A fixed gas detector and a corrosion coupon are both measuring correctly, and they are not measuring the same thing. The detector watches an occupational exposure limit expressed in ppm. Copper responds to concentrations in ppb — roughly three orders of magnitude lower. An electrical cabinet can therefore be corroding while every gas reading stays comfortably within the limit.
If you want to know whether your environment is aggressive, the coupon is the right instrument. If you want to know when and why, you need a measurement with a time axis. Those are two different questions, and neither instrument answers the other one's.
Why this comes up in an industrial environment
The pattern is familiar. Unexplained failures on boards, in cabinets or at contacts. Components giving out after three years where ten were expected. A discolouration on printed circuit boards that nobody can quite place. And throughout all of it, a gas detection system that has never once alarmed.
The reflex is to doubt the detector. That is usually unfair. The instrument on the wall was built for something else.
This is sharpest wherever hydrogen sulphide is present as a process gas or as a decomposition product: wastewater treatment and sludge handling, pulp and paper, manure processing, biogas, sewerage infrastructure, parts of the chemical industry. But the switchgear room itself is rarely where the gas originates. It sits next to it, or above it, or at the end of a cable duct.
The mechanism: why copper responds on a different scale
Under normal conditions copper carries a thin oxide layer (Cu₂O). Hydrogen sulphide adsorbs onto it dissociatively, and the only bulk corrosion product of consequence is copper sulphide (Cu₂S). It nucleates, spreads laterally, coalesces into a closed film, after which growth becomes diffusion-limited through the sulphide layer that has by then built up.
Three properties of that process explain why a gas detector structurally cannot see it.
It is cumulative. In the gas-supply-limited regime, film thickness scales linearly with accumulated exposure — ppm·hours — rather than with the peak value. A detector is built to catch a peak. A cumulative process does not need one.
It is strongly humidity-dependent. At around 5 % relative humidity sulphidation is close to negligible; at 95 % it proceeds substantially faster. The acceleration correlates with the number of adsorbed water monolayers on the surface, rising from roughly three layers upward. The same concentration does something quite different in a cool, damp room than in a warm, dry one.
It migrates. This is the property that costs the most in practice. Solid copper sulphide does not stay where it forms. It travels across the surface of a printed circuit board without requiring any electric field, and eventually bridges traces that are not meant to be connected. The literature calls this creep corrosion. It is not a cosmetic problem and it is not tarnish — it is a short circuit. Lead-free boards proved more susceptible than the older leaded assemblies, which is why the subject returned to the agenda after the RoHS transition.
A gas detector set to an occupational exposure limit sees precisely none of these three properties.
Three instruments, three questions
This is where the misunderstanding sits. Three measurement methods circulate around this problem, and they are continually compared as though they were competitors. They answer different questions.
1. The fixed gas detector — "is this dangerous for the people here?"
At European level, the indicative occupational exposure limit value for hydrogen sulphide is 7 mg/m³ (5 ppm) as an eight-hour time-weighted average, with a short-term limit value of 14 mg/m³ (10 ppm) over fifteen minutes. These were set by Commission Directive 2009/161/EU, the third list of indicative values under Council Directive 98/24/EC.
One point matters for anyone reading this outside a single jurisdiction. An indicative value is not itself the number that applies to you. Member States establish national occupational exposure limit values "taking into account" the Union values, and national limits can be stricter. In Belgium, for example, the Codex on Well-being at Work sets hydrogen sulphide at 5 ppm as an eight-hour average with a fifteen-minute value of 10 ppm — the same figures. Elsewhere they differ. Check the limit that applies in the country where the installation stands; do not work from the EU indicative value alone.
What none of these values are is a statement about equipment. An occupational exposure limit is not a safe threshold for copper, and it was never intended to be one.
2. The corrosion coupon under ISA 71.04 — "how aggressive is this environment?"
The reference method for classifying an environment is ANSI/ISA-71.04-2013, still the current edition. You expose a copper and a silver coupon, typically for thirty days. A laboratory then determines by coulometric reduction how much metal has been converted — expressed in ångströms per thirty days — and, this being the strongest part of the method, which corrosion product it is. Sulphide behaves differently from chloride. You get a number and a causal family.
The severity levels:
| Level | Copper (Å/30 d) | Silver (Å/30 d) | Description |
|---|---|---|---|
| G1 | < 300 | < 200 | Mild |
| G2 | < 1,000 | < 1,000 | Moderate |
| G3 | < 2,000 | < 2,000 | Harsh |
| GX | ≥ 2,000 | ≥ 2,000 | Severe |
Silver was added as a required metric in the 2013 edition, precisely because copper alone cannot separate the causal families sharply enough.
3. Continuous concentration measurement — "when, where and driven by what?"
A continuous measurement of gas concentration with a time axis gives you no metal loss and no classification. It gives you a course of events: at which hours, in which room, during which kinds of moments the concentration rises, and how that relates to temperature and humidity in that same space.
It is the only one of the three you can base an intervention on.
About those ppb figures you may have been shown
Here a clarification is needed that is missing from almost all commercial communication on this subject, and that matters if you are assessing a quotation or a piece of advice.
ANSI/ISA-71.04-2013 classifies on measured metal loss, not on gas concentration. In the current edition the discussion of corrosive contaminants was moved out of the normative part of the standard and into an appendix. The ppb values you encounter are inherited from the 1985 edition.
That has a visible consequence. The concentration table circulates today in at least two mutually incompatible versions. The variant used by the specialist air filtration firms puts G1 below roughly 3 ppb H₂S and GX from roughly 50 ppb. Another published version, in a corporate engineering whitepaper, places the same classes at 125, 2,000 and 10,000 ppb. That is a factor of about forty, and both tables are passed on in good faith.
We have been unable to check either against the standard text itself — the preview pages for both the 1985 and 2013 editions keep the tables behind the paywall. What is established is why two versions exist: the concentrations are no longer normative, and a figure nobody has to comply with stops being corrected.
What to take from this: do not ask for a ppb target and do not design to one. The standard does not classify on it. If a supplier presents a ppb threshold to you as a requirement of the standard, that is an appendix from 1985, not a requirement.
And to make the scale concrete: even the lower of those two tables places the severest equipment class at around 50 ppb. The European indicative eight-hour limit of 5 ppm is 5,000 ppb. The air in which your electronics reportedly cannot survive sits at roughly one percent of the occupational exposure limit. Your detector is staying silent entirely correctly.
One detail makes this tangible for anyone working in the space. The odour threshold for hydrogen sulphide is reported in the literature at somewhere between 8 and 130 ppb, with some sources placing it lower still. That range runs straight through the severest equipment class. People can often smell the air that is attacking their switchgear, while every gas measurement remains well within the limit.
What a thirty-day coupon does and does not resolve
A coupon is the reference method, and it remains so. This is not a criticism of the instrument. It is what a cumulative measurement is.
Work it through. A month of quiet, low exposure containing one heavy six-hour excursion can produce the same accumulated dose as a month of uniformly elevated background. So the same coupon, the same class, the same advice. That is not a shortcoming of the laboratory. That is what addition does.
Then there is turnaround. The number comes back weeks later. By then the month is over, and with the month everything that happened in it: the breakdown, the shutdown, the work on the adjacent installation, the wind that sat differently for two days, the door that stood open for three weeks during an overhaul.
You now know your environment was aggressive. You do not know when, and you do not know what drove it. A number without a moment attached is not something you can act on.
That is precisely where a continuous measurement adds something, and precisely the limit of what it can do. It measures concentration, not metal loss. It is not a classification and it does not replace the coupon. What it does is give the coupon's number a time axis, so the question shifts from "is it bad here" to "what happens here on a Tuesday night".
And because sulphidation is humidity-dependent, the combination is more informative than the gas measurement alone. The room with the highest H₂S concentration is not automatically the room with the most corrosion. A warm, very dry technical room with a lot of gas and a cool, damp room with less gas can be each other's inverse. Which of the two carries the higher risk is a real question that only the combination of parameters answers — and that a single-parameter measurement structurally cannot.
What to do with this
Four steps, in this order.
1. Reconstruct one coupon period. If a coupon has ever come back at anything other than mild, try to establish what happened during those thirty days. Maintenance, breakdowns, process changes, construction work, weather. If you cannot, you have your answer to whether one number a month is enough for your situation.
2. Establish what scale has ever been measured. Not: does the detector alarm. Rather: has anyone ever measured on the scale copper responds to, or ever hung a coupon? In many installations the answer is no. That is not negligence — the instrument on the wall was built for something else.
3. Measure humidity and temperature in the same room. Without relative humidity, a gas concentration in this context is half the story. It is also the cheapest parameter on the list.
4. Build your investment case on your own numbers. Failure frequency, replacement cost, downtime hours, failed components per board. A corrosion case resting on a supplier's figures will not survive an investment committee. One resting on your own failure history usually will.
What this is not about, and where you do not need it
This is not about your people. At the concentrations copper responds to, there is no health issue. The exposure limits in this article are here to show the difference in scale, not to imply a risk. Where an exposure question genuinely does arise, it is a separate matter with its own instruments: the statutory measurement and the advice of the occupational health and prevention service. Continuous monitoring complements those and does not replace them. (For completeness, because it belongs in the same breath: above roughly 100 ppm the odour warning from H₂S disappears entirely. That is a safety fact, and it stands apart from everything said above about equipment.)
Measuring does not prevent corrosion. A measurement produces evidence. People and interventions — sealing, pressure regime, filtration, relocating a cabinet, a revised maintenance interval — prevent corrosion. Anyone selling you a measurement as the solution to the corrosion problem itself has skipped a step.
And the most likely answer for most buildings: you do not need this. If you have no failure history and no suspicion of a source, one coupon a year is probably all you will ever need. Not every building has a problem, and a continuous measurement installation in an environment with no indications produces a handsome graph and nothing else.
The question is not whether continuous monitoring beats a coupon. The question is whether you want to classify something or solve something.
InsightAir provides continuous air quality monitoring in critical environments. What happens between two point measurements is usually the most interesting part.
