A virus is far smaller than the gaps in an N95. The mask catches it anyway, and it catches it better than a bigger particle

The argument turns up every time masks are discussed, and it sounds unanswerable. A coronavirus particle is about 100 nanometres across. The fibres in a mask are tens of micrometres apart. Expecting that mesh to stop that particle, the argument goes, is like expecting a chain-link fence to stop a mosquito.

The reasoning is clean, the numbers are right, and the conclusion is wrong — because it assumes a filter works by sieving, and a filter does not work by sieving. In fact a well-made filter captures a 100 nanometre particle more reliably than it captures a 300 nanometre one. The hardest particle to catch is neither the biggest nor the smallest but one in the middle, and understanding why explains masks, air purifiers, vacuum cleaners and industrial clean rooms all at once.

A filter is not a sieve

A sieve separates by size: anything wider than the hole stays behind, anything narrower passes. That is one mechanism, and in a fibrous air filter it is close to the least important.

A fibrous filter is a deep, disordered mat of fibres with a great deal of open space — typically well over ninety per cent air by volume, which is why you can breathe through it at all. Air threads its way between the fibres along curving streamlines. The question is never whether a particle fits through a gap. It is whether the particle manages to stay on the airstream all the way through the mat without touching a fibre, because a particle that touches a fibre stays there, held by van der Waals attraction. There is no bouncing off and continuing.

So filtration is not a question of geometry but of whether a particle's path departs from the path of the air. Four mechanisms cause it to depart, and they behave very differently with size.

Interception. The particle follows its streamline faithfully, but that streamline passes within one particle-radius of a fibre, so the particle brushes against it. This depends on the particle's size relative to the fibre spacing and matters most in the middle of the range.

Inertial impaction. The particle is heavy enough that when the air curves sharply to flow around a fibre, the particle cannot turn in time and carries straight on into it. This dominates for large particles, above roughly a micrometre, and it works better when the air moves faster, because a sharper turn is harder to follow.

Brownian diffusion. This is the one that overturns the intuition. A very small particle is light enough to be visibly knocked about by collisions with individual air molecules, so it does not travel in a straight line at all — it wanders, jittering randomly across streamlines as it goes. The smaller the particle, the more violently it wanders, and the more likely it is to blunder into a fibre. Diffusion dominates below about 0.1 micrometres and it works better when the air moves slowly, because a slow-moving particle spends longer in the mat and has more time to wander into something.

Electrostatic attraction. Many modern filter media carry a permanent electric charge on the fibres. Charged and even neutral-but-polarisable particles are pulled out of the airstream and onto the fibres regardless of which of the mechanisms above would otherwise apply. This one is worth its own section below, because it is the reason a comfortable mask can also be an effective one.

The size a filter is worst at

Now put the two dominant mechanisms side by side. Impaction gets weaker as particles get smaller — less mass, less inertia, easier to follow the air. Diffusion gets weaker as particles get larger — more mass, less jitter, easier to stay on the streamline.

One fails going down. The other fails going up. Somewhere between them is a size at which neither works well, and that is where a filter performs worst.

For typical filter media and airflows, that crossover sits at roughly 0.3 micrometres, and it has a name: the most penetrating particle size, or MPPS. It is not a coincidence that filter standards specify testing at about this size — the number quoted for a filter is deliberately its performance at its worst case, not its average.

This single fact dismantles the chain-link-fence argument completely. An N95 respirator is rated to capture at least 95% of particles at the size it is least able to capture. A 100 nanometre virion is three times smaller than that, deep in the region where Brownian diffusion is strong, and is therefore filtered more efficiently than the 0.3 micrometre test particle — not less. The same is true of a HEPA filter rated at 99.95% or better at MPPS.

There is a second thing wrong with the original argument, which is that a virus in the air is very rarely a naked virion. It travels inside a respiratory droplet or, once that droplet has partly evaporated, inside a dried residue — objects measured in micrometres, which are in the range filters handle easily.

A filter's rating is a promise about its worst case. Particles smaller than the tested size are not the ones that get through — they are the ones the filter is best at.

The charge that does the quiet work

If capture depended only on mechanical mechanisms, a mask efficient enough to be useful would be dense enough to be unbearable to breathe through. The compromise is resolved by charge.

Most N95-class media are electrets: polypropylene fibres carrying a quasi-permanent electrostatic charge, applied during manufacture by corona discharge or by triboelectric charging. The charge attracts particles across streamlines that would otherwise have carried them safely past, adding a large amount of capture without adding a single fibre. Efficiency goes up; breathing resistance does not.

The catch is that a charge, unlike a fibre, can disappear while the material looks exactly the same.

• Water and humidity dissipate it. This is why washing an N95 ruins it, and why a mask that has been soaked with sweat or breath moisture for a long shift is not what it was at the start.
• Alcohol and other solvents are worse — spraying a respirator with sanitiser is one of the most reliable ways to destroy its performance, and it was done very widely.
• Oils in the air neutralise it, which is precisely what the N in N95 means: Not resistant to oil. R and P ratings exist for oily environments.
• Time and handling degrade it slowly even in storage.

The uncomfortable part is that none of this is visible. A de-charged respirator looks, feels and fits exactly like a working one, and it will still stop the large particles that mechanical mechanisms catch. What it has lost is most of its efficiency in the difficult middle of the range — the part that the rating was about.

Fit beats filtration almost every time

A respirator has two failure routes: air can go through the medium, or it can go around it. The second is nearly always the bigger problem.

A gap at the bridge of the nose or along the cheeks is a hole with no filtration at all, and air is lazy — it preferentially takes the low-resistance path, which is the leak. This is why a well-fitting cloth or surgical mask can outperform a poorly fitted N95 in practice, why formal fit testing exists in occupational settings, and why a beard is not a minor detail: facial hair crossing the seal prevents a seal from forming at all, and no amount of filter quality compensates.

For anyone actually choosing a mask, the priority order is therefore: does it seal against your face, is the medium genuinely rated, and only then which brand. A simple check — cover the mask surface with your palms and inhale gently, feeling for air arriving anywhere other than through the fabric — takes five seconds and finds most leaks.

The same physics in your room

HEPA filters in air purifiers, vacuum cleaners and building systems work by exactly the mechanisms above, at a different scale. A true HEPA filter is defined by performance at the MPPS — typically 99.95% or 99.97% depending on the standard — which is why the term "HEPA-type" or "HEPA-style" on a product means nothing at all. Either it meets the standard and can say so, or it does not.

Two practical points follow from the physics rather than from the marketing.

Airflow rate is half the product. A purifier's clean air delivery rate is the filter efficiency multiplied by the volume of air it actually pushes through. A superb filter on a weak fan cleans a small room slowly. Matching delivery rate to room volume matters more than the filter's last decimal place.

Air has to actually go through it. A purifier in a corner behind a sofa, or one whose intake is against a wall, recirculates the same small pocket of clean air while the rest of the room stays as it was. This is the appliance equivalent of a mask leak, and it is common.

For Indian homes during high-pollution months, this is worth being concrete about: PM2.5 is by definition particles below 2.5 micrometres, spanning the region where filters are weakest through to the region where they excel, and a correctly sized unit running continuously in a closed room measurably works. What it cannot do is keep up with an open window onto a bad street.

Where nanofibres genuinely change the trade-off

Everything above involves one unavoidable tension: any filter can be made more efficient by adding more fibres, and every fibre added increases the pressure drop — the resistance to breathing or the load on a fan. The whole engineering problem is efficiency per unit of pressure drop, and this is where nanomaterials make a real contribution rather than a decorative one.

Electrospun nanofibres — polymer fibres drawn to diameters of tens to hundreds of nanometres by an electric field — improve that ratio for a reason that is genuinely counterintuitive. Ordinarily air sticks to a solid surface, and the drag on a fibre reflects that. But when a fibre's diameter becomes comparable to the mean free path of air molecules, about seventy nanometres at atmospheric pressure, air begins to slip along the surface instead of gripping it. Drag falls, and pressure drop falls with it, while the fibre's ability to intercept particles does not.

That same seventy-nanometre number appeared in our article on heat, where confining air in pores smaller than its mean free path is what makes aerogel insulation work. It is the same physics used for a different purpose, and it is a good illustration of why the mean free path of air is a number worth carrying around.

Nanofibre layers are now standard in high-performance filtration, and they also make filters less dependent on electret charge — an advantage precisely because charge is the fragile part. As with every material this site has covered, the laboratory claim and the manufacturable product differ: uniformity over large areas, durability under airflow, and cost per square metre are the constraints, not the physics.

Why it matters for students and researchers

Aerosol science is taught far less than its public-health importance warrants, and the MPPS is one of the most elegant teachable results in it — two mechanisms with opposite size dependence, producing a worst case in the middle that neither would predict alone. Any student who can explain why a filter is worse at 0.3 micrometres than at 0.03 has understood something that a great many confident public commentators have not.

The reasoning generalises. Whenever two mechanisms trade off against a variable in opposite directions, expect an optimum or a worst case somewhere in between, and expect it to be counterintuitive to anyone reasoning from only one of them. That habit is worth more than the specific number.

The open problems are practical. Filter media that maintain efficiency without relying on a decaying charge would remove the single largest source of unreliable real-world performance. Reusable and genuinely decontaminable respirators remain an unsolved need, as the pandemic made painfully clear when improvised methods destroyed the media they were meant to sanitise. Low-pressure-drop media matter for energy consumption in building ventilation at national scale. And the interaction of filtration with humidity, in the conditions much of India actually has, is under-studied relative to how much it changes performance.

Frequently asked questions

So can a mask stop a virus or not?

A properly fitted, genuinely rated respirator substantially reduces how many virus-carrying particles you inhale, both because airborne virus mostly travels in droplets and residues that are easy to filter, and because even a bare virion is smaller than the most penetrating particle size and is therefore captured efficiently. What no mask does is reduce exposure to zero, and the limiting factor in real use is nearly always fit rather than the filter medium.

Can I wash or sanitise an N95 and reuse it?

Do not wash it, and do not spray it with alcohol. Both destroy the electrostatic charge that provides much of its efficiency, and the mask will look and feel unchanged afterwards, which is what makes this dangerous rather than merely wasteful. The practical approach for repeated use is rotation — several masks used in turn with days of dry rest between wearings — and replacement once straps loosen, the seal degrades or the mask becomes visibly soiled or damp.

Is a cloth mask useless then?

No, but it is a different product. A cloth mask has no electret charge and a much looser structure, so its efficiency at small sizes is considerably lower — though not zero, because interception and diffusion still operate. Its main contribution is reducing the droplets a wearer emits, and a well-fitting cloth mask genuinely outperforms a gaping respirator. It is a meaningful measure and not an equivalent one.

What about masks with an exhalation valve?

The valve makes exhaling easier by letting breath out unfiltered. That protects the wearer as normal and offers little protection to anyone else, which is the opposite of what is wanted in a shared indoor space. In an industrial setting where the hazard is dust rather than other people, a valve is entirely reasonable.

Do air purifiers actually reduce PM2.5 indoors?

Yes, when the unit's clean air delivery rate is matched to the room, the room is reasonably closed, and the intake is not obstructed. Independent measurement with a monitor is worthwhile, bearing in mind that low-cost sensors have their own limitations. What a purifier cannot do is compensate for continuous infiltration from outside, so sealing obvious gaps and running the unit in a closed room matter as much as the specification does.