In Q2 2023, a batch of conformal coating failed on our production line. The boards passed visual inspection, went through reflow, and then—during the first thermal cycle test—the coating cracked and started delaminating from the PCB surface.
We lost two days of production. The rework bill came to $18,000. And the worst part? We had ordered exactly what engineering specified: Dow Corning conformal coating 1-2577. Same part number we'd been buying for two years.
I'm the purchasing coordinator for a 220-person manufacturing company. I manage MRO and indirect materials—roughly $780,000 annually across 14 vendors. I report to both operations and finance, which means when something like this happens, I hear about it from two directions. That coating failure was the first time I understood that "buying the right part" and "managing the right process" are not the same job.
What we thought went wrong
Our first assumption was counterfeit product. It happens more than people admit—grey-market silicone materials with a Dow Corning logo printed on generic packaging. So we checked the COA (Certificate of Analysis), compared the batch number against the distributor's records, and looked at the packaging details. Everything verified.
Then we assumed a specification mismatch. Maybe we ordered the wrong viscosity or the wrong cure type. We pulled the TDS and compared it against the engineering spec. It matched.
That's when the conversation got uncomfortable. If the product was genuine and the spec was correct, then the problem was somewhere else—somewhere nobody in the room had been tracking.
The first hidden cause: part numbers don't freeze chemistry
Here's what I didn't understand back then. A part number identifies a product family. It does not guarantee that every batch behaves identically.
Dow Corning—now part of Dow after the 2016 merger, though the brand still appears on certain product lines—manufactures silicone materials across multiple facilities. The 1-2577 conformal coating you buy this quarter might come from a different production line than the batch you bought last year. The nominal chemistry is the same. The datasheet specs are the same. But small shifts in catalyst concentration, solvent purity, or cure kinetics can change how the coating wets out on a specific board finish.
In my first year in purchasing, I made the classic rookie mistake: I assumed "same part number" meant "interchangeable batch." It cost me a week of arguing with our quality team before I understood that a COA is a snapshot, not a guarantee.
The same logic applies to gasket adhesives. The bond strength you measure on a lab coupon depends on surface prep, humidity during cure, and substrate temperature. If a supplier adjusts a filler or tweaks the catalyst ratio without flagging it, the adhesive might still pass a tensile test in isolation—but fail in your actual assembly.
And it gets riskier with bonding magnets in electric motors. Those adhesives have to survive thermal cycling, vibration, and years of service. A subtle formulation change can shift the glass transition temperature by just a few degrees (which, honestly, sounds trivial until you see fatigue cracking at the bond line after 18 months in the field).
Where the Dow Corning logo fits in
I'm not suggesting you distrust every shipment. But the Dow Corning logo should be a starting point, not an ending point. On legitimate products, the logo, the batch coding, and the COA should all be consistent. If you're buying through distribution—which most mid-size companies do—ask for the chain of custody. A genuine product from a reputable channel is one thing. A genuine product that sat in a warehouse for two years past its shelf life is another.
We now log the COA reference number for every critical material we buy. It takes about 30 seconds per order. I can't prove it has prevented a failure—I'm not 100% sure we'd ever capture that data—but it has made supplier conversations a lot more specific.
The second hidden cause: internal changes nobody tells purchasing about
Back to the coating failure. After three weeks of investigation, we found the real cause: a combination of two changes.
First, the supplier had adjusted the solvent system in a newer batch. It was a minor tweak—still within the claimed spec—but enough to affect wetting on our specific board finish.
Second, and more importantly, our engineering team had switched to a different flux chemistry three months earlier. They updated the BOM in the PLM system, but nobody told purchasing. I was still ordering against a PDF that had been emailed to me in 2022.
I assumed that when engineering changed a process, purchasing would get a notification. I was wrong. In a lot of companies, the BOM lives in one system and the purchasing records live in another. The gap between them is where failures breed.
What pharma gets right about change control
This is where I started reading about change control in pharma. If you've ever wondered what change control in pharma actually means: it's a formal process that requires any change to materials, suppliers, or manufacturing processes to be documented, risk-assessed, and approved before implementation. ICH Q10—the pharmaceutical quality system guideline—treats change management as a core element of quality, not an afterthought.
Pharma companies do this because the stakes are obvious. Change a tablet coating without testing, and people get sick.
Industrial adhesives are not that different. If you change a conformal coating, a gasket adhesive, or a motor magnet bonding adhesive without verifying compatibility, you get field failures, recalls, and customer complaints. You just don't read about it on the evening news.
Most manufacturers don't have a pharma-grade change control process (ISO 9001:2015 clause 8.5.6 touches on it, but it's a light touch). You don't need the full bureaucracy. You need the habit.
What it costs when you skip this
The $18,000 rework was the visible cost. Add two days of lost production—roughly $9,000 in deferred revenue—plus overtime to catch up, plus the customer who asked pointed questions about our quality process. Then add the intangible: the next time engineering specifies a Dow Corning product, someone says "remember what happened last time?" And purchasing gets treated like order-takers again.
The real cost of a change-control gap isn't the invoice. It's trust. Once operations decides that purchasing "just buys what we tell them," nobody loops you in on the decisions that actually matter.
The checklist I now use
I'm not a quality manager. I don't have the authority to rewrite engineering specs or audit suppliers. But I do control the purchase order, and that gives me three levers.
One: a change-tracking card for every critical material. Nothing fancy—a spreadsheet with four columns: product name, supplier, COA reference, and "last verified" date. For our conformal coating, gasket adhesives, and magnet bonding adhesives, I check it every quarter. It takes about 45 minutes total.
Two: a supplier notification clause. We added a line to our standard PO terms: "Supplier must provide 30 days' written notice of any change to formulation, manufacturing site, or packaging." Most vendors agreed without pushback (surprise, surprise—the ones who hesitated were the ones with something to hide).
Three: an internal BOM-change email loop. I asked engineering to copy purchasing whenever they update a spec for a material we buy. It's not a formal process. It's just an email. But it closed the gap that caused the original failure.
None of this is glamorous. But the 45 minutes I spend each month on change tracking has already caught two near-misses since 2024—once when a distributor quietly switched to a different production facility, once when engineering updated a spec without telling anyone in purchasing.
Five minutes of verification beats five days of correction. If you're in purchasing and you've never asked your supplier how they notify you about formulation changes, that's probably the first question worth asking. Not because the product is bad. Because the process around the product is where the failures live.
This worked for us at a 220-person company with fairly stable demand. If you're running a smaller operation or a business with seasonal spikes, the cadence might need to be different. I can only speak to what I've seen in industrial manufacturing—your mileage may vary.