Most neutral cure silicone sticks to clean glass and to mill or anodised aluminium with no primer at all. The only way to know about your metal is to put a bead on it and pull it off. The test costs ten minutes of work and a week of waiting, and it is the cheapest week on the job.
Quick answers
Does silicone sealant need a primer?
Usually not on clean glass or bare aluminium. It often does on porous substrates such as concrete, masonry and some stone, and on coated or painted metal — because there you are not bonding to the metal at all, you are bonding to the coating. Coatings vary between suppliers and between batches, which is exactly why the question cannot be answered from a data sheet.
How do I test adhesion on site?
ASTM C1521 sets out the field procedure. In short: run a bead on the real substrate, let it cure fully, cut a tail free and pull it back on itself. What matters is not whether it comes off — everything comes off if you pull hard enough — but where it breaks.
What is the difference between cohesive and adhesive failure?
Cohesive failure means the sealant tore through itself and left silicone on both faces. That is a pass. Adhesive failure means it peeled away clean and the substrate is bare. That is a fail, however good the bead looked before you pulled it.
Can I test it the next day?
No. A part-cured bead fails a pull test that the same bead passes once it is through. Give it the cure time on the technical data sheet — for a normal bead that is a week, not a day. Ask us for the sheet and we send it.
The test, step by step
- Use the real material. An offcut of the same alloy, the same coating, ideally the same delivery. Not something similar from the rack.
- Prepare it the way the crew will. If the wall gets a solvent wipe, the sample gets a solvent wipe. If the wall gets nothing, the sample gets nothing. You are testing your process, not an ideal one.
- Run a bead about six inches long, at the depth the joint will actually be.
- Let it cure fully. Same temperature and humidity as the site, as near as you can manage.
- Cut along both bond lines with a sharp blade, then cut across one end so a tail comes free.
- Grip the tail an inch back from the bond and pull it at a right angle to the surface, stretching it well past any movement the joint will ever see.
- Look at what is left on the substrate.
Reading the break
| What you see | What it is | What to do |
|---|---|---|
| Silicone left on both faces, torn through the middle of the bead | Cohesive failure | Pass. Go ahead. |
| Bead peels away clean, substrate bare and shiny | Adhesive failure | Stop. Primer, a different preparation, or a different substrate. |
| Holds in places, peels in others | Marginal | Treat as a fail. A facade does not average out. |
| The coating comes off with the silicone | Substrate failure | The sealant is fine. The coating is not bonded to the metal — that is a conversation with whoever supplied it. |
When a primer is the answer — and when it is not
A primer is a genuine fix for a porous substrate, and often for a difficult coating. It is not a fix for a dirty one. Priming over dust, site grime, a fingerprint or a release agent left from fabrication simply glues the dirt down harder, and the joint fails at the same place a few seasons later.
It is also not a fix for the wrong chemistry. Acetoxy silicone on coated aluminium will keep failing whatever you prime it with, because the acetic acid it gives off as it cures attacks the finish. Why neutral cure is the only answer on aluminium.
Four things that make a good sealant fail this test
- A fingerprint. Skin oil is a release agent. Handle the sample area the way you would a lens.
- Release agents from fabrication. Extrusions and panels often arrive with a film on them that nobody mentions. It wipes off — if you wipe.
- Testing a different alloy or a different coating batch. The result is then true about something you are not building.
- Three-sided adhesion in the real joint. The bead can pass this test on the bench and still tear itself apart on the wall, because it is stuck to the back of the joint as well as the two sides. That is what the backer rod is there to prevent. The five ways joints fail.
How to get the sample
Ask, and we send a sausage and the laboratory report before you order anything. Run it on your own offcut, pull it, and decide from the break rather than from our description of it. If it does not hold on your metal, do not buy it — that is the whole point of sending it.