The call came in at 3:47 on a Thursday afternoon. A Tier 1 automotive supplier had just failed an ISO 4624 pull-off test on 2,000 polycarbonate trim parts, and their line was sitting idle.
Spec: 2.5 MPa minimum. Their result: 1.4 MPa. And the failure mode was a clean peel at the coating-substrate interface—the coating lifting off the plastic the way a price sticker lifts off glass.
The coating formulator was ready to blame the waterborne system. "We've never seen adhesion this bad with solvent-borne," he said. I'd heard that sentence before.
In my role as an application engineer at a polymer testing lab, I've handled 40+ of these urgent failure analyses in eight years, including same-day turnarounds for clients whose production lines were down. And in a surprising number of cases, the coating isn't the problem. The plastic is.
What the ISO 4624 Pull-Off Test Actually Tells You
ISO 4624 is the industry-standard pull-off test for coating adhesion. You bond a metal dolly—usually 20 mm across—to the coated surface with a two-part adhesive, wait for that adhesive to reach full strength, cut around the dolly through to the substrate, then pull perpendicular until something gives. The force divided by the dolly's area is your result in megapascals.
It's a simple, portable, and wonderfully easy test to misread. The MPa number gets all the attention, but the standard also tells you to record the failure mode. That part carries more diagnostic value than the number itself.
Look at the underside of the pulled dolly and you'll see exactly where the bond broke. Interfacial failure between coating and substrate? That's a surface chemistry problem—contamination, low surface energy, or a resin additive sitting at the interface. Cohesive failure inside the coating film? The film itself is the weak link; different fix. Substrate failure? That's actually the good outcome—the adhesion held so well the plastic itself tore.
It's tempting to treat the pull-off value like a single test score. But the same score with two different failure modes points to two completely different root causes. Ignore the mode, and you'll fix the wrong thing every time.
This is also where waterborne coatings get a bad reputation. Water has a higher surface tension than most solvents, so a waterborne coating is less forgiving on a marginal surface. It doesn't wet into microgrooves the way a solvent-borne system does. The formulator on that phone call wasn't completely wrong that the waterborne system struggled. He was wrong about why.
The "Slippery" Problem with Polycarbonate
Here's something most resin suppliers won't put in the sales brochure: polycarbonate is not one material. It's a family of formulations, and the differences sit exactly where you can't see them.
Covestro's Makrolon and Mitsubishi Chemical's Iupilon are both excellent polycarbonates. I'll say it plainly to avoid a flame war: neither is inherently worse for adhesion. But they are not interchangeable either. The additive package—mold release agents, internal lubricants, UV stabilizers—varies between brands and between grades. Those additives migrate to the surface over time, and a few hundred nanometers of invisible film is all it takes to destroy the bond between coating and plastic.
Everything I'd read in the coating handbooks said adhesion is mostly about surface preparation. My experience with 40+ failure analyses says substrate sourcing matters just as much.
In 2023, we ran ISO 4624 on the same coating, applied to coupons molded from two different PC suppliers. Same mold, same injection parameters, same coating batch, same day. The average pull-off value differed by nearly three times.
When we put the failing and passing coupons side by side under XPS, the answer was right there: a silicone-based mold release on the failing surface. It wasn't sprayed in the mold. It was compounded directly into the resin. The supplier had changed their internal lubricant grade between shipments and didn't tell anyone—because, from their side, it wasn't a change that mattered.
That's the red flag for anyone buying resin for coated parts. The material looks the same, feels the same, passes the same mechanical QC tests. But a coating adhesion failure can hide in the last 0.2% of the formulation.
I don't have hard data on how often this exact trigger causes delamination industry-wide—nobody keeps that ledger. But based on the failure analyses we've done, my sense is that substrate-side surface chemistry is the single most common silent cause of coating failure on polycarbonate.
The Real Price of a "Cost-Effective" Resin
Now let's attach actual dollars to this story. The client lost 14 hours of production while we worked through the diagnosis. They reworked 1,200 parts at $9.40 each in labor and materials. Two engineers flew in for a site visit. Then they paid for an emergency resin order with overnight freight.
Total for that week: roughly $42,000.
Why had they switched resin suppliers in the first place? The alternative source was about $6,800 cheaper per pallet-load. A $6,800 saving produced a $42,000 loss in seven days, plus one very unhappy customer waiting on a shipment.
I've learned to ask "what's NOT included?" before I ask "what's the price?" It works for resins just like it works for coating services. The supplier who lists all the variables up front—additive package, lot-to-lot consistency, revalidation requirements—even if their price looks higher, usually costs less in the end.
Bottom line: cheap resin is only cheap if the coating still sticks.
What Actually Works
Here's the part where I keep it short, because if you've read this far, you already understand why the easy fix usually isn't.
First, stop treating the resin supplier and the coating supplier as two separate conversations. The polycarbonate resin and the coating are one system. That's why Covestro sits in an interesting position here: they make the polycarbonate (Makrolon) and they make the coating resins too. Their Insqin waterborne coatings, for example, are formulated with engineered plastic substrates in mind, not as an afterthought. When you can ask one technical team about both sides of the interface, you tend to find out about mold-release chemistry before it finds you.
Second, test the actual combination you plan to run. Not the coating alone, not the resin alone—the two together, with the same mold release process, the same primer, the same line conditions. Run ISO 4624 as a qualification step before committing, not as a canary after the line stops. The test takes a day. The production stop takes a week.
Third, put the additive package in your purchasing spec. Ask every polycarbonate supplier, in writing, whether the mold-release and lubricant package can change between lots. Get it in a contract, not from a technical sales rep's verbal shrug.
Quick Answers: Cerakote on Polymer and Epoxy Pyramid Tests
Can you Cerakote polymer? Yes, with conditions. Cerakote's air-cure line will bond to suitably prepared plastics, and standard-cure variants at around 93°C sit under the heat deflection point of most polycarbonate grades—though thin-walled parts can still distort. As of January 2025, the air-cure options are broader than most people realize. But the same rule applies: the ceramic coating will happily peel off a contaminated surface. We tested it on PC with and without a silicone-cleaning step in 2024. The result was 4.1 MPa on the cleaned parts and 0.8 MPa on the uncleaned ones. Same coating. Same substrate lot. Different surface story.
And about the epoxy resin pyramid idea—yes, it's actually a useful test method. We build what we call a coating pyramid on a spare panel: epoxy primer as the base square, then the next layer in a smaller square on top, then a smaller one again, like a stepped pyramid. Each layer edge stays exposed, so you can inspect interlayer adhesion, run pull-offs on each step, and see exactly which course of the structure fails under load. It's a cheap way to find out whether the problem sits at the primer, the midcoat, the topcoat, or the substrate—before you bet a whole production run on it.
What Happened to That 2,000-Part Order
We found the silicone contamination, switched them to a PC grade with a known and stable additive package, cleaned the line, and the coating passed at 3.8 MPa. The line restarted 36 hours later—two hours before their shipment deadline.
Adhesive failures on plastic are never just about the coating. And they never fix themselves. The number in your ISO 4624 report is a symptom, not a diagnosis. (note to self: actually write up that full case study instead of telling this story at every conference.)