You qualified the adhesive. You never qualified the surface.

Ford's adhesive bonding recall landed last week, and I guarantee you the PPAP folder for that joint is immaculate. Cure temperature logs, neatly initialled. Tensile test coupons, passed. Control plan signed off, FMEA risk priority number acceptable, launch approved. Not a single measurement of surface energy anywhere in the file — because nobody asked for one.

I have run this exact play. At WITTE Automotive I inherited control plans for adhesive-bonded latch assemblies that tracked oven setpoint, belt speed, and bond gap to three decimal places. The surface those adhesives were bonding to? Cleaned by an operator with an isopropanol rag and a visual check. At SNOP, building a 900-person greenfield plant from scratch, I wrote surface preparation into the PFMEA because I'd seen what happens when you don't. Standard automotive APQP documentation does not require it. You can submit a flawless PPAP package and never once verify that the surface you're bonding to is bondable.

A special process treated like a standard one

Here is the uncomfortable part. Adhesive bonding is a special process under IATF 16949 — your output verification is destructive, delayed, or statistical. You cannot fully inspect bond strength on a part you intend to ship. Welding gets this treatment. Heat treatment gets this treatment. Adhesive bonding, which is more sensitive to process variation than either, routinely gets a control plan that looks like it was written for a machining operation.

The variable that governs whether your adhesive holds is surface energy. Not cure temperature. Not bond gap. Not the datasheet lap-shear value the supplier handed you during qualification. If the substrate's surface energy is lower than the adhesive's, or if contamination drops it below the threshold between cleaning and bonding, you get a weak interface. The adhesive cures perfectly. The joint fails anyway.

At AIRBUS, this is not a debate. Under AS9100 process controls for structural bonding and composite joining, surface energy verification — dyne testing, contact-angle measurement, or equivalent — is mandatory before bond. You do not proceed without it. The aerospace industry learned this through in-service delamination events that killed people. Automotive is working through the same curriculum, one recall at a time, and the tuition is going up.

The adhesive never fails the spec sheet. The surface fails the adhesive.

The PFMEA listed delamination as a failure mode. It missed what causes it.

I have reviewed hundreds of PFMEAs across automotive suppliers in Europe and North America. The pattern is consistent. Delamination appears as a failure mode. The cause listed is "inadequate adhesive application" or "cure temperature deviation." The control is "visual inspection" or "oven temperature monitoring." Risk priority number calculated, box ticked, document filed.

The actual cause of delamination in structural adhesive joints is an uncontrolled surface. Silicone mould release carried over from the stamping die. Drawing compound residue that cleaning didn't fully remove. A batch of substrates with lower surface energy because the supplier changed their release agent. A humidity shift that changed the contamination profile. These are the real failure modes, and they are invisible to a control plan that treats "surface preparation" as a single line item verified by a glance.

This is where the gap between writing a PFMEA and understanding one gets expensive. I have sat in 8D reviews where a cross-functional team spent three days chasing cure parameter variation before anyone asked whether the parts were actually clean. The chemistry was fine. The process window was fine. The surface was never bondable to begin with, and nobody had measured it.

What aerospace validates at the surface. Automotive discovers at the recall.

The structural shift in automotive is making this worse, fast. Lightweighting programmes push adhesives into joints that used to be welded — body-in-white, battery enclosures, composite closures, mixed-material architectures where you cannot weld even if you wanted to. Load-bearing adhesive bonds are safety-critical in vehicles that didn't exist five years ago.

Aerospace went through this transition decades ago with primary structural composite bonding. The response under AS9100 was surface energy verification, process control documents that specify surface preparation as a measured variable, and regular process audits that confirm you're doing it. The control plan treats surface as the process, not the adhesive.

At SNOP I built the quality system for a plant where adhesive sealing and bonding were part of the core process flow. We tracked cure parameters diligently. What we did not have — and what no automotive plant I've audited since has had — is a dyne pen on the line with a documented measurement frequency and a reaction plan tied to it. That equipment costs less than €300. The recall Ford is now processing will cost more per vehicle than equipping every adhesive station in the supply chain.

The Kia and Hyundai headlines running alongside Ford's this week are not the same failure, but they share the same architectural weakness. Automotive quality systems were built around mechanical joints and destructive testing of metal. The industry is now bonding things that carry structural load, and the control philosophy has not caught up to the joining technology.

Ford's adhesive passed every test it was given. The tests stopped one variable short of the one that mattered. Every automotive quality professional reading that recall notice should be opening their adhesive PFMEAs right now and asking a single question: where is surface energy in my control plan? If the answer is nowhere, the next recall in this category already has a production date, and it is coming off your line.