Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.702834
Title: On the prediction of partition coefficients using the statistical associating fluid theory underpinned by quantum mechanical calculations
Author: Hassan, Abdihakim
ISNI:       0000 0004 6059 3004
Awarding Body: Imperial College London
Current Institution: Imperial College London
Date of Award: 2016
Availability of Full Text:
Access through EThOS:
Full text unavailable from EThOS. Please try the link below.
Access through Institution:
Abstract:
The thermodynamic modelling of phase equilibrium is of central importance in chemical engineering applications. The design, operation and develop- ment of new chemical processes is based to a large extent on the knowledge of the equilibrium that occurs between co-existing fluid phases. Where re- liable experimental data at required process conditions is unavailable, an understanding of the molecular description of condensed phase matter is key to predicting the thermodynamic properties of these fluid systems. To this end, numerous models and theories have been developed that seek to link microscopic intermolecular interactions with bulk macroscopic thermo- dynamic properties. In this thesis, two such constructs for the prediction of phase equilibrium are considered. The empirical linear solvation energy relationship (LSER) that relates specific/unspecific intermolecular interac- tions to infinite dilution solute properties, and equations of state (EoS) for the prediciton of vapour-liquid and liquid-liquid equilibrium. The LSER model utilises hydrogen bond acceptor/donor parameters (A and B) alongside polarisability (S), volume (V) and molar refraction (E) param- eters to describe various solute properties. In this study, the prediciton of solute infinite dilution partititon coefficient is of particular interest. While the V and E parameters can be obtained from molecular structure calcula- tions that account for the number of atoms and bonds in a molecule, the re- maining LSER parameters are usually derived from chromatographic experiments. However, successive studies have successfully correlated and pre- dicted the hydrogen bonding parameters from quantum mechanical (QM) calculated molecular properties, enabling the rapid calculation of infinite dilution solute properties in the so-called QM/LSER approach. In this the- sis, two independent linear regression relationships that relate theoretically calculated hydrogen bond stabilisation energies at a donor and/or acceptor site(s) to experimental hydrogen bonding ability of a solute molecule have been determined. Once obtained, the solute hydrogen bonding parameters are used in conjunction with dispersion and volume parameters in the LSER to obtain solute partition coefficients. Using this approach ,the octanol/wa- ter partition coefficients of various molecules have been estimated, of this, the absolute average error of a sub-set of straight chained, mono-functional solute molecules has been determined to be 23.04% when compared to ex- perimental data. The second approach to modeling condensed phased matter is based on the statistical associating fluid theory (SAFT), a molecular-based equation of state with a foundation in statistical mechanics. Here, a recently devel- oped group-contribution version i.e., SAFT-1 is considered. The SAFT-1 EoS has been successfully applied in the prediction of the octanol/water patition coefficients of a range of solute molecules that include n-alkane, n-alkene, 2- ketone and n-amine molecules. Where the average absolute error of SAFT- 1 predicitons when compared to experimental data is found to be 13.20%. However, as with other EoS, SAFT-1 is dependent on experimental data re- quired to parameterise the various groups that make up the fluid/fluid mix- ture under investigation. The aim of this work is to increase the predictive ability of SAFT-1 by reducing dependence on experimental data, whereby in- stead of equilibrium data, solute partition coefficients estimated using the QM/LSER method are used to parameterise the relevant molecular groups. In the final part of the thesis, the proposed hypothesis of combining the QM/LSER and SAFT-1 methods is tested with the aim of predicting the phase behaviour of binary mixtures. The method relies on the calculation of partition coefficients using QM and LSER, the calculated partition coef- ficients are then used to parameterise the unlike group-group interactions required for the prediction of binary mixture behaviour in SAFT-1. This methodology has been validated using the n-aldehyde and 2-ketone chemi- cal families, where using QM/LSER to parameterise SAFT-1 has been found to achieve results that are comparative to the classical empirical approach of parameterising the SAFT-1 EoS when predicting binary phase behaviour. The unlike group interaction parameters for the SAFT-1 EoS have been suc- cessfully parameterised using partition coefficient data estimated from the- oretically calculated quantum mechanical molecular properties. However, the solutes considered in this study are limited to linear mono-functional molecules. The reason for this limitation is two fold. Firstly, predicting hydrogen bond parameters of multi-functional molecules is unreliable mainly as a consequence of polarisation of H-bond sites due to the proximity of functional groups. Therefore a better understanding of how polarisation affects hydrogen bonding is required. Secondly, within SAFT-1 the major- ity of available groups are for modeling linear mono-functional molecules. However there is continuing work to model both branched and multifunc- tional molecules. Once both of these concerns are effectively dealt with, the proposed methodology can be used to characterize a wider range of SAFT- 1 groups and predict thermodynamic behaviour of molecules based on QM molecular calculations.
Supervisor: Adjiman, Claire ; Galindo, Amparo ; Hunt, Patricia Sponsor: Engineering and Physical Sciences Research Council
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.702834  DOI: Not available
Share: