| Ion-exchange chromatography is one of the most useful techniques for recovery and purification of biomolecules. In preparative applications, chromatographic performance is affected by the complex interplay of many factors. To design an efficient and reliable process requires a fundamental and quantitative understanding of the underlying mechanisms. The objective of this thesis is to correlate systematically adsorption isotherm data of proteins for single-component and binary systems, and to apply this understanding of adsorption behavior, especially for nonlinear adsorption, to the study of column performance, where the adsorption behavior also has an impact on transport properties.; The isocratic retention factor, k', was measured on two sets of charge-variant adsorbents in linear chromatography. No clear trend in k' was observed when the charge density is higher than a threshold value, suggesting that adsorbent charge density is a relatively unlikely factor to use in manipulating protein retention.; Single-component adsorption isotherms of three basic proteins were measured on cation exchangers under various solution conditions and used as the basis for developing a predictive approach for correlating adsorption behavior using a mechanistically based isotherm model. The model explicitly considers the contributions of protein-surface and protein-protein interactions, and decoupling them allows each to be correlated with different experimental measurements. Specifically, protein-surface interactions were related to k' values, while protein-protein interactions were analyzed based on high-coverage isotherm data on an arbitrary stationary phase. Data analysis within this framework revealed a high level of consistency. The model was also used to facilitate prediction of adsorption isotherms on other ion-exchange media using isotherms on one adsorbent.; For binary systems, competitive adsorption can differ substantially for different protein pairs for one pair under different solution conditions. Adsorption patterns largely followed the same trends as the intrinsic protein-surface interactions. Several exceptions observed were related to the contribution of protein-protein interactions, especially protein-protein cross-interactions. A model based on colloidal energetics was proposed to predict binary adsorption using only single-component isotherm data. The consistency of the predictions was generally better at higher pH and salt concentrations, where protein interactions were weaker. In addition, the model described some systems better than others for which less repulsive protein-protein interactions were involved.; The mechanistic understanding of adsorption and transport was integrated and applied to column behavior. Column modeling with no adjustable parameters facilitated prediction of various chromatographic processes. The modeling showed consistently the critical role of mass transfer kinetics in controlling chromatographic profiles (e.g., width and asymmetry) under a wide range of salt concentrations. The adsorption properties have a direct effect on the effective transport properties through the coupling of adsorption and transport. For isocratic and gradient elution, reasonably good predictions were achieved using constant effective pore diffusivities. For frontal loading, however, where the isotherm non-linearities are most pronounced, the mass transfer coefficient was evaluated based on a parallel diffusion mechanism, which provided good predictability in single-component systems. |
