PHOTO: IBM

Richard L. Garwin is an IBM Fellow Emeritus and a fellow of the IEEE, the American Physical Society, and the American Academy of Arts and Sciences. He earned his doctorate in physics under Enrico Fermi at the University of Chicago in 1949, played a key role in designing the first hydrogen bomb, at Los Alamos National Laboratories, and worked at IBM from 1952 to 1993. He has taught at Columbia University and Harvard’s Kennedy School of Government, and he has advised the Pentagon and the government for 40 years. He served from 1993 to 2002 as chairman of the U.S. State Department’s nonproliferation advisory board and in 1998 as a member of the Rumsfeld commission on emergent ballistic missile threats.

IEEE Spectrum sounded Garwin out on his views about the presumed North Korean nuclear test of 9 October. Four days later the U.S. government detected radiation, and on Monday 16 October it confirmed that a nuclear test had indeed occurred. —William Sweet

IEEE Spectrum: Dr. Garwin, are you confident that a test in fact took place—that it was not a fake?

GARWIN: There was an explosion. I think it’s highly probable that it was a nuclear weapons test.

It was a fraction of a kiloton. They could easily and affordably have made an ammonia nitrate/fuel oil explosion of that magnitude: 500 tons [of material] is just enough to fill a substantial room, 10 meters by 10 meters by 5 meters high.

But why in the world would they have done that? And why would they have told the Chinese just 20 minutes before the test that they were getting ready to test a 4-kiloton device, as CNN has reported, if they knew it was going to be a lot less?

I gather from what you’re saying that you accept the mainstream estimate of about a half kiloton?

Yes, more or less, with an uncertainty range of about a factor of two.

What do you conclude then about the test? If they were shooting for 4 kilotons but got a half kiloton, what do you suppose happened?

They were shooting for four or even more. If I were getting ready to test, I wouldn’t give people my central estimate, I’d tell them what I’m really quite confident of achieving. So perhaps they were really aiming for more like 5 to 15 kilotons, as the Russian foreign minister has reported.

But let’s assume [conservatively] they were going for 4 or 5 kilotons and they got a half kiloton, maybe. There are two possible reasons: they got the design wrong, or they just had bad luck.

How could they have got the design wrong?

If they had just copied the U.S. design of 1945 they would have had an 8000-pound [3629 kilogram], 5-foot diameter bomb that would have given them 20 kilotons, and that would have been hard to miss unless they misfired some of the explosive detonators, or whatever.

But probably they were trying to make something in the 500- to 1000-kg class that would fit on their SCUD missile, for threatening South Korea, or on their Nodong [missile], for delivery to Japan. Then they would be in uncharted territory, and they may just have done the calculation wrong, for example in the amount of plutonium or explosive that’s required to obtain the design yield.

But they also could have just been unlucky.

Remember, when the Los Alamos group was readying the first U.S. test bomb, Robert Oppenheimer wrote to General Groves telling him that there was a chance of a stray neutron triggering a fizzle. The plutonium itself releases neutrons spontaneously—which is why you can’t make a gun-type nuclear explosive from plutonium, something that was discovered very late in the game at Los Alamos. Oppenheimer said there was a 2 or 3 percent chance of a fizzle, and that there might be a substantial reduction in yield—but that it could never go lower than 5 or 10 percent of the expected yield. [In a gun design, a bullet of fissile material is shot at a target of the same material to form a critical mass, which is necessary for detonation. Stray neutrons, however, could trigger a small, “predetonation” explosion, or fizzle, that forces the bullet from the target, forestalling the process.]

So North Korea could have been shooting for 10 kilotons, but got just a half kiloton because of bad luck.

Right.

I gather you completely exclude the possibility of their deliberately trying to make a very low-yield weapon, which would be very difficult, would it not?

There’s no chance they were trying to do that.

And it must have been a plutonium device? They couldn’t possibly have obtained enough highly enriched uranium to, for example, build a primitive gun-type bomb?

I don’t know where they would have got it, and that would have taken about 60 kg of fissile material, as opposed to 5 or 6 kg of plutonium for an elementary implosion device, in which multiple explosive charges compress a ball of plutonium into a ball of higher density and thus lower critical mass.

How significant is this test, given that the mainstream intelligence view for several years has been that North Korea already has several nuclear weapons?

I said that myself, as a member of the nine-member Rumsfeld commission. We judged in 1998 that they had one or two nuclear weapons.

This gets pretty speculative, but do you think they were just testing something they had sitting around for some time, or were they trying to do something new or better?

It’s to be assumed that they have about 40 kg of separated plutonium now, enough for another six implosion weapons. But they might have taken plutonium from one of their earliest bombs and inserted it into something slimmer [to make a more deliverable weapon]. We’ve done that kind of thing.

On balance, given that we’ve been assuming the North Koreans have a small nuclear stockpile, and considering that this test was not as successful as they were looking for it to be, how would you assess the significance of this event?

I would say that they are not quite as far along as we thought and probably they thought. But it won’t take them long, so I’d expect another test within a few months that’s likely to be a successful 4- or 5-kiloton device.

Of course each time they test they use up 5 or 6 kg of that valuable stockpile, so there’s a cost to testing.

But on balance you think they need to mount a more successful test, and that they will.

Yes.