(292c) Prediction of Liquid-Liquid and Solid-Liquid Equilibria of Isomers

Authors: 
Goetsch, T., TU Dortmund University
van den Bongard, R., TU Dortmund
Köhler, A., TU Dortmund
Enders, S., KIT
Zeiner, T., TU Dortmund University
In chemical industry there are several reactions which produce a linear product and a number of branched side products simultaneously. One example is the hydroesterification of long-chain unsaturated esters of fatty acids where linear and branched diesters are produced. For application only the linear ones are of interest resulting in a demand to separate these isomers. This separation is a challenging task since distillation is very elaborative because of small separation factors and extraction often does not meet the high purity demands. Solvent crystallization is a promising alternative because linear and branched molecules show different solubilities in a solvent. The challenge is the appearance of a superposition of a liquid-liquid equilibrium (LLE) and a solid-liquid equilibrium (SLE) during the separation process. Since this superposition can affect the product properties, it has to be understood in order to build up suitable crystallization processes.

The aim of this contribution is the experimental and theoretical investigation of the mentioned superposition of LLE and SLE for linear and branched molecules. To be certain that the final crystallization product shows the desired properties, one has to avoid a liquid-liquid demixing during the crystallization process. Therefore, it has to be known in advance whether a demixing will take place or not under the defined process conditions. To reduce the experimental effort thermodynamic modelling is commonly applied. In the course of modelling one has to fit pure component parameters to experimental data. For branched molecules, it is often not possible to gain experimental data, because they are barely commercially available in a high purity. Therefore, phase equilibria of systems containing branched molecules can often not be modelled applying the common procedure. To overcome this limitation a model-based experimental design, which allows for the prediction of phase equilibria of branched molecules, was developed. Since in the above mentioned hydroesterification a lot of different isomers are produced that differ only in the position of the side group, it is crucial that our methodology is able to cover all configurations of isomers. For the validation of the developed approach different binary and ternary mixtures of linear and branched alkanes dissolved in an alcohol were investigated.

The basic idea of the developed methodology is to perform only experiments with linear molecules and use the gained results for the prediction of phase equilibria of branched molecules. For this purpose, it is mandatory to use a thermodynamic model that considers the architecture of the branched molecules. The lattice cluster theory (LCT) was developed to consider the architecture of the molecules without any additional adjustable parameters [1]. The LCT, therefore, allows for the prediction of thermodynamic properties of branched molecules based on pure component parameters of linear molecules. To consider the self-association of the alcohol the LCT is combined with the chemical association lattice model (CALM) [2]. The combination of the LCT with a chemical association model was already successfully applied for the calculation of liquid-liquid equilibria of hyperbranched polymers [3] and for the calculation of the superposition of liquid-liquid and solid-liquid equilibria of hyperbranched polymers dissolved in an alcohol [4].

The developed methodology can be divided into three steps. First of all experiments with linear molecules have to be performed. In this contribution LLE measurements of binary systems of a linear alkane and an alcohol were performed. These experimental data were then used to fit the required model parameters. In the framework of the LCT there is only one adjustable parameter for a binary system, namely the interaction energy Î?ε. Additionally, there are two adjustable parameters to consider the self-association of the alcohol. In the second step the fitted parameters are used for the prediction of the LLE of binary systems including a branched alkane and an alcohol. To validate the predicted phase equilibria LLE measurements for these systems were performed. In the third and last step LLE and SLE of ternary systems containing a linear alkane, a branched alkane and an alcohol were predicted based on the binary subsystems. The predicted phase equilibria were again validated by experiments.

Using the developed model-based approach one is able to predict the superposition of LLE and SLE for binary and ternary systems containing branched molecules by only knowing the binary LLE of the corresponding linear molecule. The predicted phase equilibria are in good agreement to the experimental data for all investigated systems.

References

[1] J. Dudowicz, K. F. Freed, W. G. Madden, Macromolecules 23 (1990) 4803 â?? 4819

[2] D. Browarzik, J. Mol. Liq. 146 (2009) 95 - 104

[3] T. Zeiner, C. Browarzik, D. Browarzik, S. Enders, J. Chem. Thermodyn. 43 (2011) 1969 - 1976

[4] T. Goetsch, P. Zimmermann, S. Enders, T. Zeiner, Chem. Eng. Process. 99 (2016) 175 - 182