Predicting virus filter performance using an advanced membrane structural model

Virus filters are essential in bioprocessing for safe therapeutic protein production. Understanding the correlation between virus filter structure and performance is important for efficient process design. By applying a multilayer structural model comprised of theoretical layers having pore size distributions, to published visualization data, we demonstrate in the present study that virus removal performance and filtration behavior can be quantitatively calculated. Virus filter structure can be classified into asymmetric high porosity, symmetric high density and laminated structure types. All types exhibit sufficient virus removal properties when used in normal processes. Laminated structure filters show extremely high virus particle removal performance, achieving the most robust virus removal even in filtration with process pauses, although flux decay tends to occur during filtration of solutions containing even very small amounts of aggregate. Conversely, asymmetric high porosity filters may show lower virus removal in processes involving multiple pauses that exceed practical conditions but shows stable filtration behavior with lower increase in pressure and lower flux decay even for the filtration of solutions containing aggregates. Such filtration behavior and virus removal properties can be quantitatively expressed and predicted by numerical calculations using a model based on the theoretical multilayer structure.