Separation of an adsorption azeotropic binary gas mixture by PSA: Propane and carbon dioxide on H-Mordenite

Abstract

The separation of an adsorption azeotropic binary gas mixture (CO₂ and C₃H₈ on H-mordenite) by pressure swing adsorption (PSA) was studied via simulation using the dynamic adsorption process simulator (DAPS), with the perfect negative dual process Langmuir (PN-DPL) model describing the mixed gas adsorption equilibria. Numerous PSA simulations, carried out to the periodic state using a 4-bed 4-step Skarstrom cycle at five feed flow rates (6 to 10 SLPM) and nine feed compositions (20 and 80 vol% C₃H₈), showed the following trends: Adsorption azeotropy did present a barrier to separation of this binary gas mixture by PSA, just like in distillation. It occurred at one particular feed composition of ~ 30 vol% C₃H₈, where no separation occurred. However, the corresponding selectivity s₁₂ was equal to 1.75 throughout the bed, not s₁₂ = 1.0, as expected for azeotropic behavior. This feed composition was also outside the pressure dependent locus of adsorption azeotropic compositions (93.0 to 43.2 vol% C₃H₈) where s₁₂ = 1.0. When the feed composition was different than 30 vol% C₃H₈, two slightly enriched LPs and HPs were easily produced relative to the feed composition; however, the LP and HP compositions were restricted to both being either above or below the no separation feed composition of 30 vol% C₃H₈, just like in distillation. An enriched LP could also be achieved by either component, but the enrichment of the other component in the HP was limited by the no separation feed composition of 30 vol% C₃H₈, also just like in distillation. The dominant gas species in the LP or HP transitioned from one component to the other component as the feed composition transitioned through 30 vol% C₃H₈ where no separation occurred, thereby exhibiting selectivity reversal; but this selectivity reversal surprisingly occurred outside the locus of adsorption azeotropic compositions. It was also possible for one of the components to be more selective during adsorption cycle steps and for the other component to be more selective during desorption cycle steps; but this selectivity reversal did not seem to enhance the separation in any way relative to the other feed compositions. Unexpectedly, the adsorption azeotropic point (s₁₂ = 1.0) was able to move freely through the bed during a periodic state PSA cycle without hindrance and without presenting a barrier to separation. Based on the culmination of these findings, a new definition of adsorption azeotropy for a PSA process is proposed, along with a new definition of selectivity. This new definition of adsorption azeotropy is based on the discovery of the non-azeotropic non-zeotropic condition of no separation coined an achorismic point in the feed composition that does not necessarily fall within the locus of adsorption azeotropic compositions. The corresponding selectivity \(\:{s}_{12}^{P}\) for a PSA process operating at this achorismic point is defined as \(\:{s}_{12}^{P}=\frac{{y}_{2}^{*}\left(d{n}_{1}^{*}/dP\right)}{{y}_{1}^{*}\left(d{n}_{2}^{*}/dP\right)}=1.0\). In this study, the corresponding \(\:{y}_{1}^{*}\) was determined serendipitously during simulation with DAPS (\(\:{y}_{C3H8}^{*}\) = 0.300) and analytically from the PN-DPL model using chord (0.308) and differential (0.302) approaches, thereby confirming the existence of this achorismic composition. Achorismic behavior shifts to adsorption azeotropic behavior only in the special (unrealistic) limiting case when the low pressure in the PSA process is set equal to zero.