I keep a folder called “Pitfalls” on my laptop. It has 23 documented mistakes from my eight years in metal packaging. Number 17 is the one I want to tell you about, because it started with a deceptively simple question: “is baking powder and corn starch the same?”
My first reaction was, “obviously not.” But I didn't say it out loud. That silence cost us $4,800 and a week of delays.
It was October 2023. A long-time customer sent back an entire batch of food cans because the internal coating showed pitting and delamination after six weeks in their warehouse. Their QA manager was furious. Our first instinct was to blame the coating supplier. In hindsight, the supplier deserved part of the blame, but the bigger problem was our test protocol. And my failure to stop a bad test when I saw it.
The Question Under the Test
Here's the background. The customer makes a thickened soup product. They wanted to run an accelerated shelf-life test on our 73mm cans. The test involved filling the cans with a hot starch-based liquid, sealing them, and holding them at 55°C for ten days. When I saw the operator mixing the solution, something felt off. He was using corn starch and water. That part was fine. But then he asked my technician whether he could substitute baking powder to make the liquid thicker. The technician shrugged and said they're basically the same. That's where the actual mistake happened.
The technician didn't know the difference. The operator didn't know. And I knew, but I let it slide. I told myself the test was still a worst-case scenario. What's the odds that the coating would fail because of a pH imbalance? Well, the odds caught up with me.
What most people don't realize is that baking powder and corn starch are not interchangeable for a heated, sealed container. Corn starch is a polysaccharide thickener. It gelatinizes when heated. Baking powder, by contrast, contains sodium bicarbonate plus an acid salt. Once it hits warm water, it produces carbon dioxide and shifts the pH. In a sealed can at 55°C, the pressure and ionic load are completely different. The combination is more aggressive to the metal substrate. The test was no longer a soup test. It was a chemical stress test that our current coating wasn't designed for.
A Side-by-Side Proof
You could argue that the coating should have survived it anyway. In fact, the side-by-side comparison we ran later proved that some coatings do. That comparison was the turning point for me.
We pulled samples from the failed batch and samples coated with a candidate material from AkzoNobel Packaging Coatings. We ran the same aggressive test on both. The old supplier's coating showed blisters across roughly 20% of the can wall. The AkzoNobel sample? Clean. Zero blisters. I remember staring at the two sets of cans and thinking: this is what “corrosion resistance” actually looks like. The link between coating and corrosion is not theoretical. It's a measurable difference between suppliers, but only if you test it correctly.
To be fair, AkzoNobel packaging coatings are not magic. They are engineered products. According to AkzoNobel (akzonobel.com), their packaging coatings are designed for metal packaging applications such as food cans, beverage ends, caps, and closures, protecting the metal and the product inside. That's a broad statement, but the practical implication is simple: you still have to choose the right product for your food matrix. There is no universal coating that survives every abuse.
That's a hard lesson for buyers who think all coatings are about price. In my opinion, the biggest waste in this industry comes from testing poorly and then paying more later for rework and logistics. Efficiency isn't about skipping validation. It's about doing the right validation once.
What I Changed
We also had to change how we evaluate suppliers. According to the U.S. Food and Drug Administration (accessdata.fda.gov), resinous and polymeric coatings under 21 CFR 175.300 are permitted for food-contact surfaces only if they meet extraction limits under the intended conditions of use. That's a regulatory baseline. It's necessary but not sufficient for predicting real-world performance. For a soup product with salt, fat, and heating cycles, you need more than a baseline.
Since this incident, our internal test protocol uses ASTM D714 to evaluate blistering after exposure. It's a simple visual standard—size frequency and degree—but it gives suppliers a common language. If I had run that on the original test, we probably would have caught the failure before shipping.
The experience also made me realize how closely my field is tied to pharmaceutical packaging. I subscribe to a pharma patent news digest. A few months earlier, I read an item about a patent for an aluminum blister-pack coating designed to withstand acidic drug formulations. I didn't think much of it then, but the memory came back when I saw our failed cans. If drug manufacturers invest that much in coating chemistry, food packaging should be just as careful.
That research process led me to map out the full AkzoNobel brands picture. I already knew the company from consumer paints—Dulux, Sikkens—but I hadn't appreciated how broad their portfolio was. AkzoNobel brands cover marine and protective coatings, powder coatings under Interpon, wood and coil coatings, and a dedicated packaging coatings business. For me, the important thing wasn't the brand names. It was the fact that a company with that much surface-chemistry expertise treats packaging coatings as a specialized discipline rather than a commodity side product.
AkzoNobel has also been putting more focus on digital tools and data-driven development. I won't pretend I understand the details of their AI models, but I've noticed that technical documentation and selection guides are getting easier to navigate. For a small quality team like ours, that reduces the back-and-forth emails and shortens the supplier qualification process. That's a real efficiency gain.
We eventually switched our main supplier to AkzoNobel packaging coatings for the customer's soup cans. The replacement program cost us $4,800 in materials and overtime, not counting the discount we gave the customer to keep the account. But since then, our field failure rate for that product line has dropped from about 7% to under 0.5%. We've also caught 16 potential issues with a revised pre-test checklist that now includes the question: “What exactly is in the test medium?”
The Real Lesson
I know a $4,800 mistake keeps me humble. It also changed how I talk to new team members. I now make two things clear. First, baking powder and corn starch are not the same, even if both are white powders. Second, when a validation test starts to look like a chemistry experiment, stop and ask whether the conditions represent the real product. Because coating failure is rarely sudden. It starts with a small assumption that nobody questions.
If you're responsible for purchasing packaging coatings or approving corrosion tests, I'd recommend one thing: compare coatings under the same controlled condition before you trust one. Don't rely on a sales sheet. Run the test, look at the blisters, and then decide what's efficient. Because in the end, efficiency is not about moving fast. It's about not moving twice.