Helium test detectability

A mass spectrometer leak detector is calibrated in helium. Its quoted minimum detectable leak rate — say 1×10⁻⁷ mbar·L/s in sniffer mode — assumes the leak is emitting pure helium at the reference pressure difference, normally 1 bar. On real plant the system is rarely charged with pure helium at 1 bar: it is more often a 5–20% helium-in-nitrogen mix at several bar. What the detector sees is only the helium fraction of the total flow, so the smallest real leak that will trip the instrument is the instrument's own floor divided by the helium partial pressure expressed as a multiple of the reference differential.

Helium partial pressure is the concentration times the absolute pressure: 10% helium at 3 bar absolute is 0.3 bar of helium, three tenths of the reference differential, so the detectable leak is 1/0.3 = 3.3 times worse than the datasheet figure. Raising the test pressure buys sensitivity in the same proportion as raising the concentration, which is often the cheaper option when the item can take the pressure.

Run the calculation the other way to size the charge: the minimum helium concentration needed to prove a given acceptance limit is the instrument floor divided by the acceptance limit, scaled by the reference and test pressures. If that concentration is not achievable, the test cannot demonstrate the criterion however carefully it is performed, and the procedure has to change — a hood or vacuum-envelope technique, a lower instrument floor, or a different method entirely.

Two practical limits sit under the arithmetic. Atmospheric air already contains about 5.2 ppm of helium, and a poorly ventilated space with a recent helium spill contains far more, so the working background is usually well above the instrument's stated floor. And the response of a sniffer depends on probe standoff, scan speed and dwell — a demonstrated response to a calibrated leak on the actual item is worth more than any of these numbers.

p_He = concentration × absolute test pressure
sensitivity scale = p_He / reference differential
Q_min = detector MDL / scale
minimum concentration = 100 × MDL × p_ref / (Q_allow × p_test)