Accident investigators have stumbled on a potentially catastrophic shortcoming in the explosive bolt attachment system used to latch a space shuttle's boosters to its external fuel tank. While there is no evidence Columbia was struck by falling debris from a faulty "bolt catcher" during its launching Jan. 16, corrective actions almost certainly will be required before shuttle flights resume.

Members of the Columbia Accident Investigation Board said today a foam strike during launch remains the most plausible explanation for the breach in the shuttle's left wing leading edge that ultimately doomed the orbiter and its crew during re-entry Feb. 1. One of the more intriguing aspects of the failure scenario is the rapidity with which a plume of super-heated air managed to eat its way inside during the first moments of Columbia's descent into the discernible atmosphere.

The accident board today released dramatic video of a test using an 8,000-degree arc jet that vividly illustrates just how fast a small breach can deteriorate into a gaping hole. Engineers fired the jet of super-heated air at a circular one-inch-wide hole in a strip of aluminum representing a wing spar, the metal structure directly behind the carbon composite panels making up the leading edge. Like a match flame burning through tissue paper from below, a one-inch hole in the plate grew to six inches across in just 20 seconds. The plume also burned through a thick wire bundle in less than one minute, in generally good agreement with the way the plume that burned into Columbia's wing quickly ate through sensor wiring once inside.

"It's really very impressive. You cannot imagine the destructive power of the gases that would flow in through that hole," said Douglas Osheroff, a Nobel laureate in low-temperature physics and a member of the accident board. "It's really pretty scary to see."

During a normal shuttle entry, the maximum temperatures experienced by the wing leading edges and the ship's carbon composite nose cap is 3,000 degrees. The temperature is moderated by the way air smoothly flows across the skin of the shuttle, setting up what is known as a boundary layer that acts as an insulator of sorts. But in a breach, no such protection exists and temperatures can soar much higher, "so hot that in fact the oxygen molecules are split into individual atoms," Osheroff said. "That makes them much more highly reactive."

"It was just very impressive to see how quickly that kind of an atmosphere, simulating as best they could the conditions on re-entry, would slice through aluminum in particular," he said. "It was just absolutely amazing."

Osheroff is the second Nobel prize winner to sit on a shuttle accident board. Richard Feynman served in that capacity for the 1986 Challenger disaster, performing a memorably simple experiment during an early hearing that showed how booster O-ring seals are affected by cold weather. He dipped a small O-ring into a glass of ice water, held it up and showed how it had stiffened, a characteristic that would have reduced a real O-ring's ability to work properly during cold weather. Challenger, of course, was destroyed by an O-ring seal failure during launch in cold weather.

Today, Osheroff discussed a modest experiment of his own, carried out in his kitchen with about $100 in equipment, that suggests NASA has never understood the mechanism responsible for external tank foam shedding. It also suggests, to borrow a phrase from Apple Computer, that Nobel prize winners "think different."

"I became rather rapidly interested in the properties of this foam," said Mac-user Osheroff. "It's really fascinating stuff. It's highly anisotropic, that is to say it's mechanical properties depend upon which direction you squeeze it, for instance. So I decided to do some experiments on the foam."

Osheroff, with the help of a graduate student, had a metal plate machined to serve as a stand-in for the skin of the external tank. He then glued a cube of BX-250 to the plate, a sample of the same type of foam that came off Columbia's tank. A small hole in the plate allowed him to apply pressure at the interface between the foam and the metal.

"The idea was to try to understand, as I increased the pressure, how ultimately this resulted in some sort of a fault that propagated through to the surface," Osheroff said. "This is important because for many years, people at NASA assumed, in fact, that one of the main mechanisms for foam shedding from the external tank was that liquid nitrogen (from trapped air) would somehow condense in a void or something inside the foam near the metal surface. And as you started getting aero heating (during launch), this liquid nitrogen would warm up, pressure would build up and it would throw foam off of the external tank.

"What I found was that, in fact, the mechanism by which the liquid expands is not consistent in any way with ejection of foam from the surface. What it does, it tends to make a two dimensional, rather flat crack or fault, which propagates up to the surface and it meets the surface normal to the surface in almost every case. This has to do with anisotropic properties of the foam."

Translation: Expanding gas in a pocket, or void, near the ultra-cold skin of an external tank produces linear surface cracks in the foam, it doesn't blow out the overlying material. In addition, calculations show heating due to atmospheric friction would not have had time to turn any pockets of liquified air into a gas by 81 seconds into flight.

"That's less than 30 seconds into aero heating," Osheroff said. "Is it possible for the heat to propagate through the foam and actually boil off the liquid nitrogen that might have condensed in that brief period of time? ... The thermal relaxation times are much too long for that."

"So the conclusion I have reached, and that independently the people at Marshall Space Flight Center have reached, is that the process by which foam is ejected is undoubtedly a very complex one involving more than just cryo condensation and ejection.

"I dare say, in fact, that these sorts of experiments which I have done, which were actually done in my kitchen at home for about a hundred dollars, are the sorts of things I think we need to see more of done. Specifically, experiments to try to understand the physical mechanisms (behind) why the foam behaves the way it does."

In a revelation that generated quite a bit of media interest, board member John Barry revealed investigators are looking into apparent problems with the system used to capture and restrain pieces of the exploding bolts used to hold a shuttle's boosters to its external tank during launch.

The boosters separate two minutes and five seconds or so into flight when explosive charges cut the massive 80-pound bolts in half. The upper half of a bolt is blown upward and captured in the dome of a fully enclosed "bolt catcher" assembly on the tank. The bottom half is captured by a bolt catcher on the booster.

The bolt catchers are designed to prevent any debris from a blown bolt from getting into the airstream and possibly impacting the shuttle.

"The problem we found was the original certification was done without the real flight hardware in 1979," Barry said. "The other thing we found out, this bolt that was used in STS-107 (Columbia) was done with a new vendor and the NDE, the non-destructive evaluation, wasn't done as well as it should have been.

"They did some bolt static tests that resulted in this dome fracturing at a lower pressure than was anticipated," he said. "In fact, it was below a 1.4 safety margin. So this dome is made of aluminum and covered with ablative. If that comes loose, with or without that half of the bolt in it, it still can cause some serious risk to the orbiter. So this is a possible return to flight issue that we're examining."

Complicating the picture, radar data from Columbia's launching shows unidentified debris separating from the shuttle 126 seconds into flight, right about the time of booster separation. Debris has been seen on radar during past missions, presumed chunks of ice, for example, and there is no evidence the debris seen during Columbia's flight involved the bolt catcher. Certainly there's no recorded data or telemetry from Columbia suggesting any kind of an impact 126 seconds into flight.

But engineers cannot rule out the possibility debris may have been released into the airstream.

"What we have here is a possibility that we have found another source of debris," said CAIB chairman Harold Gehman. "We don't have any evidence that it was a source of debris except the radar tracking of the Columbia indicated that at the time of SRB separation, 126 seconds, at a time when there's not supposed to be any debris, it noticed a piece of debris. We don't know what that was."

He said engineers are trying to determine if a bolt catcher failure could "create any debris which might fall back on the wing?"

"We're just at the front end of this and we're not ready to make any statements about how this affects the process. The question, then, is we have a potential piece of debris here now. Or in a future flight. But in this particular case, STS-107, it's a potential that we've got a piece of debris.

"Then we go back to the radar of the launch. ... And lo and behold, at 126 seconds after launch, at the time of SRB separation, something is seen on the radar which indicates that there's a piece of debris ejected from the separation. It could be the bolt catcher. We can't prove that. But in an effort to positively close out the fault tree - and you multiply this times a thousand - you can see why this investigation has taken five months. So here's one we can't close out."

But he and other board members downplayed the possibility that a 40-pound bolt fragment could have hit Columbia without being detected by the shuttle's myriad sensors. The real issue is fixing the problem before flights resume.

"There's no indication, in either the (recorded) OEX data that we know about or any of the telemetry, that something hit the wing past 120 seconds," Barry said.

"What we're trying to couch here is it's very important that we understand all potential debris. We're not changing our working scenario, it's still pretty evident foam came off and hit the wing. But we also have to take into consideration any other future debris elements that could be potentially catastrophic to the orbiter."