The Cold-Stiffened O-Ring

How Fluoroelastomer Glass Transition Defeated Dual Redundancy on Challenger

Sub-freezing Arctic air poured across Florida's Kennedy Space Center on the night of January 27, 1986, hours before the scheduled launch of Space Shuttle Challenger (Mission STS-51-L). From Morton Thiokol headquarters in Brigham City, Utah, staff engineer Roger Boisjoly and his seals team pleaded on an emergency teleconference with NASA managers to halt the countdown. Six months earlier, after inspecting charred O-rings from prior shuttle boosters, Boisjoly had written a formal company memorandum warning that a failure of the Solid Rocket Booster (SRB) field joint seals would destroy the vehicle and cost the crew their lives. During the teleconference, Boisjoly presented telemetry from a January 1985 launch at $53^\circ\text{F}$. In that flight, cold had stiffened the primary synthetic rubber O-ring enough for black combustion gas to blow past it. He warned that launching at the forecast $26^\circ\text{F}$ to $29^\circ\text{F}$ fell far outside the booster's safe operating envelope.

After NASA's Marshall Space Flight Center managers challenged the engineering recommendation, Thiokol senior executives overruled their own engineers and signed the launch waiver. Challenger lifted off at 11:38 AM on January 28, 1986, at an ambient air temperature of $36^\circ\text{F}$. Heavy ice coated the launch structure, and infrared sensors measured the shaded right-hand SRB aft field joint at $28^\circ\text{F}$ ($-2.2^\circ\text{C}$). Six hundred milliseconds after ignition, puffs of black smoke spurted from the right aft field joint. At $T + 58$ seconds, a blowtorch plume of $6,000^\circ\text{F}$ aluminized flame burned through the casing and severed the lower strut attaching the booster to the external liquid hydrogen tank. The orbiter broke apart under aerodynamic forces at $T + 73$ seconds, killing all seven astronauts.

The field joint failed through a lethal interaction between structural "joint rotation" and polymer glass transition kinetics. Each 149-foot SRB was stacked from four cylindrical half-inch D6AC steel segments joined by a tang-and-clevis connection. The upper segment's protruding steel blade (tang) slid into the lower segment's U-shaped groove (clevis). That joint was locked by 177 steel pins and sealed against the inner clevis wall by two redundant 0.280-inch-thick Viton fluoroelastomer O-rings.

Unlike conventional pressure-actuated seals that tighten under load, the SRB joint flexed backward at ignition. When combustion pressure reached $1,000\text{ psi}$ in the first 0.6 seconds, the thinner cylindrical steel case walls above and below the joint ballooned outward three times further than the thick tang-and-clevis ring. That differential bending levered the steel tang away from the inner clevis lip—a motion called joint rotation. Within milliseconds, joint rotation opened an extrusion gap of up to $0.042\text{ inches}$ across both the primary and secondary O-ring grooves just as peak combustion pressure arrived.

For either O-ring to seal during joint rotation, the compressed polymer had to rebound elastically across that widening gap in less than 600 milliseconds. In amorphous fluoroelastomers, viscoelastic relaxation time increases by orders of magnitude as temperature $T$ approaches the glass transition temperature $T_g$, governed by the Williams-Landel-Ferry equation:

As Nobel physicist Richard Feynman demonstrated during the televised Rogers Commission hearings by clamping an O-ring in a glass of $32^\circ\text{F}$ ice water, cooling Viton to $28^\circ\text{F}$ pushed the rubber into its leathery glass-transition regime. Its rebound speed dropped so sharply that the cold-compressed O-ring stayed flat for several seconds after release. Because the ambient cold affected both O-rings identically, a single environmental variable defeated both the primary and backup seals at once.