Integration of Material and Process Design for Kinetic Adsorption Separation

Abstract

Kinetic adsorption separation exploits differences in diffusion behavior of different adsorbates in materials and enables selection of the faster over the slower components despite possibly unfavorable equilibrium loading amounts. The aims of this article are to explore the material property space that leads to effective kinetic adsorption separation and to test the predictive value of the current ideal kinetic selectivity metric. This metric is the product of equilibrium selectivity and square root of their diffusivity ratio. Throughout our analysis, we use the Maxwell–Stefan diffusion formulation which has been shown to capture transient crystal diffusion characteristics with greater accuracy than a Fickian description. The study is conducted in two parts. Optimized purity–recovery pareto curves are generated for three-step PSA cycles for four Langmuir isotherms that have no equilibrium selectivity (=1), with varying diffusivity ratio and crystal radius. This highlighted the role of crystal radius and isotherm shape in bed performance objectives and not just the diffusivity ratio. A surrogate-assisted illumination (SAIL) algorithm based on single crystal uptake simulations was used to cover a wider range of equilibrium selectivities in a fraction of the time taken for full PSA simulations. We conclude that for similar diffusivities and equilibrium selectivity, steeper isotherms show better kinetic separation performance, in agreement with the existing literature. In addition, the size of the crystal also strongly impacts optimal performance, whereas the diffusivity ratio was found to have a weaker than expected impact. Therefore, using the ideal kinetic selectivity metric mentioned earlier could, in certain cases, be inadequate for comparison of materials for kinetic separation of a target gas mixture. In addition, the combination of SAIL and the single crystal uptake model was successfully able to replicate observations made using full PSA cycle optimizations showing how a multifidelity modeling approach can be used to screen materials for kinetic adsorption separation design.