Adhesion Failure on Re-Sealed Curtain Wall Joints and a Bubbling Polyurethane
On a 1990s curtain wall, a maintenance contractor is replacing failed perimeter sealant joints. The existing joints are 3/4 in wide × 3/8 in deep with closed-cell backer rod. Crews clean the joints with a solvent, apply a neutral-cure silicone directly over the existing silicone (which is still firmly bonded at the joint edges) with no primer, and tool it immediately. Two weeks later the new sealant peels away in long ribbons, leaving a clean, glossy, residue-free surface on the old silicone - a classic adhesive failure. At the base of the wall where the aluminum meets the concrete slab, the same crew used a polyurethane sealant, and it is blistering and bubbling along the joint. The specification requires a 50% movement-capable joint, and the design movement at the curtain wall perimeter is stated as ±3/8 in.
1. Diagnose the adhesive failure on the silicone joints and describe the correction, including the tests you would require.
Read the worked answer
The clue is the failure surface: clean, glossy, residue-free. The new sealant did not fail cohesively - it never bonded at all. Silicone does not bond to cured silicone. The old material is a low-energy surface, and the new bead simply sat on top of it.
Solvent wiping does not fix this. Solvent removes oil and dust; it does not create a bondable surface on silicone rubber, and wiping with the wrong solvent can even leave a residue that further hurts adhesion.
Correct preparation:
- Mechanically remove the old sealant back to the substrate - the aluminum, glass, or concrete - using a sealant-cutting tool, not a solvent. Verify with a probe that no silicone film remains; residual silicone contamination on metal or masonry is the single most common cause of repeat failure.
- Abrade and clean metal per the sealant manufacturer's written instructions, then prime exactly as specified (many neutral-cure silicones require a primer on aluminum and on concrete).
- Tool only after the manufacturer's specified time window, and use compatible backer rod.
Tests to require:
- Compatibility test per ASTM C1087 whenever new sealant touches existing or adjacent materials.
- Field adhesion test (hand-pull / peel test per ASTM C1193) on sample joints that cure under the same conditions, with the joint opened to confirm cohesive failure rather than adhesive failure. Do not proceed with production until the peel test passes.
2. Check the joint geometry and movement capability. Is 3/4 in × 3/8 in acceptable for ±3/8 in of design movement at 50% movement class?
Read the worked answer
No - the joint is undersized for the stated movement, or the movement assumption is wrong. Either way it needs resolution before more sealant goes in.
Depth-to-width. 3/4 in wide × 3/8 in deep is a 2:1 ratio, which is the correct configuration: sealant should be about half as deep as it is wide, with a minimum depth of 1/4 in and a maximum around 1/2 in so the bead can strain rather than tear. The depth/width geometry is fine.
Movement is the problem. A 50% movement-capable sealant can take ±25% of the joint width. At 3/4 in wide, that is ±3/16 in, or 3/8 in total movement across the joint.
The design demand is ±3/8 in - 3/4 in total, which is 100% of the joint width. That is double the capacity of a 50% joint at this width.
Options:
- Widen the joint to about 1 1/2 in to get ±3/8 in within a 50% class. Often not possible at a perimeter reveal.
- Select a higher-movement sealant (for example a 100% movement-class product) and verify it against the manufacturer's published data for the actual width and depth.
- Reduce the design movement by re-examining the actual structural drift and thermal movement, and get the ±3/8 in figure reconsidered in writing.
Never stretch a 50% sealant to 100% movement - it will fail at the first real cycle even if it looks perfect on day one. Confirm with the design professional in writing, since the perimeter movement originates in the structural design.
3. The polyurethane at the aluminum-to-concrete joint is blistering. What are the likely causes and what would you specify instead?
Read the worked answer
Bubbling in a polyurethane at a slab joint is almost always a moisture or substrate problem, and it is a different failure mode from the silicone adhesion issue.
Likely causes:
- Moisture in the concrete. Slabs hold water. Polyurethane curing releases gas when it reacts with water, and moisture vapor drives out through the bead, producing blisters. Check slab moisture and allow a proper dry-out.
- Uncured or green concrete and alkalinity. Fresh concrete is highly alkaline; most sealants need a primer or a cure period before they will bond.
- Solvent or cleaner trapped under the bead. If the joint is closed with a solvent wipe and sealed wet, the solvent flashes off under the cured skin and blisters it.
- Wrong product class. Some polyurethanes are not intended for the movement and exposure of a perimeter curtain wall joint.
What to specify:
- A neutral-cure silicone intended for the joint's movement class and exposure, with the manufacturer's specified primer on the concrete and on the aluminum. Silicone tolerates concrete alkalinity and UV exposure far better than most polyurethanes, which chalk and degrade in exterior vertical exposure.
- Correct backer rod (closed cell, sized about 25% larger than the joint width so it stays in compression) and correct depth-to-width ratio.
- Proper joint preparation: mechanical removal of the failed material and any prior sealant, then dry substrate before priming.
- Field adhesion testing on the same substrate before production begins.
Use one system per joint and confirm the sealant manufacturer's written compatibility with every material the bead touches, including the backer rod.