EXPERIMENTAL EVALUATION OF CONFLICTING MODELS FOR SIZE EXCLUSION CHROMATOGRAPHY

Comparisons were made among the predicted dependences of K(SEC), the chromatographic partition coefficient, on the solute radius R, for three widely acknowledged theories of size exclusion chromatography. It was found that data which appear to agree with the general form of K(R) for any one theory will also appear to support the others, because of the mathematical relationships among the three K(R) functions, over the range of K most usually reported. It must be concluded that the shape of the functional dependence of K(SEC) on R alone does not provide a meaningful test of theory. To test these theoretical relations, data were obtained by aqueous SEC on Superose columns for molecular weight fractions of Ficoll. This solute may be shown to closely approximate a rather inflexible sphere, and hydrophobic and electrostatic interactions can be neglected for this polymer/stationary phase pair. The selection of conditions thus reduces uncertainties related to the definition of R and minimizes the influence of enthalpic effects (i.e., "nonideal SEC"). The measured dependence of K(SEC) on R (with corrections made for polydispersity of the samples) is found to agree remarkably well with predictions from a treatment in which the stationary phase is modeled as a Gaussian distribution of cylindrical cavities.