), sodium chabazite (NaCHA) and lithium chabazite (LiCHA). Attempts were successful to produce a fully exchanged potassium chabazite, which is a unique product that has not been reported in the literature to date. The adsorption of nitrogen and carbon dioxide was studied since these gases are component of flue gas and the application to which chabazites were to be directed was CO2 capture form flue gas. Adsorption equilibrium isotherms for CO2 and N2 were measured at pressures up to 101.3 kPa and temperatures of 273, 303 and 333 K. More importantly, this work focused on determining fundamental properties of adsorbate-adsorbent interactions which require low pressure measurements hence low pressure isotherms, down to 0.001 kPa, were measured. These data provide valuable information to study the adsorption behavior in the Henry’s law region. The results showed that porosity characterization of KCHA using the conventional approach of nitrogen at 77 K reveals a surface area of only 17.82 m2 g-1 and a diminished pore volume (by density functional calculation) of 0.005 cm3 g-1, compared to 584.4 m2 g-1 and 0.214 cm3 g-1 using carbon dioxide at 273 K, respectively, calculated from the revised Toth model and CO2 isotherm data at 273K. These findings strongly suggested that KCHA is a highly porous adsorbent, in spite of the large size of K+ ions which block access of the N2 molecules to the pore space. It is concluded that traditional methods of characterization are not suitable in the case of pore blockage leading to incorrect interpretation of adsorbent properties. It was initially hypothesized that carbon dioxide molecules enjoy large freedom within the zeolite pores, however, virial plots developed from equilibrium isotherms showed anomalous behavior for carbon dioxide adsorption at 273 K on NaCHA and KCHA at loadings lower than 1.5 gmole kg-1, which correspond to pressure of about 0.15 kPa. This suggested that carbon dioxide molecules were not at true equilibrium condition, but rather were subject to steric hindrance due to partial pore blockage. Adsorbed phase densities calculated from van der Waals constants confirmed the pore blockage phenomenon on KCHA, however, it also reveals a slight blockage against nitrogen molecules on NaCHA. On the synthesized chabazites, carbon dioxide affinities and heats of adsorption were considerably higher than those for nitrogen for which their adsorption equilibria differ significantly due to screening against nitrogen molecules in KCHA and partially in NaCHA. Carbon dioxide entropy measurement revealed a concave trend, demonstrating an appreciable loss of degrees of freedom within increase in loading. Carbon dioxide desorption isotherms showed low pressure hysteresis at 273 K with residuals of 0.37 and 0.57 molecule cavity-1 on NaCHA and KCHA, respectively, at pressures lower than 0.05 kPa. This outcome confirmed the pore blockage occurrence suggesting a low pressure encapsulation. The combined use of the statistical theory of the radial distribution function (rdf) and the theory of the perfect 3D lattice gas to describe the encapsulation process underestimated the number of accommodated molecules compared to the experimental results. The individual implementation of Lennard-Jones and quadrupolar potentials to describe the adsorbate-adsorbate and adsorbate-host interactions, respectively, affected the performance of the models.