Macroporous ion-exchange membrane adsorbers : correlation between membrane structure, separation conditions and performance in bioseparation
In the first part of this dissertation, commercial Sartobind® porous cation exchanger membranes, based on stabilized regenerated cellulose and with sulfonic acid (S) or carboxylic acid groups (C), were analyzed with respect to their pore structure in dry, slightly swollen and wet state by three microscopic methods, conventional scanning electron microscopy (SEM), environmental SEM (ESEM), and confocal laser scanning microscopy (CLSM). The dehydration behaviour of the membranes was in situ observed at varied vapour pressure in the chamber of the ESEM, indicating some deformations of the macropore structure (largest pore diameters up to 20 µm) and significant changes in dimension and mobility of smaller cellulose fibers within these macropores, both as function of water content of the membrane. The binding of mono-Cy5-labelled lysozyme inside fluoresceine-labelled and unlabelled Sartobind® membranes was monitored by CLSM. The characteristic fluorescence intensity distributions in areas of (146 * 146) µm2 indicated that protein binding takes place predominately in a layer which is anchored to a fine cellulose fiber network and, to a lower degree, directly to thick cellulose fibers. Due to the limited thickness of this binding layer, a significant fraction of the macropores remained free of protein. Protein binding as function of concentration and incubation times was also monitored by CLSM and discussed related to the binding isotherms for the membranes Sartobind® S and C. Further, a flow-through cell for the in situ monitoring with CLSM of protein binding during the binding step was built, and the results obtained for binding of lysozyme in membranes Sartobind® S indicate that this experiment can give very important information on the dynamic behavior of porous membrane adsorbers during separation: The lateral microscopic resolution in the x,y plane enables the identification of different breakthrough times as function of the location (pore structure), and this information can help to explain possible reasons for axial dispersion (in z-direction) observed in breakthrough analyses of the same separation in a chromatography system. The combination of advanced microscopy with detailed investigations of static and dynamic protein binding will provide a better understanding of the coupling between mass transfer and reversible binding in membrane adsorbers onto separation performance, and it will provide valuable guide-lines for the development of improved membrane adsorbers. Based on the knowledge of the first part, new membrane adsorbers with carboxyl groups were prepared via UV-initiated heterogeneous grafting polymerization on Hydrosart® macroporous regenerated cellulose membranes. The dynamic performance was investigated in detail with respect to the pore size and pore size distribution of the base membranes, ion-exchange capacity, target proteins and architecture of the functional layers. Main characterization methods were pore analysis (BET and permporometry), titration, analysis of protein binding under static conditions including visualization by CLSM and chromatographic analysis of dynamic protein binding and system dispersion. The trade-off between ion-exchange capacity, permeability and static binding capacity of the functional membrane has partially been overcome by adapted architecture of the grafted functional layer achieved by the introduction of suited uncharged groups and stabilization of binding layer by chemical cross-linking. These membranes have negligible effect of flow rate. There is no considerable size exclusion effect for large proteins due to mesh size of functional cross-linked layers. Investigation of system dispersion based on breakthrough curves confirms that the adapted grafted layer architecture has drastically reduced the contribution of the membrane to total system dispersion. The optimum pore structure of base membranes in combination with the best suited architecture of functional layers was identified in this study. In the last part, the internal flow distribution of the flat sheet membrane modules was further quantified characterized based on a novel radial zone rate model. The proposed model partitioned the total void volume of the chromatography module into zones that have approximately homogeneous velocity profiles over time. The model was mathematically represented and analytically solved as a network of continuously stirred tank reactors (CSTR). An additional plug flow reactor (PFR) was connected in series with the CSTR network in order to account for a time-lag that was not associated with system dispersion. The capability of the model to describe experimental breakthrough data was compared to the often applied standard model for extra-membrane system dispersion which consists of a single CSTR in series with a PFR. The commercial CIM® module and a custom designed Sartorius cell were studied with acetone and lysozyme as test tracers at varied flow rates and for varied membrane pore sizes under non-binding conditions. In all studied cases the proposed model fits the measured breakthrough curves better than the standard model. Moreover, the minimal number of radial flow zones that were required to accurately describe the observed breakthrough curves and the estimated flow fractions through these zones provided valuable information for the analysis and optimization of internal module designs.