Two numbers govern any purity measurement, and they are easy to confuse. The compliance limit is the maximum amount of an impurity a specification allows. The detection limit is the smallest amount an instrument can reliably distinguish from zero. They are not the same thing, and the distance between them the margin, decides how much confidence you can place in a result. For a safety-critical impurity such as benzene in beverage-grade CO₂, that margin is the whole game.
Two numbers, two meanings
A compliance limit comes from the specification. In beverage-grade CO₂, the ISBT guidelines cap aromatic hydrocarbons, measured as benzene, at 20 parts per billion. It is a line drawn by the standard, and the product must sit beneath it.
A detection limit comes from the instrument. It is the lowest concentration the analyzer can measure with confidence that the signal is real and not noise. If a system's detection limit for benzene is 20 ppb, it can only just tell you whether benzene is present at the compliance limit; it cannot tell you how far below the limit you actually are. If the detection limit is 0.5 ppb, the same system can place the result with confidence well inside the safe zone and quantify exactly how much margin remains.
Why detecting at the limit is not proving compliance
Every measurement carries uncertainty. Near an instrument's detection limit, that uncertainty is at its largest — the signal is faint, and the analyzer is working at the edge of what it can resolve. A reading taken there is the least trustworthy reading the instrument can produce. So a system whose detection limit coincides with the compliance limit is being asked to make its most important judgment at the precise point where it is least reliable.
Put plainly: detecting an impurity at the compliance limit tells you that you are near the line, not that you are safely under it. Proving compliance requires resolving the impurity comfortably below the limit, where the measurement is stable and the margin is visible. The further the detection limit sits beneath the compliance limit, the more defensible every pass becomes.
Margin as the unit of trust
This is why margin is worth thinking about as the real output of a monitoring program, rather than a simple pass or fail. A benzene result of "below 20 ppb" is binary and brittle. A benzene result of "0.5 ppb against a 20 ppb limit" is quantified and robust; it carries roughly a 40-fold margin, and that margin is what holds up when a customer audits the batch, when a regulator asks for evidence, or when the feedstock changes and the process drifts. Margin absorbs the ordinary variation that every real production stream produces. Without it, a single bad day puts you over the line with no warning; with it, drift shows up as a rising number long before it becomes a failure.
One number, or the full picture?
Margin answers "how far below the limit am I?" Two more questions matter just as much: is the right technique measuring each analyte, and do you know exactly what you are looking at? A broadband method can miss or under-resolve a trace aromatic, where a method built for it — a gas chromatograph with photo-ionization detection — separates and quantifies each one. The strongest programs use the technique best suited to each analyte across the specification, rather than asking a single detector to do everything.
Speciation closes the loop. Resolving each compound in a group such as BTEX — benzene, toluene, ethylbenzene and xylene — individually, rather than reporting a single combined total, tells you which contaminant is present and where it likely came from. When a number starts to drift, knowing the species, not just the sum, is what lets you act before it becomes a failure.
What to ask of any CO₂ monitoring system
Three questions separate a system you can defend from one you merely hope is right. First, what is the per-analyte detection limit for each safety-critical impurity, benzene above all — and how far below the compliance limit does it sit? Second, is each compound resolved individually — speciated — or reported as a single combined figure? Third, is the evidence captured in an auditable record fit for batch release? A specification sheet that quotes a vague "ppb-level" sensitivity answers none of these; a per-analyte detection limit with documented margin answers all three.
The practical takeaway
A compliance limit tells you where the line is. A detection limit tells you how clearly you can see your distance from it. The two are only as useful as the gap between them. For benzene in beverage-grade CO₂ — regulated, carcinogenic, and capped at 20 ppb — that gap is the difference between a result that survives scrutiny and one that simply hopes to. Margin is not a technicality; it is the assurance.
Learn more: AirBreather resolves benzene to a 0.5 ppb detection limit — around 40× below the ISBT limit — on a dedicated GC-PID that speciates the full BTEX group → https://astg.com/pages/beverage-co2-benzene-monitoring
For the full specification context, see our guide to the ISBT CO₂ quality guidelines
FAQ
- What is the difference between a detection limit and a compliance limit? The compliance limit is the maximum an impurity is allowed by the specification; the detection limit is the smallest amount the instrument can reliably measure. Confidence lives in the gap between them.
- Why isn't detecting at the limit good enough? Measurement uncertainty is largest near the detection limit, so a reading taken there is the least reliable. Proving compliance needs margin below the limit.
- What is a good benzene detection limit for beverage CO₂? Well below the 20 ppb ISBT limit. A 0.5 ppb detection limit gives roughly 40× margin and absorbs normal process variation.