Series 01 · EP 01 / 09 Why It Failed Updated 2026-11-05 8 min read

Why overmolding passes at T1 and delaminates after autoclave

A 2K part that passes at T1 and peels after autoclave failed at the substrate edge, not at the bond face — steam found a path the dry peel test never opened. Peel between 3.5 and 8 N/mm on a dry sample is a process check, not a service-life number.

In short

The autoclave is the first test that puts heat, steam and pressure on the bond at once.
Peel reads 3.5–8 N/mm dry; conditioning changes the surface, not the gauge.
We do not bond untreated PBT or PA66, and no T1 peel predicts steam life.

The T1 peel is measured on a dry part

First article inspection answers one question: can we make this part to the drawing, today, on this tool. The bond coupon that usually travels with it is peeled within hours of demolding, on a bench, at room temperature. That coupon is worth having. It proves the primer went on, the platens reached temperature, the cavity filled. It says nothing about what the same interface does when it meets 134 °C and saturated steam for the first time — and that is the only condition the autoclave cares about.

What a T1 peel measuresWhat it does not

Whether a bond formed at allWhether that bond survives a steam cycle

Primer coverage on a clean, dry substrateWhat the primer does after it hydrolyses

Force at the moment of testing, 3.5–8 N/mmForce after sterilisation, on the same joint

Settings that ran on that shiftMoisture the housing took up in storage

Two rows are process data and two are service data. A moulder who reports only the first two has not lied to you; he has answered a different question. The fix is administrative before it is chemical: put the conditioning step on the drawing, so the bond test cannot be run on an unconditioned part.

When the failure does come, look at where it starts. It is almost always the substrate edge rather than the middle of the bond. Silicone moves far more than PBT or PA66 across −60 to 230 °C, and every cycle leaves a small shear at that boundary. A dry peel never loads the edge. Steam reaches it first.

Where the water actually gets in

There are three paths, and they produce three different signatures. We sort them in this order because the first one is the cheapest to rule out.

SymptomMechanismFirst action

Peel collapses after steam, starting at the edgeWater enters the cut edge of the bond line and travels inwardDry the housing, then repeat the steam cycle on ten parts

Blisters under the silicone, spread across the faceMoisture absorbed by the substrate flashes at 134 °CAdd a drying step the resin tolerates; verify on your part

Clean peel at T1, nothing left after one cyclePrimer hydrolysed; the coupling agent is no longer thereChange primer chemistry, not primer thickness

Bond holds, the plastic tears insteadSurface energy is adequate; failure moved into the substrateStop tuning the bond; check wall section and gate position

Surface energy is the number worth putting on the drawing. Silicone needs around 42 mN/m to wet a substrate properly, and most untreated engineering plastics sit well below that line. Measure it on incoming housings with a dyne pen. It takes a minute per part and it settles arguments that a peel test only prolongs.

Blisters are the second signature and the most misread. A blister is not a bonding problem; it is a gas problem. The water was already inside the plastic before the silicone arrived, and 134 °C turned it into steam faster than the bond could let it out. Drying the housing before molding is the only gate that removes this failure, and the drying window depends on the resin and the wall — a typical range, verified on your part, not a number we can promise from a catalogue.

Three gates that survive a steam cycle

The gates are sequential. Skipping one does not halve the result; it restores the failure you were trying to remove.

Gate one: dry the substrate before molding. Ask what the housing absorbed between the injection press and the overmolding cell. If nobody knows, that is the answer. A drying step the resin tolerates — hours, not days — removes the blister path. Typical range; verify on your part, because a glass-filled grade and an unfilled one do not dry at the same rate.

Gate two: prove surface energy on the parts you received. 42 mN/m on a test plaque means nothing if the production housing was demolded with a release agent or handled bare-handed. Measure the housing. Record it per lot. This is the cheapest control in the whole process and the one most often replaced by assumption.

Gate three: move the bond test after conditioning. A bond test before the autoclave measures the process; a bond test after it measures the part. Run both, report both, and write the second one on the drawing with the cycle it was run at. Once the conditioned number is on the drawing, a delamination argument stops being a matter of opinion.

None of this needs a new tool. It needs the drying step scheduled, the dyne pen used, and the test order fixed. Those three changes are usually enough to move a part from “passed T1” to “passed the cycle it was bought for.”

Where this stops

We do not overmold onto untreated PBT, PA66 or glass-filled nylon, and we do not quote a steam-autoclave lifetime from a peel number. If a housing cannot be dried before molding, no primer saves the edge — move the joint to a mechanical seal or change the substrate before the tool is cut.

What to do next

Measure three incoming housings with a dyne pen before the next run. 42 mN/m is a line to sit above, not an average to hit.
Steam ten parts at 134 °C after the drying step, then peel them. Ten parts settle the question more cheaply than any tool change.
Write the conditioning step on the drawing. We quote a per-lot bond test when the cycle is specified, because that is the number the lot is judged on.