Sterilizing-grade filters

A sterilizing-grade filter is the last physical barrier between a process fluid and a batch that cannot be terminally sterilised. The most useful thing to understand about it is that its rating is not geometric. “0.2 micron” does not assert that the membrane contains pores 0.2 µm across; it asserts that the membrane passed a defined bacterial challenge.

The rating is a performance definition

The convention is a challenge with Brevundimonas diminuta, a small, well-characterised organism, at a high loading per unit of filter area, with sterile filtrate required. The organism is chosen deliberately: grown under nutrient-poor conditions it becomes small and deformable, which makes it a demanding test. The rating therefore encodes an outcome under a specified challenge, and it follows that a filter is only “sterilizing” for fluids resembling that validation. Viscosity, surfactants, high protein load, unusual pH or long filtration times all change the answer, which is why product-specific bacterial challenge validation exists at all.

Retention is more than sieving

Membranes are tortuous three-dimensional structures with a pore-size distribution, not a screen. Retention combines three mechanisms: size exclusion, where the particle simply cannot pass; adsorptive capture, where an organism is held by electrostatic and hydrophobic interaction with the pore wall long before geometry alone would stop it; and, for gas service, inertial impaction and diffusional interception, which is why a hydrophobic air filter retains particles far smaller than its liquid rating implies.

Adsorptive retention explains two things at once — why filters usually outperform their nominal geometry, and why retention can degrade. Adsorption sites saturate. High ionic strength screens the electrostatic contribution. A long, heavily loaded run is not equivalent to a short clean one.

What an integrity test actually measures

Integrity tests do not measure sterility; they measure a physical property correlated with it during validation. In a bubble-point test the membrane is wetted and gas pressure raised until gas displaces liquid from the largest pore — the Young–Laplace relationship links that pressure to pore radius through the liquid’s surface tension and contact angle. A diffusive-flow test measures gas dissolving through the wetted membrane below that pressure, an area-proportional quantity sensitive to a small defect. Hydrophobic membranes, which cannot be water-wetted, are tested by water intrusion instead.

Each is an inference. Its meaning comes from the correlation to bacterial retention established during qualification — which is why wetting fluid, temperature and test parameters cannot be varied casually, and why testing after use matters as much as before: a filter can be integral at the start and damaged by the run.

Where the barrier does not hold

Sterilizing filtration removes bioburden. It does not remove everything alive. Mycoplasma lack a cell wall, are pleomorphic, and can pass a 0.2 µm membrane; media at risk are filtered at a tighter rating. Viruses pass freely — virus clearance is a different unit operation with different physics. Endotoxin is a molecule, not an organism, and is not retained by size, though membrane chemistry may adsorb some of it. Prions are outside the scope entirely.

Two further physical limits belong on the same page: protein and preservative can adsorb to the membrane, so filtration can change the fluid it passes; and every polymeric membrane and housing has an extractables and leachables profile that has to be characterised rather than assumed.