Elucidating the 3D microstructure of virus filtration membranes and its impact on membrane performance
Restricted (Penn State Only)
- Author:
- Brickey, Kaitlyn
- Graduate Program:
- Chemical Engineering
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- November 01, 2024
- Committee Members:
- Andrew Zydney, Co-Chair & Dissertation Advisor
Enrique Gomez, Co-Chair & Dissertation Advisor
Hee Jeung Oh, Major Field Member
Bruce Logan, Outside Unit & Field Member
Robert Rioux, Professor in Charge/Director of Graduate Studies - Keywords:
- virus filtration
virus filter
LRV
filter fouling
pore morphology
membranes - Abstract:
- The purification of biopharmaceuticals that are produced from human plasma or genetically modified mammalian cells requires high levels of virus clearance, which can be accomplished using a combination of low pH inactivation, solvent-detergent inactivation, chromatography, and / or virus filtration. Virus filtration is ubiquitously regarded as a critical downstream processing step because it provides a robust size-based removal of at least 99.9% of all virus particles, equal to a log retention value (LRV) of 3, complimenting other strategies that clear virus by inactivation or adsorption. This level of removal is typically achievable under a variety of processing conditions, but there is now extensive experimental evidence that virus removal can be compromised with extended processing, low pressure operation, and process disruptions. To the detriment of virus filter performance, the origins of these phenomena are still poorly understood. Within this thesis, the 3D pore space and virus capture behavior were quantitatively evaluated in two commercial virus removal filters with varying morphology, the highly asymmetric Viresolve® Pro and the relatively homogeneous Pegasus™ SV4; findings were then related to the current understanding of virus capture phenomena. Novel characterization methods were established to obtain and study the 3D microstructure using focused ion beam scanning electron microscopy (FIB-SEM) tomography. The results for the average pore sizes determined from the 3D microstructure showed that 2D estimates were limited and largely underestimated. Internal porosity and interconnectivity were also examined for the first time within these filters, which is impossible from 2D methods. Additionally, capture performance was examined and validated using a filtration-specific computational fluid dynamic software, GeoDict®, demonstrating the applicability of 3D characterization methods to study virus filtration performance. The effects of filtrate flux on virus retention were also explored using GeoDict®, including the contributions of Brownian motion. Model calculations incorporating process pauses were studied to better understand why LRV is compromised in some filters but not others. Visualization of particle trajectories showed particle travel/diffusion during low filtrate flux and / or pressure pauses. The impacts of protein fouling on filtration performance were examined for both membranes with human immunoglobulin G as a model foulant. Membranes were fouled to varying degrees and the pore morphologies evaluated using the developed FIB-SEM methods. Fouling with the Viresolve® Pro membrane appeared mostly uniform in a band near the filter exit, similar to what has been reported previously. The Pegasus™ SV4 filter showed fouling that was inhomogeneous in nature and appeared in large clusters. Further characterization with confocal imaging supported the previously noted inhomogeneous fouling. Simulations with the Pegasus™ SV4 membrane showed a sharp decline in the predicted LRV after filter fouling, well in line with previous observations. Using small angle x-ray scattering (SAXS), decreases in filter performance during processing of highly concentrated mAb solutions were explored. SAXS data were combined with independent DLS measurements and fouling experiments to show the impact of intermolecular protein-protein interactions on filter performance in highly concentrated solutions. Lastly, two case studies were highlighted in which FIB-SEM virus filter reconstructions discussed throughout this thesis were used to validate other research models, highlighting the utility of this characterization method in the membrane research space.