MIL-STD-883 Method 1014: Fine and Gross Leak Testing Explained for Sealed RF Parts

A sealed module comes back from the test house with a helium reading of 8 × 10⁻⁸ atm·cc/s and one line of paperwork: "seal test per MIL-STD-883 Method 1014." Engineering wants a verdict, quality wants a signature, and both are staring at a standard that refuses to hand out a single number.

The practical answer first: Method 1014 tells you how to run a seal test, not what your part has to survive. The reject limit comes from the device drawing, from the device specification, or from the volume-based limits inside the method itself, and it shifts with the internal cavity volume of the package. Get that distinction wrong and you either scrap good hardware or ship a leaker with a clean-looking test report.

What Method 1014 Covers, and What It Leaves to You

MIL-STD-883 is the test-method document for microelectronic devices, and Method 1014 is its seal test. It has two halves that barely overlap. Fine leak testing uses a tracer gas, normally helium, to find leaks in the region of roughly 1 × 10⁻⁹ to 1 × 10⁻⁵ atm·cc/s. Gross leak testing picks up everything above about 1 × 10⁻⁵ atm·cc/s, where the leak is large enough to show as a bubble stream or to register as penetrant weight gain.

Fine leak testing is then split into fixed and flexible conditions. A fixed condition applies the same bomb pressure and dwell time to every part, which is reasonable for small cavities and misleading for large ones. The flexible condition lets the laboratory calculate pressure and dwell from the internal volume of the package, so a measured reading can be converted back into a true leak rate.

Every one of these conditions is a technique, not a limit. The limit is a separate decision, and it is the one that decides whether your reading of 8 × 10⁻⁸ is a pass or a rework ticket.

Fine Leak Before Gross Leak: the Sequence Is Not Optional

Method 1014 expects the fine leak test to run before the gross leak test, and the reason is physical. The fluorocarbon used for bubble testing is heavy, wets surfaces, and can sit inside a pinhole long enough to hide it from the mass spectrometer. Run the tests in the wrong order and a leaker can come back looking tight.

Helium testing itself is a two-step process. The part is pressurized in helium so the tracer is forced into any leak, then moved to a mass spectrometer that measures the helium escaping back out. What the detector sees is not the true leak rate. It is a measured rate that depends on how much helium went in, how long the part sat before it was read, and how quickly the gas escapes from the cavity.

Cavity volume is the variable that catches people out. A large cavity absorbs more helium, which pushes the measured rate up, while a tiny cavity gives a weaker signal and needs a tighter threshold to be meaningful. That is why a single blanket leak rate quoted in an email is worth very little.

Two parts with identical leaks can therefore produce different readings, and two parts with different leaks can produce the same reading. The flexible condition exists precisely to strip that ambiguity out of the data.

Where the Seal Test Meets RF Hardware

In RF hardware the hermetic barrier is rarely a ceramic lid. More often it is a glass-to-metal seal: a glass bead fused into a metal housing with a centre pin running through it, or a sealed connector body built around a glass insulator that carries the conductor through the wall without losing pressure integrity.

Those parts fail in predictable places. The glass-to-metal interface comes first, because the bond is mechanical and chemical at the same time, and any void, bubble or incomplete wetting becomes a leak path. Plating porosity is second: a pore that reaches the base metal creates a slow leak that a bubble test will never see. The third is the joint itself, where an incomplete solder or weld fillet leaves a gap that opens further under thermal cycling.

Cavity volume inside a coaxial feedthrough is small, often a fraction of a cubic centimetre, so the applicable fine leak limit sits at the tight end of the range. A void that would be cosmetic in a large hybrid package can push a small connector body straight out of specification.

RF Glass Insulator for Hermetic FeedthroughsGlass-sintered 50Ω RF insulator for compact hermetic feedthroughs, relevant where small cavity volume makes fine leak limits especially tight.View Product →

Running the Test Without Burning a Week

Most seal-test delays come from preparation rather than from the instrument. A few habits keep the schedule intact:

• Bake parts dry before bombing. Moisture inside the cavity competes with helium and produces readings that drift between runs.
• Check the ambient helium background before every batch. A cylinder change or a leaky fitting nearby will lift the baseline and turn good parts into failures.
• Keep the transfer time from bomb to detector short and identical for every sample. Method 1014's conversion assumes a known wait period; a long coffee break invalidates it.
• Verify the mass spectrometer against a calibrated helium leak of known value at the start and end of the shift.
• Handle parts with clean gloves and padded tweezers. A scratch across a glass bead is a new leak path that did not exist when the part was built.
• Record the condition letter, bomb pressure, dwell, wait time and instrument ID on the data sheet, not just the final reading.

That last point matters more than it looks. When a customer disputes a failure two years later, the condition letter and dwell time are usually the only evidence that shows whether the test and the specification actually agreed with each other.

What to Put on the Drawing Before You Order

Purchasing risk on hermetic parts is almost always specification risk. The supplier builds to the drawing, the test house tests to the purchase order, and if those two documents disagree the argument lands in your inbox. Write the following into the drawing and the order:

• The method and condition, for example "seal test per MIL-STD-883 Method 1014, condition A2".
• The reject limit in atm·cc/s, together with the cavity volume it was derived from.
• The test sequence, fine leak before gross leak, plus the sample size and whether it is lot-by-lot or qualification only.
• Any exclusions, such as parts that cannot be exposed to fluorocarbon because of an internal coating or an absorbed material.
• Whether the certificate must include raw readings, calibration data and the instrument background level.

Where the sealed interface is a blind-mate RF joint rather than a flat lid, the seal has to survive mating cycles on top of the pressure differential. That is a different design problem from lid sealing, and it belongs in the specification conversation before the first part is machined.

SMP-J Hermetically Sealed RF ConnectorGlass-to-metal sealed SMP-J connector for panel or tube-shell mounting, suited to blind-mate RF joints where sealing and repeated mating cycles both matter.View Product →

Failure Modes and What They Usually Mean

A failed seal test is a symptom, not a diagnosis. The table below maps the common signatures to the first place worth looking.

Frequently Asked Questions