The Elastomers and What Attacks Each One
Why this matters
A seal that fails in weeks was almost never the wrong size. It was the right size in a compound that the fluid, the heat, or the grease you put on it during assembly was quietly eating. The tell is a repeat leak at the same joint after a correct-looking repair, and shops chase that for months because every individual repair passes inspection. Learning elastomers by what attacks them, rather than by what they are made of, is what turns that run of callbacks into a one-visit fix.
A run of callbacks that all looked like good repairs
A shop had five leaks over four months on the same type of joint across different customers. Each time the seal was replaced with the correct part number, each time it held on the pressure check, and each time it wept again within a few weeks.
Before any of those seals came out, the line was isolated, pressure was relieved to zero at an open vent rather than by feel, and the system was drained. A seal that is weeping is still a joint holding stored energy, and the fastener you are loosening is the only thing between you and it.
Three explanations got tested and eliminated:
Wrong size. The removed seals were measured against the new ones. Same nominal size on the box, and the new parts matched print. This one was killed early, but not cleanly, and that turned out to matter.
Bad joint preparation. The groove faces were inspected on the fourth callback and found clean, undamaged, and free of the score marks that cause a leak path. Two different techs had done the work. Preparation was consistent and not the variable.
A bad lot of seals. The seals came from two different suppliers across the five jobs. A lot problem would not span both.
What the measurement showed
On the fifth callback, the removed seal was measured properly instead of eyeballed. New cross-section was 0.139 inch. The seal that came out measured 0.158 inch.
That is an increase of 0.019 inch on 0.139, which is 13.7 percent on the linear dimension. Swelling is roughly uniform in all directions, so the volume change is that linear factor cubed: 1.137 cubed is about 1.47, or roughly 47 percent volume swell.
Put that number next to the general shape of seal acceptance limits and it is not close. A static seal will tolerate meaningfully more swell than a dynamic one, and both treat shrinkage of even a few percent as an immediate reject because shrinkage removes the squeeze that makes the seal work. Nobody's chart tolerates 47 percent. Get the actual acceptance numbers from the seal manufacturer for the specific compound, because they vary by compound and by whether the seal moves.
A swollen seal seals beautifully at first. That is why every one of those repairs passed its pressure check. Swell increases squeeze before it destroys the material, so the failure is always delayed past the point where the tech has left.
What was in the joint that nobody had written down
The seals were EPDM, which is correct for the water service they were in. The assembly grease on the truck was a general-purpose petroleum-based product.
Petroleum oil is the specific attacker for EPDM. It is not a marginal pairing, it is the classic incompatibility, and a smear of it at assembly is enough. The fluid in the system never touched a compound it disliked. The technician did, in the two seconds before the joint closed.
The correction was one line on the truck stock list: a non-petroleum assembly lubricant approved for the seal compound in use, and a note on the job template naming the compound so the next tech does not have to infer it.
The attacker-first table
Look up the environment, not the material. This is the direction the field question actually arrives in.
| What is present | Compounds it attacks | Compounds that hold up |
|---|---|---|
| Petroleum oils, greases, fuels | EPDM, butyl, natural rubber, most silicones | NBR, HNBR, FKM, neoprene |
| Hot water and steam | NBR, polyurethane (hydrolysis), many FKM grades | EPDM (peroxide-cured), PTFE |
| Ozone, sunlight, weather | NBR, natural rubber | EPDM, neoprene, FKM, silicone |
| Ketones, esters, glycol-ether brake fluid | FKM, NBR, polyurethane | EPDM, butyl |
| Strong oxidizers and strong acids | NBR, EPDM, neoprene | FKM, PTFE |
| Amines and ammonia | FKM, some neoprene | EPDM, butyl |
| Chlorinated water and pool chemistry | NBR, EPDM over time at high residual | FKM, PTFE |
| Sustained high temperature | NBR first, then neoprene, then EPDM | FKM, silicone, PTFE |
| Sustained low temperature | Standard FKM grades stiffen | Silicone, low-temperature-specific grades |
Two entries in that table trip people who know the materials well. EPDM and petroleum are opposites, even though EPDM is the default for water and is on every truck. And FKM, which most techs think of as the compound that survives everything, is poor against ketones, brake fluid, and amines, so it is exactly the wrong upgrade in those services.
Reading the families by temperature
Typical service ranges for general-purpose compounds, useful for ruling things out rather than for final selection, because a specific compound within a family can sit well outside these:
| Family | Typical range | The property it is chosen for |
|---|---|---|
| NBR (nitrile) | about -30 to 250 degrees F | Petroleum oils and fuels, low cost |
| HNBR | about -25 to 300 degrees F | Oils plus higher heat, common with modern refrigerant oils |
| EPDM | about -40 to 300 degrees F | Water, steam, weather, glycol |
| Neoprene | about -40 to 250 degrees F | General-purpose outdoor, moderate oil |
| Butyl | about -50 to 250 degrees F | Very low gas permeability, vacuum service |
| Silicone | about -80 to 400 degrees F | Extreme temperature range, food and medical grades |
| FKM | about 0 to 400 degrees F | Broad chemical resistance plus high heat |
| PTFE (not an elastomer) | about -300 to 500 degrees F | Near-universal chemical resistance |
Read the temperature column with the chemical column, never alone. A compound rated to 400 degrees F in air can fail at 200 degrees F in the wrong fluid, because heat multiplies chemical attack rather than replacing it. As a rule of thumb, reaction rate roughly doubles for every 18 degrees F of temperature rise, which is why a marginal compatibility rating that survives at ambient falls apart on a hot line.
Assembly lubricants and cleaners are a compatibility decision
The callback run above is the common version of this, and it has three more forms worth knowing:
- Silicone grease on silicone rubber swells it. The general-purpose answer for EPDM is the specific wrong answer for silicone seals. Fluorinated or PTFE-thickened lubricants are the usual route there.
- Solvent cleaners get used to prepare a groove and then the seal goes in before the solvent has flashed off. The seal spends its first hour in a chemical nobody would have specified.
- Leak-detection sprays, thread sealants, and anti-seize all end up in contact with seals. Each one is a fluid the seal has to survive, and each one carries Section 10 of its own safety data sheet naming the materials it is incompatible with, which the hazard communication standard at 29 CFR 1910.1200 requires the supplier to give you.
The practical rule: anything that touches the seal is part of the fluid list. Not just what flows through the joint in service.
Two failure shapes that are not chemical attack
If you go looking for an incompatible fluid every time a seal fails, you will misdiagnose these two.
Compression set. The seal has taken a permanent shape and no longer springs back. The removed part looks flattened into an oval or square section rather than round, and it measures under-height without measuring under-width. That is heat and time doing their normal work, or a groove that squeezed too hard, not chemistry. The fix is a higher-temperature compound or a corrected groove, not a different chemical family.
Extrusion and nibbling. The seal has been pushed into the clearance gap by pressure and has a shaved or ragged edge on the downstream side. That is a mechanical problem: too much clearance, too much pressure, or a compound too soft for the duty. A backup ring or a harder durometer fixes it. Switching compounds for chemical reasons fixes nothing.
Never burn-test an unknown seal to identify it. Fluoroelastomers and PTFE release hydrogen fluoride and other toxic decomposition products when overheated, and the amounts from a small sample in an unventilated space are enough to hurt you. Identification comes from the parts documentation, the system's fluid, and physical measurement, never from a flame.
How to verify you got this right
Before the joint closes:
- Name the compound, not the part number. A part number identifies a size and a supplier. Two seals with the same dash number can be different compounds, and that difference is the entire subject of this card.
- List every fluid the seal will touch, including the assembly lubricant, the cleaner used on the groove, and anything sprayed on the joint for leak checking. If the list has one entry, it is incomplete.
- Measure the removed seal before you throw it away. Cross-section over nominal points at swell, under nominal points at shrinkage or extraction, flattened section points at compression set. That measurement is the only evidence you will get, and it costs fifteen seconds.
- Check the temperature against the fluid, together. A compound that passes on both lists separately can still fail where they meet.
- Write the compound into the job record. The next tech on this joint inherits either your finding or your guess.
References
- Seal and o-ring manufacturer compatibility documentation and acceptance limits for volume swell, shrinkage, and compression set for the specific compound
- Safety data sheet, Section 7 (handling and storage) and Section 10 (stability and reactivity), for any fluid, lubricant, or cleaner entering the joint, per 29 CFR 1910.1200
- Trade-standard practice for o-ring groove design, squeeze, and backup ring use
- See related: Gasket + O-Ring Material Reference; How to Tell Which Seal Material You Are Holding