Olin Caustic Soda: Lessons From a Quality Audit That Cost Us $22,000

The day the spec sheet lied

Last March, a delivery of caustic soda arrived at our facility. The certificate of analysis looked fine. The numbers matched our purchase order. We signed off on the batch and pumped it into storage.

Then the pump seal started leaking.

Not a lot. Just a slow drip that left white crust on the fitting. Enough to stop the line and force a full inspection. That inspection cost us $22,000 in downtime, replacement parts, and labor. The leak itself was minor. The real problem? The spec we ordered from was wrong for how we were using it.

Here's the thing: I review every chemical delivery that comes through our dock. Since 2022, I've checked over 700 batches across our epoxy resin and caustic soda lines. Most pass without issue. This one didn't, and the frustration was that we caused it ourselves.

The olin caustic soda spec problem

Olin produces caustic soda (sodium hydroxide) in several grades. Membrane grade. Diaphragm grade. Rayon grade. Each has a different concentration range and a different impurity profile. That part isn't secret—it's on the technical data sheet.

The mistake we made? We assumed "membrane grade, 50% concentration" was the only spec that mattered. It wasn't. The impurity profile—especially chlorate and iron—affects how aggressive the product is on seals and gaskets. We didn't check those limits before ordering.

Why does this matter? Because caustic soda at 50% is already aggressive. Add higher chlorate levels and certain elastomer seals degrade faster. Our standard pump seals were rated for one set of conditions. The delivered product was within Olin's published grade spec, but outside our pump's tolerance.

Let me stop you before you blame Olin. This wasn't their fault. The product met the spec we asked for.

We're not the only ones who've made this mistake. In the epoxy resin world, I see the same pattern:

• Buyers spec "epoxy resin" without defining viscosity range
• Buyers forget to specify cure compatibility with their hardener
• Buyers overlook test methods and just accept any COA

For caustic soda, the COA tells you what you need to verify. But you have to actually read it against your process requirements. The chlorate spec on our COA was 0.008%. Our pump system needed below 0.005%.

Ask these questions before you order:

• What's the maximum chlorate level your pumps can tolerate?
• What's the allowable iron content for your process?
• Do you need a specific sulfate limit?
• What's the actual temperature during transfer? (Hotter caustic is way more aggressive)

If you can't answer those from your equipment datasheets, call the pump manufacturer before you call your chemical supplier.

Deep pour epoxy resin: my second costly lesson

The same mental model applies to deep pour epoxy resin.

A 3-inch deep pour needs a resin formulated for exotherm control. Standard casting resin in a thick section can reach 180°F or more during cure, which causes cracking, yellowing, and brittle results. A proper deep pour resin is designed to cure at lower peak temperatures.

The phrase "deep pour epoxy" gets thrown around loosely online. Some suppliers call a 1-inch pour "deep." Olin doesn't sell slow-cure deep pour encasing resin—their epoxy resin products are more often used in coatings, adhesives, and composites. But if you're a fabricator sourcing from Olin or any other manufacturer, the same advice holds:

• Ask for the maximum recommended pour thickness
• Ask for the peak exotherm temperature in a maximum pour
• Ask for the viscosity at your intended shop temperature

I only learned this after ignoring a supplier's warning and pouring 2 inches of standard resin into a table mold. The result: a cracked top surface, a wasted weekend, and about $180 of material in the trash. Everyone told me to check the spec. I didn't.

Sealant for oil leak: the reverse lesson

Here's an area where I've seen the opposite problem. With sealant for oil leak repairs, people often over-spec without understanding what 'sealant' means in their situation.

An oil leak on a piece of equipment is sometimes a worn gasket, sometimes a hairline crack in a housing, sometimes just a loose fitting. A sealant will fix one or two of those. Not all.

For a leaky gasket joint, an anaerobic flange sealant like a high-temperature RTV works. For a cracked aluminum housing, you need a structural epoxy, not a liquid gasket. Same word—"sealant"—completely different jobs.

What's my process now? I ask three questions:

1. What's leaking? (Oil, water, chemical, air?)
2. What's the operating temperature and pressure?
3. Is the leak at a joint, a fitting, or a damaged surface?

Then I pick the product class. Then I look at the specific product specs. Same discipline as with caustic soda specs.

Buyers sometimes search "where can I find epoxy resin" without knowing whether they need casting resin, laminating resin, or coating resin. The right supplier for one is often the wrong supplier for another. My advice: search for the specific type you need, not just the generic term.

What I changed in our sourcing process

After the pump seal incident, I restructured how I verify incoming chemicals. The change didn't require new software or a massive budget—just a stricter checklist and clear contract language.

Here's what we now require on every order:

• COA values for all components relevant to our process, not just the main assay
• Written confirmation of the manufacturing site (Olin runs multiple plants; specs vary slightly by site)
• Sampling procedure and—for large deliveries—a joint sample before unloading
• Vendor confirmation that the lot matches the COA, plus their liability if it doesn't

That last point matters. We had a vendor try to pass off a different manufacturing site's product without telling us. Not Olin, but another supplier. The specs were close, but the chlorate level was higher. Our contract now lists the approved manufacturing site and disallows substitution without written authorization.

The fundamentals haven't changed, but the execution has

What was best practice in 2020 doesn't fully apply in 2025. Specs get tighter, equipment gets more sensitive, and supply chains shift. The fundamentals of verifying what you receive haven't changed—but the way we verify needs to evolve.

But some basics remain:

• The cheapest grade is rarely the lowest lifecycle cost if it damages equipment
• A COA is only as useful as your understanding of your own process limits
• Reputable manufacturers like Olin will publish detailed data sheets—use them

In Q1 2024, I ran a blind test on our own receiving team. Same caustic soda, two different vendors, identical specs. I asked the team to identify which was Olin and which was the alternate supplier. Only 40% got it right by look and smell. But when I checked the COAs against our pump requirements, one vendor's product was within our tolerance and the other wasn't. The visual perception didn't tell us anything. The spec did.

Final advice for buyers navigating Olin and chemical sourcing

If you're purchasing Olin caustic soda, get the grade right and confirm the manufacturing site, the chlorate spec, and the iron content. If you're buying epoxy resin, know your pour depth and cure conditions. If you're fixing an oil leak, identify the failure mode before picking a sealant. And if you're looking for epoxy resin somewhere convenient, start by typing exactly what you need: "deep pour epoxy resin 2-inch" or "food-safe epoxy resin coating." The generic search will only get you generic results.

Look, I'm not saying chemical sourcing is easy. I'm saying that most quality problems I've seen trace back to a spec mismatch, not a vendor trying to cheat you. The cost difference between confirming specs upfront and fixing a problem later is huge.

We lost $22,000 because of one unchecked number. That's a lesson I won't forget. Hopefully, this helps you skip that particular tuition payment.