Early on in the 16-day mission, NASA and its contractors assumed what they saw hitting Columbia was dry foam, something akin to a Styrofoam cooler or a boater's life jacket. The assumption colored the engineering analysis that deemed dry foam was too light to do enough damage to endanger Columbia or its crew.

If the debris included ice, it could be heavier, a possibility raised by engineers inside and outside the agency.

"Think bowling ball," said former NASA engineer Gregory Sakala of Titusville.

The makeup of the debris has become a major line of inquiry for the Columbia Accident Investigation Board, which has assigned at least a dozen teams to one job: "Follow the foam."

The analysis to determine what the debris is made of may be finished as early as this week.

"I think that that's still an open question as to whether or not there might be ice in there or not," said inquiry board chairman, retired Adm. Harold Gehman.

Conditions show recipe for ice

In an effort to resolve the ice question, Florida Today obtained weather readings from launch pad 39A and compared the conditions with past weather data, NASA ice research and inspection reports from past shuttle missions. A computer-assisted analysis of the data, and interviews with shuttle and weather specialists, indicates a recipe for moisture, frost and ice on the big tank.

Columbia made the three-mile journey from the Vehicle Assembly Building to the launch pad Dec. 9. The tank sat outside for 39 days, exposed to nearly 10 inches of rain and almost daily humidity above 90 percent.

In the overnight hours of its last day on Earth, the shuttle was aglow on Pad 39A, towering out of a soupy fog blanketing the Cape.

It was jacket weather in Florida, moist and temperatures in the mid-40s Fahrenheit on the ground. Sixty feet up, on the launch platform, the temperature hovered below 50 degrees Fahrenheit. The humidity was almost 100 percent when the launch team gave the go-ahead to start pumping super-cold liquid fuel into the massive tank.

Anywhere else, that's just nippy weather. Up on the pad, it's a different world. The presence of a half million gallons of liquid hydrogen, at minus 423 degrees Fahrenheit, and liquid oxygen, at minus 298 degrees Fahrenheit, changes everything.

That's one reason why manufacturer Lockheed Martin sprays an inch-thick layer of polyurethane foam onto the tank. The insulation stems the growth of frost and ice that could come off the tank and pelt the shuttle during launch. But the deep freeze inside means temperatures on the insulation surface can be 10 to 30 degrees cooler than the air outside.

So with temperatures as high as 60 degrees, and high humidity, condensation can turn to frost and ice. That's especially true where the foam is thinner, cracked or somehow altered, according NASA-sponsored research.

Thick ice can form in warm weather

In 1983, the U.S. Army Corps of Engineers recreated the wide range of atmospheric conditions the foam must endure, including varying temperatures, humidity and wind. Then they watched what happened when moisture formed on the insulation.

Droplets trickled down the test foam. Even at air temperatures far above freezing, the drips pooled and froze on thinner foam as well as inside cracks or tiny defects in the surface. The researchers found ice grew dangerously thick in conditions warmer than NASA's models had predicted. Experiments also showed ice patches could linger for hours even as temperatures rose.

"A potential hazard to the orbiter tiles which has not been previously identified could occur during relatively cool and humid ambient conditions as a result of extensive frost formations," the report said.

". . . The avalanche of frost at liftoff could be large enough to be of concern."

The Florida Today analysis compared the Army experiments, weather and ice reports from past missions and the conditions at Pad 39A to determine ice likely formed on Columbia's tank Jan. 16.

The weather that day was similar to a past Columbia launch.

That day in 1990, temperatures stayed under 54 degrees, with 100 percent humidity. Inspectors touring the pad three hours before launch saw condensation running down the tank and forming patches of ice and frost. They reported ice at the pad did not violate safety rules and, once things warmed up, Columbia blasted off.

Once in orbit, the tank tumbled away. Pictures showed divots, including one as wide as 28 inches, in the tank near the same spot where foam came off on Columbia's last flight. When Columbia landed, inspectors counted more than 100 tiles hit by debris and measured one gouge 2-by-3 inches. The damage was deemed less than average.

The ice formation that day was not unique. Some frost or ice forms on almost every tank, even during hot, sunny Florida summers.

Treatments allow moisture to penetrate

NASA discounts any suggestion of water or ice-laden debris.

The fuel tank foam is closed-celled. That means individual cells are tightly packed together so other molecules, even water or gas, can't get inside. The bulk of it is sprayed on the tank at a plant near New Orleans, mostly by robots. The outer layer hardens into a sort of rind, an orangish skin that further protects it from the moist air outside. This is the kind of foam shuttle program manager Ron Dittemore showed the news media in the days after the accident to bolster his point that the material is lightweight and impervious to moisture.

But the foam suspected of popping off Columbia's tank is different. It doesn't have that protective outer layer. It's called "close-out" foam because it's applied near the end of manufacturing, by workers using their hands, molds and tools. Some of the work is done in Louisiana; some at KSC where crews attach the tank to its orbiter and solid rocket boosters.

A perfect example of that kind of foam are the bipod ramps, the triangular blocks that fell off during Columbia's launch and at least four previous launches. Workers pour the foam for the twin ramps into place near metal struts that attach the tank to the orbiter's nose. They use tools to cut the foam to an aerodynamic shape.

In that general area, workers shave or sand other close-out foam. They also use what looks like a wire brush to poke tiny holes in large tracts of nearby foam. The process called venting was meant to let gas trapped inside escape instead of expanding and blasting the foam off the tank.

These treatments can provide a path for gas and moisture to get inside the foam. The workers are slicing open the walls of those closed cells and removing the polyurethane's hard skin.

"The presence or the absence of that skin has a dramatic effect," said Gordon Nelson, professor of chemistry at Florida Tech in Melbourne, Fla.. Nelson studies how polymers and similar materials behave under different conditions.

Experiments by companies that make similar closed-cell foams, which are used for everything from airplanes to roofing, show the skinless foam absorbs moisture during prolonged exposure to humidity. A study in the late 1990s by Huntsman Polyurethanes showed the skinless foam could triple in weight after 30 days in a very moist environment.

In the case of Columbia's final mission, the formula could mean the left wing was hit by a chunk of foam/ice weighing up to 71/2 pounds instead of a 21/2-pounds.

But NASA says such tests were conducted on a slightly different kind of foam.

Neil Otte, deputy manager of the NASA program that oversees design and manufacture of the tanks, said the agency tested the foam's resistance to moisture by exposing it to 125-degree temperatures and 95 percent humidity for seven days. The foam absorbed a little water, but never gained more than 1 percent in weight, Otte said. NASA tested both intact and shaved foam.

Otte conceded shaving the foam on the bipod ramps, for example, slices open a layer of cells several millimeters deep. He acknowledged that area could absorb moisture that could freeze into an icy crust under certain atmospheric conditions. But he said that's too thin to be dangerous.

NASA has been redesigning the bipod ramps since last fall, when foam from that area fell off during launch. The foam redesign and efforts to preclude foam loss altogether are on Dittemore's official checklist for preparing the shuttles for return to flight.

Cracks, dents increase ice risks

Other common defects in the foam, such as cracks, dents or shoddy repairs, also can cause problems made worse by moisture.

Moisture can accumulate in the tiniest crevice, freeze into ice and form dangerous projectiles during launch. The temperature can plummet hundreds of degrees as the moisture gets fractions of an inch closer to the tank's metal surface.

Freezing material can expand and aggravate a phenomenon called cryopumping. Gas or moisture gets into air pockets or voids between the insulation and the tank, expanding as the temperature rises during launch. If the resulting gas can't escape as quickly as it needs to, the pressure can blow foam off the tank.

Otte said that's why NASA treats cracks or other defects very seriously.

Florida Today's review of ice team inspections for dozens of past missions indicates some cracks are deemed acceptable, but only in certain locations away from the shuttle's belly or when there is no dangerous ice buildup.

The assessment depends on judgment by inspectors and launch managers.

NASA has delayed launches because of fears about ice. The delays have ranged from several hours of waiting for the air around the pad to warm up, to weeks for repairing cracks or holes in foam.

In two examples in the 1990s, NASA rolled Discovery back from the launch pad to the Vehicle Assembly Building to touch up holes drilled in the foam by woodpeckers and hail.

The reason? NASA engineers determined any holes an eighth of an inch in diameter posed a danger. Ice one-sixteenth of an inch thick can be deemed dangerous enough to scrub a launch, depending on where it forms.

Launch conditions deemed safe

The conditions on Jan. 16 did not violate NASA's weather standards meant to minimize ice. TV monitors, computer temperature models and the human inspection before launch are the last line of defense against ice and debris. A squad of hawk-eyed veterans climbs up and down the launch tower three hours before launch. They look for possible debris, including cracked insulation, condensation, frost and ice. The inspectors use temperature scanners and binoculars to spot problems.

A 130-page checklist the team used during its Jan. 16 inspection of Columbia, obtained by Florida Today under the Freedom of Information Act, shows inspectors saw no ice buildup that violated safety rules.

Handwritten notes indicate frost was spotted on at least three regions of the tank. The team also noticed frost and ice coming off the shuttle stack during liftoff. A more detailed report like those from past missions may not be prepared about STS-107, NASA said Friday.

The checklist indicates the team was allotted 10 minutes at the 195-foot-high level of the launch tower to inspect the bipod ramp and at least seven other areas. There's no indication they looked again at those places.

Otte said NASA data challenges the Florida Today conclusions about ice, but would not elaborate. He suggested interviewing ice team members, but KSC has denied requests to interview the inspectors.

Shuttle and ice experts, including workers who've performed the final ice inspection, say there is no way to see everything as they scan the 15-story tank from platforms 75 feet from the tank.

Cracks in the foam, even if spotted by the team, could harbor moisture and ice that would not be spotted. Ice can worsen in the three hours between the inspection and liftoff, even if conditions at the pad warm up.

NASA engineer Gregory Katnik, an 18-year veteran of the inspections, said despite not being able to see everything, the team knows what it's looking for.

"It gets to the point where you can tell when something is wrong," Katnik said of the inspection team. "You have looked at so many vehicles over the years, you can tell something is out of the ordinary. There are pages and pages of checklists. . . . You know there should not be liquid dripping here and there should not be a protrusion there."

B.K. Davis, a retired NASA external tank manager who now lives in Cocoa Beach, Fla., trusts the inspection and other processes that make sure there is no dangerous ice before launch.

"Everyone talks about ice, but that is a red herring," Davis said.

Still, as part of the investigation, NASA weather experts are analyzing rain, humidity and other data from Columbia's stay on the pad and previous launches. They're analyzing tile damage reports to see whether there's a relation between atmospheric conditions and debris, weather officer John Madura said.

If investigators conclude there was ice, it would raise new questions about how NASA managers and their contractors analyzed possible damage while Columbia still orbited Earth.

The analysis assumed the debris was light foam only. The mission managers concluded that foam could damage the wing but not badly enough to destroy the ship during re-entry. The analysis said small changes in the weight of the debris could cause more serious damage.

In the end, the space agency's managers apparently rejected the possibility of icy debris. Not all engineers liked that assumption.

In a pre-Feb. 1 e-mail to colleagues debating the damage, engineer Dan Mazanek of Langley Research Center noted if the debris were solid ice, it could be 30 times as heavy as foam.

"That would be the equivalent of a 500-pound safe hitting the wing at 365 miles per hour."

Published under license from FLORIDA TODAY. Copyright © 2003 FLORIDA TODAY. No portion of this material may be reproduced in any way without the written consent of FLORIDA TODAY.