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.
Please contact [email protected] with any feedback.
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.
In footnotes or endnotes please cite AIP interviews like this:
Interview of Richard L. Garwin by Ken Ford on 2012 December 20,
Niels Bohr Library & Archives, American Institute of Physics,
College Park, MD USA,
For multiple citations, "AIP" is the preferred abbreviation for the location.
Autobiographical profile of the experimental and theoretical physicist, Richard L. Garwin, focusing first on his contributions to thermonuclear weapons (1950-1952), supplemented by observations on various of his Los Alamos colleagues at the time, and extending to related and unrelated work during his later career at IBM. Subjects covered include: Werner Heisenberg, Edward Teller, Teller-Ulam report, Stan Ulam, Enrico Fermi, Hans Bethe, John von Neumann, radiation, Marshall Rosenbluth, SEAC, Cornelius Everett, Frederic de Hoffmann, Leo Szilard, Robert Richtmyer, Conrad Longmire, John Wheeler, Carson Mark, Norris Bradbury, Marshall Holloway, Rod Spence, George Cowan, transition to IBM, work on computer memory, gravity wave detection, Andre Landesman and John Tukey.
Well, Dick, just to repeat the date, it’s December 20, 2012. And thank you very much for being willing to talk with me. I know you’ve been interviewed a great many times, and this will cover ground you’ve talked about before. There are three broad things I want you to talk about. One is your own work on the H bomb. Second is what you know about other people’s work [on the H bomb], and third is what you remember about the role of computing [in the H-bomb development].
OK. I went to Los Alamos first in the summer of 1950, probably arriving June or so and stayed until September. I was an instructor in the Physics Department at the University of Chicago, and I had just gotten my Ph.D. degree in December 1949 with Enrico Fermi [for] an experimental thesis on beta-gamma angular correlation following beta decay. I had used the reactor at the Argonne National Laboratory to prepare my samples. Lois [Garwin’s wife] tells me that I had told Fermi some ideas for nuclear weapons, and Fermi told me, well, he couldn’t talk about them there, but I should come to Los Alamos. So he arranged for me to be a consultant in the summer of 1950, and I went with my wife and our six-month-old son and lived in the Chapel Apartments, which are on the Los Alamos Mesa not far from… well, I was actually working the first summer in P Division, Physics Division, headed by Jerry Kellogg. First I went to the classified report library and I read in the first week all of the progress reports during the war from 1943 and afterwards up to 1950, so I was pretty current on what people were doing. I was interested first in getting better values for the reaction cross sections of D-D and D-T because they [researchers in P Division] were working with 1939 data. So I started building an apparatus to accelerate deuterons up to 120 kilovolts with a variable acceleration voltage and also a target that was at the high negative potential. The ion source was grounded, and then we had an analyzer, which was connected to the ion source so that we could measure the energy of the deuterium ions — the deuterons — that traversed the gas cell, so we could measure the energy loss in a front window — a very thin plastic window — and the rear window of the cell, and the energy loss while the cell was filled with deuterium or tritium gas because the cross section was such a steep function of the energy that we really had to measure the energy very well. So I designed that experiment and started building it. After I went back to Chicago, the Laboratory decided that was a good thing to do, and they created a team to do the experiment. They brought Jim Tuck, who had gone back to England after his couple of years at Los Alamos during the war, brought him back to head the experiment. And then it was Arnold, Sawyer, Stovall, and Tuck, I guess, who published the results in 1954. [[W. R. Arnold, J. A. Phillips, G. A. Sawyer, E. J. Stovall, Jr., and J. L Tuck, Los Alamos Scientific Laboratory, University of California, Los Alamos, New Mexico, “Cross Sections for Reactions D(d, p)T, D(d, n)He3, T(d, n)He4, and He3(d, p)He4 below 120 keV,“Phys. Rev. 93, 483-497 (1954).]] But I was doing some other things. I was looking at diagnostics, at the fluorescent response of air to ionization so one could use fast cameras, framing cameras, 4 per microsecond, and streak cameras, that I helped the people in the optical engineering part of Los Alamos to improve, in order to do detailed diagnostics by fast photography of nuclear explosions, looking right at the case of the bomb or at…
I want to interrupt with one quick question, Dick, excuse me. In the summer of ‘50, was the Greenhouse [Test] Series for the following May already in the planning stage?
Well, yeah, sure it was in the planning stage. And so there were Greenhouse Item, which was the first boosted fission bomb, and Greenhouse George, which was a thermonuclear burning experiment. I was much involved in 1950 talking to the people who were doing diagnostics on Greenhouse George and understanding how that was supposed to work. I talked to people from the Radiation Laboratory at Berkeley and Naval Research Laboratory because they had to build the apparatus. These were about gamma-ray diagnostics of nonequilibrium thermonuclear burn, and neutron diagnostics—neutron and gamma-ray both. So I did a lot of work on that. And I devised a couple of techniques myself. One was the stable isotope addition where you would put a little pellet or, if you were looking for some integral, you would alloy some material. In this case it was arsenic and… I’ll get it in a minute… arsenic and nickel, two of the particular tracers. And so if you were interested in an average, then you could put a number of these things around, and after the explosion they would have been activated by the neutron fluence, some of them with threshold detectors so that they discriminated between fission neutrons and [neutrons from] thermonuclear burning, particularly 14-MeV neutrons. That became a very useful technique for future experiments. It was used in Greenhouse George in order to determine the characteristics of burn — not only the total amount of tritium burned, but also, I guess I can say, its compression. So there was a lot of that involved, and I understood Greenhouse George thoroughly and the radiation transport and stuff like that. In fact, I later published a paper, not about that but about something else, because in radiation transport you have a highly non-constant linear behavior of the medium, and its opacity decreases rapidly as you go to higher temperature; in stellar atmospheres it is called Rosseland mean. [[R. L. Garwin, “Calculation of Heat Flow in a Medium the Conductivity of which Varies with Temperature,” Review of Scientific Instruments 27, 826-828 (1956).]] But it turns out… and so you can integrate the opacity, or the inverse of the opacity, with respect to temperature and find the temperature analytically, or easily, anyhow, as a function of position in the steady state. And as an experimenter working in cryogenics, I needed to calculate heat transport from room temperature down to low temperatures. The thermal conductivity of materials varies rapidly with temperature. I remember [later] Charlie Townes was driving me to work from Riverdale to Columbia one day and telling me that his boil-off of helium reduced very substantially when he went to a lower temperature, probably [he said] because the thermal conductivity of the stainless steel dropped. And I said, no, no, it can’t be, because it’s the integral of the conductivity with respect to temperature precisely which determines the heat flow. And since nobody knew that, I wrote it up and published it. And then it turned out that a mechanical engineering friend of mine had published it in a different journal a year or two earlier. So, anyhow, I worked on those things. And I was familiar with the Classical Super, which I characterize as a long cylinder of liquid deuterium. Of course the Classical Super has the problem that there’s a certain potential energy generation per unit volume, the potential reaction energy of the deuterium. And then there’s the energy loss from the hot ions, if you manage to get them hot one way or another, to electrons, and from electrons to the radiation bath, except it’s not really a radiation bath — it’s not in thermal equilibrium, you hope. And then those X-rays escape. And it’s better to have them escape because if they don’t escape then there are a lot of them, and the electrons are degraded in temperature because of their frequent collisions with the photons.
The so-called inverse Compton effect?
Yeah, except it’s not really inverse, but anyhow… So the whole idea was to have a bare cylinder of deuterium. If it’s too big then the photons collide on the way out, and that’s not good because they can’t escape, so it has to be built quite small. If it’s too small, then the energy escapes, and it won’t [burn]… Anyhow, it’s a big problem. And in 1950, I shared an office with Enrico Fermi for the whole summer —
If I could just back up for a moment… I don’t like to stop the thread of your conversation, but…On Greenhouse George, how much can you say about the actual physical arrangement that’s not classified? I’ve forgotten exactly.
I don’t think you can say anything except it was a cylindrical design and a very, very big high explosive [that] used an awful lot of highly enriched uranium. But it’s still classified, and [it] has never been revealed exactly how the deuterium-tritium was heated and compressed by the fission explosion. So I can’t talk about that.
Well, there was a second test at Greenhouse called Item, I believe.
Item was a spherical normal implosion weapon. It was a boosting system, so it had only a small amount of tritium…deuterium-tritium. And that was the first boosted fission explosion. You can say a good deal about boosting. Boosting has a charge of D-T, typically. And of course the fission explosion is linear; that is, once you have a supercritical mass of plutonium or highly enriched uranium and you put in a neutron, then, you know, you’ve been able to calculate since 1942 or earlier what would happen. You have a chain reaction, and the exponentiation rate for the chain is something of the order of 100 per microsecond because that’s the collision rate of neutrons. And you need also…they knew the number of neutrons per fission. Of course the critical mass is determined by the escape probability of the fission neutron before it causes another fission. And that’s what Peierls and Frisch got right and what Heisenberg got wrong. Heisenberg said, well, you know, you have to have enough fissile material so that the neutron stays in there for the fifty generations or so that’s required to go from one neutron and one fission energy to fissioning the whole mass, so an exponentiation of 1024 or so, [which is] 80 [doublings]. The fact is that all that neutron escape does is to reduce slightly the number of neutrons per fission. So the neutron has to stay in there just for one generation, not for eighty. The radius is nine times smaller — that is, the square root of eighty, eighty-one, times smaller — and the mass is nine times eighty-one, seven hundred thirty times smaller. So that’s why Heisenberg had tons for his critical mass, and Peierls and Frisch had that divided by seven hundred, perhaps [a] few kilograms, which was also wrong, but… [it was actually] about sixty kilograms for the bare critical mass of uranium-235. So all of that was known, and there were [a] few people who were working with Edward Teller on the Classical Super, but it wasn’t a big program at the lab. It was interesting, especially in Theoretical Division. So, now I go to summer of 1950. I shared an office with Enrico Fermi in T Division because P Division had kicked me out. They decided that [chuckles] they would rather have me in Theoretical… in T Division. That was fun. And I kept our classified notebook; Fermi would occasionally write in that because he didn’t want the problems of caring for the classified notebook. We had a safe to which we both had the combination. Now in May of 1951 (actually April 27-May 23, 1951), I had just been for a month in Japan and Korea with Joe Mayer from the University of Chicago because the Air Force had created the Tactical Air Command, and they wanted to know what kind of… what their advisors — Zacharias, Oppenheimer, Rabi, Charlie Lauritsen — they wanted to know what kind of technology would be useful for the Air Force and what kind of laboratories they could build. So I went over there and looked at applications of flexible radios, pulsed-light rangefinders, bomb sights, fan-jet engines — a number of things — image intensifiers. We gave a report when we got back and then soon after I went to Los Alamos again. There I asked Edward Teller, whom I knew very well because he was at the University of Chicago, too, and we would go to seminars and talk about things… I asked him what was new, and he told me about the Teller-Ulam radiation implosion concept and then pointed me to the March 9, 1951 Heterocatalytic Detonations…uh, let’s see, what was the subtitle… anyhow, it was Hydrodynamic Lenses and Radiation Mirrors or something like that. [[“On Heterocatalytic Detonations I. Hydrodynamic Lenses and Radiation Mirrors”]] So its title is unclassified. It’s a secret document. So I read that. And Teller wanted me as an experimenter to devise an experiment that would be absolutely persuasive that this would really work because the Classical Super had just gone along, and nobody could show it would work… or, well, wouldn’t work… and it would be a big thing to try. You could always make it work by putting in enough tritium. But we didn’t have the tritium. So now when Stan Ulam… and we know more about this now because in 1979 Edward Teller dictated what’s called his “testament,” twenty or twenty-one pages of typescript that was only transcribed in 1983 or so… he had called Jay Keyworth to bring a recorder, a tape recorder, and he wanted to straighten out the history of the atomic bomb and the hydrogen bomb. And so he talked about Seth Neddermeyer and the implosion approach for the fission bomb and the radiation implosion approach for the thermonuclear weapon. So I pieced together from Edward’s testament and from his memoir that Stan had come to him in February of 1951 and said, “Look, Edward, I have a good idea. Let’s use a fission explosion…the energy from a fission explosion… to compress material.” And he wasn’t very specific as to whether this would be a way to bring a large amount of fissile material, supercritical, or fusion material. But he was going to use the shock from a nuclear explosion to do this. And he would use hydrodynamic lenses, which are what they used in the Trinity-Nagasaki weapon. So that 1945 nuclear weapon was a charge of uniform, high-performance explosive, and around that there were thirty-two lenses. The lenses were a combination of slow explosive, baratol… barium-loaded explosive… and fast explosive. You can calculate the shape of the heavy… the slow insert in order to get the spherically diverging waves from each of the detonators and explosive boosters converted into a segment of the spherically converging detonation wave that would then ignite the surface of the sphere of high-performance explosive. That was brought from Britain [but] was really not the best way to do it, by far. And there is a document… an official document… called “RDD-7” — seven is the most recent from the year 2001 — which is the official Department of Energy record of declassification decisions (maybe it’s “RDD” and the seventh edition). [[RESTRICTED DATA DECLASSIFICATION DECISIONS—1946 TO THE PRESENT (RDD-7); January 1, 2001; U.S. Department of Energy, Office of Declassification.]] And so it says that the United States has used ring lenses, which are much smaller in radial extent, and air lenses and other systems for getting the spherical, or approximation to a spherical, detonation wave. But to go from one fission explosion with an outgoing spherical shock wave to one converging spherical explosion… [it’s] not obvious how to do that. And if you could have thirty-two fission explosions that all went off at the same time, then you could have the shock waves… you could imagine having the shock waves… all form a spherical explosion. But you’re not going to have an explosive material. You really are using passive focusing, not a detonation wave, but sound focusing… shock focusing. Edward had two responses. One, compression won’t help with thermonuclear fuel. He had a “theorem,” which went back to the early 1940s… Edward was a fine theoretical physicist, very ingenious… you know, when you could pin him down to one idea at a time, he could really do a lot of working it out. So, he said to Stan Ulam, “Look, I’ll show you… I’ll write it up.” And the reason it doesn’t work is that everything, in his theorem, is bimolecular. So if you compare liquid deuterium at normal density and deuterium at a thousand-fold density, well, the reaction… everything per unit volume goes as the square of the density, and the time then goes inversely with the density because [of] the amount of material per unit volume. So the reaction goes to completion a thousand times faster, the transfer of energy from a hot ion to an electron goes a thousand times faster because there are a thousand times as many electrons, and the transfer from the electron to the X-ray, since it’s bimolecular, goes a thousand times faster. That was Edward’s theorem.
Dick, do you know if he ever wrote that up?
He never wrote it up… he says he never wrote it up.
There’s no report, as far as you know?
… explaining that…
… that principle…
His false theorem?
No, no, no. And he says he never wrote it down.
And when he was asked to write it down — by Konopinski, I guess — he pulled out a blank sheet of paper because, you know, something was telling him that this theorem wasn’t right. And he wasn’t going to… Edward is not quite so intuitive as I am… [that] is, he could actually write things down, calculate things… but he wouldn’t do it unless he saw the end was appropriate. So he never wrote it down. And this was the time when he was going to write it down to show Stan Ulam that his idea wouldn’t work. And, first of all, Edward says… there’s a very limited range of refractive indices that is available for shock waves because everything is fully ionized. And so the sound velocity is just dependent upon the number of electrons per atomic mass. And that goes for hydrogen, you know, from two particles per unit atomic mass, to uranium, in which case there are ninety-three particles for two hundred thirty-eight. So 2 to 1 up to 2.5 to 1, something like that. That’s not a very big bending of a shock wave that you could imagine to try to convert. It’d be a very big thing that you’d be talking about, a long focal length of this hydrodynamic lens. On the other hand… let’s see… Stan Ulam really wanted in that era to do something with a nuclear explosion. And so Edward said, well, how about the radiation? You’ve got all this radiation coming out. And there I think everybody, including Edward, had goofed. Any of us, in my opinion, could have — should have — thought about this if we weren’t so wrapped up with the Classical Super. And everybody knew that things would go faster. There are two things involved here. One is the pressure of radiation, and the other is the energy flow of the radiation. And they’re, of course, very simply related by the velocity of light… and, like pressure and energy for… even for gases, which are related by factor gamma, gamma minus one. So what should have been apparent to everybody is that if you increase the density, then you increase the potential reaction energy per unit volume. And you can increase that so that even at thermonuclear burning temperature for the ions, which might be 5 kilovolts or so, the energy in the radiation bath is σT4 times 4/3 or whatever, 4/π, divided by c, is the energy per unit volume. And everybody can calculate that. And as you ask to increase the density when you don’t need to react very much deuterium in order to satisfy the radiation bath, and if you compress more, then radiation content at 5 kilovolts is no bigger, but the reaction rate and the reaction energy are larger. So that’s one thing that says, yes, it does help to compress. It can solve your problem. And the other is how to compress. That’s what is in the now — unclassified term “radiation implosion.” There the radiation itself has sufficient pressure when it’s incident on something that either absorbs it or reflects it. And anything… absorbs soft X-rays, of course, and reflects… it depends on the temperature distribution. If the material is sufficiently opaque, then it will come to a high temperature on its surface and reflect, and that’s a factor two in the pressure that drives the material. Now you didn’t want to heat the fuel, so you have to surround it with something heavy and use the radiation or its consequences to compress. So, I talked to people from the time I got there in the summer of 1951. And I asked myself how I would demonstrate this. And of course everybody knows about radiation pressure, and you can calculate this whole thing, more or less. But you really couldn’t calculate for sure what goes on at fission explosion radiation densities. You cannot demonstrate it unless you have a fission explosion. And they’re limited in their size. You can’t have a tiny one and have high temperatures. So I decided that the best approach would be to design it full size. That would be most convincing. And of course the people at Los Alamos… the capabilities at Los Alamos were so enormous… and I’d been at Chicago, I had worked with the cyclotron there, and I had liquid deuterium and liquid hydrogen targets that I’d made myself. So [it] wasn’t any problem to me to design a test that would have liquid deuterium and liquid hydrogen. So I just went ahead and designed what seemed to me to be the simplest to understand and had the highest probability of working. And, you know, I had talked with everybody about their views, and there were only a few people, really, who had views of a practical configuration. I include the cryogenics when I said a practical configuration. And so I wrote up a paper. It’s title is unclassified. It’s “Characteristics of a Sausage Based on Ideas Current at…” [added by RG, post-interview: July 25, 1951, LAMD-7416, “Some Preliminary Indications of the Shape and Construction of a Sausage, Based on Ideas Prevailing in July 1951.”]
Are these ideas unclassified now?
The document is classified, but the title’s not classified.
It’s a four-page memo, July 25th, 1951, with a big foldout, maybe… because the thing was a cylinder, and I drew it to scale… maybe three eight-and-a-half by eleven-inch pieces of paper. And so that was the idea. And there was at that time the Theoretical Megaton Group that would meet every week or so.
Do you know if it was a successor to the Family Committee or in addition to the Family Committee.
I think it was in addition to the Family Committee. I think [the] Family Committee was smaller than [the] Theoretical Megaton Group. I have the minutes of a couple of meetings of the Theoretical Megaton Group. They’re redacted, but I came across them when I was trying to reconstruct some of this timeline. So Fermi was there that summer , and he and I went to the TMG. Hans Bethe was in charge [at] that time. I don’t know whether Hans had seen the memo first or whether we just used foils for projecting and that was his first acquaintance with it; I don’t remember that. But he could see that it would work, and I could explain how it worked pretty well and magnitude of… you have to have a radiation case (“radiation case” is [an] unclassified term)… and so there was a radiation case that kept the radiation in. It’s not really a radiation mirror, but indeed it does get up to a high temperature on its inside so…
I would assume the outer case is steel, is that correct?
Yeah the outer case was steel, but then you would line it with lead or uranium or whatever to do a better job of keeping the radiation in. And, of course, everything is vaporized at the several-hundred-volt, or kilovolt, temperature that you get by dividing the energy of the primary by the volume of the space accessible to radiation. And then you have to calculate how the radiation flows in the space between the radiation case and the thing that you want to compress.
In your initial design, did you envision it as empty or full of some lightweight, low-atomic-mass material.
At first, it was going to be empty, but then there was a concern that the radiation case or the tamper would vaporize and impede the flow of radiation. So I had liquid hydrogen. I could calculate it with hydrogen better than anything else. I knew about other things. I was using Styrofoam in my experiments at Chicago, so I knew about that. Styrofoam has carbon in it and maybe a little bit of nitrogen. But I was in a hurry, so I used liquid hydrogen. Then, of course, you have a problem because liquid hydrogen is cold. Good, because liquid deuterium has to be kept cold. But liquid hydrogen has a vapor pressure, one atmosphere at… at its normal boiling point, and so you have to work out a boundary, a closure at the top… the whole thing is vertical so you need a closure for the liquid hydrogen at the top of the liquid, and that would keep the fission bomb from getting cold. You can’t explode high explosive even at nitrogen temperature. And so I designed that, too. I had beryllium plates, despite all the problems you have with beryllium. Now it turned out that this was accepted, as Edward Teller says in his testament, and built just as I had designed it. And it worked. At this meeting, Hans Bethe worried about the case being thick enough. I pointed out that it only had to be half as thick as you might imagine because there’s a shock wave which is driven by the radiation pressure on the inside, and the ablation, it goes through the radiation case and blows off the outside of the structural steel. But nothing happens. Nothing is visible from the inside until the rarefaction gets back through this steel. Hans didn’t take that into account, and it’s a factor two in the thickness and so ran up the weight a lot [but] it wasn’t worth arguing about.
Now, were the dimensions that you visualized very close to the final dimensions of Mike?
Yes. And I calculated the heat leak and the evaporation rate of liquid hydrogen. I worked with Ferdinand Brickwedde of the National Bureau of Standards, which took the responsibility of building liquefiers in Boulder, Colorado for both hydrogen and deuterium, and the transport vessels, and refrigerators, and all that. And he told me later that the heat leak had been somewhat less than I calculated, so that was a reasonable job, too. So that’s what I did in the early part of the summer of 1951. But I learned some other things in preparing for, I think, the fiftieth anniversary [of the lab] in 1993… For the Sixtieth Anniversary, 2003 [yes, sixtieth]… there was a panel on the early days of the thermonuclear weapon. Harris Mayer and I and Conrad Longmire and Michael Bernardin, who’s head of Theoretical Nuclear Weaponry now (to some approximation, anyhow) had a classified panel discussion before a packed auditorium. And I wanted to get it right because I knew only what I had done. I wasn’t in doubt what I had done, but I didn’t know what other people had done… and vice versa, especially, since I was only a summer consultant. So somebody, in preparing for the sixtieth anniversary, wrote me a letter. And he said, “You should look at a classified handwritten letter from Stan Ulam to John von Neumann.” This was a secret letter… I wrote it down someplace… of February 23rd, 1951. So Stan had written Johnny, who was a good friend of his, and Klari [John von Neumann’s wife] also, that Stan was anxious to see him because he had had an idea, and he was worried now about the validity of the idea because it seemed to be popular. It was to use an auxiliary nuclear weapon to prepare a main charge, and…
Is that the letter in which he said, to von Neumann, “Edward likes this, so it must be wrong”?
That’s right. “Edward likes it, so I have my doubts.” That’s right. Exactly. In fact, that letter had been translated from the handwriting by somebody who didn’t know all that much about the subject, so that was a typed version. I was able to decrypt a few more words from the handwritten version. But I knew that he had written this to Johnny. And, of course, Edward eventually says that Stan doesn’t deserve credit for this — radiation implosion — why? Because he didn’t support it. That is, after the program began, Stan wasn’t enthusiastic about actually building the hydrogen bomb and so doesn’t deserve credit for it. Now, my own view is that this could have happened any time over the years, especially after 1945 when people had more leisure. And the fact that it didn’t was just a blind spot on the part of all of us. Hans Bethe says that it’s a kind of miracle, this radiation implosion, this idea. Carson Mark wrote that anybody who would have sat down to ask what were the consequence of Stan’s suggestion would have found that the radiation got there first, and it would have been obvious. So Carson says [that] Teller didn’t do anything. Bethe says it’s a wonderful idea. Teller says Stan didn’t do anything. And I think that it wouldn’t have happened for, oh, a year, two years, whatever, if Stan hadn’t come to Edward and said, “I know that it would help to compress.” And Edward says, “You can waste a lot of time talking to Stan, and so I decided to work it out.”
There’s an interesting remark in Ulam’s memoirs in which he indicates that he spent about two hours at that meeting with Edward and that Edward took a while to come around to realizing that implosion was a good idea, provided one did it with radiation.
Yeah, yeah. I think that’s right. And in the actual document, the March 9, 1951 document, the entire writing is by Edward Teller. Stan couldn’t have worked out any of that. Stan was not a physicist, he was a mathematician. He was a very smart guy. But it’s all Edward, you know, ranging from the propagation of the shock waves on the hydrodynamic lenses side, the calculation of the containment time, and the loss of radiation into the solid materials on the radiation mirror side. In fact, the title is Heterocatalytic Detonations: I, as if there was going to be a second; but [there] never was. So that’s the [story of the] design of what would become Mike.
Just one or two more questions, if I could. You have mentioned the outer steel case in your design, and I would assume an inner uranium cylinder, and inside of that, deuterium. How much of that is unclassified, and are you able to speak about the actual dimensions of the actual arrangement?
No, no, I don’t even remember the dimensions. But that’s it. That’s all that can be said. There was an inner cylinder of uranium — it might have been depleted uranium — surrounding the deuterium. There might have been a liner just to keep the deuterium away from the uranium. And there was the outer steel structure and an inner layer [on the steel] of what was going to be uranium, but was, I think, eventually lead. Now, Richard Rhodes’ book [[Dark Sun: The Making of the Hydrogen Bomb (New York: Simon and Schuster, 1996).]] has a lot detail in it, probably provided by Jay Wexler. But that’s not officially declassified, and I can’t comment on it. I can’t comment on what’s in the deuterium, either.
Right. OK. Well, two more questions. One is a simple human one, which is why do you think that Edward turned to you, a twenty-three-year-old guy, I guess you were? Did he know you well enough personally…
…to know you would be a good choice…
…to try to do this?
In fact, there was a memo — I don’t know whether I sent it to you — that Edward had sent Bradbury in 1949, when Edward was trying to round up the team.
Oh, I saw that, yes. Fermi says here…[you were] the best graduate student he ever had.
That…that’s right. So…
And Teller had seen that.
Yes. Well, Teller and Fermi were together at Chicago, and I was at Chicago since ‘47. And so by then we’d had a good deal of interaction. I was an experimenter, but I knew enough about the theory that I could be useful there.
My second question is, was lithium-6 deuteride an option in your design, or was it not even considered?
Well, I thought about it, but (a) it’s complicated, and (b) we didn’t have it. I wanted something that I could really figure. I couldn’t have done all those [other] things myself. I could figure deuterium, I could figure hydrogen, but I couldn’t do the others. But we really knew that we could use other fuels. I learned later from Herb York, who sent me the manuscript of a book that he was writing, maybe it was in [the] 1970s, on Oppenheimer, Teller, and the Superbomb [[The Advisors: Oppenheimer, Teller, and the Superbomb (San Francisco: W. H. Freeman, 1976).]], that the AEC had taken… Well [here’s the story.] After July 25th , I had some time at Los Alamos, and so I designed a deliverable, cryogenic version of Mike that would lie down rather than stand up. You know, for a physicist, that makes no difference, but this thing has to be supported. And the insides weigh a good many tens of tons, so you need long steel bolts, long so that they have less heat leak. You could have bolts that go straight through holes bored in the case, or whatever, but that would conduct too much heat. And so you have, in this steel case, long holes that are drilled, and diagonal bolts that go down. Not only do they have to support the inner parts, the uranium cylinder against gravity, but they have to support it against eight times gravity, which is a design requirement for airplanes in case the thing is bumpy on takeoff or has to land with the bomb inside. You worry about the heat leak because you can’t have a refrigerator connected all the time, so what is the pressure rise in the deuterium if you valve it off, and all that. So I designed that, too. And they built six of them—the Emergency Capability Weapons (named “Jughead”) — that were ready before, well about the same time, I guess, that Mike was fired. They were ready before that. They were available, and they’d been seen on airplanes, ready to take off and drop these things untested on the Soviet Union. But, yes, so in 1952. [But first] what else did I do in 1951? Well, I worked a lot on diagnostics for Mike and other thermonuclear weapons and these included things like boring holes part way through the outer case and using fast photography there or electronic detectors to look at the breakout of radiation-driven shocks so that you could tell what the temperature was on the inside, in the channel, and determine the equation of state for the steel or whatever you would like to put in this little experimental hole, which had a planar shock coming through it. I consulted with other people who had come to the Lab about diagnostics that they were preparing — again, the Berkeley group and later to become Livermore — Harold Brown, Herb York, Hugh Bradner… and then people from NRL [Naval Research Lab] — Ernst Krause and Montgomery Johnson, I remember. And the radiochemists, because by then the stable isotope addition for determining fluence and spectrum at various points, which I had published (in Secret documents) in 1950, had become standard. I had to do the cryogenic design for Mike by myself. I went to the head of the Cryogenics Group there, Ed Hammel, who said that they were all really burned out because they had all worked on George — Greenhouse George — so hard that they had to go back to physics and publishing papers and things like that. So that was just an additional task. I was fully capable of doing that. Now in the summer of ‘52 I went back [to Los Alamos]. Of course, we didn’t know exactly when Mike would be tested, but it was tested on November 1, 1952. I worked on some other things, but I can’t talk about them, for ordinary nuclear weapons. And then I went to Washington with Marshall Rosenbluth. We worked nights on the SEAC, which had mercury delay line memory. And we were doing calculations of alternative fuels and alternative dimensions for…
Do you remember whether this was June, July, August, what part of the summer?
No. Lois and I and our son Jeffrey had gone from Los Alamos to Cleveland and dropped off Lois and Jeffrey with my parents there while I was in Washington. I could probably reconstruct that. I think it was probably July. [Added after by Garwin: I still haven’t found Lois’s letters to my mother, but I will.] So we were working on alternative fuels, and Marshall was master of the code. I contented myself with helping him decide what examples we should use and which potential fuels. I remember I liked ammonia — nitrogen trideuteride, ND3, instead of NH3 — which has a lot of deuterium in it and works quite well. But, of course, it doesn’t have the contribution of lithium — we did lithium, of course, too. And there you have a choice. You have normal lithium deuteride, you have depleted lithium deuteride, lithium-7 deuteride, and lithium-6 deuteride. In fact even as late as January 1952, according to Herb York, the Atomic Energy Commission hadn’t made up its mind whether they were going to fuel Mike with liquid deuterium or with lithium deuteride. And, of course, thinking about lithium-6 deuteride: When you have the DD reaction, half of which goes to a proton and tritium, and 4 MeV of reaction energy, and half of which goes to a neutron and helium-3, then the helium-3 has such a big cross section for neutrons, even neutrons in the 10 kilovolt range, that it eats up the neutrons and it regenerates tritium. Lithium-6 does the same thing because lithium-6 plus a neutron gives you tritium, which is how we make our tritium anyhow. So that’s a good thing to do. [There’s] a lot more lithium-6 to begin with than there is helium in there, which is a reaction product itself.
Did you help Marshall do the actual programming for the SEAC or just advise on the physics?
No, no, just the physics.
It is so startling to me that I don’t remember this, because I was working that summer on the SEAC at night, the graveyard shift, and only with deuterium as fuel. We were just trying to…
When were you there?
Well, it was about a six-month period — early 1952, late winter, up until probably one or two months before Mike. So it’s quite astonishing that somehow Marshall was there, and I was there, yet we didn’t interact with each other or… if we did, I don’t remember.
Well you remember Marshall, right?
Oh, sure. Is he still living?
No. He died about four years ago [on September 28, 2003] from pancreatic cancer.
Oh, too bad.
He was diagnosed, and he was then very sick very quickly, but he lasted several years and in quite good health. He and his second wife, Sara, would travel all over the world to plasma physics conferences because Marshall was highly regarded as the dean of plasma physics. So the summer of ‘52, well…
And you spent days? Or weeks? How long were you in Washington?
Two weeks. It was pretty hot in the summertime… I’ll find out when I was there. I’ll let you know. I would walk from the Bureau of Standards out on Connecticut Avenue, I guess, to my hotel (the Park Central, 17th and G NW) in downtown Washington. It wouldn’t be safe in some of the intervening years, but it was OK then. Anyhow, it turned out all right. So I was in Los Alamos essentially only in the summers because I was an academic, or quasi-academic, and that was enough. And besides I wanted to take the family there, who all have fond memories of Los Alamos. So we were there every summer through 1958, and then in ‘59–‘60 we were at CERN so weren’t in Los Alamos either summer… and then for a couple of summers in the 1960s. I worked on other things, but they weren’t so exciting. I worked on the nuclear rocket, some, but it wasn’t so exciting as the hydrogen bomb. And in 1954 I wrote the first paper — it’s still secret but presumably doesn’t have any reason to be — on electromagnetic pulse from nuclear explosions. I started from the beginning. The problem is that if you have a nuclear explosion in vacuum, and you have a spherical explosion, then you don’t get any transverse sources for electromagnetic waves, which are transverse waves. Yet people from the very beginning had seen electromagnetic pulses, radio pulses, from nuclear explosions and from ordinary high-explosive explosions, too. And you [think about] this, and you can ask how it happens. Well, the explosion itself may be spherically symmetric, but the propagation for electromagnetic waves may not be. For instance, you may have a non-spherical case, you may have a nuclear explosive that’s sitting on the ground, and so you only get Compton electrons going up; the ones going down are shielded. Then, of course, you have the problem of the atmosphere because the atmosphere is ionized at some point, and so the propagation is damped. Take a space nuclear explosion, and now if you have an asymmetric case, even for a spherical nuclear explosion, then you can’t see the electromagnetic source, the Compton electrons that are inside the aluminum-foil case. And so you can have a lot of asymmetry as a result, not just a tiny amount. It’s going to be a big asymmetry. But even so — this was 1954 — as you calculate, you don’t get a very big signal. Why? Because in space when you drive Compton electrons out from the case there’s nothing beyond that, so the case becomes positive — all the electrons are being driven out — and after it gets up to a few million volts, then the electrons can no longer get out. They’re projected by the Compton effect, they start off at, say, two million volts, then they turn around because they’re in this field of the positive case with respect to infinity. So they may go out a meter or something like that. So the source is limited to an MeV-meter, that’s the dipole source, and in the radiation zone it goes down as 1/r because [there’s] constant energy flux in the radio waves of successive spheres. And so if you have an MeV per meter at one meter, then at a megameter, a thousand kilometers, you’re going to have one volt per meter. Well, we had some space nuclear explosions in 1958 — two kilotons, three of them — [over] the South Atlantic. But I don’t think anybody measured the electromagnetic field, the electromagnetic pulse. Then in 1962, Starfish, Starfish Prime, was 1.4 megatons at 400 kilometer altitude [over] the Pacific, and people were set up to measure the EMP at a distance of 1,000 kilometers — Hawaii, for instance.
Is this the Christmas Island tests? I’m trying to remember.
I don’t remember. If you just look up Starfish on the Web, it tells you all the details. But in fact, Hans Bethe apparently advised the experimenters as to how they should set their attenuators, amplifiers, and what not. And they looked. I think he must have repeated the calculation I had made because there was a similar calculation by Karzas and Latter [[W. J. Karzas and Richard Latter, The RAND Corporation, Santa Monica, California, “Detection of the Electromagnetic Radiation from Nuclear Explosions in Space,” Phys. Rev. 137, B1369–B1378 (1965).]] just before the tests in the context of detecting space nuclear explosions for verification of treaties. And they had the same value. But when people actually did the test, and they recorded the signal on the ground or in airplanes, it was not 1 volt per meter, it was 10 kilovolts per meter. That factor of 104 in voltage is a factor 108 in power in the radio wave. This was a great puzzle for many months. Conrad Longmire was worrying about it and was giving some lectures in 1963 in Albuquerque at the Air Force lab there. And it suddenly came to him what was happening — that is, these radio waves seemed to come from the bomb, but they don’t. They come from the upper atmosphere at 60 kilometers or so where the gamma rays from the bomb are converted into Compton electrons, mostly forward. And then another miracle happens because that doesn’t give any radio signal because it’s a longitudinal polarization, and you need a transverse polarization. Another miracle happens. The tiny Earth’s magnetic field, 1 gauss, bends all these trajectories the same way. So you have an acceleration of this pulse of electrons. And another miracle happens because the conversion region is spread out over 10 kilometers, and 10 kilometers involves 30 microseconds rather than hundredths of a microsecond. So how do you get a sharp pulse in the multi-megahertz range from something which is converted gradually over 30 microseconds? Well, it’s because in proper time at the observer, who is straight ahead on the radius, all of these things add up. The speed of the gamma ray is the same as the speed of the radio wave in this rarefied atmosphere. And so when you figure it, that’s what you get. I still work on that, actually, because there was the EMP Commission — the Electromagnetic Pulse Commission — in the early 2000s that was worried about such things. And it can be a big effect. I didn’t work much with others. I observed Fermi and Ulam in 1950, doing calculations on the Classical Super — the infinitely long cylinder of liquid deuterium. You know, my desk was here, Fermi’s desk was there, we were facing one another, a window behind him, [the] door to the room behind me. And Stan Ulam would come in in the morning and pull up a chair and pull out the slide on the desk. First they’d look at graphs. They would graph the previous day’s work that had been done by a computer — the computer was Miriam Caldwell, and she would take away the spreadsheet that Fermi had filled in the first few lines of to make sure that it worked. And he would use a slide rule for multiplication, division, exponentials and logarithms, and a motor-driven Marchant desk calculator for addition and subtraction.
By the way, she became Miriam Caldwell in the fall of 1950, having been Miriam Planck before that.
That’s right. That’s right. I’d forgotten her name. But she was Caldwell then. I knew her through David Caldwell as well. And so every day they would have [this session]. It isn’t what Fermi did all day because people would come in and talk to him about other things. But that’s what they would do in the morning. Then they would decide what other cases to run, what parameters to run. Fermi would fill out the spreadsheet — time going down the spreadsheet, radius going across.
This was summer of ‘50? or ‘51?
‘50, yes. (It was clearly 1950. I did not share an office with Fermi in 1951, and he and Stan would not have been calculating the Classical Super after the March 1951 Teller-Ulam paper.) At this point, RG examines the following written question provided by
“What do you remember about the contributions of others in 1950-52?”
Ulam and Everett
Ulam and Fermi
Fermi, alone and with Miriam Planck
Wheeler and his gang
Foster and Cerda Evans
OK. So I knew Cornelius Everett; I would see them [Ulam and Everett] together sometimes. We’d go to the Ulams’ home and he was often there. Ulam and Fermi I’ve told you about. Ulam was a very ingenious person. He invented a lot of things like iterated functions. You would have the sort of thing that Mandelbrot and others popularized, and strange attractors, or whatever. But he didn’t have any graphics. He would apply a function to a starting point and get a curve; then you’d have other curves [and so on]. Then Fermi. Well, I knew Fermi very well. And when I was learning about nuclear weapons, for instance, I said to Enrico, well, we all know, they’re very sensitive to neutrons, particularly the gun. Which is why in 1944, when plutonium started arriving from Hanford, and they were measuring its properties — it was the first time that they had multiple grams of plutonium — Emilio Segrè was measuring its properties, and he found that it was a copious emitter of neutrons, more than they had imagined from the cyclotron production of plutonium-239. And that’s because when you produce it by neutron capture in a reactor, you get the uranium-239 that goes to neptunium-239, and, in two days, it’s plutonium-239. But then the plutonium-239 sits there, and it has a substantial cross section for capture. So some of it gets burned up in fission, and some of it — about one-quarter — of what gets burned up goes to plutonium-240, which has a few (actually 6.5) thousand years lifetime… plutonium-239 is twenty-five thousand years or so. Plutonium-240 doesn’t decay significantly faster — 8,000 years — but it has a big, spontaneous fission probability — a short spontaneous fission lifetime. So they had designs for Little Boy, which was to have 60 kilograms, more or less, of uranium-235, but within a tamper that gave you more than two critical masses. And the critical mass of plutonium, instead of being 60 kilograms, is about 10 kilograms. And so they had Thin Man. Thin Man was the plutonium gun that instead of using 60 kilograms would have used 10 kilograms, and so it could be a lot smaller. But they had to abandon that because it would have given only a tiny fissile yield of only a few tons of high-explosive equivalent. And that’s where the implosion came in for assembly of plutonium. Little Boy and Fat Man started, in the beginning, in the Berkeley summer study in June . This is in the Los Alamos Primer, in Bob Serber’s lecture notes, which were put together really on the basis of the summer study. And so there was the gun-type assembly, and then there was an implosion assembly. And that can work with higher neutron background because the assembly time is shorter, because you have many hundreds of kilograms of explosive pushing on a small amount of heavy metal arranged in a shell. What counts is the thickness of the explosive and the thinness of the heavy metal. So the velocity is a lot higher than is the smokeless powder assembly of a large mass, where you have a little bit of powder pushing on a large mass. And it makes all the difference, even though the probability of pre-detonation — pre-initiation by background neutrons in the Trinity and Nagasaki weapon — was on the order of 10 percent.
10 percent, that high?
Yeah. There’s a memo from Oppenheimer to Groves, which has been declassified, and that’s what he says. He also says that when you get a fizzle yield with plutonium, with high-explosive assembly, that the fizzle yield is never below 1 kiloton — that’s if you initiate just as it’s passing first critical. And there’s a publication, oh, in the early 2000s — Carson Mark and Frank von Hippel and somebody else were involved with that. [[J. Carson Mark, Frank von Hippel, and Edward Lyman, “Explosive Properties of Reactor-Grade Plutonium,” Science & Global Security 17, no.’s 2-3, 170-185 (2009).]] So Miriam Caldwell’s involvement was that she would come in and take the spreadsheet and then fill it up and bring it back. She was a very attractive young woman.
We’re all agreed on that point.
And then Teller. I would see Teller and he would be interested in these diagnostic techniques — and maybe even ask me what you could learn with photography from a distance. De Hoffman. I guess there had been a lot written about de Hoffman — he was a perfectly competent physicist from Harvard, I guess…
Did he stay in Los Alamos when Teller went back to Chicago?
Yeah. Teller wasn’t at Los Alamos all the time. I don’t know whether he was on leave from Chicago or whether he just shared his time. I don’t know exactly when Freddie was in Los Alamos. But I remember at a cocktail party at Marvin (Murph) Goldberger’s house in Chicago — it must have been, probably, the winter of 1951-‘52. Leó Szilárd was there. I knew Leó Szilárd. I would visit him and look at his experiments and talk with him about things. He and Edward were talking, and then Edward said to me, “Well, you know how I use Freddie de Hoffman to work out my ideas.” He says, “That’s how Leó Szilárd uses me.” So Leó was even more ingenious and less likely — or capable — of working things out than Edward himself. [RG skips Wheeler and Mark, perhaps inadvertently. Returns to them later in the interview.] Now, Harris Mayer. You know, Harris is older — in fact, he’s 92, and I’m only 84-and-a-half. But he’s in very good shape. And so we (I and two video documentarians, Prof. Richard Breyer of Syracuse University and Anand Kamalakar of Brooklyn) talked with him for several hours in Los Alamos a few days ago. I knew him there from 1950. He was pretty quiet. But he really talked to everybody. And [it] turns out he’s a “person person” — he knew a lot about people, understood them very well, cared about them. He and his wife, Rosalie — late wife, she died, maybe 20 years ago — he and his wife, Rosalie, were very good friends of ours in Los Alamos. Then he left in ‘56 or ‘57 and went to work for Aerospace Corporation… oh, I guess he went to work with Plesset Associates in Los Angeles. And Ernie Krause, he was there, also Montgomery Johnson, some other smart people. He [Mayer] had worked on opacity during the war for Maria Mayer (no relation) at Columbia University. He finally got his Ph.D. from Edward Teller at Chicago in 1947 for the work he had done during the war and joined Los Alamos in the Theory Group [T Division]. I used to talk to him about equations of state and opacities and such like. Marshall Rosenbluth had been a student… a graduate student at Chicago. I got my degree in ‘49. I think he got his in ‘48. He was a very smart guy. Jack Steinberger was a graduate student at the same time. And Jack says that Rosenbluth was probably the smartest of all of us there. I worked with Marshall at Los Alamos, not so much on his implosion weapons… implosion calculations and explosion calculations on the thermonuclear secondaries, but I worked with him on the pinch effect, the Z-pinch. We have a paper on the snow-plow model of the pinch. And I knew his wife, Arianna, and his second wife, Sara; we were very good friends. We worked together a lot in JASON on all kinds of things. He was always a joy to work with. And Johnny von Neumann. I never worked with him. I met him several times, including at a cocktail party in Chicago at the Fermi’s house. I remember once really very vividly: Fermi was in his office… it must have been 1952, I think, because it was not the one he shared with me, but across Trinity [Drive] — T Division was on both sides of Trinity, and there were two bridges that went over Trinity.
Just as a side question, do you remember the names of those buildings? I remember Gamma Prime as one building — a wooden building — next to Trinity Drive. There must have been a Gamma, since there was a Gamma Prime.
No, I don’t remember. But I’ll look because we were at the Los Alamos Historical Society, which had some pictures… no, at Los Alamos Historical Museum… but it’s the Historical Society which has the archive. And I saw plans of Los Alamos during the war with the buildings on them, and probably they have their names. So I’ll find out. Your pen please, thank you. So Fermi was writing on his blackboard, and I asked him what he was doing. I had been with him to visit some experimenters who were working on Taylor instability. They had a water tank, maybe about six feet high; they would admit air at the bottom, and a big air bubble would then rise, and it would have a kind of rounded cap on it as it rose. So this is highly developed Taylor instability because it’s the light material which is rising through the heavy material. So Fermi had written on his blackboard. He was wanting to look at the evolution of Taylor instability at an interface. Here, for instance, you have a cup of water, and if you have water in the ordinary way, in a cup, you tap it and get ripples on the surface, the wave equation. And if you turn it over, gravity is reversed, and the wave equation no longer corresponds to a propagating wave, but an exponentially growing wave [instead]. The velocity… and the k value… is just this… the time behavior… frequency as a function of k… is just the same, except it’s imaginary. So you have it growing as eαt, you have the amplitude as A0eαt. And the t has the gravity and the pitch (spatial period or wavelength of the initial sinusoidal disturbance) and so on. All right, so anybody knows that, everybody can calculate that. Now what happens… a later step. You would imagine that the amplitude grows and the thing is symmetrical. But it can’t be, because pretty soon exponential growth means that the velocity is going to exceed that of light, it means that the energy in the wave is going to exceed the mass times gravity times the distance, and that can’t be. So what happens? Fermi schematized it. He said, “Well, this grows eventually to a narrow tongue [RG drawing] and then back here compensated because it’s incompressible by a wider but lower amplitude deformation. And so this whole thing can be schematized as the fraction of the pitch which is characterizing the tongue and the position of the front as a function of time. So he had an equation, which you can write down, but he didn’t see immediately how to solve it. Then John von Neumann came around — this I think was about 4:45 PM — and Johnny asked him what he was doing, and Fermi said, very briefly, what he was doing. And Johnny said, “Well, I’ll think about it and see you later at the cocktail party” or whatever. But he came back in about 15 minutes, and he wrote the answer on the board, and Fermi said to me, “That man makes me feel as if I know no mathematics at all.” And Fermi, you know, was not a powerful mathematician, but he was extremely good at solving problems. He could actually transform that problem… he would have transformed it if [it was] something that he knew the solution to. Now Richtmyer I didn’t know very well. I went fairly early on to a lecture by him in the auditorium. People at the end stood up and applauded because he had done the first yield calculation on the digital computer. It had taken many months of running time. So I guess he had assumed in the explosion calculation… he had assumed a static configuration of compressed plutonium in the Christy gadget, in the implosion assembly of a plutonium shell… that was the original idea. So you have a thin shell, and you move it together, and then the hollow [core] gets squeezed out of it, and that’s the implosion. But when they tried to do that with steel or lead simulants, then it was an awful mess. They could take photographs, especially in two dimensions — rather than 1D or 3D — two dimensions to compress the tube. And so you look through the tube while the explosives are compressing it. And you have Taylor instability, so multiple L values around the periphery, and those grow because the explosion is not perfectly symmetrical. They were afraid to do that with the plutonium… and so deep gloom on what to do to have explosive assembly. Then it may be von Neumann and Teller… who knows who it was… who started asking themselves questions like this. They said, well, what is the pressure in the plutonium as the assembly progresses? Because if you imagine that you just start with no explosive at all, but the explosive has set the shell into uniform motion… and now the velocity geometrically increases, very rapidly because you have a thin layer in one of the sectors, and it comes in along a geometrically defined wedge, or cone, and so it has to go much faster in order to stay ahead. So it can’t be incompressible. The pressure that you calculate exceeds the doubling density of plutonium if you estimate it. And so this gave the idea… then Bob Christy, I guess, said, “Look, we can change the density with high explosive. And so start with a solid ball and squeeze it with high explosive.” And, of course, people say, what good does that do? If you didn’t have a critical mass to begin with, you don’t have a critical mass now. But what counts for critical mass is σρR = 1, approximately, where σ is the macroscopic cross section for fission; macroscopic is per gram of material, so square centimeters per gram, and then you have the density ρ here. And so when the probability of a neutron escaping gets to be half or less, then the fact that the υ is 2.5 or 3.5 for plutonium — that’s neutrons per fission — means that you have a chain reaction going on. So, what is the mass required to make σρR = 1? And if I change the density, and the mass then is ρR3 times (4/3)π… So, for a fixed mass, just critical, if I change the density from… let me just write this a little… so the R is ~1/ρ, and the mass involved is ~ρ(1/ρ)3. Well, then I want ρR, so I have mass is ρR3 x (4/3)π and ρR is a constant — ρR is 1/σ. And so the critical mass for a given density is going to be ρR3, so it’s 1/ρ2. If I can double the density, then I only need one-quarter the mass to make a critical mass. Or, if I have just less than a critical mass, and I double the density, then I have a number of critical masses, which is four critical masses.
So, this is in reference to the so-called Christy gadget?
Yes, which is a solid ball, not a hollow ball. That was Christy’s point.
Oh, I see.
Previously, people said, well, we’ve got this metal and we’ll assemble it, push it all together then squeeze out the void and reduce the neutron leakage that way. And Christy said, probably on the basis of observations by von Neumann and Teller, but who knows… Christy said, “No. We can start with a solid ball, and if we can increase its density, we can get multiple critical masses that way.” Well, anyhow, in 1950 (I think) Bob Richtmyer had done the explosion calculation on the IBM computer on Madison Avenue, 590 Madison Avenue. It had taken him, I think, six months.
Yeah. And Richtmyer, of course, was one of the people who introduced the artificial viscosity in the calculations. Up to that time people handled shock waves by putting them in by hand. You had shock conditions, and you would determine where the shock would be… and you had constant conditions on one side of the shock and constant [conditions] on the other side, and you knew the velocity. And so you would reposition the shock in the numerical calculations. But he – Richtmyer — and von Neumann, as I understand it, introduced artificial viscosity to take care of that. And, of course, that required fine zoning. And then Conrad Longmire. I didn’t know [him] so well as Marshall Rosenbluth at that time… I knew him socially. He and his wife had five children [eventually seven, perhaps some after he and RG interacted]. He was a very capable person. I would talk to him… especially in the later years when I was at Los Alamos in T Division… and always found him to be helpful. But at our 2003 conference, the panel on the early thermonuclear weapons, he didn’t know of my work on the hydrogen bomb.
Yeah. He was astonished to see the memo. I’d gotten from the archivist the memo and sketch that projected the… I think he had just come back from getting his Ph.D. at Cornell, and so he and Marshall were working closely together on the burning calculations for the radiation-imploded secondary. The only reason that anybody knows of the work I did was that in 2001, I think it was, Jay Keyworth — George Keyworth, who had been President Reagan’s Science Advisor — sent Bill Broad of the New York Times this 20-page Teller testament. And there Teller has several pages about my work on the hydrogen bomb. Bill Broad, being a science writer for the New York Times, is somebody whom I knew. He would ask me about various things. He sent me a few paragraphs; he was going to write an article about this. And I said, “I haven’t seen the testament that you’re talking about. I can’t comment on this until I see it, you know, both for reasons of classification… and, besides, you may not have all this right.” So he sent it to me, and I read it with great interest, and you have a copy.
I have a copy, yes.
When it was transcribed, it was obvious that the person who transcribed it didn’t know any of the physics or any of the people because many of the names were wrong or confused — Kistiakowsky and Konopinski are interchanged in some cases and what not. So I [thought of writing to] John Browne, who I guess was Director at Los Alamos, and I said, “Look… here I have this thing, and I’m talking to Bill Broad… and I don’t want this to be personal, I want it to be about the topic.” But when it finally was published there was a lot of personality in it, and Bill Broad said, “Well, that’s my editor. You know, they won’t publish things that aren’t of human interest.” I then wrote John Browne, and I said, “Look, I didn’t say that Los Alamos was not able to do the cryogenic work. I said that they told me that they were burned out from having done so much before. And so I had to do it.” I don’t know that I ever got an answer to that. But now if you look on the Los Alamos Web site, it does say that Teller and Ulam wrote this paper, and I converted it into a practical design. Now, Walter Goad, whom I knew, of course, in T Division. But I didn’t know much about his work. Was he mostly in neutronics, Walter? I know that he wanted to… biology…
I don’t even remember, except he did collaborate with the Wheeler group in ‘50-‘51. He went over some of our calculations, but I don’t remember any details.
I remember Walter; my impression was he [was] a very sound person. Now Foster and Cerda Evans. Yes, I knew them. And I knew one problem that they were working on; they may even have worked on something I suggested, but… who knows? And this was to ask what happens when you have inhomogeneous media. I wanted to have foils of uranium separated by deuterium, liquid deuterium. The idea is, if you look at the shock relations, that macroscopically, in a strong shock, the kinetic energy and the potential energy behind the shock are equal. Now the potential energy is the thermal energy, the energy of heat. And the kinetic energy is just (1/2)ρv2, where v is the material speed behind the shock. This comes out of the Hugoniot equations. And, I said to myself, I want high temperatures, because I want to burn something, and so why don’t I have a medium which is laminated… uranium foils and… the shock is going this way… uranium foils with liquid deuterium between them. And so, especially if I have not very much uranium…or maybe if the whole material has rather low density,… maybe I’ll even put in some spaces in there. Then I’ll have a high shock velocity when being driven by any initial piston velocity… high shock velocity. And then because it’s vibrating, some of the time the deuterium will be compressed between adjacent uranium foils because I’m guaranteed that the internal energy — of vibration in this case — is going to be equal to the kinetic energy. I can arrange it so that the internal energy every once in a while ends up in the deuterium, and even though that’s a small fraction of the time, it’s going to help. So, they worked on such things. And Nick Metropolis… of course he built…
Well, I don’t want to take too much more of your time, but… [interruption for refreshments and to change to new recording track]
That’s OK, let me finish this last story. Foster and Cerda Evans… you know, they were very nice people… they worked hard. They were people who, I guess, had been assigned to work with Edward Teller. Edward was, you know, very unhappy after the June 1942 meeting that Oppenheimer convened and directed. Most of that meeting… like a day… was how you make fission bombs, nuclear weapons, because people had arranged for the uranium-235 from Oak Ridge, from gaseous diffusion and electromagnetic separation, and the plutonium to come from Hanford. ‘41 was Pearl Harbor. In ‘42 Fermi moved from Columbia to Chicago, and he was building the pile, but it didn’t go critical until November… no, let’s see, December, early December. December 7th 1941 was Pearl Harbor I think criticality was December 2, 1942. So in June ‘42 people assumed that the pile would work, and they would get plutonium and [it would] be just like uranium-235 and what not. And what would you do with it? They talked about this for about a day. You would have a gun that would shoot them together, and you would need an initiator to give the neutrons…turned on, you know, at the right time. And how do you switch on a nuclear reaction? Well, that was (α,n) reaction, the polonium-210/beryllium initiator. Then plutonium would just be the same [they thought]. But other people had ideas… an explosive assembly of shells. But they didn’t have the idea for the Christy gadget — that is, squeezing the metal to increase the number of critical masses. And then at Los Alamos… Of course Teller had played such a role in the June ‘42 conference — after the first day or so they went on to hydrogen bombs. And so suppose you had one… what would you do next? But Oppenheimer had the good sense to realize that you would never get a hydrogen bomb during the war — and you didn’t need it, either. Besides, people had the wrong idea, which I’ll get to, about hydrogen bombs. So Teller wanted to be put in charge of the Theoretical Group at Los Alamos and to have a big effort on thermonuclear weapons. Oppenheimer would have none of that. Oppenheimer recognized that Teller was not a person to run a group. He was a person to inspire groups — multiple groups — but not to run a group. So Teller did play a role — but I’m not quite clear what it was — in the fission weapon. He [actually] worked mostly on the Classical Super during the war, as I understand it, and Foster and Cerda Evans, I think, had been assigned to work with him from the very beginning. But that’s a matter of record; I just don’t know. Now Nick Metropolis was involved with the computers there. I knew Nick socially quite well. And, of course, there’s the Metropolis method in statistical mechanics or in solving… Well, now take the equation of state. You want to calculate the energy as a function of volume for some system where you know the molecular interactions and the energy is the sum of the kinetic energy of the individual particles at a given temperature, so we’re talking about thermal equilibrium. But you don’t know where the particles are and if you would just throw them in at random into a box and do a sum over states, because all of those would be equally probable… But the probability would be so low because of the interaction energy between the molecules… atoms. exp(-E/k), where E is the energy for the state, would be the probability, and you multiply that by the energy, E. And so that’s the energy versus density that you’re looking for at a given temperature. But you never find an acceptable state that way because the probability of accepting every state is so very low. I think it was Marshall Rosenbluth who actually invented the approach. But I think Rosenbluth, von Neumann — maybe both von Neumanns — and Metropolis [have their] names on the paper. So what you do is to put the atoms into a lattice so they’re guaranteed to have a low enough energy to be acceptable. And then you move one particle at a time an arbitrary distance, but not too big because the odds are if you move it too far that you will get into a highly repulsive region. And you accept the move by recalculating the energy. You don’t have to calculate very many because they’re all short-range interactions, so just the neighborhood. And if the energy is reduced… the repulsive energy is reduced… then you accept the move. If it is increased, then you accept it with a probability, e to the minus ΔE/kT, where T is the stated temperature. So in this way you always have a reasonable acceptance probability, and you can show that you map out all of the phase space, and you get the right answer for the energy. This is sometimes called the Metropolis method. But Marshall Rosenbluth, who was really quite a generous person, said that Nick just insisted on having his name attached to it because he had supplied the computer.
Would that method have required a computer in order to work out?
This was done on… well, by the time people had thought of it, I think. It was an invention, I think a really major invention. But they had the MANIAC at Los Alamos, and Nick was in charge — he had built it. Fermi used the MANIAC. He was fascinated because he had done these computations for the Classical Super with Ulam, but Fermi did all the work using the Marchant calculator on his desk for additions and subtractions and a slide rule for multiplications, divisions, and logarithms, and exponentials. And so he really knew how hard it was to calculate anything. And something that would do tens of thousands of multiplications per second compared with a couple of multiplications per minute was fascinating to him. So he knew about the applied computations. But he was looking to do some research. He’d always been interested in the ergodic theorem and how it was that systems explored phase space. So he said, well, let’s have a chain of finite oscillators all of the same energy coupled together. I think there were 16 or 64 masses.
This is Fermi, not Metropolis, you’re talking about?
Pure Fermi… so coupled by springs. They think he probably coded it himself. But eventually John Pasta, who was a mathematician who came to Los Alamos — I think from DOE, AEC headquarters, at the time — worked with Fermi on this and took over the coding. And sure enough, you reproduce all the normal modes of the system, so about 16 — and there are 16 normal modes — and you put in the distortion from equilibrium, and it just continues, and any other distortion. They’ve put in an arbitrary displacement of one particle from equilibrium. Then it pushes on its neighbors, and things got complicated and what not. But you can see at any time [that] you just have each of the normal modes not interacting with one another — but of course they get out of phase. OK. Then you add a small cubic energy to the spring. So instead of the energy of the spring being proportional to the square of the deviation, it has a little cubic term. And, obviously, this is not going to dissipate energy because there’s nothing which is dependent on velocity, so it can’t dissipate energy. But it will mix the normal modes. So if you have only a small cubic component, then you can still represent the system as normal modes, but you have transfer of energy from one mode to the other. And, sure enough, you can estimate how long it will take for a mode to lose its energy, and then it goes off and gets “uniformized” — made uniform among all of the normal modes. Why not? When they did it, they found that 95% of the energy came back into the original configuration after some hundreds of cycles, given the magnitude of the cubic term that they put in. And even more after 60,000 cycles, or something like that, 98% of the energy came back after having 95% repeated every few hundred cycles. And this was totally unexpected, and of course has a lot to do with solitons and things like that, which were later discovered. Fermi came back to Chicago after one summer in Los Alamos… his students say… and taught a little course in programming.
Before we leave the cast of characters, is there anything you recall about Wheeler… John Wheeler… or your interactions with him, if any?
I didn’t… I really had no interaction with John. I don’t know whether I even knew him at the time. Lois, do you remember Johnny Wheeler? I mean, of course we knew him at the Philosophical Society and things like that.
He was at our house, in Scarsdale. He must have come to IBM to give a talk.
Oh, that’s right. You’re right. Thank you.
He was in Los Alamos in the summer of 1950, but not ‘51. He had gone back to Princeton to set up Project Matterhorn.
Yeah, well, in ‘50 I was in P (Physics) Division; I had a little office in P Division and wouldn’t have seen much, probably, of theoreticians. So that’s probably how I missed him. Yes, he came to talk about gravity or foam or black holes or something like that at Yorktown… right. I might have even arranged for him to come, so I was probably his host. I didn’t know him. I got to know him later and learned that his brother had been killed during the war. Now, two other people who should be mentioned are… [rather] three other people… are Norris Bradbury and Marshall Holloway and Carson Mark.
Oh, I think I put his name here.
Oh, Mark. That’s right. I’m sorry I skipped over him. I got to Mayer. Thank you. So Carson [Mark] really was the spirit of the T Division. He was a wonderful manager. He really understood everything that was going on. He was Canadian by origin and had come, I guess, during the war. I would talk to him once in a while… [but] not so much. Sometimes I’d just go to his office and chat. Once he told me about the 1958 to ‘62 or whenever…moratorium on nuclear weapon tests. So he told me — this was obviously after that — about the fact that Los Alamos had carried out these hydronuclear tests that President Eisenhower had declared would not be considered as nuclear explosion tests. So there were 20 shallow wells around Los Alamos where they had detonated critical masses of plutonium…