homogenously distributed over the complete filter surface. Second, the transmembrane

pressure drop can be controlled independently from the cross-flow velocity [53].

Both the cells and particles/cell debris in the broth are exposed to a multitude of

various forces: gravity force, axial force caused by impeller rotation, centrifugal force

created by the filter rotation, and radial force due to the perfusion flux [52]. In case

external filters are used, additional secondary flows (so called Taylor vortices) are

formed, which are supposed to produce abrasion effects preventing fouling, although

this has been shown to play a minor role in the filtration performance [63].

Implementation of spin filters resulted in high cell concentrations in a variety of ex-

periments. However, short perfusion times and premature termination of many runs

highlight the major concerns regarding the use this technology: fouling and clogging

[43,53]. Since replacement of internal units is impossible, spin-filters should be de-

signed and operated to reduce both risks [52]. Fouling is mainly caused by deposition of

dead cells and nucleic acid on the filter [19], while clogging can occur either if cells

accumulate in the pores when the filtration flux exceeds the retention capacity of the

filter or when pores are increasingly narrowing down due to cell growth on the filter

surface [43]. Several points have to be considered for efficient operation of spin filters:

Hydrophobic plastic instead of stainless steel should be used as a filter screen material

due to lower binding of proteins, nucleic acid, and cells [52,66]. By using larger pore

sizes (up to 50 μm) the filter longevity can be increased while allowing a selective

retention of viable cells [52]. Moreover, this could allow virus particles to pass through

the membrane. As for cross-flow filters, the risk of filter fouling increases with cell

concentration and perfusion rate [53,67]. Transmembrane pressure differences in the

range of 0.5–1 bar were found for maximum perfusion fluxes. At smaller (reduction of

the driving force) or at larger differential pressures (improved cake layer formation),

less favorable filtration conditions are present. In particular, it is critical not to use too

high perfusion rates. Henzler et al. calculated maximum perfusion rates in relation to

specific membrane areas based on literature data compiled by Castilho and Medronho

and Voisard on cultivations with spin filters [43,52,63]. They showed that high per-

fusion rates can only be achieved with very large filter areas. This is highly unfavorable

for intensified virus production processes and scale-up, as high cell concentrations

require high perfusion rates. Fouling caused by high perfusion rates could be partially

compensated by increasing the tangential rotation speed. However, too high rotation

speeds were shown to decrease cell retention efficiencies [52]. Nevertheless, successful

scale-up was demonstrated up to 500 L at a rate of 1 RV/d [67].

While spin filters are mainly used for the production of monoclonal antibodies,

some studies investigated their use for the production of an experimental rabies

vaccine using suspension BHK21 cells [32,54]. Despite much progress regarding

the design, the scale-up and the operation of spin filters over the last decade, fouling

or retention problems still persist which could be particularly problematic for the

production of viruses.

6.6.2

TANGENTIAL FLOW FILTRATION

Originally developed as a rapid and efficient DSP method for separation and pur-

ification of biomolecules, tangential flow filtration (TFF) can be applied in a wide

Process intensification

153