Richard Garwin - Session II

Notice: We are in the process of migrating Oral History Interview metadata to this new version of our website.

During this migration, the following fields associated with interviews may be incomplete: Institutions, Additional Persons, and Subjects. Our Browse Subjects feature is also affected by this migration.

We encourage researchers to utilize the full-text search on this page to navigate our oral histories or to use our catalog to locate oral history interviews by keyword.

Please contact [email protected] with any feedback.

Image not available
Interviewed by
Dan Ford
Interview date
La Jolla, California
Usage Information and Disclaimer
Disclaimer text

This transcript may not be quoted, reproduced or redistributed in whole or in part by any means except with the written permission of the American Institute of Physics.

This transcript is based on a tape-recorded interview deposited at the Center for History of Physics of the American Institute of Physics. The AIP's interviews have generally been transcribed from tape, edited by the interviewer for clarity, and then further edited by the interviewee. If this interview is important to you, you should consult earlier versions of the transcript or listen to the original tape. For many interviews, the AIP retains substantial files with further information about the interviewee and the interview itself. Please contact us for information about accessing these materials.

Please bear in mind that: 1) This material is a transcript of the spoken word rather than a literary product; 2) An interview must be read with the awareness that different people's memories about an event will often differ, and that memories can change with time for many reasons including subsequent experiences, interactions with others, and one's feelings about an event. Disclaimer: This transcript was scanned from a typescript, introducing occasional spelling errors. The original typescript is available.

Preferred citation

In footnotes or endnotes please cite AIP interviews like this:

Interview of Richard Garwin by Dan

Ford on 2004 June 27,Audio and video interviews about the life and work of Richard Garwin, 2004-2012Niels Bohr Library & Archives, American Institute of Physics,College Park, MD USA,

For multiple citations, "AIP" is the preferred abbreviation for the location.


In this interview Richard Garwin discusses topics such as: hydrogen bombs,  Los Alamos Scientific Laboratory, Enrico Fermi, nuclear weapons, uranium enrichment, nuclear reactions, Edward Teller, thermonuclear burning, Greenhouse George, Stanislaw Ulam, Soviet Union, Richard Rhodes, Marshall Rosenbluth, International Business Machines (IBM), IBM Watson Scientific Computing Laboratory, Project Lamplight.This interview is part of a collection of interviews on the life and work of Richard Garwin. To see all associated interviews, click here.



I wanted to start talking about the work on the hydrogen bomb, and I guess part of my question was, before you went to Los Alamos, what did you know?


In April 1947, I married my wife, Lois Levy, a childhood friend, and began my studies at the University of Chicago in physics. I finished my graduate work in December 1949. Lois helped me by running the calculator, a desk calculator, for my thesis. She had had two jobs by then to support us. She worked for Blue-Cross Blue-Shield in Chicago and then, in the summer of 1948, when we were home in Cleveland for the summer, she got a job with the Ohio Bell Telephone Company courtesy of my Uncle Nate [one of my mother’s brothers, Nathan A. Schwartz], who was comptroller there.

Then she could get a job with the Illinois Bell Telephone Company, which was a better job than the Blue-Shield job.


When I finished my degree, Enrico Fermi had persuaded the Physics Department at the University of Chicago to hire me as a faculty member — an instructor, the lowliest position — and I began my independent work in physics. The University of Chicago paid its faculty nine months of the year, and for the summer they were expected to get government grants or some other way of support. But my family ate 12 months of the year, and I really didn't know how to get government grants.


Did you have your children by this time?


So Fermi suggested that I could be very useful if I were a consultant at Los Alamos. He had gone to Los Alamos after the war as a consultant. I guess I had made some comments to him about nuclear weapons, and he thought that we could discuss those only after I had my weapon clearances and began work on the subject.

Our first child, Jeffrey, was born November 1949. We have had two more: Thomas, born March 1953, and Laura, born July 1957. When we went to Los Alamos in June, probably, 1950 for the first time, we drove from Chicago and had a nice little rental apartment on the second floor of a garden complex around a green lawn. It was just big enough for the two of us and our baby, who was six months old by then.

We had a good time. The apartment was rented to us by the Zia Company because there was no private ownership of houses or land in Los Alamos at that time. Zia owned all the furniture too, so it was easy for them to provide us with the furnishings that were required: A truck drove up and delivered an apartment-full of furnishings.

We were very happy there. The first thing I did was to go to the Classified Report Library to read all the secret documents about nuclear weapons. It had the weekly progress reports from all the groups at Los Alamos, beginning in March 1943 right up to June 1950.



You said you went to the classified reports laboratory?




Library. Were these things segregated on a need-to-know basis, or could you just read through everything?


The wonderful thing about Los Alamos is that they had set the rules from the very beginning. General Groves, who had selected Robert Oppenheimer to head the laboratory, wanted of course categories of information — different people having different access to information — but Oppenheimer, supported by his governing council of senior people at the Laboratory, insisted that everybody who had a laboratory Q clearance would have access to all secret information with a very few exceptions. The number of nuclear weapons was, for a while, among those exceptions, but all of the scientists had access to all of the same information. They could talk freely to one another. And that was very important for the success of the Laboratory. Groves also wanted to commission the scientists as military officers so that he would have more control over them. Oppenheimer and the governing council refused to do that, and that was good also.

It took a week, or two weeks or so, for me to read all that stuff. I had a notebook like everybody else — a classified notebook that had to be locked up in the safe at night. It was the first time I had a safe of my own, and I was to have a new safe — a different safe — every summer that I went to Los Alamos as a consultant. The first thing you do with a safe is to set a combination for it, so I set a combination which had nothing to do with my birthday, telephone number, or whatever.

One summer I got the safe, I set a combination, I went home, and lo-and-behold, Lois had gotten a telephone installed, and the telephone number was the same as that I had set on the safe — but I had set the number first [laughter]. Nobody would have believed me, so I went back right away and changed the safe combination to something else.

I was fortunate the first summer also to share an office with Fermi. It was a room about, I would say, 18-feet by 8-feet wide. My desk was at the corridor door end of the room, and Fermi's was by the window. There were two desks. We faced one another over our desks. And there were a couple of chairs in the room, and of course the indispensable blackboard with real chalk — not a whiteboard, such as you have today.

I began to think what I could do. At that time, of course, we had only fission bombs — that is, bombs that were based on uranium-235 or plutonium. The idea of a fission bomb is that you have enough of this fissile material suddenly assembled so that it is super-critical. That means that if a neutron causes fission in one of the nuclei and the two or three or four neutrons come out, more than one of those neutrons will cause another fission in one of the nuclei.

If the metal is too spread out, then it doesn't happen: the neutrons escape, so that less than one additional fission is caused per initial fission, and so you'll have, on the average, 100 fissions initially, 90, 81, 72, whatever, and it will go down to zero. But if you have enough material compactly spaced, then you have one fission causes two, causes four, causes eight, causes 16, and in less than one-millionth of a second, you have so much energy produced locally that the material blows itself apart very rapidly. That amount of energy is thousands of tons of high-explosive equivalent. In fact, one kilogram of uranium or plutonium completely fissioned gives 17 kilotons — 17,000 tons — of high-explosive energy release.

Of course some of these things are totally unclassified, but in my reading, I saw all the details of the first nuclear weapon that had been tested — the Trinity test in July 1945 at Alamogordo, New Mexico. That used six kilograms of plutonium made in the Hanford reactors in the State of Washington.


I saw the design of the gun-type weapon that was used against Hiroshima — that used solid rings of uranium-235, and a projectile of uranium-235, and the various innovations that had been made after 1945, in which — and this is all unclassified now or I wouldn't be saying it — the plutonium or the uranium was arranged in a hollow shell with explosives outside it to make it safer and to be able to use less of this precious metal than was the case with the first bombs.


One question I had about the atomic testing program: Were there any duds?


There were very few duds. Maybe one or two. I knew at the time, and there are some reports on this. In fact, Fermi used to say that people were less venturesome than they should have been because there was such a big penalty associated with a test that did not achieve its predicted yield. I learned also about all of the tests that had been made early on to see whether nuclear weapons would work. These were tests underwater, tests ultimately in the atmosphere. Of course, the initial tests — then underground tests — and finally tests in space, which we had not made any by that time.

Pretty soon people had more detailed diagnostics, and there were, in addition to the photography — there was very rapid photography with speeds up to a million frames — a million pictures — per second or more. There were radiochemical tests, first simply to collect samples of the radioactive cloud, to pick up fission fragments and residual uranium and plutonium to determine the yield of the nuclear explosive. Then as people began to try to exploit the energy in fusion instead of fission — that is, the energy that comes from two light elements fusing together to make a heavier element — there were more diagnostic tests to look at those high-energy neutrons — 14-million volt neutrons compared with an average of 2-million volt [neutrons] from fission.

There were, in addition, plans for experiments to explore aspects of the fusion reaction. These experiments were to take place in 1951 at the so-called Greenhouse test [series] in the Pacific. There we were to test a number of explosives. One was simply the largest fission bomb ever tested, which was 500 kilotons compared with the 13-kiloton yield of the Hiroshima bomb and the 20-kiloton yield of the Trinity and the Nagasaki bombs.

There was also a famous test called George — Greenhouse George — because it was the seventh, or whatever letter in the alphabet George is, in the sequence of tests for that series. This was really the first experiment on thermonuclear burning. It had liquid deuterium and tritium, the two heavy isotopes of hydrogen. Deuterium is present one part in 5,000 atoms in every hydrogen sample around, and it is not too costly to get it out of [separate it from] the normal hydrogen that has an atomic mass of one — simply one proton — whereas the deuterium has a photon and a neutron, so it has an atomic mass of two.

This is done typically by electrolysis, or there are other means in which one has essentially distillation — taking advantage of the tiny difference in vapor pressure — of heavy water or, in this case, HDO, because one in 5,000 molecules in water is HDO – hydrogen, and a deuterium, and an oxygen — and only one in 25 million is D20, simply because it is unlikely that there would be two deuteriums in a molecule by accident. These can be separated by multistage distillation or other chemical equilibrium, so it's easy to get heavy water or pure deuterium gas and to make liquid deuterium of that.

It's a different matter to make tritium because it has a half-life of 12 years, and any tritium made at the beginning of the universe is long, long gone. There's a tiny bit made by cosmic rays, and in the modern era, there's about a kilogram of tritium made per year by the heavy-water reactors that Canada operates, simply because of parasitic capture of neutrons in the heavy water moderator.

We made tritium in nuclear reactors especially for these tests by an obvious means — that is, by including in the reactor some lithium. The lithium-6 component of natural lithium — about 10% I think, but I have to look it up — has a very big avidity for neutrons. It captures a neutron to make highly excited lithium-7. Lithium has three protons, and with lithium-7 it has four neutrons, so the breakup of this excited lithium-7 is into a helium-4 — normal helium — and a triton — that is, the nucleus of tritium, one proton and two neutrons. This is a very efficient way to make tritium, but if you are short of neutrons in your reactor or short of neutrons in your production facility, you only get, for one neutron, one tritium atom instead of one plutonium atom. So it really cuts into your plutonium production to make tritium.


The lithium they used, that's the same thing that they use to treat manic-depressive people?


Yeah, this is metallic lithium, or lithium hydroxide, or whatever. Lithium compounds, yes, are used for treating manic depression, but we have to look up what compound is used. I don't think that it is ordinary lithium hydroxide, which is a lye, or lithium carbonate that does it. But we'll look it up. That was a major discovery, in fact, by a person at Brookhaven who died not long ago.


I just find it amusing that…


It is, it is. And it would be a good point to make. So in this one atom of plutonium, when used in a bomb — would give, if it were totally fissioned, 150 million electron volts of energy [“150 MeV”]. That's a unit of energy. It's in my book. We can get to it. The tritium, if it were fully reacted with deuterium — that is, with the heavy hydrogen — would give 17 million electron volts. So this one neutron has its energy potential reduced by a factor-10 almost by being absorbed into lithium to make tritium.

From the very beginning of the nuclear weapons program in 1942, I guess, it had been imagined that one could tap the energy of light elements in this way. Edward Teller relates a conversation he had with Enrico Fermi in New York where, on a walk, Fermi said to him, "You know, if we had an atomic bomb, then we could use it to heat this material, deuterium and tritium, to a high temperature and in that way get energy out of it."

When one is using a fission explosive, there is a limit to the amount of energy you can get because you have to have more than a critical-mass when it is assembled, but less than a critical-mass before assembly. So a gun-type weapon that puts a projectile into a stationary amount of plutonium or uranium can never have much more than two critical-masses when assembled. That means that, since a critical-mass is about 60 kilograms, if you could get 100% efficiency, that would be about a megaton of energy release. But it turns out that the material not being compressed — the efficiency is only a couple percent for a gun-type weapon, and that accounts for the 13-kiloton yield of the Hiroshima bomb.

You can do better with plutonium — or even with uranium — but in an implosion-type weapon, where in advanced weapons you have a large, thin shell of material, that's noncritical — or subcritical — for as much material if you like, if you make it big enough, simply because it has to be thin enough so that a neutron can pass through it with the probability less than 50% percent or so of causing a fission. So if I want to have a whole ton of plutonium, no problem. I just make the shell bigger with the same thickness so that it's subcritical.

Another problem comes in because then it takes a long time to assemble all that stuff into a compact, metallic mass at the center, and during that time, if a stray neutron were introduced either from cosmic rays or from the nuclear reaction of the plutonium alpha particles — it's intensely radioactive with alpha particles — it has a decay time of 26,000 years or so compared with 12 years for tritium and 1,800 years for radium. But a gram of radium is a very intense source for radiotherapy, and we're talking kilograms of plutonium — six kilograms at least for a critical mass. So that's equivalent to 600 grams of radium in its radioactivity. Some of those alpha particles will strike oxygen impurities or beryllium impurities and give a stray neutron, and that would pre-initiate the bomb and cause a very low yield.

It was realized from the beginning that, if you could have large amounts of heavy hydrogen — heavy hydrogen is what powers the sun. That's been known since the 1930s. Hans Bethe received a Nobel Prize for that. And in fact the deuterium from the sun has long ago been exhausted, but there's a little bit that is being regenerated all the time because of the very slow proton-proton reaction. In any case, people were well aware of the deuterium-deuterium reaction, and Fermi suggested it to Teller in that walk in New York.

When people began seriously during the Second World War to organize the Manhattan Project to produce nuclear weapons, the first step was to demonstrate the chain reaction in order to ensure that one would have a reaction that would work. Then there were two approaches.


Excuse me, but did Fermi ever talk to you about that conversation with Teller?


No, he didn't. And in fact, before I went to Los Alamos, Fermi and I never had a classified conversation, although I had a security clearance — a Q clearance — because I needed to use the reactor at the Argonne beginning in 1948 for my work using radioactive materials, and to have access to the Argonne Laboratory, I needed a Q clearance. I still have the same Q clearance number from the Department of Energy that I had from the Atomic Energy Commission beginning in 1948, and it is — it's called a Z number now. There's the 'Q' for Q clearance, and my Z number was 32122, by which I deduced that I was the 32,000 person to have Q clearance.


The first step was to demonstrate the chain reaction and to understand quantitatively how many neutrons there would be per fission. The two approaches to building a nuclear weapon were, once fission was understood, to separate the uranium-235 from natural uranium, and that was regarded by many as an impossibility. The other approach was to produce an isotope — really a new element, plutonium-239 — from uranium in a nuclear reactor. But for that you had to have a nuclear reactor. That is, something that would take, in principle, natural uranium and have such good neutron economy, as we discussed yesterday, that a neutron born in fission at 2 million volts energy would come down to the low energy where it is most efficient in being captured in U-235 to cause another fission.

Of those neutrons, some of them would then be inevitably captured in uranium-238 and would cause the production of plutonium-239. That occurs by first producing uranium-239, which then decays to neptunium-239, and that decays to plutonium-239. One of those decay lifetimes is 20 minutes, and the other one is two days, so you don't have to wait a long time. One knew immediately that a reactor producing a megawatt of heat would produce about a gram per day of plutonium for nuclear weapons, if plutonium could be used.

All that was known in 1942. Really by the time Fermi successfully demonstrated the nuclear chain-reacting pile under the west stands in Chicago, which was operated at a steady power of about two watts, so it did nothing to produce material for nuclear weapons. But because fission and the behavior of neutrons is a linear process, it could be scaled up. People were very sure that a reactor that had 50 tons of uranium in it and was critical and operated at a few watts would, if you just pulled out the control rods and let the population climb operate at a megawatt or 100 megawatts, and the only requirement for such high-power operation would be to have sufficient cooling that the reactor would not overheat, vaporize the uranium, or vaporize the graphite, or whatever. That was why the production reactors were located at Hanford in Washington State on the Columbia River, and the river itself was just diverted — part of the river — through the reactors to cool the canned uranium assemblies.

In the summer of 1942, people began to think seriously how they would actually make nuclear weapons once the materials were available. Plutonium was to come from Hanford, and uranium was to come from the isotope separation — or isotope enrichment — plants at Oak Ridge, Tennessee. Many approaches were considered for enrichment at the beginning of the Second World War. A natural one would be to use a liquid or gaseous compound of uranium, and what's used now is uranium hexafluoride, a uranium atom and six fluorines. This is a solid at normal temperature, but at slightly elevated temperatures it's a gas. So one would heat this to the liquid or gaseous phase, put it in fast-spinning centrifuges, and take advantage of the fact that uranium-238 is three units — about 1.5% — heavier than uranium-235.

Even when combined with six fluorines, each of which has an atomic mass of 19 — which adds then 94, so it's 3 units out of 300, or about 1% difference in mass. So there would be a slight enrichment if one spun the centrifuge fast enough, maybe a tenth of a percent or so depending on how fast one could spin the centrifuge — of the uranium-235 hexafluoride at the inner portion of the centrifuge. And the idea would be that you would have plumbing that could pipe uranium hexafluoride from one spinning centrifuge into another — into a whole bank of centrifuges. And the denser material, which is still 99.9% as full of uranium-235 as is natural uranium, so that would be driven to other centrifuges which would try to strip the uranium out of it.

You would start with 0.7% uranium-235 and have a cascade of centrifuges of the enrichment side — would gradually increase by a tenth of a percent fractionally, so 1,000 stages would be required to have a doubling of enrichment to 1.4% and another 1,000 stages to another doubling. [This greatly understates the effectiveness of the gas centrifuge and, especially, the regenerative Zippe machine]. [brief off-topic]. The other part of the cascade is a stripping cascade, so that when you… What was done at Oak Ridge, not by centrifuge enrichment, but by another process which works in a similar fashion, is to enrich up to 95% for the bomb material and to strip down to 0.25%t. So material comes in, natural uranium purified and converted to hexafluoride, enriched, and now about 30% of that uranium-235 coming in is thrown out as tailings. Discarded uranium hexafluoride is a toxic compound and dangerous as well. It burns. So it's kept in tanks, and we have lots of tanks of uranium hexafluoride tails.

The technique actually used at Oak Ridge were two — one was gaseous diffusion. In this approach, uranium hexafluoride is used in gas phase, and the trick is to have a microporous barrier with tiny pores. If the pores are small compared with the mean free path in the gas, then molecules will enter the pores individually, and the uranium-235 hexafluoride is moving faster by half a percent than the uranium-238 hexafluoride. So the material coming out the other side of the barrier is enriched by half a percent — that is, half a percent of a 0.7%. Again, it would take 200 stages to convert this 0.7% material into 2% material — a factor-3. Because after one stage of membrane separation the enrichment is so little, the resulting rejected material has to go through, again, stripper stages of gaseous diffusion separation so that one has a cascade just as in the case of centrifuge. Rather than spinning centrifuges driven by motors, one has big compressors.

Uranium hexafluoride is very corrosive. If you have a little bit of water present, then it makes not only uranium oxide, but hydrofluoric acid. Hydrofluoric acid attacks almost everything, so one needed to have stainless steel systems in this case. To find out how to weld stainless steel for the materials to have coated stainless steel or gold-coated compressors, to have a barrier material which is still a secret material — it's said to contain nickel, which would stand up to the uranium hexafluoride. Because there were rotary materials and one had to have joints, the new materials were created for the gaskets and seals. Teflon was created for this purpose. It's a fully fluorinated hydrocarbon, no hydrogen left, and so it's inert to the action of fluorine and uranium hexafluoride.


I always thought Teflon was part of the space program.


The other approach used at Oak Ridge was the calutron, or so-called electromagnetic separation process. This is just a mass spectrometer which had been used since the 1920s and 1930s in physics. You heat a source of material that you want to analyze. You ionize the material. You accelerate, and then you have a magnetic field so that the particles going through the magnetic field bend more if they're lighter particles than if they are heavier particles. It's the same thing that you have in your television tube, except there they are electrons coming through. They're accelerated, they're bent by a magnetic field, and they — in fact, the magnetic field varies with time, which is why the electron beam scans across the tube. One set of coils sweeps it back and forth at 15,000 times per second horizontally. Another set of coils moves it down the tube in 1/60th of a second. They're alternate frames, so 1/30th of a second paints a complete picture, and your eye — the persistence of vision — converts this into a steady picture. But if you had two masses of electrons, different by 1%, then you would get two pictures, one of which is half-percent bigger than the other.

That's exactly what happens in the mass spectrometer, except you don't have the sweeping; you have a steady magnetic field. The lighter molecules don't go so far — are bent more by the magnetic field — and the heavier molecules are bent half a percent less, so you put two collectors up there in the vacuum. One starts collecting the uranium — and it's not uranium hexafluoride, it's uranium metal ions, the uranium-235 ions and other the uranium-238. The big problem with calutron — named for California. It was perfected by E.O. Lawrence, and that's where Herb York got his first contact with the weapons program, although nominally he and the people at Oak Ridge didn't know what they were doing this for. But they were enriching uranium, and it was going to go to Los Alamos, where it would be made into nuclear weapons.

The big problem with enrichment by calutron is that you have to ionize and accelerate all of the material — the 99.3% that isn't going to be used, ultimately. If you accelerate to 100 kilovolts, which is a reasonable voltage for a calutron, a tenth of a million volts, then for every atom of uranium-235, you had 140 of the others. So you've already paid 14 million volts per uranium-235 atom that you get. If you're going to be 2% efficient, which is what the Hiroshima bomb was, you only get out — well, you need 50 times as much, so for 150 MeV achieved in fission, you would have spent 50 times 14 — 700 MeV. Your bomb would not be an energy producer; it would just be a storage battery. You would store the energy at Oak Ridge coming from a coal-fired power plant or hydropower, and you would liberate it in a millionth of a second over Hiroshima. In fact, gaseous diffusion is also quite inefficient, and it takes about 5 million volts per uranium-235 atom rather than the 14 million volts or more. That's also a storage battery, but that's what we did.

In contrast, the plutonium production creates energy in production which is thrown away, warming the Columbia River, and then creates more energy when it is finally used. Since you get about one plutonium per fission, there's about an equal amount of energy rejected to the Columbia River — or ultimately in Georgia or South Carolina, where another production plant was located — as is liberated over the target.

But I digress. I was saying people were interested in thermonuclear fuel. That was because you could have an infinite amount of energy. The idea was — this was Teller's idea from the very beginning — that you would have a long cylinder of liquid deuterium. An atomic bomb — that is, a fission bomb — at one end would heat the cylinder so hot that there would be rapid reactions — deuterium on deuterium — and this happens to go two ways. Half of the reactions produce a neutron and helium-3; the other half of the reactions produce a proton and tritium. In either case, you start with two protons — one from each of the deuterons — and two neutrons — one from each of the deuterons. You have four particles all together, and the neutron plus helium-3 means that you end up with two protons and two neutrons, one in helium and one free.

The proton plus tritium means that you end up, again, with two protons: one free, one in the tritium. And two neutrons in the tritium. This gives about 4 million volts of energy either way. When you have the tritium, it then has a much bigger reaction rate than the deuterium, so it reacts with some of the original deuterium and it gives you the aforementioned 17 million volts of energy release. But we already decided that that wouldn't pay back the energy required to produce the tritium. It's a big loser if you need tritium throughout this long cylinder of deuterium, so the idea was to have a pure deuterium cylinder. You might have some tritium at the beginning to make it ignite better. And that's what Teller wanted to work on from the beginning of the nuclear weapons program, and that's what he was still working on up to February 1951, when Ulam and he wrote their paper on radiation implosion.

Thinking of how actually to make a nuclear weapon began in the summer of 1942 at a summer study at Berkeley chaired by Robert Oppenheimer. This was before the Los Alamos Laboratory had been defined or its location chosen, and before Oppenheimer had been selected as director. Many people, mostly theoretical physicists, traveled to Berkeley. This is in Teller's book. It was Teller, and Emil Konopinski, Bob Serber, Hans Bethe… I don't think Fermi was there. He was working too hard designing reactors — Hanford reactors. Fermi didn't go to Los Alamos, in fact, until 1944 — I think the summer of 1944. He stayed there until December 31st, 1945.

This theoretical study group at Berkeley was secret. It was under the Manhattan Project — looked at how you would actually make nuclear weapons. They focused on gun-type nuclear weapons in which uranium metal projectiles would be shot into a set of uranium rings or something like that — and a neutron injected at the right time to start the chain reaction at the time of maximum criticality. Or it was recognized that you could arrange to stop the projectile, and then you'd just wait for a cosmic ray neutron. That was the approach for the plutonium gun as well. So that was the baseline approach, as they say.

Teller got bored. There was nothing more to do. It was mechanical engineering, so he tried to hijack the whole study to think about hydrogen bombs. He succeeded. It's very interesting to think about hydrogen bombs — if you had an atomic bomb, how you would make a hydrogen bomb. I don't know exactly at what time during this study that was brought in and how much effort was put on it. It's a matter of record, but I just don't know.

Then people disbanded and got together again primarily at Los Alamos beginning in March 1943. It was really a stroke of genius for General Groves — Leslie Groves — to select Oppenheimer as director of the Laboratory, because Oppenheimer was a charismatic theoretical physicist. He had a school at Berkeley, a school at Caltech. He spent half his time in each place, and a lot [i.e., of followers] of his graduate students would follow him from one place to the other so they could learn physics from him and from one another.

When the Los Alamos project opened, Bob Serber gave a set of lectures to each new group of people who arrived, most of whom had no idea what they were going to work on, and many of whom had had no contact at all with this area. Even if they were nuclear physicists, it was very strange to them to be working on these particular nuclei, and they didn't need much nuclear physics. They needed to do design. They needed to understand things. People like Feynman were there. Fermi came ultimately.

I think that Oppenheimer said that he needed about 13 people in his laboratory, so they figured maybe 70 would be enough. I'm just making it up, but it's written down. The laboratory now has 7,000. I think during the war it grew to a couple thousand because there was really a lot of work to be done in the metallurgy and understanding all of the details of plutonium, which is a devilishly complicated metal with five or seven different crystal structures — so-called allotropic forms — that is really of crucial importance. A nuclear physicist wouldn't think of it. But when you cast plutonium, it solidifies at some temperature, like that of aluminum for instance — although plutonium's a very dense material.

However its two principal low-temperature crystal structures are the alpha and the delta phase. The alpha phase has a density of 19 grams per cubic centimeter compared with steel or iron, which is 8 grams per cubic centimeter, and lead, which is 11. Mercury is 13, I guess. Uranium's about 18. So plutonium is extremely dense. That's the alpha phase, but the delta phase has a density of only 15 grams per cubic centimeter — about 20% less. So if you cast it and it solidifies into a nice shape in the alpha phase and then it cools and becomes delta phase, it expands 20%, and it's unlikely to expand uniformly because it's a crystal. The same thing happens with tin when you cool it. When tin was used to can things — pure tin was used by Napoleon, I think, to package things — they took it to Russia and it got tin disease. That is, by changing its crystal structure, it turns into powder. You can do that. I forget whether a freezer is enough. We can look it up.

Anyhow, there were terrible problems with plutonium. Then, when people assembled at Los Alamos, they began to do detailed measurements. It was realized that the plutonium coming from Hanford was creating a lot more neutrons all by itself than the tiny plutonium samples that had been made in the cyclotron — or for that matter, in the Fermi reactor.

What was happening here, it was thought at first, was the plutonium {alpha, N} — that is, alpha particles from the plutonium radioactivity striking light impurities such as oxygen or beryllium and creating a neutron which is loosely bound to these light materials. But when they purified the material, the neutron production didn't decrease. At least, it didn't go to zero. Then it was realized that this material from Hanford contained a few percent of plutonium-240 because plutonium-239, like uranium-235, is fissioned by slow neutrons, but it also captures some of them. About a quarter of the slow neutrons, instead of causing fission, cause capture, and plutonium-239 becomes plutonium-240, which has a half-life of 6,000 years. So it was present in the samples.

Now, plutonium-240 has a much shorter spontaneous fission lifetime — a much higher rate of disintegration by spontaneous fission. Not 6,000 years or something like that, but millions of years [actually 200 million years]. But it produces enough neutrons to cause pre-initiation of the bomb. So Emilio Segrè, one of Fermi's students and colleagues at Rome who had come to the United States, was looking at this spontaneous fission of plutonium in a little laboratory at Los Alamos at the end of one of the mesas, where all the equipment is operated by batteries because he had to count these very rare events — because they didn't have a lot of plutonium. He established beyond doubt that it was spontaneous fission, and that threw the lab into a turmoil in 1944 because they could no longer make a plutonium gun. It would have negligible yield. They had to have an implosion or other device.

There was a lot of discussion in the laboratory, and Seth Neddermeyer from the University of Minnesota is generally credited with the implosion method, but it's there in the Serber notes, which were presented to everybody who came in — written up actually by Ed Condon. He took notes on what Serber wrote on the blackboard, so from the very beginning one has had this first, formal document from Los Alamos — LA1, Los Alamos 1 — which is the Los Alamos Primer on nuclear weapons.


Was that classified or declassified?


It was classified for a long time, then it was declassified. In fact ultimately, after it circulated for a long time just in Xerox or mimeograph copies — photocopies — Bob Serber added modern notation to it, and it was published by the University of California. So Serber's Los Alamos Primer is available as a regular book.


The gun-type weapon made of uranium didn't need to be tested. People could make measurements on each of the separated uranium masses to make sure they were almost critical and then move one toward the other very carefully to ensure that the reactivity went up some. But of course they stopped far short of making a supercritical mass. There were at Los Alamos some so-called critical assembly devices in which one moved the metal parts ever so carefully and securely toward one another to measure the rate of increase of reactivity and to determine the effectiveness, fission effectiveness, of different batches of material. You didn't want to have material coming from Oak Ridge and having a lower uranium-235 concentration than was stated, or be in some other way defective. So these were very important.

Then there was an experiment called Dragon. When Dragon operated, Fermi always took his group skiing because he didn't really think it was safe. Dragon was a system that became supercritical, but very slightly and for a very short time.


Fermi or Oppenheimer took people skiing?


Fermi took people skiing. Fermi had his own division at Los Alamos called F Division. There were other divisions: T Division for theory, P Division for physics, G Division for gadget — which was actually in charge of making the bombs. Fermi was just himself and a few other people. He was really a consultant to others at the Laboratory.

George Kistiakowsky, chemist from Harvard — later Harvard — was head of the G Division. He was an explosives expert, so he was actually in charge of casting the explosives, deciding what explosives to use, testing them, machining them, holding big chunks of explosives between his legs while he filed off or melted more explosives in to fill in pin holes or big cavities or whatever. I knew George after the war really very well. Then there was C Division for chemistry, people who did radiochemistry.

I should say that fission has also delayed neutrons, which are extremely important for nuclear reactors and have no importance at all for nuclear weapons. For uranium-235, about 0.65% of the neutrons don't come out promptly — that is, in much less than a trillionth of a second. They come out over a period on the order of 10 seconds, ranging from a few seconds to 100 seconds. This was discovered before there were chain reactions. They come from fission products. Some of these fission products, which are produced by the breaking of the uranium or plutonium nucleus into two… They're all highly excited. They have more neutrons than normal nuclei of their mass, so they convert those neutrons into protons.

It's hard for neutrons to get out [of the nucleus]. It's hard for protons to get out. The neutrons are converted into protons by beta decay, by the emission of a fast electron from the nucleus. So a neutron goes to a proton plus an electron — a negative electron. That means that the initial-zero charge for the neutron — neutral, that's why it's called a neutron — still stays zero but is split up into two. The proton has a positive charge; the electron has a negative charge. And incidentally, there's a neutrino, a tiny massless particle, that comes off. Known since Pauli named it and Fermi did the theory for it in 1931.

Occasionally some of these very highly excited fission products will give rise to a neutron instead of to a beta decay. Those different fission products contribute to these delayed neutrons, and that means that you can have a nuclear reactor — which is so-called delayed critical, but not prompt-critical. If a nuclear reactor were prompt-critical, then the neutron population would double in a millisecond or a microsecond, depending on whether they were fast reactors or slow reactors, where the neutron has to slow down. You'd have to have a very different control mechanism, whereas for delayed critical — that is, if there's 99.7%, 99.9% as much uranium as needed for prompt-critical — then it will be delayed critical, and the neutron population will rise, maybe doubling every 10 seconds or so. Plenty of time to move control rods and to hold the neutron population constant and the output power of the reactor constant.

People of course were very used to reactors, which operate on delayed criticality, but prompt criticality is something else. Now, if you have something with two critical masses — so one neutron produces two, produces four — the time between neutron generations in metals is about a billionth of a second — a nanosecond. Actually a little more. And it's determined by how far the neutron has to go to strike another nucleus. That's typically five centimeters or so. How fast the neutron is going — goes at a couple percent of the speed of light, so about 6x10^8 centimeters per second. So a few nanoseconds is the time required between one fission and the next fission. No time for control there, unless you had very violent motions to begin with — of the speed and energy of nuclear weapons.

But if you only have something which is a tenth-percent supercritical, then in this 10 nanoseconds (let’s say) the population grows by only a tenth percent. So it takes 1,000 of those multiplications, 1,000 of those generations, to even grow by a factor-3. You start with only a few neutrons. You can stand having billions of neutrons. Not billions and billions of neutrons; that begins to add up. But billions of neutrons. Instead of going by a factor-3, you could grow by a factor — by 20 generations. So the time available is not 10^-5 for growing by a factor-3, but 10^-4 or so. And you could have a piece of material shot through another piece or falling from a considerable height, and it's only critical for a centimeter or so. It only has to go 100 meters per second or so in order to have a modest rise in temperature. A substantial burst of neutrons come out. Nice to work with, but not enough that it would melt, destroy, or vaporize the material.

So that was the Dragon, and you have this slug of material which goes through it — very carefully measured. If it got stuck there, you would be in serious trouble, and that's why Fermi took his team skiing instead of having them around. There was never an accident with the Dragon. There were accidents with the critical assemblies, and a couple people at Los Alamos were killed. Louis Slotin and half a dozen people were seriously irradiated. Two of those were Jane Hall and Dave Hall. Jane became Deputy Laboratory Director. They didn't suffer too much from their irradiation. The irradiation levels were actually determined by the radioactivity of the gold in their teeth — in their fillings — and other things.

Greenhouse George was the first experiment on thermonuclear burning. It occurred in 1951. All of the dates and names of the shots are publicly available, but the design of Greenhouse George and exactly what it did is still secret. But it used liquid deuterium and tritium, and it produced a very successful burn of these materials.

There was a lot of diagnostics done on Greenhouse George, which was a very substantial fission bomb — it was a uranium-235 fission bomb — to get the energy and temperature required to do this thermonuclear burning experiment. People devised systems to look at the high-energy neutrons, the 14 million volt neutrons coming from the deuterium-tritium reaction. These were, like laboratory experiments, collimators — things that had apertures in them close by — and pinhole cameras. We helped on some of these things… I helped design some of this instrumentation in 1950. It had long tunnels filled with helium — plywood tunnels maybe six feet on a side — miles long so that the neutrons and gamma rays could go that distance without being absorbed or scattered by the air.

There was fast photography that I helped devise as well. And wonderful experiments, gamma ray experiments. The pinhole experiments were called Pinex. The gamma ray experiments were called Ganex. There were people from the Naval Research Laboratory — Ernest Krause and Montgomery Johnson were two of them. People from University of California at Berkeley, especially including Herb York. And of course the Los Alamos team, too. The Los Alamos cryogenicists — the low-temperature physicists — were much involved with the provision of low-temperature facilities for providing the liquid deuterium and tritium for Greenhouse George.


Greenhouse George was exploded where? In the Pacific?


In the Pacific in 1951. It was 250 kilotons in yield, but only a tiny, tiny fraction of that came from the thermonuclear yield.

There was another thermonuclear shot in the Greenhouse series. It was Greenhouse Item…


Excuse me, this Greenhouse George was sort of a hybrid nuclear and thermonuclear…?


No, it was an experiment.


An experiment?


Yes. There was a not very practical, very powerful uranium implosion bomb, and that was used to provide the heat temperature to ignite a thermonuclear charge. There were many, many later experiments. Some of the experiments were done with fission bombs. Some of them were done — no, with fission bombs. But these later experiments have been declassified in considerable part. They were done in the era of underground testing, when we were no longer testing in the atmosphere. Their purpose was to investigate thermonuclear burning, in part for nuclear weapons, but in part for obtaining energy from so-called inertial confinement fusion.


Were you involved as a designer of Greenhouse George.


I was not involved as a designer of Greenhouse George. I was involved very much in talking to people about the experiments that they were proposing to do — diagnostic experiments on Greenhouse George. I devised an important technique, brand new, for such experiments. That was to use stable isotopes, ordinary materials, to insert into the experiment so that they would be made radioactive by the neutrons and could then tell us, after everything had been vaporized and a tiny portion collected by air sampling — as was done routinely… But that was done in order to determine the yield of the nuclear explosion. I said, "Well, let's look at some details — what happens at a particular place. Let's be very specific as to whether there are low-energy neutrons or high-energy neutrons at this place." So by appropriate choice of materials, one can find out very precisely what’s happening in a tiny portion of the nuclear explosion. If you have several materials that are equivalent, they can chemically separated from the debris so you can determine what's happening at several places within the nuclear explosive.

Now, I can't say exactly what — am not allowed to say exactly what we were measuring there, but that technique was used. In fact, I remember a later visit to Livermore, probably 10 years ago or so. One of the Livermore folks had received an E.O. Lawrence Award. I think there are 5 or 10 E.O. Lawrence Awards given every year for contributions to nuclear weaponry. I don't remember what the definition of the E.O. Lawrence Award is. Somebody was introduced as having received it for the contribution of this technique, of stable isotopes for diagnosing nuclear weapons. And I said, "Yes, that's a very good technique, and it was very good when I used it on Greenhouse George." So there was a lot of incredulity, and I was able to give them the title of the Los Alamos memorandum that I had written about this. So they had it faxed by secure fax, and sure enough [laughter]…


This was essentially seeding the bomb with substances, ordinary substances, that you could then see how they were transformed?


How they were activated, yes. I was saying that later on, much later after we had two-stage thermonuclear bombs by radiation implosion… And radiation implosion is achieved — this is all in my book — by having a fission explosion in nuclear weapon primary — a nuclear explosion of the primary — and the energy from the primary is largely given off in the form of soft X-rays, which have not only a lot of energy, but they have a lot of pressure, because energy and pressure are very closely related. Pressure has the dimensions of energy per unit volume, and most forms of energy have very similar pressures. So whether I have a kiloton of fission yield present in the form of hot hydrogen in a certain volume or present in the form of just X-rays in a certain volume, the pressure is almost exactly the same. But the X-rays move very rapidly, and they can be contained for a time by a dense, heavy — that is, high-Z, high atomic-member material.

Most of our radiation cases for this purpose are made of uranium alloy. You have one of these things, a primary in a radiation case, and the idea is that you're going to have a thermonuclear secondary which is compressed by the primary energy, by the X-rays. It may have uranium in it; it may not. It's compressed and heated. Under those circumstances, you can have very substantially more energy liberated by the nuclear weapon secondary than from the primary. And you can do that with using cheap deuterium rather than a lot of tritium. But if you don't have a secondary, you can put in tiny little pills — some this big, some that big, some millimeter size — of things that are shaped like the inertial confinement fusion pills that you want to use in your laser-induced fusion or whatever in the NIF… NIF… I forget what the 'N' is, but the 'I' is the inertial and the 'F' is fusion. National. National Ignition Facility. That's what it is. The NIF.

People wanted to know how much laser energy converted into X-rays would be required in order to have a thermonuclear burn that would repay the energy put into the laser. So people had the idea that they would use this readily available source of very powerful X-rays — no problem about getting much more than you're going to have in NIF, and instead of trying to creep up by having more and more powerful lasers, each of which cost you a lot of money and takes years to build, will creep down. We will do experiments using collimators, these Pinex experiments or radiochemical experiments, to see what the yield of each of these little pellets is. That's the sort of thing that you do with a nuclear explosion source of energy when you're investigating thermonuclear burning.


Greenhouse Item was a very important test. It was the first use of boosting. This is very different from thermonuclear burning, although there's a little of that. A booster in a nuclear weapon is an amount of deuterium, or deuterium and tritium, present in the form of compressed gas or perhaps solid placed in the center of the weapon. If one had a solid core weapon, there could be a hollow there with the booster cavity. If one has a shell — [brief off-topic] — amount of deuterium and tritium, or deuterium. It could be compressed gas, or it could be solid.

If one has a hollow shell like the modern US nuclear weapon primaries, there would be deuterium and tritium introduced into this so-called pit just before the implosion is started. Then the hydrogen gas would be compressed, and when the fission reaction started and the material was heated and further compressed, a thermonuclear burn would produce a burst of neutrons. Only some grams of material are used, so the energy produced is negligible compared with the fission energy. But because one gets one neutron per atom of tritium burned up, the number of neutrons can be very significant, and the result is that the energy output of the plutonium grows exponentially, and then all of a sudden when the thermonuclear burn takes place, it jumps up to a higher level and continues exponentially, and then the thing disassembles.


One of the things I wanted to ask is, when you looked through the whole history of the bomb making process — obviously there was a lot of new things, et cetera — but was there anything particularly surprising to you?


Well, there were a lot of things I had never suspected. To see all these numbers and the people who had done the work, and the amount of material that had been accumulated, not only physical material, but analyses and formulas, was really very interesting to me. So no, I can't say that I thought of all these things before in any way.


Did they include things like the debates people had about different [overlapping] the final report on things?


In fact, there wasn't even much of that. These were progress reports, for the most part. People would have debates, and I guess there would be minutes of some of these meetings, but mostly I would be reading the technical reports of the individual groups. I needed to infer what had gone on, for the most part.


Things like the whole political debate on whether to build a Super — I guess they called it — or not. Was that public at the time? Or was that just something that was going on within the government and within the labs like Los Alamos?


No, the question of the Super was quite public because Teller was disaffected with Oppenheimer from the very beginning for not having given him his own big team at Los Alamos to work on the Super. Then at the end of the war, Oppenheimer was further disaffected with Norris Bradbury, Oppenheimer's successor, because he felt that Norris had made him a commitment that, since the atomic bomb was in hand, now there would be time to work on the Super. But the problem was that Teller's approach really wouldn't work, that this question of burning a cylinder of deuterium just was technically infeasible.

That's an interesting story. There's some of it told in my speech in Pisa in 2001, I guess it was — the 100th anniversary of the birth of Enrico Fermi. There were celebrations all over the world. My talk was “Working With Fermi at Chicago and Los Alamos,” so I had some of that. But the problem with burning a cylinder of deuterium was explored again in 1950 when I shared an office with Fermi. In fact, Stan Ulam would come in, and they would discuss the physics involved. Fermi had made a spreadsheet, the kind of thing that you do on a computer, but accountants did before.

His problem was to ask, what happens when you have a long cylinder of deuterium? It has a radius. There are radial zones, and there are longitudinal zones, and in order to do some calculations, you need to divide up the material into these zones – chunks. You only need the radius and the length, because it's assumed to be symmetrical about the axis. Each of these zones then is assumed to have a mass, a temperature, and a compression. In addition, because things happen very fast, the temperature of the electrons is different from the temperature of the ions, and there's also the temperature of the radiation field, which is important — as I indicated — because it has a lot of energy and a lot of pressure.

Fermi made a spreadsheet for this purpose. He wrote the equations for how a segment — a zone — heats as influenced by its neighbors and things that are farther away. He then showed how you progressed from one time step to the next by these little equations that he wrote for each zone. He did the first few time steps, then they would call in their computer, who in the 1950s era was a young woman typically. She would come and have the spreadsheet explained to her. She would take it back and, overnight, work on it with her desktop Marchant calculator. So she would do the numbers, and she would fill in the new zone figures for temperature compression and so on — and now position of the zones because they're moving. This young woman's name was Miriam Caldwell.

This happened for maybe a week or more. In the morning, Fermi and Ulam would plot up the results by hand on pieces of graph paper and decide what the next experiment should be — how big a nuclear explosion should they use to start heating it. But it would always die out. The problem is, that if the cylinder is too small in diameter, then there's too much loss by radiation and expansion before the thing burns. If it's too large in diameter, then the electrons get heated by the ions, the radiation bath gets heated by the electrons. That cools the electrons, and that puts a big drag on the ion temperatures, and things never get hot enough to react. So you get very little out of it. That was the problem all along with the burning of deuterium.

Anyhow, I want to go back first to Item. That was a boosted primary, and that is used in just about every US nuclear weapon now. I had nothing to do with it, except I read about it. Since then, of course, I have had a lot to do with the nuclear weapons. Now, the burning of the long cylinder of deuterium is called the classical Super. That was really the only game in town except for some excursions, which were called, in Los Alamos, the Alarm Clock and, in Russia, independently developed by Sakharov — I think he called it Layer Cake. That was presumably a kind of close mixture of uranium and thermonuclear material, as in the case of a booster. That's what the Russians exploded in 1953.

That is, the United States did Greenhouse George in 1951, and the booster Item. Then in 1952, we had a test series, and that had the 11-megaton Mike shot — that is, the first true thermonuclear explosion that I had a lot to do with designing. Then in 1954, we had our thoroughly weaponized — instead of using liquid deuterium, liquid hydrogen, we'd used solid lithium deuteride fuel — that is, one atom of lithium-6, one atom of deuterium, in something which is colloquially known as salt and is a kind of waxy, white material — reacts with water to form lithium oxide and hydrogen, so you have to keep it dry, but it's a lot easier to handle than liquid hydrogen and liquid deuterium.

Then in 1953, the Russians had their fourth nuclear explosion — so-called Joe 4. Joe 1 was the first nuclear explosion in 1949, which was a copy of our Trinity/Nagasaki bomb — a copy really with the design obtained by espionage. But the Russian designers and intelligence people have spoken a lot about this over the last two decades. They had a better design that was their native design; however, they were afraid to test it because Beria [Lavrenti] if there had been a failure, it would have killed them all. So they were persuaded by the one scientist who really had the intelligence information — who was… I'm sorry. The head of their nuclear weapon effort. Began with a 'K.' I'll remember his name. I know it very well [Kurchatov].

Anyhow, he had the information provided to him by the KGB, or the GRU. Every time people came in to talk to him about progress on their atomic bomb, he would say, "Well, have you considered this?" Or, "Why don't you make it a little smaller or a little bigger?" — until it was exactly what the United States had tested. But in 1951 — and I just participated last year, I guess it was last May, in a meeting on the — it must have been 2003 — 60th anniversary of Los Alamos on the origins of the early hydrogen bomb. So I and Conrad Longmire, who with Marshall Rosenbluth was one of the people who really did calculations on the radiation implosion — and another person who had been involved to a lesser extent, Harris Mayer, discussed this in a classified session.

There was a February 1951 letter from Stan Ulam at Los Alamos to John von Neumann, who was coming to Los Alamos on one of his many visits. Stan wrote Johnny that he had had a new idea, and it must be wrong because Edward Teller liked it. But the idea he had was to have an auxiliary bomb, and that would prepare the main bomb, which would be of thermonuclear fuel. Apparently the Ulam idea was that the auxiliary bomb would compress the thermonuclear fuel so it would have a higher reaction rate, because the rate of energy production is proportional — overall energy production — not only to the amount of material that you have, but also to the square of the density, because these things react by colliding with one another, and if you double the density, then there are four times as many collisions per unit volume. There's only one-half as much volume, but the overall energy production rate is then doubled. Per unit volume, it's multiplied by four.

Teller, in his memoirs, says that — and there's something else which I will send you. In 1979 he had a heart attack, and he thought it was possible that he was going to die. So he thought he would take the opportunity, since he was okay at the moment, to dictate to his friend, Jay Keyworth, who brought a tape recorder, his reminiscences about the implosion design and about the first thermonuclear weapon. He did talk a little bit — as I told you — about the implosion design. I think he gives the credit to Seth Neddermeyer.

This was recorded in 1979. It wasn't transcribed I think until 1987 or so, and that's the first time Teller had seen it [It was transcribed immediately but not sent to Teller until 1987]. Many of the names were wrong. The person who transcribed it was a secretary who didn't know any of the names, so she got George Kistiakowsky confused with Emil Konopinski, and things like that. I went through it and straightened it out, but the first time I saw it was when Bill Broad had, for some reason in 2001, been sent a transcript of this by Jay Keyworth. Keyworth had been President Reagan's Science Adviser, I guess. I knew him. But he had sent this to Bill Broad of the New York Times, and Bill Broad called me and asked what did I know about this. And it mentioned me prominently.

Teller said that Stan Ulam had come in to talk to him, and he said you could waste a lot of time talking to Stan, but he listened to him. Stan said that you could compress the fuel, and then it would burn much better. And Teller said that he had a theorem that, if you couldn't make a hydrogen bomb with deuterium liquid at normal density, you couldn't make it with liquid at a 1000-times normal density. That's because, even though the reaction rate would go up, all the loss mechanisms would go up by the same factor. So the ions would produce more energy, but they would lose energy at the same enhanced rate to the electrons. The electrons would lose energy at the same enhanced rate to the radiation bath, and so on.

He said that he had this theorem, so he didn't pay much attention to what Stan was saying. But then he got to thinking about it, and he realized after all that he had made a mistake. Then he says — especially in his book — he says he realized this in the Fall of 1950. I knew Edward Teller very well, and I knew Stan Ulam very well. If Edward Teller had realized this in the Fall of 1950, then he would not have waited until March 9th, 1951 to publish this paper with Stan Ulam — a secret paper.

The idea was that even though the energy loss rate would go up, the radiation bath that has all of this pressure and energy doesn't increase its energy content as you squeeze. It's just what there is. It's temperature, and it's the temperature of vacuum or whatever. It's full of soft X-rays, and it has just so much energy. So when you compress the liquid, the thermonuclear fuel, you increase the potential amount of energy that can be produced, but you don't do anything to increase the amount of energy present in this loss mechanism in the radiation bath.

So even though Stan Ulam had no idea of what he was proposing — that this was the key to it — Teller soon saw it, and Teller then said also, "Well, instead of using the shock from the primary — that is, the mechanical motions" — in this paper he actually calculates how you would use it — "you should use the X-rays themselves. They get there faster. They're easier to make uniform, and whatnot." So that's their paper of March 9, 1951, the details of which are still secret. It's called "Hydrodynamic Mirrors" — no, "Hydrodynamic Lenses and Radiation Mirrors." The hydrodynamic lenses were the Ulam idea, and the radiation mirrors were the Teller idea. Then the people at Los Alamos, who had been really negative about Edward and his classical Super, all said this would work. If you could do it, it would work. The question is how you go about doing it.

But as for the debate, none of these papers had anything about — none of these secret documents — had anything about those debates. Those debates were conducted in Washington in the General Advisory Committee of the Atomic Energy Commission in the Defense Department and political circles.


Did you have your own feelings at the time about the issue?


The idea of hydrogen bomb and its prominence had leaked, so there were public debates. Many of the people who had worked on atomic bombs thought they were plenty big enough and there should be no more hydrogen bombs. What we needed were smaller and fewer ordinary bombs. Hans Bethe, who had led the theoretical work at Los Alamos to produce the atomic bomb, was very much publicly against the hydrogen bomb. Then Truman, as President, ignored the advice of the General Advisory Committee and decided to move forward with the hydrogen bomb. The GAC was negative. A subcommittee — that is, a subset — I.I. Rabi from Columbia University and Enrico Fermi — were even more negative. They said that this is evil in and of itself because it could have unlimited power. Truman decided nevertheless to go ahead.

Once Truman decided that — and I think that was in 1949 — Bethe, being a loyal citizen, said, "Well, I will help. I was against making this decision, but if the country's going to make a hydrogen bomb, I'll help make a hydrogen bomb." He made no bones of the fact that he really hoped it wouldn't be possible, but he threw himself into doing it. When the idea of radiation implosion came along, Bethe realized that this was the solution. Teller of course wanted to be in charge. Teller should not be in charge of anything because he has so many ideas that one could never proceed and build anything, because he's always having a new idea which, in his memoirs, he says 90% of them or more are wrong, but some of them are good — which is true.

Eventually Bethe was put in charge of the theoretical work on the hydrogen bomb. He chaired a Theoretical Megaton Committee at Los Alamos, of which I attended a few sessions. I was not on the committee. It was mostly division heads and people who were involved ex officio. So that was the step forward in March of 1951, and when I arrived at Los Alamos for my second summer — probably in May of 1951 — I asked Edward Teller what was going on, and he told me about this and asked me to devise an experiment that would determine whether this radiation implosion would really work to make a hydrogen bomb.

I thought about it for a while, and asked what would be a convincing experiment, and decided that one didn't need a preliminary experiment. These things are easier to do in the large sizes — so that I would just design a full-scale hydrogen bomb. In my physics work at Chicago — I'm an experimental physicist — I had been building liquid hydrogen and liquid deuterium targets for use in these external beams of the Chicago cyclotron. I was no stranger to cryogenics, to the liquid and solid hydrogen, and had no problem designing this thing that would hold cubic meters, cubic yards of liquid hydrogen to keep it cool and liquid deuterium for thermonuclear fuel.

I just took the ideas that were floating around at Los Alamos in June of 1951 and made decisions as to how they should be put together to make a particular hydrogen bomb. It had to have a primary. It had to have a radiation channel to move the soft X-rays from the primary to compress the secondary. It had to have a secondary design that would have the thermonuclear fuel and some uranium preferably to use some of the neutrons that were generated by the thermonuclear fuel, and various still-classified aspects. So, the end of July I had a sketch and calculations as to how it would do. The Theoretical Megaton Committee reviewed that and decided that it would work and they should go ahead and do it. Nobody in Washington had to approve the design. I guess they had to approve the creation of the joint task force that would go out and test it. And 16 months after I had heard of the idea, the thing was built, shipped to the Pacific, and exploded. It worked just as was proposed. In fact, it worked a little better.

I had designed also a flight model. Even though this stationary model weighed — I don't remember — 40 tons or so, the radiation case was much too conservative, so I had done my best to thin it and to reduce the weight and provide something that could be carried horizontally by an airplane. My test was a vertical, cylindrical device. It turned out — I read later in the draft of Herb York's book The Advisers: Oppenheimer, Teller, and the Superbomb that they had actually made five or six of these so-called Jughead liquid deuterium thermonuclear bombs to be carried by our Strategic Air Command.

Of course, when in 1954 the solid fuel designs came in, they scrapped the few liquid designs that they had. But we really had those. The Russians say they tested the first true, practical hydrogen bomb with their solid fuel design, but it wasn't a two-stage weapon. They never made another like that. It was very limited. It did get some of its energy from thermonuclear burning, but mostly from the uranium.

In fact, we've had more than 1,000 tests ourselves, and some of them were very high yield. I guess our biggest yield was 20 megatons. Most of the nuclear weapons we have now are half-megaton or less because once people realized that 20-megaton weapons, 50 megatons, 100-megaton weapons would destroy the largest city and waste their energy, they decided it would be better to use that weight to produce many smaller weapons to be carried on missiles or airplanes that could be more flexibly allocated.


When you said, by the end of July, you had made a sketch, is this a page, or five pages?


My sketch was several pages, and it had a little bit of explanation. There were several pages of explanation as to what it was. But it was perfectly clear. It was an engineering sketch, not an artist sketch. I'm no artist.


Were there other people who assisted with this sketch?


No. Well, I said I had taken the ideas that were current, so the idea of radiation implosion, the idea of what you do with the radiation channel, what the materials are… There were various proposals, and you just have to make a kind of engineering decision as to which ideas you use here and there. Edward recounts this story — of course, nobody knows the whole story. I don't know the whole story. I only know what I did, and I don't know what other people did.


You mentioned Marshall Rosenbluth?


Yeah, a very capable physicist. He and I were graduate students together at Chicago. I think he got his degree in 1948 on some abstract theory about the behavior of the nucleus. Then he went to Los Alamos and worked on plasma physics. When Edward Teller had the radiation implosion idea, he recruited Marshall Rosenbluth to work on the actual machine — numerical calculations of radiation implosion. So Marshall Rosenbluth, who died last year — he was a professor here after a stint as professor at University of Texas at Austin — and Conrad Longmire, who's still alive, did a lot of this work on actual calculations of how radiation-driven implosions would heat the fuel, and what would be optimum ratios of the thickness of the thing that pushes on the fuel, and how it would burn. I spent two weeks with Marshall in the summer of 1952, which was after the Mike shot had been designed…

[brief off-topic]


Have you read the Richard Rhodes —


Sure, I read The Making of the Atomic Bomb, and then I read Dark Sun.


Did he interview you for —


No, he didn't. I told Marshall Rosenbluth… I'd heard from Marshall that Richard Rhodes was doing this. Then I even contacted Rhodes, whom I know reasonably well now, but I didn't then. He said he was finished interviewing, so I wrote him a little note. I said, "Well, I had had somewhat different involvement." But he didn't do it.


It was strange. In the article that Broad published in the Times about Teller and this tape recording and all of this… If I remember correctly, they were quoting some Harvard historian saying that the description of your role seemed to fit what they understood. But they quoted Rhodes as saying something 'no.' And it seemed very strange why he would say that. I was wondering whether he had interviewed you.


No. I was a consultant at Los Alamos, so I was there three months the first year and four months the second and five months the third year, and then more or less three months throughout most of the '50s. But I would start something, and then other people would finish it. In this case, Marshall Rosenbluth and Conrad Longmire, who worked very closely together… I was very good friends with Marshall, not only at that time but even recently.

Conrad, who was at this meeting in Los Alamos a year ago on the early history of the hydrogen bomb said that he didn't know of my involvement. He knew I was there, but he and Marshall were doing this work and they didn't know of my involvement at all. Edward had asked me to do it, and it was not part of the Laboratory organization. So I talked to these folks to find out what they were doing. I talked with everybody else, and I just went ahead and put the whole thing together.

That's why I say I know what I did, and other people know what they did, but nobody really knows the whole story. I was glad to see that Edward, with whom I have had big disputes over policy — not over technical things, but over policy — over the years… I was glad to see that in 1979 he happened to have the occasion to recount this, not because he heard it from me, but because that was the way he remembered it.

Our early disputes were over the missile defense, ballistic missile defense, when Hans Bethe and I wrote our article published in the March 1968 Scientific American called "ABM Systems.” Edward has always been in favor of all defense, all technology, civil defense. He was younger than Gregory Breit, a theoretical physicist from Yale, but also than Eugene Wigner, a theoretical physicist from Princeton.

I remember when we had the first of these nuclear weapon, nuclear war, meetings in Erice, Sicily in 1981. Teller was there, and Wigner, and I, and Harold Agnew, Spurgeon Keeny — a person with whom I've worked for many years in Washington. A very smart, very capable person. Wigner had been passionate about civil defense. Teller treated Wigner like a son, or maybe an old parent — standing up for him when Wigner couldn't make his points very well.

But yes, people have different views of the particular elephant, and nobody apparently mentioned to Rhodes — who's a very good reporter and very smart person — my involvement at all. This was a little strange because Rhodes goes into great detail. Apparently one of his principal informants was — it's a person whom I know very well [Jay Wexler] — but he was in the low temperature business in Los Alamos. My problem in designing Mike was that I did the nuclear design, sketched the nuclear design, and the cryogenics design because I'm really good at that and I had had all this experience with liquid hydrogen in Chicago. Now of course, this was a much bigger thing. The parts were much heavier. I had to hold a lot of uranium or whatever in the low-temperature region. So my support of this vacuum insulated vessel had to be very strong, and especially when I was designing the ones for airplanes, they have to survive not just one-times the acceleration of gravity, but eight-times the acceleration of gravity — is a standard requirement for airplane design.

I wanted to enlist the low-temperature physicists at Los Alamos in this activity. I went to see them — Ed Hammel was the group leader — and met with a couple of them. This must've been in July or August 1951. And they said that they were worn out. They had been distracted from their work with the design for Greenhouse George, and they wanted to go back to their physics. I thought that was terrible, so I just said, "All right, I'll do it myself." So I did.

They had a consultant, Ferdinand Brickwedde, who was from the National Bureau of Standards — now called the National Institute for Standards and Technology. When you have a big program like this actually to build the hydrogen bomb, a lot of requirements — not only for the testing that I went into, diagnostic testing — but also for manufacturing the thing. Norris Bradbury, the director of Los Alamos, chose Marshal Holloway for this job. Holloway was a very good engineer and obviously did a very good job working with industry in order to fabricate the steel and uranium parts, and the test procedures, and all that.

Then we had to build liquefying plants to make enough liquid hydrogen and liquid deuterium. So the Bureau of Standards agreed to do that, and Ferdinand Brickwedde was in charge. I had shown him my designs for the cryogenics part, the low-temperature parts of the bomb. The main calculation that you make is what will be the heat leak into the materials, so the evaporation rate, or how much refrigeration you have to provide. There was a lot of criticism of that too, but Brickwedde, when the thing was built, told me that it was even a little less heat leak than I had calculated. So that worked very well.

As I say, the person who spoke so fully to Richard Rhodes didn't mention my work at all. But maybe he didn't know about it, because by the time the actual development and manufacture was taking place — which must've been from September 1951 until April or May 1952 — I wasn't involved. I don't know that I went back to Los Alamos even once during that interval.


Was Bethe aware of your sketch?


Oh sure. Bethe was.


And the sketch itself is, presumably — still remaining?


Still classified.






Is there any way of getting the cover page of it or something that's not classified?


Oh sure, yeah.


Just because since there was the article in the Times going back and forth with people who have different opinions of what your role was, something documentary… Just because there's been all of this discussion, having something documentary page would be useful.


Yeah, I can get the cover page.


Did it have a title?


I have the title someplace. I think it was something like "Ideas Current" — or, "Sketch of the" — and there's a word I won't use here. "Based on Ideas Current at Los Alamos in July 1951." But I'll get the title.


Is Brickwedde — is he still around?


Oh, I doubt it. Bethe is, but he's not very well. I'll have to talk with Kurt Gottfried to see whether it would be useful to talk with him. But Harold Agnew is. Harold was much involved in such things. In fact, I don't even know how much, because this wasn't Harold's responsibility at that time. He came in later, and he was a key person on one of the solid fuel bombs for 1954. So I don't know really how much he knows about this. And Conrad Longmire's still around.


That was very quick work, between May and July.


Yeah, time passes quickly when you're having fun.


In terms of the way that you worked, it's a little bit hard to discuss this because it's classified. The end product is classified. Is there some particular method that you used?


Well, I told you earlier that you want to do everything two ways so that you have some comparison. You can choose the better. And I said you want to calculate the same thing two ways so that if it differs and it's supposed to be the same, then you can find out which one or both are wrong. But there's another reason why you want to do things two ways. When I was Director of Applied Research at IBM in 1965, 1966, I was delighted to have two teams working for me that I would put on different approaches to the same problem. I would look at them, and one would be better. I would choose that. And they both benefited from the competition because they knew that there was competition, and I benefited from being able to choose the better and also from being able to choose neither, because if you have people working on one approach, then their general tendency is, "Well, I've done this. Now why doesn't it go into production?" But if you have two, then everybody knows that you're not going to produce both of them, and you have a better stance for being able to reject both by saying there's no need, or they aren't big enough advances, or go back to the drawing board and make it work better.

I had comparisons for different ways of doing things. I made these choices and figured out more or less how well it would work with approximations. I didn't do any machine computations.

When I was with Marshall Rosenbluth working in Washington for two weeks, it was with the digital computer of the National Bureau of Standards. We worked nights, and we would have different implosion systems for hydrogen bombs.


This was after Mike?


No, it was after Mike was designed, but before it was shot. So we could hardly influence Mike at all. He would have the results of these implosions, calculations, and I would — I worked out an approximate formula that I felt should give me the results. In this case I didn't have experiment to compare with, but I had computer calculations, and in fact it worked pretty well. I could predict what the result would be of a 10-hour run, and I could choose different fuels for the thermonuclear device other than lithium deuteride or liquid hydrogen. So it's just something that I'm interested in doing, and once you have worked out the approximate formulas, which you do either theoretically or by approximating the results of real calculations, then they're easier to play with than actual computer calculations.


Are these what Henry Kendall used to refer to as back-of-the-envelope calculations?


That's right. I even did one on the back of an envelope just to show it could be done.


He used to explain — I can't remember what the context was — but he said he was strongly convinced that, if you couldn't solve something with a sophisticated back-of-the-envelope calculation, then going onto an elaborate models and whatnot wasn't going to do you much good.


There's something to that. Occasionally you make discoveries from real calculations, and you find that the back-of-the-envelope calculation had left out something important. But mostly, either before you do a real calculation or after you do a real calculation, you gain a lot of insight from the simplified calculations.


I think that's enough for the moment about Los Alamos and whatnot. If you want, we can go on. The next topic is how you ended up at IBM.


Now for the IBM era. After I received my PhD degree in December 1949 and was made an instructor and then assistant professor at the University of Chicago, I began to build equipment for use on the accelerators there — the betatron and the cyclotron. I mentioned the coincidence counters that I had devised the previous couple of years, and now I built them into systems that would be used for these particle beams, rather than just to detect a beta ray and a gamma ray coming from a nuclear decay.

These were very complicated and useful electronic systems. They were mounted in relay racks of equipment 6-feet tall and 19-inches wide. Of course these were vacuum tube rather than transistor electronics. I devised scintillation counters for detecting the passage of particles in the beam. In the fringing field of the cyclotron, you need to have shielding against the magnetic field. So these had rather heavy steel shields over them. When people were mounting them on the aluminum channels that defined the path of the beam, they'd come in the morning and have one of these counters that cost hundreds of dollars — would be lying on the floor, broken, because when they screwed them onto the channel, the temperature would change and the screws would come loose.

I devised a very simple system, which anybody could do, which didn't have any screws, so there was nothing to come loose. It had this channel, which is about four-inches wide and an inch high, lying with its top up — so it looks like that — and goes for 20 feet with occasional supports. The counter now is at this height above the channel, where the beam comes. Back here is the offset — is the steel pipe which is magnetic shielding. And then there is a support, a steel support, that comes down over to the channel, and that's where the screws were.

I took off the screws, and I just added a little lip at the bottom. So to mount one of these counters, all you did was to hook its baseplate in the little lip over the channel and just let the weight of the thing hold it down. So it was perfectly safe. There was no way it could come off. It was just resting there. To move it, you didn't have to unscrew anything. You just picked it up and moved it to where you want it. So it took only a few days for that to propagate to all of the other experimenters.

I did only one experiment with the cyclotron: I looked for the new particles, so-called strange particles, and found that they were not produced directly from the proton collisions in the cyclotron — that they must be produced in an associated production way. But I didn't like the sociology of working at the cyclotron. There was a limited amount of time and space, so to do experiments, you really had to team up with three, or four, or five other people and decide what to do, and then have a program, go ahead and do it, devise an experiment apparatus.

It isn't the way I wanted to do things. It would require scheduling times weeks ahead, and, as I said, working with six people. And I wanted to do things the next day, which wasn't even time enough to coordinate with anybody. So I decided I would leave that field of work and look for something else to do. I decided that after the Second World War, the field of low-temperature physics — that is, the behavior of liquid helium and liquid helium-3 and superconductors — really hadn't made much progress, and that would be a good field to work in.

Also, Chicago was a very dangerous place to live. All kinds of muggings and robberies, rapes and whatnot. The University of Chicago was in a terrible area — South Side, right. And we had a baby — Jeffery had been born in November of 1949, so in '52 he was three years old — so we decided it didn't have to be at Chicago. It didn't have a very strong low-temperature physics department or group, anyhow.

I looked around, and people at the University of — I guess Florida State University asked me to come and head their physics department, but I visited them and it was too hot there. And I didn't want to head a physics department, anyhow. Then the people at IBM wanted to create a solid-state physics laboratory in their IBM Watson Scientific Computing Laboratory at Columbia University. They had asked — well, the person who was head of this IBM little laboratory, Wallace J. Eckert, was a person who, in the 1930s, had introduced the punched card to scientific computing. He was an astronomer, and he headed the IBM Watson Scientific Astronomical Computing Bureau, or whatever, on the top floor of Pupin Laboratory, which was mostly physics but a little bit of astronomy.

Then during the war he went to the Naval Observatory and was in charge of computing the Nautical Almanac that people on ships and airplanes used for navigating. So you would have your sextant. You would determine the altitude of star, or whatever. And your chronometer — and find out where you were. It was essential that this be done accurately, and many, many copies were printed for navigation purposes. These were all produced by Eckert using his punched-card computing techniques — and other innovations. In fact they were typed by typists, but they were typed twice and proofread [cross-checked by machine] in order to make sure there were no errors.

Then at the end of the war, he was asked by Tom Watson Sr. to create a new laboratory which would make use of some of the electronic computing technologies that had been developed at the Radiation Laboratory at MIT. So they hired a number of engineers from MIT and had the IBM Watson Scientific Computing Laboratory at 612 W 116 St., between Broadway and Riverside Drive.


It's up near Columbia?


At Columbia. Well, he was a professor of astronomy, and some of these folks had adjunct appointments. When I came in 1952, I was an adjunct associate professor, or something like that, at Columbia in the physics department. We had chemists.

So they had built several computers, and they taught computing instruction for the Columbia community. But by 1952, Eckert's little advisory committee, which consisted of I.I. Rabi and Polykarp Kusch from the physics department, and probably somebody from astronomy, had said that they ought to look into solid-state physics. So he took that advice. They hired a number of new PhDs from Columbia University, and they looked for a director of the laboratory, because Eckert was an astronomer.

They talked with Emilio Segrè, who was on sabbatical from Berkeley that year at Illinois. Segrè visited and was impressed. He asked me to come down to Illinois to talk to him about it, to see whether I would be interested in joining the laboratory in New York. I saw him and found it interesting. In the end Segrè decided not to go, but I decided I would. So I began work there in December of '52 on work on liquid and solid helium and helium-3 — the light isotope of helium, not radioactive — and on superconductors.

It was a very good establishment. Eckert continued as director. He was a very good director. I was associate director. I don't know whether I was then, or somewhat later. And a lot of good work went on. Almost as soon as I got there, the management of IBM asked me to spend a year at MIT on an air-defense study, and I told them that I would not do that. I'd just come to do physics, not air defense. But I decided this would be interesting, so I spent half-time. I would go up I guess maybe Tuesday-Wednesday-Thursday, something like that, and then come back. So there I worked on Project Lamplight for extending to the sea lines of approach of Soviet bombers, the air-defense system that we and the Canadians were deploying across the country, the distant early warning line for radars, the pine tree line for another kind of radar, the surface-to-air missile interceptors, the fighter aircraft, the direction centers, and so on. And how you would do this with airborne radar to cover the oceans, line of approaches, or picket ships.

So I learned a lot about radar, and I learned also about nuclear war. At Los Alamos I knew about nuclear weapons and weapons testing and effects of nuclear weapons, but it stopped there. I didn't have any good understanding of the damage that they would cause in a city. These folks at MIT had good connections in Washington. They were doing this study for, I guess, the Air Force probably, though I don't recall. So we went to Washington, and we had briefings from the intelligence community about the characteristics of Soviet bombers. Of course they imputed to the Russians the nuclear weapons capabilities we had. So what would a 10-megaton bomb do to a city, and how many bombs they might expect.

We finished the Lamplight study. I told Jerry Wiesner, who was co-head of it, and also Jerrold Zacharias — Jerry was an electrical engineer from MIT, and Zacharias was a physicist from MIT — that we were working on the wrong topic. By the time that anything that we could do about airplanes was put in place, the threat would be missiles and not airplanes. I remember Zacharias said, "Well, let's solve this, and then we'll work on the missile defenses later on."

I learned a lot from those people and Zacharias. When you're working with a group, it's different from working by yourself. You have to learn from the others and teach the others, so there's a lot of documentation. You write down ideas. I learned there that engineers really do design things. They work out theoretically what the performance should be, whereas physicists and their apparatus will just be conservative and they'll make something big enough or strong enough.

So Jerrold Zacharias insisted that the operative motive was 'don't get it right, get it written.' Then it would be available for improving, criticizing, and whatnot. It was through that introduction that I became involved in Washington, especially with the President's Science Advisory Committee. Both Wiesner and Zacharias were members of this Committee, which was in the Office of Defense Mobilization and chaired by I.I. Rabi.

[Abrupt end of recorded material]