ISO 14644-1 class limits, explained

Why the table has blank cells

Open the classification table in ISO 14644-1:2015 for the first time and the thing that looks like a printing error is the empty space. Whole regions of the grid carry no value at all. Engineers hit those gaps, assume something is missing, and go looking for the number elsewhere. There is no number elsewhere.

The gaps are not omissions. Every blank cell carries a footnote letter, and the footnotes say the same kind of thing three different ways: at that particular combination of class and particle size, classification is not the right instrument to reach for.

• At the very high concentrations in one corner of the table, a concentration limit is simply not applicable.
• At very low concentrations, sampling and statistical limitations make classification at that size inappropriate. There is not enough signal to draw a defensible conclusion from.
• Above one micrometre, sample collection itself becomes the limitation. Particles that size can be lost inside the sampling system on the way to the counter, so a count taken there would be measuring the apparatus as much as the room.

A blank cell is therefore a statement rather than a hole. It says the standard declines to classify that combination, and it means that a certificate claiming a class at one of those sizes is claiming something the standard does not offer.

The five micrometre figure that will not go away

The most common single error in cleanroom content on the web sits exactly on this point. ISO 14644-1:2015 does not permit an ISO 5 classification at 5 µm at all. The widely circulated figure for it descends from the 1999 edition's table, and before that from the federal standard lineage the 1999 edition inherited. It has been quoted forward ever since, into specifications, into vendor pages, and into procurement documents that are still in circulation.

It matters more than a stale citation usually would, because it is precisely the change that forced the European GMP annex for sterile products to be rewritten. A quality manager in a regulated plant will spot it on sight. If a specification handed to you asks for that classification, the honest answer is not to quote a number for it. It is to say that the current edition does not classify at that combination, and to agree what is actually being asked for.

Why the numbers look the way they do

The class limits were derived rather than chosen. That is the whole explanation for why they look so strange, and it is worth knowing because it stops people treating the odd-looking values as typographical errors and rounding them off in their own documents.

Each limit follows from the class number and the particle size through a fixed relationship the standard sets out, and the result is rounded. Nobody sat down and picked the values. They fall out of the relationship, which is why they land where no human would have put them and why they are not tidy multiples of each other.

The practical consequence is short: do not round a class limit to something that reads better, and be suspicious of any document that has. A limit that has been tidied up is no longer the limit.

A class number on its own is not a classification

If your certificate says ISO 7 and nothing else, it is incomplete. This is the single most useful thing on this page for anyone about to accept a report.

ISO 14644-1:2015 clause 4.4 requires a classification to be designated with three elements, not one. It needs the ISO Class number, the occupancy state the class applies to, and the particle size or sizes that were considered. The standard's own worked example of a complete designation is written as "ISO Class 4; at rest; 0,2 µm, 0,5 µm". A class number by itself is a third of that.

The occupancy state is the element most often dropped, and it is the one that changes the answer most:

• As-built means the room is complete with all services connected and functioning, but with no equipment, furniture, materials or personnel in it.
• At-rest means the room is complete with equipment installed and operating in a manner agreed upon, but with no personnel present.
• Operational means the room is functioning in the specified manner, with equipment running and the specified number of personnel present.

People are the dominant particle source in most rooms, so the same room can comfortably hold a class at rest and fail the same class operationally. That is not a contradiction. It is the entire reason the standard defines three states instead of one, and it is why a class number with no state attached does not tell you what you need to know.

Where more than one particle size is reported together, the sizes have to be far enough apart to be measuring different things: each larger diameter must be at least 1.5 times the next smaller one.

Every sampling location has to pass on its own

This is the change most likely to surprise someone who has been buying certification for a long time, and it is the one with a real consequence attached.

Under ISO 14644-1:2015, clause 5.3, the particle concentration at each sampling location is assessed against the limit on its own, and every location has to comply. The 1999 edition did not work that way. Where the number of sampling locations was small, it applied an upper confidence limit calculation across the set, and that calculation could carry a room past a single marginal location on the strength of the others. The current edition removed it.

The consequence is worth stating plainly, because it catches people out: a room that was certified under the older method may not pass under the current one with exactly the same air in it. Nothing has to have got worse. The rule for deciding what counts as a pass changed, and a room that was carried by the calculation before is now judged on its weakest location.

If that reads like an argument for finding out where your own weakest location actually is, it is. How often the verification itself should be repeated is a separate question with a separate answer, and it is covered in how often a cleanroom should be certified.