The spread of nuclear know-how

In the early 1970s, at a factory in the Dutch town of Almelo, the governments of Britain, West Germany, and the Netherlands were perfecting a secret uranium-enrichment technology: the ultracentrifuge.

The machines were made of precisely crafted tubes of metal that spun at fantastic speeds.

The centrifugal force this spinning created was so great it could physically separate the different isotopes of natural uranium.

Naturally, this technology was housed in a factory that was supposed to be secure.

But in practice the atmosphere at Almelo was relaxed. The centrifuge building housed a snack shop, and workers without full clearance routinely filtered through - including a well-liked Pakistani metallurgist named Abdul Qadeer Khan.

Other workers thought nothing of their repeated sightings of Dr. Khan walking through the centrifuge facility, notebook in hand.

Fast forward to 2004. Khan, who became the father of Pakistan's nuclear-weapons program, has admitted to peddling nuclear know-how for profit - and the secrets of the centrifuges of Almelo have leaked all over the world.

The characteristics of the machines can be as distinctive as fingerprints. Parts and plans related to centrifuges have proved crucial clues linking Iran, Libya, and Pakistan together in a web of nuclear proliferation.

Their dissemination is particularly dangerous because they can solve the most daunting aspect of building nuclear weapons - acquiring the fissile core.

"There is no secret to making a nuclear bomb," says Matthew Bunn, a nuclear expert at Harvard University's John F. Kennedy School of Government. "The hard part is getting the [fissile] material."

Libya has admitted to International Atomic Energy Agency (IAEA) officials that it first bought centrifuges and centrifuge parts in 1997. This initial batch - enough for at least 220 machines, according to IAEA documents - was similar in design to the first centrifuge model produced by the British-German-Dutch Urenco consortium.

Beginning in 2000, Libya set its sights on a more advanced centrifuge. This design, dubbed "L-2" in IAEA documents, was itself based on a second- generation Urenco centrifuge that uses super strength maraging steel instead of aluminum for rotors. Libya ordered parts for 10,000 L-2s. These components began to arrive in large quantities in December 2002.

Iran, for its part, has some 920 centrifuges of the less-sophisticated aluminum rotor design, according to the IAEA. It declared ownership of these machines to the IAEA last year.

But further investigation - including interviews with ex-Iranian officials - led international inspectors to suspect that Iran knew more than it was saying about advanced steel rotor centrifuges. This January Iranian officials admitted that they had received blueprints from foreign sources for advanced "P-2" machines.

The Iranians said that they decided they weren't capable of making the finely machined rotors out of steel, and instead had tried to make them from carbon composites. This failed. So they did what any backyard inventor frustrated with a balky whiz-bang might do - they threw the whole thing in the garbage.

According to Iran, after June 2003 "all of the [P-2] centrifuge equipment was moved to the Pars Trash Company in Tehran," says the IAEA's recent Iran report.

The IAEA - not to mention the Bush administration - isn't buying this part of the story. They want the Iranians to talk more about what they really have in terms of P-2 equipment.

But Iran continues to insist that its nuclear program is meant only to produce electricity. Squeezing them too hard at this point might be counterproductive, say some experts. They're like someone hauled in by law enforcement for an interview who can leave at any moment, since they haven't officially been charged with a crime.

"We want them to continue cooperating with the police," says Daryl Kimball, executive director of the Arms Control Agency.

The development of gas centrifuges was an attempt to solve a problem which has dogged scientists since the beginning of the atomic age: the tremendous expense and energy involved in refining fissionable material.

The enriched uranium thus produced can be used in nuclear power plants. Urenco began its own work so that Western Europe would not have to depend on the US to supply reactor fuel, for example. But centrifuges can also produce higher enriched uranium. This, or plutonium, is the material necessary for the core of a bomb.

The presence of centrifuges doesn't establish intent to make weapons. But combined with other clues they can be a powerful indicator of an intention to develop a home-grown arsenal.

"The technology is inherently dual-use," says Corey Hinderstein, a senior analyst at the Institute for Science and International Security (ISIS). Thus Western intelligence agencies were suspicious of Iran as early as 13 years ago. In 1991, Italian intelligence reported that Sharif University in Tehran had ordered a sophisticated ring magnet from the Austrian firm Tribacher, according to an ISIS article.

The magnet in question was suitable for use in the upper bearing of a Urenco-like centrifuge.

Centrifuges work by spinning at a very high speed - close to or surpassing the speed of sound. Uranium gas is pumped inside the spinning cylinders.

The gravitational force is so strong that the heavier molecules of U-238, the most-common isotope in natural uranium, move toward the outside. The lighter, much rarer, and highly fissionable isotope U-235 collects closer to the center.

A stream of gas slightly enriched in U-235 is withdrawn and then fed into the next of a train of centrifuges, and so on until it becomes more than 90 percent pure.

This is simple in theory but highly difficult in practice. The precision necessary to keep intact a rotor moving at more than 100,000 revolutions a minute is at the limits of modern engineering methods.

In fact, the Urenco P-1, the base design for the first machines acquired by Libya and Iran, was never all that great, according to David Albright, head of ISIS and a former international weapons inspector.

Thus the proliferation network which provided them may have been selling off the centrifuge equivalent of bug-ridden version 1.0 software.

"The P-2 - now that worked like a charm," says Mr. Albright, former international weapons inspector.

• Faye Bowers contributed to this report.