Rendering: a dim lab leak-test station with a coupling sealed inside a small chamber on the bench and a mass-spectrometer unit beside it on a vacuum hose

Helium · sccm · bands

Ask for the band, not the verdict

A leak claim without a unit is not a claim — it is an adjective.

Sensitivity reachable · industry-published benchmark 1×10⁻⁶ sccm, to 1×10⁻⁹ at best

High-precision acceptance band · industry-published benchmark 10⁻⁶–10⁻⁴ sccm

Standard acceptance band · industry-published benchmark 10⁻⁴–10⁻² sccm

Why helium

It is the smallest tracer gas in practical use for this work: the paths it can pass through, other gases can pass through too. A test with it therefore speaks about a path that exists, not about a molecule that happens to be small.

Why not something else

It is inert, non-flammable and does not condense at the temperatures a lab works at, so a tested part does not have to be flushed or dried before it goes back into a build.

What a bubble test says

Mass spectrometry resolves rates far below what a visible bubble can indicate. A visible bubble and a measurable leak rate are not the same measurement, and one cannot be converted into the other without a method statement.

Every band and sensitivity figure on this page is an industry figure, not a whole-unit leak-rate promise from this site.

The unit will not convert itself

sccm · atm·cc/sec · the 590 factor

Rendering: a macro of a leak detector inlet with a capillary line running to a small sealed chamber holding one coupling, and a calibrated leak standard beside it — Where the number is measured inlet · capillary · reference

Anchor

1 sccm ≈ 1.69×10⁻³ atm·cc/seca unit definition, so it carries no source level of its own. It is the whole conversion, and the reason two datasheets can describe the same leak with numbers that look nothing alike.

Band endpoints, converted

10⁻⁶ sccm = 1.69×10⁻⁹ atm·cc/sec · 10⁻⁴ sccm = 1.69×10⁻⁷ · 10⁻² sccm = 1.69×10⁻⁵ · the 1×10⁻⁹ reach = 1.69×10⁻¹². derived by this note from the anchor

Why it matters

The same leak written in the two units differs by a factor of about 590 (1 ÷ 1.69×10⁻³). Convert before you compare two statements, or compare nothing. derived by this note

Upper scale10⁻⁹ → 10⁻² sccm Lower scale1.69×10⁻¹² → 1.69×10⁻⁵ atm·cc/sec Markedone graduation, both labels

Put the unit in the purchase order. Without it, “low leak rate” is three words, not an acceptance criterion.

Where the line gets drawn

High precision · standard · choose by consequence

High-precision band

10⁻⁶–10⁻⁴ sccmthe tighter acceptance band, an industry figure rather than a specification requirement. industry-published benchmark

Standard band

10⁻⁴–10⁻² sccmthe looser of the two published bands, and still an order of magnitude below anything a bubble test can resolve. industry-published benchmark

Choose by consequence

Pick the band from what a leak does in that loop, not from the part's price. Where the consequence of a leak is not recoverable — a fluid that damages what it touches, a service window that cannot be reopened — take the tighter band.

Write it down

The band belongs in the acceptance clause, and the drawing or spec sheet states the method that produced it. A band with no method attached cannot be checked by a second party.

Left bracket10⁻⁶–10⁻⁴ sccm, high precision Right bracket10⁻⁴–10⁻² sccm, standard Marked endthe tight end of the tight band

Rendering: a coupling sealed in a bell jar beside a calibrated leak standard and a vacuum hose, no readable numerals anywhere in frame — A band is only as good as its reference bell jar · calibrated leak

Both bands span two orders of magnitude, so “inside the band” is not one answer — it is a range you have to place a number inside. That is why the calibration chain matters more than the headline sensitivity: a detector reading against an expired reference is reporting a number with an unknown meaning. derived by this note

Two orders of magnitude is the width of the band. Do not read a band as a single pass mark.

Four questions for a leak claim

Method · unit · sampling · calibration

Method?

Helium mass spectrometry, pressure decay or immersion bubble test — three methods whose verdicts are not interchangeable, and whose sensitivities are orders apart.

Unit and band?

Ask for an interval in sccm rather than the words “low leak rate”. Convert both statements to one unit before you compare them.

Per piece or sampled?

Testing every unit and testing a sample are two different supply commitments with two different failure modes. Whichever one is meant, write it into the order.

Calibration?

The validity of the instrument and of the reference leak it is checked against. A sensitivity figure is a statement about a calibrated instrument, not about a model number.

A detectable sensitivity of 10⁻⁶ sccm sits four orders of magnitude below the top of the standard band. The band your part is accepted against is a purchasing decision, not a property of the instrument.

The four-order comparison is this note's own subtraction of two published figures; it says nothing about what any particular unit is tested to.

The hardware behind the four questions

Different methods · different benches

Rendering: five test-bench units presented as one composite sheet, two across the top and three below — ① Method — the bench line-up five units · one sheet

Rendering: a coupling body clamped between two fixture plates, its seal face turned toward the camera — ③ Sampling — one part, clamped fixture plates · seal face forward

Rendering: a coupling clamped in a pressure fixture with a coiled high-pressure line running out of frame, gauges rendered unreadable — ② Method — pressure, not helium pressure fixture · coiled line

Rendering: a pressure test rig with a fixture head and a manifold block, its gauges rendered unreadable — ④ Calibration — the rig it is checked on fixture head · manifold block

A bench is not a band. Which bench a part crosses says which method was used; only the acceptance clause says what number the part had to meet.

Leak specification

Ask for the band, not the verdict.

Send the part number. The reply gives the method and the band for that part number, with the quote rather than as a file to chase.

Request the leak spec One business day · English quote + a one-page spec sheet

Send the part number

Reply carries method + unit + band

Scoped by part number

Turnaround 1 business day

A blue-panelled cabinet with white framing, its upper doors carrying mesh-backed viewing windows — Craft note · seals and grooves before the test cabinet unit

The seal groove and the sealing face are produced to the drawing revision the order was placed against, and the part is cleaned before it reaches any chamber. What a leak test measures is the result of that work — a test cannot add tightness to a groove, it can only report what the groove already is.

Published 2026-09-24 · when a figure in this note changes, the note is rewritten and the change is dated here. Bands and sensitivities quoted above are industry-published figures; the conversions and the four-order comparison are this note's own arithmetic.