Quite often one meets people with excellent knowledge and experience in analytical HPLC, and while much theory and underlying equations can be used on a preparative scale, a lot may differ. The main difference is the scope of the separation. In analytical chromatography you are most certainly looking for information in some way, either identification or quantification (or both) of one or several peaks in one chromatogram. However, in preparative chromatography the isolation of a single substance at a certain purity is usually the main goal. This difference has a large impact on how to develop the chromatographic method in order to optimize factors such as yield, purity and loadability.
The strategy outlined below will give you an insight in how a preparative HPLC separation can be developed.
A general methodology can be summarized in the following steps:
Define purification goals. Most often, a certain productivity and yield at a defined purity is the overall goal. Referring to the picture, these parameters can be seen as corners in a triangle and it becomes evident that one of these parameters must grow at the cost of another for a given separation. Performing a separation at these three corners at the same time simply is not possible.
Early on, it is important to try to establish if the separation should be performed in normal phase (NP) or reversed phase (RP) mode. Critical factors to look upon are solubility in candidate mobile phases and loading capacity. Generally, small polar molecules are separated in normal phase mode which is the main reason for the fact that the pore size in normal phase stationary phases is smaller (typically 60 Å, compared to 100-120 Å in reversed phase). Smaller pores gives a higher specific surface area which, as long as it is accessible, increases loading capacity. For larger molecules, peptides and proteins reversed phase is the most common mode for separation. In normal phase preparative chromatography bare silica is the most commonly used stationary phase. However, several derivatized phases are available, as shown in the figure.
Choose a few analytical columns packed with preparative material for your method development. For normal phase, SIL or DIOL is usually the best choice while C18, C8, C4 or Phenyl is the best choice for reversed phase applications. In this initial step, it is very important to choose analytical columns from a vendor that can provide the stationary phase in large quantities, defined by the scale of the project. Switching vendor during a scale-up process is not recommended since stationary phase properties can vary considerably.
Carry out method development by screening selected columns under different mobile phase conditions (including buffer systems, pH and organic modifiers). At first, focus on selectivity under analytical conditions. Selectivity is the most important parameter for any preparative separation, and production rate increases rapidly with increased selectivity! Depending on the interaction with the stationary phase it is important to choose isocratic or gradient conditions at this step. Generally, the need for gradient conditions increases with the size of the molecule, as the interaction behavior changes from partitioning to adsorption/desorption.
Continue with overloading experiments of the most promising stationary phases and evaluate the results by collecting and analyzing fractions analytically.
Calculate productivity (defined as mpure product/(mstationary phase*tcycle)) using yield from collected data. With the equations given above final column size in order to cope with the overall productivity goals can be calculated and evaluated.
Tune in by optimizing important scale-up parameters such as flow rate, particle size, bed length and loading amount. The optimal running point is usually at a point where flow rate, particle size and bed length are balanced so that the back pressure is at the highest, acceptable level. This will give the highest productivity, and lowest cost.
Ajax content requested but not retrieved...
Something went wrong. This is not a picture.