Guides / Design
Joining Ceramic to Metal Without Cracking It on the First Cycle
July 6, 2026
A ceramic part that has to be sealed, mounted or electrically connected usually ends up joined to metal. That joint is where most ceramic assemblies fail, and almost never because the ceramic was weak.
Two routes to a joinable surface
Moly-manganese metallising is the classic. A molybdenum and manganese paste is applied and fired in a wet hydrogen atmosphere at high temperature. The manganese reacts with the glassy grain boundary phase in the alumina, which migrates into the metallised layer and locks it mechanically. A nickel plate goes on top so that a standard silver copper braze will wet it.
The important consequence: it needs a glassy phase to work with. That is why 94 to 96% alumina is the traditional metallising grade and why very high purity alumina is harder to metallise without adding a glass frit. If your part is a brazed feedthrough with no plasma or high voltage requirement, this argues against over specifying purity. See which alumina purity do you actually need.
Active metal brazing skips metallising. The filler alloy itself contains a reactive element, usually titanium, which reduces the ceramic surface and wets it directly in a vacuum furnace. It works on high purity alumina, aluminium nitride, zirconia and silicon nitride, and it is a single step rather than three. It costs more per joint and the process window is narrower.
The real design problem is expansion
Every ceramic to metal joint is assembled hot and used cold. The metal contracts more on cooling, so the joint ends up stressed, and ceramic is weak in tension.
Three levers control it:
- Pick a metal with matched expansion. Kovar sits near 5 ppm/K, close to alumina at 7 to 8. Molybdenum and tungsten are even closer. Stainless steel at 16 to 17 ppm/K is a long way away.
- Use a ductile interlayer. A copper or nickel interlayer yields and absorbs the mismatch instead of passing it to the ceramic.
- Choose geometry that puts ceramic in compression. A metal sleeve shrinking onto a ceramic cylinder is a far happier joint than a metal flange pulling on a ceramic face.
Assemblies that survive thermal cycling almost always use at least two of the three.
Hermetic feedthroughs
The common application is a feedthrough: a ceramic insulator carrying a conductor through a metal wall, sealed to vacuum or pressure.
Things worth specifying:
- Leak rate requirement, usually in mbar l/s or atm cc/s, since that decides the process and the inspection
- Operating temperature range and number of cycles
- Voltage and creepage distance, which sets the insulator geometry
- Whether the assembly will be bakeable and at what temperature
- Cleanliness class, if the part goes into a vacuum system or a medical device
What to put on the drawing
- Which surfaces are metallised, with a boundary dimension and a tolerance on it
- Plating type and thickness
- Braze alloy, or the requirement the braze must meet
- Surface finish in the metallised area, since a rough surface changes adhesion
- The leak rate and cycle requirement
Practical notes that save assemblies
- Metallised areas need a ground surface, not an as-fired one, when the joint is critical
- Keep the metallised band away from sharp edges, which concentrate stress and chip
- A radius under the joint is worth more than extra material around it
- Say if the part will later be autoclaved or steam sterilised, especially for zirconia. See zirconia ageing
Brazing shops and grinding shops are not always the same business. Say on the RFQ whether you need the ceramic only, or the finished assembly, so we can match a shop that either does both or has a brazing partner already.
Send your drawing and we match it with up to 3 shops. Reply within 24 business hours.
Frequently asked questions
What is moly-manganese metallising?
A molybdenum and manganese paste is fired onto the ceramic in a wet hydrogen furnace, where it bonds through the glassy phase at the grain boundaries. It is then nickel plated so a conventional braze alloy will wet it.
When is active metal brazing used instead?
On high purity aluminas, aluminium nitride, zirconia and silicon nitride, where there is little or no glassy phase for moly-manganese to bond to. The filler alloy contains titanium, which reacts directly with the ceramic.
Why do ceramic to metal joints crack?
Almost always expansion mismatch. The joint is made hot and cooled, so the metal shrinks more than the ceramic and puts the joint into stress. Managing that with the right alloy, geometry and a ductile interlayer is the whole design problem.
Need a quote for this part?
Send the drawing. We match you with up to 3 shops that grind this grade and reply within 24 business hours.
Get matched with a ceramic shop