Diffusion Coefficients in Dense and Supercritical Fluids
authors Magalhães, Ana L.
supervisors Carlos Manuel Santos Silva; Francisco Avelino Da Silva Freitas
thesis type PhD Thesis
author keywords Diffusion coefficient, transport property, modelling, correlation, prediction, dense fluid, supercritical fluid, chromatographic peak broadening
abstract Diffusivities (D12) are fundamental properties both at research and industry levels, but the lack of experimental data and the non-existence of reliable and accurate equations to estimate them in compressed and condensed phases constitute important shortcomings. The main objectives of this work comprise: i) the compilation of a large database of D12 values in gas, liquid and supercritical systems; ii) the development and validation of new models for tracer diffusivities, applicable over wide ranges of temperature and density, for systems containing very distinct components in term of polarity, size and symmetry; iii) the installation and test of an experimental set-up to measure diffusion coefficients in liquids and supercritical fluids. Concerning modelling, a new accurate expression for tracer diffusion coefficients of hard sphere fluid was developed and validated using molecular dynamic data (average absolute relative deviation, AARD = 4.44%). The binary diffusion coefficients of real systems were also studied. An extensive database of diffusivities in gas and dense solvents was compiled (622 binary systems performing 9407 data points and comprehending 358 molecules) and used to validate the new models developed in this thesis. A set of new models were proposed for tracer diffusivities using different approaches: i) two molecularly-based models with one system-specific parameter that are applicable to gas, liquid, and supercritical systems, where the nature of solvent is limited to non-polar or weakly polar (global AARDs in the range 4.26-4.40%); ii) two molecularly-based models with two parameters, applicable in all physical states, for any solutes diluted in any type of solvent (non-polar, weakly-polar, and polar). Both models provide global errors between 2.74% and 3.65%; iii) one correlation with one parameter devoted to D 12 coefficients in supercritical carbon dioxide (SC-CO2) and liquid water (AARD = 3.56%); iv) nine empirical and semi-empirical correlations involving two parameters, dependent on temperature and/or solvent density and/or solvent viscosity. These models are very simple and provide accurate results (AARDs between 2.78% and 4.44%) in liquid and supercritical systems; and v) two predictive equations for diffusivities of solutes in SC-CO2 where the global deviations for both are inferior to 6.80%. In the whole, it may be emphasized that the new models cover the large variety of systems and molecules generally found. The results achieved are consistently better than those obtained by well known models and approaches taken from the literature. In the case of the 1- and 2-parameter correlations, it has been shown that such parameters can be fitted to a very small set of data, and subsequently used to predict D 12 values far from the original set of points. A new experimental set-up to measure binary diffusion coefficients by chromatographic techniques was designed and tested. The equipment, experimental procedure and analytical calculations to obtain the D 12 values by the chromatographic peak broadening technique were assessed by measuring diffusivities of toluene and acetone in SC-CO2. Then, the diffusivities of eucalyptol in SC-CO2 were determined in the ranges 202 – 252 bar and 313.15 – 333.15 K. The experimental data were analysed using D 12 predictive and correlation models.
year published 2013
link http://hdl.handle.net/10773/12215