AS568 · ISO 3601 · MIL-PRF cross-references, 15 compounds

Low-Temperature Sealing: What −40 °C Really Means for an O-Ring

Tg, TR-10 and brittleness are three different numbers, and none of them says your seal will work on a cold morning. How low-temperature performance is specified, ranked and designed around.

10 min read 1,265 words Updated 2026-11-17

"Good to −40 °C" is one of the least informative statements in a datasheet. It can mean the compound does not crack at −40 °C, or that it recovers 10 % of its stretch within a minute at −40 °C, or that a seal made from it passes a low-temperature leak test after a soak. Those are three different claims with three different test methods, and only the third one describes a seal that works.

A stainless valve body with frost on the flange
A seal at −40 °C is a different material from the one on the datasheet, measured at 23 °C.

Three numbers, three meanings

Glass transition temperature (Tg). The point where the amorphous phase of the polymer changes from rubbery to glassy. Below Tg the material is not a seal, it is a rigid plastic. Measured by DSC or DMA, and usually reported as a single number for the compound. It is a material property, not a seal performance limit.

TR-10 (ASTM D1329). A specimen is stretched, frozen, released, and allowed to warm. TR-10 is the temperature at which it has retracted 10 % of its original stretch. This measures low-temperature elasticity, and it is closer to what a seal needs. A compound with TR-10 at −40 °C is flexible at −40 °C in the sense that it will move — slowly.

Brittleness temperature (ASTM D746). The temperature at which a specimen breaks under impact rather than deforming. This is the lowest of the three and the least useful for sealing, because a seal does not need to survive impact; it needs to conform.

The gap between the three is not small. A compound can have a brittleness temperature below −50 °C, a TR-10 of −35 °C, and still leak at −25 °C in a joint with a wide clearance, because conformability at low temperature is about modulus and recovery time, not about whether the material cracks.

Why a cold seal leaks and then stops leaking

The common field signature: the joint weeps during a cold start, then seals once the system warms up. This is low-temperature stiffening, not damage. The ring is too stiff to follow the groove and the shaft at the moment of pressurisation, so fluid passes; as it warms, it softens, conforms, and seals.

Two design responses:

  • Give the ring time. If the system starts cold and is pressurised immediately, the seal has no warm-up. A short idle or a pre-heat changes the outcome.
  • Reduce the demand on conformability. Lower squeeze is not the answer — the ring needs stored energy. The better answers are a smoother surface (less need to conform), a narrower clearance (less unsupported area) and a compound with a lower TR-10.

The ranking, from the data this site uses

CompoundContinuous low limitComment
FVMQ−73 °CThe low-temperature answer with fuel resistance
IIR (butyl)−59 °CLow gas permeability, poor at high temperature
EPDM−57 °CWidest low end of the common compounds; not for oil
VMQ (silicone)−54 °CExcellent cold flexibility, weak mechanically
NR / SBR−50 / −46 °CGeneral rubber, not for oil or heat
AU/EU, ECO, AEM−40 °CThe polyurethane and specialty mid-range
NBR−34 °COil resistance with a modest cold limit
HNBR−32 °CHeat resistance costs cold flexibility
FKM / FFKM−26 °CThe price of chemical and thermal resistance
ACM, FEPM−21 / −9 °CSpecialty compounds with narrow windows

The pattern is the trade: every step up in heat and chemical resistance costs low-temperature flexibility. There is no compound that is excellent at both ends except FFKM, and FFKM's low limit is no better than FKM's.

Plasticisers, and why low-temperature grades age badly

The usual way to improve low-temperature flexibility of a compound is to add plasticiser. Plasticiser lowers Tg and softens the compound at low temperature — and it is also the component most likely to be extracted by the media, or to evaporate in hot air service.

A low-temperature NBR that has been plasticised to reach −45 °C may lose that capability after a year in hot oil, because the plasticiser has migrated into the oil. The seal then behaves like a standard NBR with a −34 °C limit, in an application designed for −45 °C.

This is why low-temperature performance should be specified after ageing or after immersion, not only on a new part. A datasheet number measured on a fresh specimen is an upper bound.

Designing the joint for cold service

Four decisions improve cold performance without changing compound:

  1. Keep squeeze at the lower end of the static band — around 15–18 % — because a stiff ring in a deep groove needs more force to seat than the cold material can generate quickly.
  2. Narrow the clearance. A cold ring is less able to bridge a gap, and extrusion risk is lower at low temperature anyway. A tighter joint helps.
  3. Improve the surface finish. A cold ring cannot fill a rough surface; 8–12 µin Ra helps more at −40 °C than at 20 °C.
  4. Avoid sharp groove corners. A stiff ring does not drape into a corner; it bridges it and leaves a leak path.
The specification that means something:

"TR-10 not above −40 °C per ASTM D1329, and no leakage in a −40 °C soak test at rated pressure after 4 hours". The first half is a material property; the second half is the only one your customer cares about.

Testing that reflects service

A cold leak test is worth more than any single material number. The procedure that catches most problems:

  1. Assemble the joint to production torque.
  2. Soak at the minimum service temperature for at least 4 hours — long enough for the metal and the rubber to reach equilibrium, not just the chamber air.
  3. Pressurise to rated pressure while still cold, and hold.
  4. Observe for leakage over a period comparable to the cold start in service.
  5. Repeat the cold/hot cycle several times. Many cold failures appear only on the second or third cycle, when the ring has taken a partial set at low temperature.

Frequently asked questions

Is silicone the right choice for low-temperature service?

For flexibility, yes — VMQ is good to about −54 °C and FVMQ to −73 °C. But silicone has poor mechanical strength and swells in many oils and fuels, so it is rarely the right answer for a pressurised oil joint even when the temperature says yes.

Can FKM be used below −26 °C?

Special low-temperature FKM grades reach about −40 °C, at some cost in chemical resistance and price. Standard FKM at −30 °C is already stiff enough to leak in a marginal joint.

Why does my seal leak only on cold mornings?

The ring is too stiff to conform at the moment of start-up, then softens as the system warms. It is not damaged. Check the compound's TR-10 against the actual start-up temperature, and consider a slower pressurisation or a pre-heat.

Does low temperature cause permanent damage?

Usually not by itself. The damage comes from what happens while the ring is stiff: it can crack if it is forced over an edge during a cold assembly, or take a set if it is left compressed at low temperature for a long period. Cold flexibility is a temporary condition, not a permanent one.

How do I specify a low-temperature seal to a supplier?

Give the minimum service temperature, the soak duration, the pressure at that temperature, and the test method you will accept (TR-10, brittleness, or a cold leak test). A single "−40 °C" in a purchase order will be interpreted as the most favourable of the three.

Scope

Figures quoted here come from published standards and compound datasheets. They describe typical behaviour, not a guarantee for a specific batch or application. Prototype and life-test before release.