Most industries grade cleanliness by looking. Data centers grade it by sacrifice. The standard method hangs small strips of bare copper and silver in the airstream, leaves them for about a month, and then measures the corrosion film the room grew on them, in angstroms, a unit so small you are counting layers of atoms.
The strips are called reactivity coupons, and the classification they feed is the one your hardware vendors care about. Under 300 angstroms of copper film per month, the room earns a G1, mild, an environment where corrosion is not a meaningful factor in electronics reliability. Above that, the grades walk down through G2 and G3 toward GX, and somewhere along that walk your equipment warranties start developing opinions.
Here is the operational punchline: a room fails this test invisibly. A data hall drifting from G1 to G2 looks identical on a walkthrough. Only the coupon knows.
The short version. The coupon converts everything wrong with the air into one number.
- Bare metal, 30 days, film thickness measured in angstroms.
- Under 300 per month on copper is G1, the class hardware makers assume.
- The eye plays no role. The measurement replaced it on purpose.
What is actually corroding the metal?
Two contamination families, working together. Gaseous contaminants, sulfur compounds and their relatives from outside air, nearby industry, or even materials inside the room, attack the metal directly. And particulate: dust is not inert fluff at this scale. It carries ionic and sulfur-bearing species, and fine particles are hygroscopic, pulling moisture out of otherwise safe air and turning a settled dust layer into a slow electrochemical event on whatever it landed on. The industry started taking this far more seriously when circuit boards moved to lead-free finishes and turned out to be measurably less forgiving of dirty air.
The coupon does not distinguish the sources. It just adds up the damage. Which is exactly why it is useful, and exactly why every party who touches the room, HVAC, filtration, and yes, the cleaning crew, shows up in its number.
Where does cleaning fit in an angstrom problem?
Cleaning owns the particulate half of the ledger. Every gram of dust that sits on a raised floor, in a cable tray, or on top of a rack is inventory: it will be re-entrained by airflow, pulled through equipment, and given its chance to settle somewhere consequential. A cleaning program that removes it with HEPA-filtered capture takes that inventory out of the room. A cleaning program with a regular vacuum and a dry mop redistributes it, finer and more airborne than before. In a room graded in angstroms, bad cleaning is not neutral. It is a contamination source with a schedule.
This is the same discipline as the zinc whisker problem, which lives one shelf over: mechanical work done wrong under a raised floor liberates exactly the particles the room cannot tolerate. Method is the product. In this vertical, the cloth and the vacuum are precision instruments or they are hazards, with very little in between.
The operator's takeaway
If you run or manage a data hall and you have never seen a coupon report, that is the first fix; monitoring is cheap against the downtime it predicts. If you have coupon data, put your cleaning program on the same page as it: cleaning events logged against the monitoring periods, methods specified in writing, HEPA capture named explicitly, and the same verification mindset we bring everywhere else. We are candid about boundaries in every industry we serve, so here too: an ATP reading speaks to organic residue on a surface, which is the right tool in a break room or office suite; inside the white space, the coupon and the particle count are the referees, and the cleaning crew's job is to be invisible in their data.
It is the most honest version of a theme that runs through everything we publish: the rooms that matter most have stopped asking how clean things look. They hang a piece of metal in the air and let it testify. Under 300 angstroms, the room is holding. No opinion required.