(325g) Optimization of the Gradient Profiles for Multi-Component Separation in Reversed Phase Column
Solvent gradient chromatography, i.e. chromatographic processes using varying strength of the solvent, is a popular tool in analytical bioseparation especially when the separating components have widely different elution times. In preparative chromatography as well, solvent gradient is being widely used to improve the yield or the productivity or both. As the adsorption behaviour of the solutes at any part of the column can be controlled by varying the solvent strength at that location, one can judiciously design the gradient profile to achieve better separation in lesser time. In industries, such profiles are developed mostly based on experience and hand optimization. In our laboratory, a step by step approach was taken to find the optimized gradient profile through both experimental and theoretical studies. The goal was to find out the optimum gradient for separating a target peptide from its heavier and lighter impurities, in a reversed-phase column. First the adsorption and transport behaviour of the peptide mixture was characterized based on experimental results; subsequently, a mathematical model (using multi-component competitive bi-langmuir isotherm and lumped kinetic equations) for the system was developed. Then, using a modified genetic algorithm, a multiobjective optimization study was carried out for maximizing the productivity and the yield of the target peptide in a batch system, by varying the gradient profiles. The product purity was taken as a constraint and three different schemes (linear, non-linear and multi-linear) for varying the gradient profiles were considered. The optimized gradients were then applied to the lab system and very close separation performance (as predicted) could be obtained. The optimization study brought out important insights about the role of gradients in overloaded conditions. A steady state recycling process of the lesser pure collects of the target peptide was further investigated, keeping the objectives, constraint and decision variables the same, and its performance compared with the batch process. All the optimization results are explained using the relationships between the design variables and the objectives.