The insulating foam that appears to have broken away from shuttle Columbia's main fuel tank at launch, possibly inflicting fatal damage to the shuttle's protective tiles, has been a long-standing headache for NASA.

The foam performs a critical insulating function before and during launch, keeping the 537,000 gallons of liquid fuel in the tank freezing cold while preventing a buildup of ice on the outside. But the demands placed upon that inch-thick layer of foam stretch the limits of modern materials science.

Most notably, it must be able to withstand temperatures as low as 423 degrees below zero Fahrenheit on the inside -- the temperature of the liquid hydrogen in the tank -- while tolerating temperatures that can soar to 1,200 degrees on the outside from the rocket exhaust and aerodynamic friction during launch.

While strained by those extremes, the foam must remain tightly bound to the shuttle's enormous aluminum alloy fuel tank -- even as the tank shrinks several inches as it gets filled with the freezing fuel. The epoxy cement used to bind the foam to the tank appears to do its job very well -- so well, in fact, that under stress, the foam has in many cases simply ripped apart, leaving a layer attached to the tank while the rest crashes against the shuttle, sometimes at several times the speed of sound, according to at least one NASA report.

But political pressures may also have contributed to the foam's tendency to break away at launch. A change in the foam's formulation, instigated in 1997 as part of NASA's effort to use more environmentally friendly chemicals, was considered a possible cause of a sudden increase in the amount of foam breaking off that year.

Scientists and federal officials yesterday said they do not know whether the foam, which apparently blew off the main external fuel tank soon after liftoff, caused a degree of damage to the shuttle that could have led to the catastrophic breakup Saturday morning over Texas. But it is "a primary area of analysis," shuttle program manager Ron Dittemore said yesterday.

The external tank on which the foam is sprayed is the largest single component of the space shuttle system, and the only part that is not reusable. At 154 feet long and almost 28 feet in diameter, the tank has a total surface area of about two-thirds of an acre.

In addition to holding the twin internal tanks of liquid fuel to power the main booster rockets, the external tank serves as a structural "backbone" for the launch bundle, which includes the tank, two solid-fuel boosters and the shuttle craft itself. All told, the external tank consists of about 480,000 separate parts, including 38 miles of wire.

The top third of the external tank contains a cone-shaped liquid oxygen tank, kept at minus 297 degrees F, and the body of the tank just below that holds a cylindrical inner tank filled with liquid hydrogen, chilled to minus 423 F. Both ingredients are gases under normal atmospheric pressure and temperatures. But when chilled to very cold temperatures they change to a highly condensed liquid form, which takes up far less space. If the same amount of hydrogen and oxygen were to be placed in their gaseous state in a tank with the same diameter as the shuttle's, it would have to be 16 miles tall from ground to nose cone.

During liftoff, the liquid oxygen and liquid hydrogen gush through fuel lines 17 inches in diameter at 1,053 gallons per second. Within 81/2 minutes, the 537,000-gallon tanks are empty.

The foam helps keep those chemicals in their liquid form by insulating them from the Florida heat. It also keeps the tank's freezing metal surface from coming into contact with the humid air, which would result in the immediate buildup of ice on the tank's outer skin. That would pose an additional risk to the shuttle, both from icicles breaking off during liftoff and by adding significantly to the craft's weight.

Even with the foam in place, some ice typically forms on the outer surface. NASA officials reiterated yesterday they are not completely certain whether the spray of debris seen during Columbia's launch was foam or ice or a combination. Dittemore said the agency is rerunning its initial calculations of what harm may have been done to the shuttle if the debris were very icy, and therefore more dense, than plain foam.

The foam is sprayed on the tanks at Bethesda-based Lockheed Martin's Michoud Assembly Facility outside New Orleans. It dries to its familiar deep orange color, company spokesman Jeff Adams said, which makes it easy to track by ground-based cameras.

In 1997, before Columbia was launched on shuttle mission STS-87, NASA switched to a new method for foaming the external tank, according to a NASA "field journal" published late that year. The goal was to eliminate the need for Freon, a volatile gas that has been implicated in the destruction of the Earth's protective ozone layer.

NASA's Gregory N. Katnik, a shuttle technical manager, noted in that report that Columbia had 308 "hits" or areas of damage to its tiles upon landing on Dec. 5, 1997, far more than the standard number, which he said was about 40. Of those damaged areas, 132 were more than an inch long, and some measured 15 inches long, with the two-inch-thick tiles pierced three-quarters of the way through.

Among the top suspects for causing that damage, Katnik wrote, was foam that splintered off the fuel tank at liftoff. It may seem strange to think that foam could damage ceramic tile, he wrote, but "when that foam is combined with a flight velocity between speeds of MACH 2 to MACH 4, it becomes a projectile with incredible damage potential."

Independent experts have confirmed the damage potential from the foam. Elisabeth Pate-Cornell, a professor and chairman in the department of management science and engineering at Stanford University, led a shuttle risk analysis in 1990 that looked at the issue.

Based on data from simulations conducted by NASA and other sources, Pate-Cornell and her colleagues constructed a mathematical model to predict the likelihood of tile damage contributing to a shuttle disaster. They concluded that 10 percent of the overall risk of an accident came from the possibility of tile failure, and 40 percent of that risk was from the possibility of the tiles being hit and damaged by debris, including external tank foam debris.

NASA's Katnik hypothesized in his 1997 report that the extra damage seen on STS-87 may have been attributable to the new foam formulation, which may not have been stable in the rugged "ascent environment." He wrote that an investigation would be undertaken by the Marshall Space Center.

Katnik did not respond to a phone message left for him yesterday. Spokesmen with Lockheed and the Marshall Center in Texas were not able to determine yesterday what the result of that inquiry had been.

Staff writer Rob Stein contributed to this report.

Employees at Lockheed Martin's Michoud Assembly Center in New Orleans pose in 1990 with four space shuttle external fuel tanks.