A network model was used to simulate the hydrodynamic dispersion, which occurs in perfusion chromatography for proteins under non-retained conditions. In order to elucidate the details of the velocity dependency of the hydrodynamic dispersion coefficient in a perfusive bed, the effects of pore-size distribution (PSD), pore inter-connectivity (PI), and solute molecular size were incorporated into the simulations. It was found that, for purification of proteins on a perfusive column, among these parameters, PI affected the dispersion results most significantly over the whole range of Peclet numbers investigated. POROS media were postulated to possibly have a very high PI, of around 4.77–5.37. PSD also played a very important role in determining the dispersion results; POROS R1 and POROS R2 behaved quite differently, although both of them possess bi-modal PSDs. At higher PI (connected-pore fraction higher than 0.8), the D L / UL curve for POROS R1 showed double plateaus, whereas that for POROS R2 gave only one plateau. Under the experimental conditions investigated, only the macropore size distribution contributed to the dispersion results in POROS R1, whereas the composite bi-modal PSD contributed to the results in POROS R2. Based on the simulation results, size exclusion was also found to affect the dispersion results, giving rise to low dispersion coefficient in the molecular diffusion controlled region and high dispersion coefficient in the convection region. However, in the case of a perfusive column, size exclusion effects were rather insignificant. It was also found that hindered diffusion resulted in sharp increase in the dispersion data at very high Peclet number ( Pe>10 4), which was more significant for POROS R2. The simulation results corroborated the experimental data very well. These indicate that the use of a network model to obtain the hydrodynamic dispersion coefficient is a more realistic representation of the column efficiency in perfusion chromatography, one which accounts directly for microscopic information, such as pore structure and solute molecular size.
Read full abstract