Computer aided development and optimization of chromatographic separations

The thesis deals with the chromatographic separation of proteins by ion exchange. Ion exchange chromatography has for many years been used for separation of proteins. The development of the chromatographic separation, however, is often done by systematic experiments and statistical methods. During the development phase modelling can be used as a valuable tool, both for process optimization and to get a better understanding of the process. Modelling has previously been used extensively for chromatography of smaller molecules. Proteins are, however, large molecules, which must be accounted for in the model. For modelling to be a valuable tool, close interaction between interpretation of laboratory results and modelling is required. Therefore emphasis has been put on comparing experimental results with results from modelling and discussion of the results. Whey proteins have been used as model system. The used whey proteins are Bovine Serum Albumin (BSA), α-lactalbumin, and β-lactoglobulin A and B. A few experiments are using aprotinin and the amino acid tyrosine. All experiments are performed using the strong anion exchangers: Source 30 Q, Q Sepharose XL, Ceramic Q-HyperD F and Fractogel EMD TMAE 650(S). The Steric Mass Action formalism (SMA-formalism) is used to describe the equilibrium. This isotherm is capable of taking the influence of salt into account. Two modifications of the SMA-formalism has been suggested: One modification which is able to handle negative protein concentrations and a method for solving small system of equations, e.g. the Craig model. It is shown that both axial dispersion and mass transfer resistance needs to be taken into
account to describe experimental results. A model has been suggested, which takes mass transfer resistance in the film layer around the particles, in the pores, and surface diffusion inside the particles into account. Both a model with diffusion inside the particles, and a linear driving force approximation for the particle side mass transfer has been derived. To avoid a double boundary value problem adding the axial dispersion with the mass transfer coefficient for the film layer has been suggested. Alternatively a double boundary value problem can be avoided by using open boundary condition at the column exit, this model has also been used. The models have been solved using orthogonal collocation on finite elements (OCFE), which allows a fast solution of the equations. The suggested method has been compared with the solution using global orthogonal collocation. The mass transfer coefficient for the film layer has been determined from a correlation. Correlations for the film layer mass transfer coefficients in packed columns at low Reynolds numbers are compared. Equilibrium and mass transfer coefficient inside the particles for a number of proteins and anion exchangers have been determined. It is also demonstrated how the equilibrium
parameters can be used to determine the gradient elution volume. A detailed example using theory and experimental results to determine equilibrium and mass transfer parameters are given for the system α-Lactalbumin and Q Sepharose XL. The example also compares modelled and experimental data.
Methods to determine the non-linear part of the isotherm has been evaluated. These are batch experiments and breakthrough experiments. The batch experiments turned out to be sensitive to impurities in the protein solutions. Additionally the resin gradually lost its capacity when reusing the resin from one batch experiments to the following. Breakthrough experiments are evaluated, too, and have been used to determine the isotherms for BSA and β-Lactoglobulin A and B with Source 30Q. Both successful and less successful experiments are presented, since this forms the basis of discussion for interpretation of the results. Due to the large amount of experimental data, a prototype of a program has been made. The program fits the experimental chromatograms, and stores these in a number of databases. Hereafter desired data can be retrieved from the databases and parameters can be fitted from these. The program is also able to simulate experiments and compare modelled and experimental data. The last part of the thesis deals with the separation of β-lactoglobulin A and B by Simulated Moving Bed chromatography (SMB-chromatography). SMB-chromatography is a continuous countercurrent process. The anion exchanger used is Source 30Q. The first part of the thesis mainly describes modelling and experiments in the linear part of the isotherm. In a SMB-plant protein concentration is normally not in the linear part of the isotherm and it is necessary to take the non-linear part of the isotherm into account. In the SMB-chapter the simultaneous use of theory and experimental work is emphasised, too. The equilibrium parameters for β-Lactoglobulin A and B are determined from pulse experiments and from breakthrough experiments on a single column. This forms the basis for a simulation of a gradient experiment in the SMB-process. An excellent agreement between simulated and experimental results from a gradient experiment has been found. A difficult method to determine the area of complete separation for components with nonconstant selectivity has previously been suggested. An alternative and easier method to calculate this is suggested.