Applied Thermodynamics and Phase
Equilibria:
Research
Ir. G.H. van Bochove, Dr. Ir. Th. W. de Loos
Contents:
Introduction
Incentives and objectives of the project
Theory
Experimental work
Modelling
Selected publications
Liquid-liquid extraction is the separation of the components of a fluid
mixture by treatment with a solvent in which one or more of the desired components is
preferentially soluble. Accurate phase equilibrium representation is essential in the
computer-aided design and simulation of this important separation technique. However, in
many chemical processes mixed solvent electrolyte solutions play an important role.
Modelling and prediction of phase equilibria of mixed solvent electrolyte systems is still
difficult from a thermodynamic point of view. Although much progress has been made in
modelling vapour-liquid equilibria of mixed-solvent systems, the modelling of
liquid-liquid equilibria of these solutions is still one of the open areas within
thermodynamics.
In this project, a model will be developed for application in
industrial extraction processes involving water + electrolyte + solute + solvent systems,
for example the extraction process of caprolactam. Current models
are usually based on the Debye-Hückel theory, which can be applied only to low
electrolyte concentrations and aqueous systems. This project aims to develop a model that
can be applied to higher salt concentrations and mixed solvents, based on statistical
mechanics. The model has to be able to give an accurate description of the liquid-liquid
equilibria that are of importance for the extraction process of caprolactam. Since
caprolactam is the monomer of Nylon 6,
a high purity is required for the caprolactam. The desired high purity is reached in a
multistep purification process. First step in the purification process is the recovery of
caprolactam from a product stream containing water, caprolactam, ammonium sulfate and
organic and inorganic impurities. Since ammonium sulfate dissociates in water, high
concentrations of electrolytes are present. In order to model the extraction process, an
accurate thermodynamic model is required for the liquid-liquid equilibria present. This
work aims to develop such a model. In addition, the development of this model requires
experiments, where the influence of solvent and electrolyte on the phase equilibria is
investigated. The experimental data will serve to test and to adjust the models.
Some theory (under construction)
Liquid-liquid extraction or solvent extraction is the separation of a dissolved component from its solvent by transfer to a second solvent. The second solvent has to be immiscible with the first solvent and preferably has a higher affinity to the transferred component. Liquid-liquid extraction can purify a component with respect to dissolved components that are not soluble in the second solvent. In the extraction device the solution of solvent 1 with the solute is contacted with solvent 2. The solute distributes between both solvents until liquid-liquid equilibrium is reached. Since solvent 1 and 2 are immiscible, the two phases can be separated and the process may be repeated at different conditions. For example, in the caprolactam recovering process, in the forward extraction step caprolactam is transferred from the aqueous solution with ammonium sulfate to the organic solvent, leaving most water soluble impurities in the aqueous phase. In the backward extraction, caprolactam is transferred from the organic solvent to water, leaving the organic impurities in the organic solvent.
To be able to make reliable predictions of the solute fractions in both solutions leaving the extraction process in a simulation program, an accurate thermodynamic model is required to calculate the liquid-liquid equilibria and the distribution of the solute between the liquid phases. Many chemical processes comprise processes that involve multicomponent electrolyte solutions. Even small amounts of salt may have considerable effects on the properties of these solutions. The development of engineering methods for the prediction of phase equilibria of complex systems containing both electrolytes and nonelectrolytes therefore is an important challenge for engineers. Damage prevention of industrial equipment requires accurate knowledge of the concentration of electrolytes through a process. Recent environmental concerns require precise control of electrolytes in final products and waste streams. Unfortunately and in spite of several decades of research, thermodynamic models for the prediction of thermodynamic properties of mixed solvent electrolyte solutions are scarce, often lead to different predictions and require many empirical adjustable parameters. However, accurate prediction of phase equilibria using thermodynamic models for mixed solvent electrolyte systems is essential to the success of computer-aided design and simulation of separation processes.
Liquid-liquid extraction is only possible when the feed solvent and the extracting solvent show a phase split. Whether a mixture with a certain composition will demix or not and how the solute will distribute between the liquid phases can be described using thermodynamics. In general, the equilibrium conditions for phase equilibria can be derived using the Gibbs energy. A closed system not at equilibrium will always go to a minimum in the total Gibbs energy with respect to all possible changes at given pressure and temperature, according to the second law of thermodynamics. In other words, at the equilibrium state differential variations can occur in the system at constant pressure and temperature without producing any change in the total Gibbs energy.
For a given liquid mixture at fixed temperature and pressure the necessary equilibrium condition is that the Gibbs energy for the system is minimum. If the mixture achieves the lowest Gibbs energy by splitting into two or maybe three liquid phases, this is what actually happens. Thus, mixtures of two solvents will only mix when the mixed state has the lowest Gibbs energy.
will be continued....
Experiments are carried out to measure liquid-liquid equilibrium data of the ternary systems water + caprolactam + ammonium sulfate and water + caprolactam + organic solvent and quaternary systems of water + caprolactam + organic solvent + ammonium sulfate at 20, 40 and 60°C. Until now, experiments have been carried out with the organic solvents benzene, cyclohexane, 1-heptanol and 2-heptanone. By studying different kinds of solvents, it is tried to investigate the influence of different functional groups on the distribution of caprolactam between the organic and the aqueous phase. A description of the experimental setup has been given by Wijtkamp et al. (1999). In the systems water + 2-heptanone + caprolactam + ammonium sulfate and water + benzene + caprolactam + ammonium sulfate, equilibria of three liquid phases were found. For the system with 2-heptanone the region with three liquid phases was studied more extensively and was used to test the quality of the models to be developed. Currently experiments are carried out on three-liquid phase equilibria for the system water + benzene + caprolactam + ammonium sulfate at 20°C and higher.
Example of a vessel with two sample points to measure liquid-liquid equilibria
At certain compositions, three-liquid phase equilbria were found!
Our first approach in modelling liquid-liquid equilibria of mixed solvent electrolyte systems has been by using the electrolyte Non Random Two Liquid (NRTL) theory. In the first part of this work, attempts were done to get some results using the primitive Mean Spherical Approximation (P-MSA), in combination with the NRTL model. Since this did not produce better results than those obtained with the electrolyte NRTL (despite the fact that the MSA theory has a better physical background), it was decided to pay more attention to the so-called extended electrolyte NRTL model.
The electrolyte NRTL expression for the activity coefficients was modified by taking into account the derivatives to the solvent composition of the physical properties and by a new Brĝnsted-Guggenheim expression. This way a more correct description of the experimental data is obtained. The model has proved to be very succesfull and is able to handle a wide variety of mixed solvent electrolyte systems: Liquid-liquid equilibria of mixed solvent electrolyte systems in literture (especially water + alcohol + salt mixtures) were correlated using the new model. In addition to ternary mixed solvent electrolyte systems, the model has been applied to the quaternary systems that are of importance in the extraction process of caprolactam: water + caprolactam + solvent + ammonium sulfate. The liquid-liquid equilibria of the ternary and quaternary systems involved and the mean activity coefficients of the salt + water systems were used simultaneously to obtain the adjustable parameters. The results were compared to the original electrolyte NRTL of Chen, and it was found that the new model gives a better description of the experimental equilibrium data, in particular for the salt concentration in the organic phase and near the critical point (plait point) of the mixture. [Figure]
Even more challenging than obtaining a good description for liquid-liquid equilibria, is to apply a model both for the description of two-liquid phase equilibria as for three phase equilibria in certain systems using the same binary interaction parameters. The transition from two liquid phases to three liquid phases and again to two liquid phases is a fascinating phenomenon that puts high demands on the consistency and power of the activity coefficient model. Therefore, hardly any literature exists on modelling LLLE with an excess Gibbs model. In this work, the modified extended electrolyte NRTL model was applied to represent equilibrium data of two and three liquid phases in the systems water + benzene + caprolactam + ammonium sulfate, water + 2-heptanone + caprolactam + ammonium sulfate and the system water + benzene + ethanol + ammonium sulfate. These system have in common that at low salt concentrations two liquid phases are present with the caprolactam or ethanol in the bottom phase. At higher salt concentrations three liquid phases can be found and the caprolactam or ethanol is present in the middle phase. At still higher salt concentration two liquid phases ar found again and the caprolactam or ethanol is found in the top phase. The modified extended electrolyte NRTL model has proven to be able to correlate the two- and thee liquid phase equilibria including the progression with from one critical tie-lines (where one of the liquid phases splits into two identical phases) to the other critical tie-line (where two of the phases become equal again).
Current modelling work focusses on using models based on the Mean Spherical Approximation theory, in order to compare the predicting and correlating capability of a "engineering" model (with many adjustable parameters) with a more fundamental, realistic model, where all the parameters have some physical meaning.
Bochove, G.H. van, G. J.P. Krooshof and Th. W. de Loos
Two- and Three-Liquid Phase Equilibria of the Quaternary System Water +
2-Heptanone + Caprolactam + Ammonium Sulfate: Experiments and Modelling.
Paper presented at PPEPPD 2001, Kurashiki, Japan, accepted for publication in Fluid Phase
Equilibria
Bochove, G.H. van, G. J.P. Krooshof and Th. W. de Loos
Modelling of liquid-liquid equilibria of mixed solvent electrolyte
solutions using an extended electrolyte NRTL,
Paper presented at AIChE's 1999 Annual Meeting in Dallas (TX), Fluid Phase
Equilibria, 171, 1-2 (2000) 45-58
Wijtkamp, M., G.H. van Bochove, Th.W. de Loos, and S.H. Niemann
Measurements of Liquid-Liquid Equilibria of Water + Caprolactam +
Electrolyte + Organic Solvent Systems.
Paper presented at PPEPPD 1998,
Noordwijkerhout, Netherlands, Fluid Phase Equilibria 158-160 (1999) 939-947
Bochove, G.H. van
Measurements and Modelling of Liquid-Liquid Equilibria of Caprolactam +
Water + Solvent + Ammonium Sulfate Systems, MSc-thesis TU-Delft (1998)
This project is supported by: 
