Filtration in microbiology separates microbes from fluids by physical size exclusion rather than by killing them. A membrane with pores smaller than the target organism retains cells, spores, or viruses while the liquid passes through. The method matters because it removes organisms without heat or chemicals, so it preserves heat-labile media, antibiotics, and proteins that sterilization would destroy.

The same principle explains why pore rating is the single variable that decides success or failure. Choose a membrane larger than the organism and the filtrate stays contaminated. Choose one small enough and you get sterility, but at the cost of flow rate. Everything else in the technique follows from that trade-off.

Size-exclusion mechanics at the pore scale

A microporous membrane works like a sieve, but not a perfect one. The pore size printed on the box is a nominal rating, meaning most pores are near that diameter but some are larger. Fluid flows through the pore network under a pressure difference. Particles larger than the pore are held at the surface by direct interception and by physical straining. Particles near the pore diameter can be captured by adsorption to the pore wall or by cake formation as retained material builds a layer.

The mechanism is depth filtration as well as surface filtration. Feeds pass through a mat of fibers or a sponge-like polymer in which organisms are trapped along a tortuous path. This is why graded depth filters such as glass fiber or diatomaceous earth can hold large bioburden loads before clogging. Membrane filters, by contrast, act mainly at the surface and clog faster once the retained layer matures.

Flow follows the Hagen-Poiseuille relationship for flow through small channels. Flow rate scales with the fourth power of pore radius and inversely with filtrate viscosity. Halving the pore diameter cuts flow through a clean membrane by roughly sixteenfold at constant pressure. This is the quantitative root of the sterilization-versus-speed trade-off.

Because retention depends on organism size, the filter must be matched to the smallest target. Sterile filtration uses 0.22 micrometer membranes because most vegetative bacteria are 0.5 to 2 micrometers in diameter. The 0.22 micrometer rating provides a safety margin, not a guarantee against everything.

Size cutoffs and performance numbers

The industry convention of a 0.22 micrometer sterilizing-grade membrane comes from a bacterial retention test rather than a geometric rule. Manufacturers qualify filters using a standard challenge organism, usually Brevundimonas diminuta at a defined challenge density. Regulatory frameworks accept a sterilizing-grade filter as one that retains this organism under validated conditions (FDA, sterile drug products guidance).

Pore ratings are nominal, and the difference matters. An absolute rating means essentially all pores are at or below the stated size. A nominal rating means a defined percentage is at or below that size, often 90 percent or more. Beverage and water filters often carry nominal ratings and are not a substitute for validated sterile filtration.

Virus retention requires much smaller pores or a different principle entirely. Ultrafiltration membranes with molecular weight cutoffs in the kilodalton range retain viruses by size and charge. Tangential-flow formats are used because dead-end flow clogs almost immediately at that scale.

The table below compares common filter types on the variables that drive selection.

Filter type Typical pore or cutoff Primary retention mechanism Main microbiology use Main limitation
Cellulose acetate membrane 0.22 to 0.45 micrometer Surface straining with some adsorption Sterile filtration of media and buffers Clogs with high particulate loads
Polyethersulfone membrane 0.1 to 0.45 micrometer Straining plus low protein binding Sterile filtration of protein solutions Higher cost than cellulose
Polytetrafluoroethylene membrane 0.2 to 0.45 micrometer Straining and hydrophobic capture Venting and gas filtration Requires wetting for aqueous use
Glass fiber depth filter 1 to 10 micrometer Depth entrapment across fiber mat Prefiltration and bioburden reduction Not sterilizing on its own
Diatomaceous earth depth filter Graded, coarse to fine Depth entrapment and adsorption Clarification of large volumes Not a sterile barrier
Ultrafiltration membrane Kilodalton molecular weight cutoff Size and charge exclusion Virus retention and concentration Rapid fouling in dead-end mode

One counterintuitive detail: tightening the pore does not always improve organism removal in a straight line. At very small pores, adsorption and charge effects can dominate, and a filter can retain organisms smaller than its rating while passing some larger ones through defects. Filter integrity testing exists because retention is a property of the whole device, not just the pore size.

Edge cases that break the simple sieve model

Mycoplasma species are among the smallest free-living bacteria, with some cells near 0.2 micrometers. A 0.22 micrometer membrane may not reliably retain all mycoplasma under every condition, and 0.1 micrometer membranes are used when mycoplasma removal is required. This is a genuine edge case, not a marketing claim.

Spores are not always smaller than vegetative cells, but they are more resistant to everything except physical removal. Filtration captures spores by size just as it captures cells. The problem is that a compromised filter or a bypass path lets spores through, and spores survive downstream conditions that would kill vegetative cells.

Mycobacteria and other organisms with hydrophobic cell surfaces can adsorb to membrane polymers and reduce apparent passage, which is helpful for retention but can foul the filter. The reverse also occurs: surfactants and high salt concentrations can disrupt adsorption and change retention behavior. Validated conditions matter for this reason.

Bacteriophages and small viruses below 0.1 micrometer pass through sterilizing-grade membranes. Removing them requires ultrafiltration or specific virus-retentive membranes with validated log-reduction values. A 0.22 micrometer filter is not a virus barrier, and treating it as one is a common and consequential error.

High-viscosity feeds, such as concentrated protein solutions, reduce flow and raise the pressure needed to drive filtration. Excessive pressure can deform the membrane or force organisms through defects. Manufacturers specify maximum operating pressures for this reason, and exceeding them voids the retention assumption.

Misconceptions worth correcting

The first misconception is that a 0.22 micrometer filter sterilizes everything. It does not retain viruses, and it may not retain the smallest mycoplasma under all conditions. It is a bacterial retention barrier with defined limits.

The second is that pore size and organism size can be compared arithmetically. Nominal ratings, pore-size distribution, adsorption, and organism shape all intervene. A rod-shaped bacterium near the pore diameter may orient and pass where a sphere of the same width would not.

The third is that filtration kills organisms. It does not. Retained organisms remain viable on the membrane and in the retentate. A filter is a physical barrier, and the filtrate is sterile only if the device was intact and properly operated.

The fourth is that a filter can be reused indefinitely if it still flows. Flow says nothing about integrity. Pinholes, seal defects, and membrane damage can open paths larger than the rating while the device still passes fluid.

Red-flag warnings to watch:

A filter that flows faster than expected after cleaning or repeated use may have a compromised membrane. Skipping integrity testing after a sterilization cycle removes the only evidence the barrier held. Using a nominal-rated filter where a validated sterilizing-grade filter is required is a documented failure mode. Exceeding the manufacturer's maximum differential pressure can deform pores and invalidate retention. Assuming a hydrophobic vent filter retains organisms in liquid service without wetting is a common error.

Practical workflow and integrity thresholds

A defensible filtration workflow has four stages: select, qualify, operate, and verify.

Select the filter by the smallest organism you must retain, not by convenience. For bacterial retention, use a validated 0.22 micrometer sterilizing-grade membrane. For mycoplasma, use 0.1 micrometer. For viruses, use a virus-retentive membrane with a stated log-reduction value. Match the membrane polymer to the feed, choosing low-binding polyethersulfone for proteins and cellulose acetate for general aqueous media.

Qualify the device before use. Perform a bubble-point or pressure-hold integrity test on the assembled filter. The bubble-point test measures the pressure at which gas displaces liquid from the largest wetted pores. A bubble point below the manufacturer's specification indicates a defect. Record the value and compare against the specification, not against a previous run alone.

Operate within limits. Keep differential pressure below the manufacturer's maximum, usually stated in the device data sheet. Prefilter high-particulate feeds with a depth filter to protect the sterilizing membrane. Avoid excessive flow rates that can force organisms through defects.

Verify after use. Repeat the integrity test on the filter after filtration to confirm it held throughout the run. Post-use testing is the standard practice in sterile pharmaceutical manufacturing, and it catches damage that occurred during operation.

There is one unit of reconciliation to make explicit. Bubble-point specifications are stated in pressure units, and the threshold depends on the liquid's surface tension and the pore size. A filter rated for a given bubble point in water will show a different value in a low-surface-tension solution. Compare measured values against specifications generated under the same wetting conditions, not across different liquids.

When filtration is the wrong tool

Filtration fails when the feed cannot be prefiltered and clogs before the run completes. It fails when the target organism is smaller than the available pore rating. It fails when the product binds to the membrane and is lost. It fails when the process cannot tolerate the pressure drop or the cost of validated single-use devices.

Filtration also cannot substitute for terminal sterilization when the product allows heat. Autoclaving at 121 degrees Celsius for 15 minutes achieves a defined sterility assurance level with established margins. Moist-heat sterilization is validated to a probability of a nonsterile unit below one in a million. Filtration achieves sterility by retention, and the assurance depends on integrity testing and validated retention rather than on a kill curve.

The practical rule is straightforward. Use filtration for heat-labile liquids, for gases, and for products that cannot tolerate terminal sterilization. Use heat when the product allows it. Use both, in sequence, when the risk justifies the cost.

Filtration techniques in microbiology are powerful because they are gentle and physically predictable. They are limited because their assurance rests on device integrity, organism size, and operating discipline. The filter does the separating, but the validation and the integrity test do the assuring.