Abstract
The primary objective of this thesis was to examine the major factors affecting the sorptive separation of simple water soluble organics, eg alcohols, aldehydes, ketones, carboxylic acids, with hydrophobic adsorbents. The model system considered is the separation of aqueous ethanol with an adsorbent formed by pelletizing micron sized hydrophobic silicalite crystals. The structural characteristics of the bi-dispersed silicalite pellets are examined in Chapter 2. The size distribution of the crystals was measured by various techniques and fitted a log-normal distribution with a median size of 3.78 um. The macropore size distribution of the pellets was determined by mercury porosimetry and showed a unimodal distribution with a mean macropore diameter of 0.6 um. The pellets had a macropore volume of 0.202 cm[3]/g, a macropore voidage of 0.26 and a density of 1.285 g/cm[3] . The sorption equilibria for the system {ethanol-water}/silicalite crystals is considered in Chapter 3. The basic difference between vapour and liquid phase sorption and the fundamental constraint caused by the inaccessibility of the adsorbed phase to direct measurement are considered first. This is followed by the development of a novel technique which for the first time enables the direct measurement of the individual adsorption isotherms for microporous solids. Comparison of the measured data with the predictions based on the available theories of adsorption from solution show considerable discrepancy. This is traced to the inherent assumption of no volume change of mixing common to all the proposed theories. For the system {ethanol-water}/silicalite the adsorbed phase exhibits as much as 30% volume change of mixing. The packing density and the total number of molecules adsorbed is therefore a very strong function of composition. Chapter 4 is devoted to a rigorous thermodynamic analysis of the adsorption equilibria for the system {ethanol-waterl/silicalite. The relevant thermodynamic relationships are derived first without any a priori assumption other than the existence of a distinct adsorbed phase. The major significance of the analysis presented is that for the very first time it is based on measured rather than predicted adsorbed phase composition. For the {ethanol-water }/silicalite system, the behaviour of the adsorbed phase is shown to be radically different from that of the bulk solution. The bulk liquid exhibits strong positive deviation from Raoult's law indicating that water-water and ethanol-ethanol molecular interactions dominate over that between unlike species. The adsorbed phase activity coefficients, however, show strong negative deviation from Raoult's law; indicating that ethanol-water interactions overshadow that between the like molecules. The major practical consequence of this observation is that surface modification of the hydrophobic crystals is unlikely to alter the selectivity significantly. The analysis presented also enables a direct thermodynamic consistency test of the adsorption data which has not been previously reported. The models proposed to describe intraparticle mass transfer are critically reviewed in Chapter 5 and the macropore-micropore diffusion model is identified as the most realistic for silicalite pellets. The intracrystalline diffusion rates were measured directly and their analysis reveals fundamental differences between uptake of pure ethanol and pure water consistent with the observed equilibrium behaviour. This information is used to determine the controlling mechanism for mass transfer between a flowing fluid and silicalite pellets. The time constants for external film mass transfer, macropore diffusion and micropore diffusion are estimated at 2.72, 82.32 and 0.07 seconds respectively. The controlling mechanism is therefore macropore diffusion with intracrystalline diffusion playing an insignificant role. This a direct consequence of the small crystal size and the relatively rapid intracrystalline diffusivity of water and ethanol in silicalite. The above information is pulled together in a mathematical model to describe the dynamics of adsorption of ethanol-water mixtures onto a fixed-bed of silicalite pellets. The model allows for axial dispersion, external film mass transfer and describes the intraparticle mass transfer in terms of a macropore diffusion model. The model partial differential equations are solved by the orthogonal collocation technique. A detailed sensitivity analysis is conducted which confirms axial dispersion and external film mass transfer coefficients can be confidently predicted from the available literature correlations. The isotherm is measured independently which leaves only the effective macropore diffusivity to be obtained by matching with experimental breakthrough curve. Such curves were obtained on a carefully designed small pilot adsorption unit with considerable attention to the distortions caused by the entrance and exit effects and the sampling procedures. The effective macropore diffusivity recovered is 4.0x10[-6] cm[2]/s which suggests a tortuosity of 3.2 for the silicalite pellets. This model and the parameter values determined independently provides a valuable tool for future scale-up and design studies.