Metal Ceramic Packages: Hermeticity Testing Methods That Prevent Field Failures

Hermeticity is the single biggest lever for reliability in metal-ceramic packages. Even trace moisture or reactive gases can corrode leads, drift optics, and poison die attach—leading to intermittent faults that only appear after shipment. The right test stack prevents those surprises.

What “hermetic” really means for metal–ceramic

Metal–ceramic packages (e.g., Kovar®/alumina) are built from brazed lid seams, ceramic feedthroughs, and plated external metallurgy. Hermeticity is the measure of gas tightness across all those joints. It’s not binary: acceptable leak limits depend on cavity volume, internal getters, expected life, and environmental exposure. Practically, your test plan should convert “hermetic enough” into quantifiable limits that correlate to years of dryness at the die surface.

Start with gross leak screening

Gross leak screens catch obvious seal defects fast and cheaply before sensitive tests:

Bubble emission (“dunk”) tests: Pressurize the part in an inert medium, then immerse it in a liquid and watch for bubbles upon pressure release. Variants include pressure-vacuum and differential temperature methods. They detect leaks roughly ≥10⁻⁵ atm·cc/s.

Fluorocarbon or isopropanol methods: Condition the part in a halogenated fluid, then move it to a hot bath; escaping vapor flags a leak. These are simple and high-throughput for production.

Best practices

Control soak and dwell times, liquid temperatures, and pressure profiles; they directly affect sensitivity.

After fluorocarbon exposure, allow adequate outgassing to avoid false “pass” signals in downstream fine-leak tests.

Fine leak testing with helium mass spectrometry

Fine leak testing finds the micro-paths that actually admit moisture over months or years.

• Bombing method (transfer): Parts are pressurized in helium (e.g., 3–5 atm for a defined time), then moved to a mass spectrometer “sniffer” or vacuum chamber to measure helium escaping from the cavity.

• Hood or vacuum-chamber method: Parts already containing helium (from backfill) are placed under vacuum and monitored for helium egress, avoiding the logistics of bombing.

The output is the equivalent standard leak rate (L) in atm·cc/s of helium. Acceptance limits commonly fall between 10⁻⁸ and 10⁻¹⁰ atm·cc/s for microelectronic packages, but the right threshold should be engineered from your cavity volume and moisture budget (see next section). 

Key parameters to control are bomb pressure, soak time, dwell before measurement, and test duration; all affect how much helium diffuses into narrow cracks and thus the measured L.

Avoiding pitfalls

• Permeation confounds: Glassy or polymeric elements (if present) can allow helium permeation that looks like a leak. Use guardbanding, proper fixturing, and material-specific dwell times to discriminate real defects from permeation.

• Calibration discipline: Run calibrated leaks daily and at shift changes; drift in detector gain can mask marginal parts.

• Cleanliness: Residual oils can temporarily seal pinholes; pre-clean and bake before fine-leak to expose true paths.

Correlate leak rate to service life, not just a number

A leak number only matters if it links to how dry the cavity stays in the field. A practical approach:

• Define the moisture limit your device can tolerate (e.g., ≤5,000 ppmv H₂O to prevent aluminum corrosion or laser facet fogging).

• Model ingress using cavity free volume, external humidity/temperature profile, and leak conductance (from L).

• Set the acceptance limit so that, with margin, the cavity remains below the moisture limit over the target life (e.g., 10 years at 25–85 °C with intermittent 85/85 exposure).

This translation turns a test result into a lifetime guarantee rather than a check-box.

Residual Gas Analysis (RGA): verify the internal atmosphere

Where fine-leak measures the pathway, RGA measures what actually sits over your die. By puncturing or valving in the package under vacuum and sampling with a mass spectrometer, RGA quantifies H₂O, O₂, N₂, H₂, CO/CO₂, and process residues.

Why it matters:

• Moisture control: Confirm the internal water vapor is below the specified limit at shipment.

• Process diagnostics: Excess H₂ often indicates getter activation; unexpected hydrocarbons point to contaminated braze or adhesives.

• Lot containment: If a heat-seal shift introduces flux residue, RGA spots it immediately, preventing field drift.

Use RGA as a periodic audit (e.g., per lot) or as a 100% screen for ultra-critical opto or high-power devices.

Specialty methods and when to use them

• Radioisotope leak testing (Kr-85): Extremely sensitive and quantitative, historically used for very small volumes. Today, it’s less common due to licensing and waste handling, but still valuable for niche aerospace/medical builds.

• Pressure decay / rate-of-rise: Useful for larger packages and lids, often at the subassembly stage, to catch seam defects before expensive components are installed.

• Dye penetrant or metallographic checks: Not hermetic tests per se, but effective for validating braze joint wetting, lid planarity, and ceramic integrity during process qualification.

Build quality in: process controls that improve test outcomes

Testing doesn’t create hermeticity; processes do. Controls that raise first-pass yield:

• Surface preparation: Tight control of ceramic metallization, plating thickness, and oxide removal prevents capillary pinholes.

• Brazing profile discipline: Ramp, soak, and peak must match filler alloy wetting windows; record with embedded thermocouples in worst-case mass parts.

• Lid planarity and clamping: Warped lids open corner leaks; use torque-controlled clamps and verify post-braze flatness.

• Getter strategy: Choose non-evaporable getters sized to cavity volume and activate at the correct temperature; verify with RGA.

• Helium backfill and seal: If using hood testing or long-term correlation to helium partial pressure, maintain stable backfill recipes and verify fill valves for seat leakage.

• Cleanroom handling: Fingerprints and silicone volatiles create pseudo-seals; enforce gloves-on, solvent-clean, and post-bake rules.