ISO 14644-1 Class 8 for a data hall: what the classification means, and how a hall is measured against it

Class 8 is the classification commonly specified for data halls and server rooms, and it is what the clean-build program measures the finished hall against. It is also the most misused term in this trade. So this section sets out exactly what the classification covers, what it does not, and what we do about it.

What the classification actually is

ISO 14644-1 classifies the air cleanliness of a space by airborne particle concentration. A class is a set of maximum concentrations, in particles per cubic meter, at stated size thresholds. Class 8 is one class on that scale. Here are its limits.

ISO 14644-1 Class 8: maximum airborne particle concentrations
Particle size thresholdMaximum concentration
≥ 0.5 µm3,520,000 particles per m³
≥ 1 µm832,000 particles per m³
≥ 5 µm29,300 particles per m³

Three things about that table that a specification sometimes gets wrong:

  • The thresholds are cumulative. The 0.5 µm figure counts every particle at or above 0.5 µm; it is not a band. A room that passes at 5 µm and fails at 0.5 µm has not passed.
  • Class 8 has no limit at 0.1, 0.2 or 0.3 µm. Those cells are empty in the standard. At Class 8 concentrations, the instruments the standard contemplates cannot measure those sizes reliably. If you are handed a Class 8 figure at 0.3 µm, whoever wrote it down invented it.
  • A classification does not have to use all three sizes. The standard classifies at one or more considered particle sizes, and the specifier picks them. A hall classified at ≥ 0.5 µm alone is fully and validly classified. Many are done exactly that way.

Two more points come up whenever older paperwork is in the room.

The second edition of ISO 14644-1, published in 2015, removed the ≥ 5 µm limit for ISO Class 5, because counting very few large particles is statistically unreliable at that cleanliness. It did not remove it for Classes 7, 8 or 9. Class 8’s 29,300 stands.

And “Class 100,000” is not ISO Class 8. It belongs to FED-STD-209E, a US federal standard canceled in 2001 that counted particles per cubic foot. The two are close by design, but they are different documents and one of them no longer exists.

Where the Class 8 specification for a data hall comes from

ISO does not specify Class 8 for data halls. ISO defines the class and the method for measuring against it. The choice of that class for this kind of room comes from ASHRAE Technical Committee 9.9, in Gaseous and Particulate Contamination Guidelines for Data Centers, which says data centers should be kept clean to ISO 14644-1 Class 8. Your own specification may name something different. If it does, your specification governs, not this page.

Four things it does not mean, each of which is regularly claimed

It is not a vendor credential
ISO 14644-1 classifies a space, not a contractor. There is no such thing as an ISO 14644-1 certified cleaning company. A contractor who describes themselves that way has just told you something useful about how carefully they read. What we do is clean a hall and measure it against the Class 8 limits above.
It is not a permanent property of a room
A classification describes the air at the moment it was measured, under stated conditions. It is a reading, not a fixture. A hall that classified on Tuesday, then had a fit-out crew and a pallet of cardboard in it on Wednesday, is not a Class 8 room on Wednesday. The certificate does not change that.
It is not one number without an occupancy state
The same room measures differently empty, finished but idle, and full of running equipment and people. A result only means something next to the occupancy state it was taken in. If someone hands you a Class 8 result, ask which state it was measured in before you treat it as evidence about the room you are about to energize.
It is not a measure of gaseous or corrosive contamination
ISO 14644-1 counts particles. Gaseous and corrosive contamination attacks copper and silver on a board rather than blocking an intake. It sits on a separate axis, covered by ANSI/ISA-71.04, and different specialists measure it with different instruments. Temperature and humidity are a third axis again, covered by the ASHRAE TC 9.9 envelope. We measure particulate. Mixing them up is the surest sign that a scope of work was written by somebody outside this trade.

Which part of ISO 14644 covers what

ISO 14644 is a family of standards. Most of the confusion in this trade comes from citing the wrong member of it.

The standards a data hall contamination specification usually touches
StandardWhat it covers
ISO 14644-1Classification of air cleanliness by particle concentration: the class limits, the sampling method and the evaluation. This is where Class 8 comes from.
ISO 14644-9Assessment of surface cleanliness by particle concentration.
ISO 14644-13Cleaning of surfaces to defined cleanliness levels.
ANSI/ISA-71.04Gaseous and corrosive contamination, the mechanism that attacks copper and silver on a board.
ASHRAE TC 9.9The temperature and humidity envelope for data centers, and the contamination guidance referred to above.

Two things are worth taking from that table. First, ISO 14644-1 does not cover surface cleanliness. That is Part 9. A specification citing Part 1 against a surface requirement is citing the wrong document.

Second, here is the real reason there is no such thing as an ISO 14644-1 certified cleaning company, and it is simpler than the argument usually made. Part 1 is not about cleaning. It is about measuring air. Part 13 is the part of the family that covers cleaning surfaces.

ISO 14644-1 is a document you can buy. Anyone writing or accepting a specification against it should work from its own text rather than from any summary, including this one.

The three occupancy states

A classification has to be stated in an occupancy state. There are three. We describe them in plain language here; the formal definitions are in the standard.

Occupancy states, and what is in the room in each
StateWhat is in the room
As-builtConstruction complete, services connected and running. No production equipment, no furniture, nobody in the room.
At-restFully fitted out, equipment installed and operating as agreed. Nobody in the room.
OperationalEquipment running as specified, with the agreed number of people present, working to the agreed procedures.

The same room, measured on the same day with the same instrument, returns three different numbers. As-built is the cleanest: nothing is generating and nothing is stirring. Operational is the dirtiest, because people are the biggest single particle source in most occupied clean spaces, and movement puts settled material back into the air.

Name the state you require in the scope of work. It changes what the reading proves. A scope that says “achieve Class 8” without naming an occupancy state has not yet said which of three different targets it means.

How we measure a hall against it

  1. Measure the room in the right condition

    An interim handheld check every 5,000 sq ft at the pre-fit-out deep clean, then new filters and a 24-hour scrubber run before the final samples. Measure a room while its filtration is still loaded from the build and you get a number that describes the filters, not the hall. The full program →

  2. Sample the room, not a corner of it

    A minimum of three particle samples, spaced evenly across the hall. The standard’s own method, and what it would ask of a hall this size, is set out above. If your commissioning authority is specifying a formal classification, say so at bid stage and we will scope the sampling to what they need rather than to our default.

  3. Hand over the readings and the conditions

    You get the readings and the baseline records at handover: the numbers, and the conditions they were taken under. That is the deliverable, and it is a measurement of your room rather than a verdict on it. We are contracted to clean to a scope and to measure, and what counts as ready is your decision. If your process needs an independent classification by a qualified testing firm, that is a separate engagement with a separate provider, and we will say so rather than let our own paperwork stand in for it.

What the instrument is actually reporting

An optical particle counter does not measure a particle’s physical diameter. It measures the light a particle scatters and converts that to an optical equivalent diameter, calibrated against latex spheres. Construction dust is angular mineral matter, so when a counter reports 0.5 µm for a gypsum shard it is reporting the size of a sphere that would scatter the same light.

We publish that because it is a limitation of the method we ourselves use and it applies equally to every contractor working with the same instruments.

Send us the scope and the date

Hall count, surfaces in scope, what phase the building is in, your access constraints, your change window, and the date you need racks in the room. That is enough to price against, and enough to tell you straight whether the date is reachable.