Please confirm you are 21 or older to continue.
This page is an editorial and informational resource about controlled environments — what an ISO class number counts, how a particle count is actually taken, how air is filtered and how quickly a room recovers, and how routine monitoring is trended against limits. It is written for professional and trade readers who read facility certificates as part of their work.
MW Labs sells nothing. It is a published editorial resource, not a shop, not a laboratory and not a testing body. Nothing here is an offer, and no order can be placed through this site. The practices described are general explanations of common industry procedure — not a substitute for the current text of a standard, or for a qualified engineer's judgement about a specific facility.
Nothing on this page is guidance on the selection, preparation or end use of any material, and no claim of any kind is made or implied about what any material does. This resource is written for readers aged 21 and over.
The ISO 14644-1 scale, what it counts, at which particle size, and why a class number alone says very little.
Sample volumes, how many locations a room needs, and the occupancy state that makes or breaks the result.
Filter grades, air change rates, recovery time, and what a pressure cascade is supposed to prove.
Alert versus action limits, what routine monitoring is for, and eight failures worth catching on a certificate.
A cleanroom is not a clean room. It is a room with a number attached, and the number is a count — how many airborne particles above a stated size were found in a cubic metre of air, under stated conditions, on a stated day, by a calibrated instrument. Everything that makes the number meaningful sits in those qualifiers, and almost every argument about facility quality turns out to be an argument about one of them.
This guide walks through the parts of that sentence in order: what the class scale actually counts, how a count is taken and how many places it has to be taken from, what the air handling has to do to keep it there, why the people in the room are the dominant variable, and how the routine monitoring programme differs from the classification itself. It is a reading guide, not a design manual — the aim is that a certificate handed to you stops being a logo and a pass mark and starts being a document you can interrogate.
The international scale is set by ISO 14644-1, which defines classes 1 through 9. The rule behind them is simple enough to hold in your head: for class N, the permitted concentration of particles at or above 0.1 µm is 10N per cubic metre, and the limits at larger sizes fall away according to a fixed relationship. The practical consequence is that each class is a factor of ten cleaner than the one below it, and that a claim always belongs to a particle size. "Class 7" on its own is shorthand; "class 7 at 0.5 µm" is a specification.
| ISO class | Max particles ≥ 0.5 µm per m³ | Roughly comparable to |
|---|---|---|
| ISO 4 | 352 | Photolithography, precision optics |
| ISO 5 | 3,520 | Critical open handling zones |
| ISO 6 | 35,200 | Support space around a critical zone |
| ISO 7 | 352,000 | General controlled processing |
| ISO 8 | 3,520,000 | Gowning rooms, staged material transfer |
| ISO 9 | 35,200,000 | Ordinary indoor air, controlled access |
Limits shown at the 0.5 µm threshold, the size most often quoted. The same class also carries limits at other sizes, and a room can pass at one threshold while sitting close to the limit at another. Read the whole row, not the headline.
Two numbers in that table are worth pausing on. The step from ISO 8 to ISO 5 is a thousandfold reduction, which is not achieved by cleaning harder — it is achieved by moving far more air through far better filters, in a different airflow pattern, with different rules about who may stand where. And ISO 9 is still a classification: unfiltered indoor air in a tidy office is roughly that, which is a useful reminder that "classified" and "clean" are not synonyms.
The same room yields three different results depending on what is happening inside it, and the standard names all three:
A class stated without its occupancy state is an incomplete claim, and the gap between at-rest and operational in the same room is routinely one full class or more. When a certificate says "ISO 7" and a monitoring log from the same week reads worse, the two are usually not in conflict — they were taken in different states. The question to ask is not which number is right, but which state the room was in when the process it supports was running.
Classification is done with a discrete optical particle counter: air is drawn through a chamber at a known flow rate, each particle crossing a laser scatters light, and the instrument sizes it by the intensity of that scatter and bins it into channels. Three practical consequences follow.
The volume at each location has to be large enough that the class limit would produce a statistically meaningful count — the standard sets it so that at least 20 particles would be expected at the limit, with a floor of two litres and at least one minute per location. In a very clean room that arithmetic gets punishing: at ISO 5 limits it means a substantial sample at every point, and at ISO 4 or better a single location can take many minutes. A survey that finished suspiciously quickly is worth a question.
The count has to be spread across the room, not taken from the most convenient corner. The 1999 edition set the minimum as the square root of the floor area in square metres, rounded up; the 2015 revision replaced that with a lookup table that generally asks for more locations and adds a confidence treatment for small rooms. Either way, locations are distributed to represent the whole space and are measured at working height — the height at which the work actually happens, not wherever the tripod stands most easily.
The older edition applied a 95 % upper confidence limit to the mean when fewer than ten locations were sampled. The 2015 revision dropped that treatment and changed the location table instead. This matters when comparing certificates across years: two documents can describe the same room, pass the same class, and have been produced under materially different statistics. A certificate that does not name the edition it was issued against is harder to compare than it looks.
The class number is an outcome. The air handling is the mechanism, and it does three separable jobs: it removes particles from the supply, it dilutes what is generated inside, and it sweeps the room clean again after a disturbance.
Filtration. HEPA filters are specified against their most penetrating particle size — the size that slips through most easily, typically around 0.1 to 0.3 µm — and are commonly quoted at 99.97 % efficiency at 0.3 µm, with ULPA grades an order of magnitude tighter. A filter's rating is a laboratory figure for the medium; what matters in a ceiling is whether the installed filter and its seal leak. That is a separate test: the filter face and frame are scanned with a probe while an aerosol challenge is introduced upstream, and penetration above a small fraction of a percent counts as a leak to be repaired, not averaged away.
Airflow pattern and change rate. Turbulently ventilated rooms dilute: air enters from ceiling diffusers, mixes, and leaves through low-level returns, with cleanliness a function of how many times per hour the volume is replaced. Unidirectional rooms sweep instead: a full ceiling of filters pushes air down as a moving blanket, conventionally in the region of 0.45 m/s, and particles are carried away rather than mixed in. The two strategies produce very different air change figures for the same class, which is why an air change rate quoted without an airflow pattern tells you almost nothing.
Recovery. Of all the performance tests, recovery is the one that best predicts how a room behaves on a bad day. A controlled disturbance raises the particle concentration, and the time taken to fall back by a factor of one hundred is measured. A room that classifies well at rest but takes half an hour to recover is a room where a single door event costs you the rest of the operation. If a certificate contains a classification and nothing else, recovery is the first thing to ask for.
Cleanliness is also maintained by direction. Adjacent spaces are held at deliberately different pressures so that leakage across a doorway always runs from the cleaner side to the dirtier one, commonly with a difference in the region of 10 to 15 pascals between grades, and airlocks between the biggest steps. Where the hazard runs the other way — work that must be contained rather than protected — the cascade is deliberately reversed, and the airlock becomes a sink rather than a source.
The failure mode is rarely the design; it is the moment the design stops being true. A propped door, a return grille blocked by a pallet, a new machine with its own extract, a filter loading up toward the end of its life — each quietly flattens or inverts a differential that a drawing still shows as correct. A gauge on the wall records the present instant. A logged differential, trended over weeks, records the room. Only one of those two can tell you when it changed.
In an occupied room, the dominant particle source is almost always the people in it. A person in ordinary clothing sheds skin scales and fibres continuously, at rates measured in the hundreds of thousands to millions of particles per minute while moving, and a substantial fraction of what a gowning regime achieves is simply keeping that inside a garment. Correct gowning cuts the figure by orders of magnitude; the same garment donned in the wrong order, or touched on the outside on the way in, gives back much of the gain.
Three behavioural variables move the numbers more than most equipment decisions:
This is also the honest answer to why operational counts diverge from at-rest ones, and why the gap is not a defect to be explained away. It is the room telling you what your process costs it.
Classification is a periodic, formal exercise against a standard. Monitoring is the routine programme that runs in between, and it answers a different question: not "does this room meet its class" but "is this room behaving the way it did last month". Confusing the two is the most common misreading of a facility file.
Monitoring is usually split into two streams. Non-viable monitoring counts particles, often continuously at fixed critical points, with alarms. Viable monitoring looks for organisms rather than particles: settle plates left exposed for a defined period to capture what falls, contact plates pressed to surfaces and gloves, and active air samplers that draw a known volume of air onto a plate so that a result can be expressed per cubic metre rather than per exposure. Temperature, humidity and pressure differentials are logged alongside, because they explain the excursions.
An action limit is a specification: exceed it and a defined response is triggered, investigated and documented. An alert limit sits below it and is a statistical early warning derived from the facility's own historical data — it says "this room is no longer behaving like itself", which is often visible weeks before anything approaches the specification. Alert limits copied from a textbook, or set equal to the action limit, throw that warning away. When reviewing a monitoring plan, the question worth asking is where the alert limits came from and when they were last recalculated.
The value of monitoring is entirely in the trend. A single excursion is an event. Three excursions at the same location in a quarter, or a slow upward drift in a background count that never once breached a limit, is information — and it is exactly the kind that a file of individual pass/fail results will hide.
Classification expires. Under ISO 14644-2 the room is reclassified at defined intervals — commonly every six months for the cleanest grades and annually for the rest — and the interval assumes the room has not materially changed in between.
Change is what breaks that assumption, and it rarely announces itself. A bench moved a metre, a new instrument with its own heat load, a replacement filter of a different make, a revised gowning supplier, an extra person added to the shift: each is small, none is obviously a facility change, and together they can move a room off its qualified state without a single document being raised. A change-control step that explicitly asks whether the environment was affected is what turns a stack of certificates into a continuous claim.
The documentation package that comes out of a requalification is where all of this becomes checkable. A complete one names the standard and its edition, states the occupancy state, lists every sampling location and its result, identifies each instrument by serial number with its calibration date, records the supporting tests — filter integrity, airflow, differential pressure, recovery — and closes with any deviations and their disposition. A one-page certificate with a class and a signature is a summary of that package. It is not a substitute for it, and a facility that cannot produce the underlying data on request has told you something.
None of the following is proof of a problem on its own. Each is a reason to ask one more question.
A practical order for working through a facility file, arranged so the cheap checks come first:
Ten questions is not an audit. But a facility that answers all ten without hesitation is a facility that keeps its own records for its own use, rather than for the visitor — and that distinction is usually visible within the first three.
About this guide. It describes general, widely published industry practice for classifying and monitoring controlled environments, and is written for trade and professional readers. Standards are revised; where a figure matters to a decision, read the current text of the standard itself. Nothing here is advice about any particular facility, and MW Labs neither classifies rooms, issues certificates nor sells anything.
Who publishes this resource, why it exists, and how to reach the editor with a correction or a question about the guide.
MW Labs is an editorial resource on controlled environments and the documents that describe them. It exists because the questions worth asking about a facility certificate are almost always the same ones — against which edition, in which occupancy state, from how many locations, with which instrument, and what has changed since — and they are easier to answer once, properly and in public, than one conversation at a time.
