Redactions

The Spy, the Human Computer, and the H-bomb

by Alex Wellerstein, published August 23rd, 2013

One of the most enigmatic documents in early Cold War nuclear history is the so-called Fuchs-von Neumann patent. It was Los Alamos secret patent application number S-5292X, "Improvements in method and means for utilizing nuclear energy," and dates from April 1946. It is mentioned, cryptically, often with heavy redaction, in many official histories of the hydrogen bomb, but also has recently surfaced as an object of historian's speculation. The most obvious reason for its notoriety comes from its authors, but its importance  goes deeper than that.

The Los Alamos identification badges for Klaus Fuchs and John von Neumann. Courtesy of Los Alamos National Laboratory.

The Los Alamos identification badges for Klaus Fuchs and John von Neumann. Courtesy of Los Alamos National Laboratory.

The co-inventors were Klaus Fuchs and John von Neumann. Fuchs was a brilliant German physicist who was later exposed as the most important of the Soviet spies at Los Alamos. Von Neumann was a brilliant Hungarian mathematician and physicist, a "ringer" they brought in especially to help manage the explosive lens program, and is generally considered one of the smartest people in the 20th century. As one of the major contributors to the invention of modern computing, it was often remarked in his time that he was much smarter than the machines he was developing — he could do crazy-complicated math in his head without breaking a sweat. And he was a vehement anti-Communist at that — a man who spoke openly about the benefits of instigating thermonuclear war with the Soviets. So on the face of it, it's an improbable match-up — the Soviet spy and the anti-Communist human computer. Of course, viewed in context, it's not so improbable: they were both talented physicists, both worked at Los Alamos, and nobody at the lab knew Fuchs was a spy.

The patent is interesting to historians because it allegedly plays a key role in answering the (still quite murky) question of whether the Soviets got the H-bomb through espionage or by their own hard work. We know that Fuchs passed it on to the Soviets — the question is, what did it contain, and how did the Soviets use it? The reason it shows up recurrently is because the patent is allegedly one of the first suggestions of the concept of radiation implosion, that is, using the radiation output of a fission bomb as a means of initiating fusion. In 1951, this would become one of the central components of the so-called Teller-Ulam design of the hydrogen bomb, on which all subsequent hydrogen bombs were based.

Record of invention for the Fuchs-von Neumann design, "Improvements in Method and Means for Utilizing Nuclear Energy."

Record of Invention for the Fuchs-von Neumann design, "Improvements in Method and Means for Utilizing Nuclear Energy." This copy is from the records of the Joint Committee on Atomic Energy in the Washington, DC, National Archives.

The contents of the patent itself is still officially secret in the United States. What is officially declassified  is little more than its title and some relevant dates — not much to go on. All descriptive aspects of it are totally classified. Which, generally speaking, makes it very hard to evaluate the aforementioned question of how useful it would have been to the Soviet Union, since we don't officially know what is in it.

But in the last couple of years, things have changed on this latter point. The patent application is still classified in the United States.1 But the contents of the patent appear to have been declassified, and published, in Russia. I've talked a bit in the past about how the Russians have declassified a bunch of information about the American bomb project that they got from espionage, despite the fact that this information is still probably classified in the United States. It would be really, really wonderful to know the back-story on why they do this, and whether there is any discussion with American classification authorities before the Russians start releasing information about old American bomb designs. The book series in question is Atomni' Proekt SSSR (USSR Atomic Project: Documents and Materials), which is cheerfully described on the inside as "intended for everybody interest in the history of the Soviet Atomic Project." Indeed!

In this case, the late Herb York told me that the late German Goncharov, one of the editors of the Atomni' Proekt SSSR series, approached him and told him somewhat informally that he thought this information should be declassified. York told me that he couldn't really officially respond to Goncharov about this, but he showed it to some people in Livermore, but they weren't very interested. Anyway, whatever the case, Goncharov apparently got the whole thing published in 2009 in volume 3, book 1 of the series.

Fuchs-von Neumann H-bomb design

The above image, supposedly the Fuchs-von Neumann concept, had appeared in a few other sources prior to that, but not with explanatory text. The only person who has published a serious analysis of it is the physicist and historian Jeremy Bernstein, who wrote about it in Physics in Perspective in 2010.2 At the time, Bernstein only had access to the diagram and its above legend, which was first seen in print in Gregg Herken's Brotherhood of the Bomb. Bernstein's caption of the above device (which he credits Carey Sublette for deciphering) is as follows:

The design for thermonuclear ignition that Klaus Fuchs turned over to his Soviet control in March 1948. The detonator (box) on the left represents a gun-type fission bomb consisting of a projectile and target of highly enriched uranium (71 kg of 70% pure U235), which when joined form a supercritical mass and produce an explosive chain reaction. The projectile is carried forward by its momentum, striking the beryllium-oxide (BeO) capsule on the right, which contains a liquid 50:50 D–T mixture, compressing it by a factor of about 3, as represented by the outer circle. The radiation produced in the fission bomb heats up the BeO capsule, producing completely ionized BeO gas, which exerts pressure on the completely ionized D–T gas, compressing the capsule further to an overall factor of about 10, as represented by the inner circle.

The interpretation is pretty good, considering the lack of additional source material! But the Russians have since released the entire document — including its original description of how it is meant to work, in the original English. Here is an excerpt:

The detonator is а fission bomb of the gun type. The active material is 71 kg of 40% pure U233 [sic].3 The plug (48.64 kg) sits in the projectile, which is shot bу the gun into the target, the remaining 22-24 kg sits in the target. The tamper is ВеО. The fission gadget has аn efficiency of 5% (calculated). The tamper, which is transparent for the radiation from the fission bomb, is surrounded bу an opaque shell which retains the radiation in the tamper and also shields the booster and main charge against radiation.  [...]

The primer contains 346 gm of liquid D-Т in 50:50 mixture, situated in the tamper. It is first compressed bу the projectile to 3-fold density. This precompression may not bе necessary. As the tamper and primer аге heated bу the radiation, the primer is further compressed, possibly to 10-fold density. (Radiation transport equalises the temperature in primer and tamper, and gives therefore rise to а pressure differential.) The compression opens the "gap" for the ignition of the primer. The primer is likely to have а very high efficiency (~80 %) of energy release.

The booster beyond the radiation shield contains D with about 4% Т. It is ignited bу the neutrons from the primer. Beyond the booster is the main charge of pure D, а cylinder of about 30 сm radius to contain the neutrons and arbitrary length.

So what's happening here is that the big piece of uranium is being shot against another piece. In the process, it rams into a bunch of fusion fuel (the 50:50 deuterium-tritium mixture), and just mechanically compresses it by a factor of 3. Just brute force. Then the fission bomb starts to detonate, using its radiation to ionize and heat the beryllium-oxide tamper. This causes it to ionize and blow off, compressing that 50:50 DT mixture, and starting a fusion reaction (they hope). This produces a huge number of neutrons, which then go and hit some more fusionable fuel (a DT mixture with only 4% tritium). The neutrons from this then go on to continue and ignite a final reservoir of pure deuterium "of arbitrary length."

The report then estimates that with 1 cubic meter of deuterium, it would have a blast range of 5 miles, a flash burn range of 10 miles, and prompt gamma radiation for 2 miles. It's not clear what values they mean exactly for those ranges (is blast 1 psi, 5 psi, 10 psi, 20 psi?), but playing with the NUKEMAP makes me think they are talking about something in the megaton range. For 10 tons of deuterium, it says: "Blast ~ 100 square miles, Flash burn to horizon оr 10,000 square miles if detonated high up. Radioactive poison, produced bу absorption of neutrons in suitable materials, could bе lethal over 100,000 square miles." Which is something in the many tens of megatons.

So was this radiation implosion? Well, kind of. The design uses the radiation energy to blow up the tamper, basically, compressing some fusion fuel. That's part of how the Teller-Ulam design would later work. But the entire thing is done in the context of the non-workable Classical Super — the idea that you can start a fusion reaction at one length of a column of fusionable material and it will propagate down the rest of it. Radiation implosion, here, is really just trying to get a better initial "spark" of energy to start the Classical Super reaction. This is very different from Teller-Ulam, where the complete implosion of the secondary is a key and fundamental aspect. All of which is to say, while this is a kind of radiation implosion (mixed in with a lot of other complicated things), it's pretty far from what is required to make a working hydrogen bomb, because the Classical Super idea just doesn't work. The fusion reaction of the sort proposed just can't sustain itself. Even Fuchs and von Neumann appear to have only perceived the importance of their invention as reducing the amount of tritium needed versus other Classical Super designs.4

The "Classical Super" design from 1946. A gun-type design is surrounded by a beryllium oxide tamper. There is a tubealloy (depleted uranium) shield to keep radiation off of the fusion fuel. The idea is to ignite a fusion reaction in a D+T mixture, which then ignites fusion reactions in a pure D mixture of arbitrary length.

The unworkable "Classical Super" design from 1946. A gun-type design is surrounded by a beryllium oxide tamper. There is a tubealloy (depleted uranium) shield to keep radiation off of the fusion fuel. The idea is to ignite a fusion reaction in a D+T mixture, which then ignites fusion reactions in a pure D mixture of arbitrary length. The Fuchs-von Neumann device is, in effect, just an attempt make the initial ignition easier, and does not question the (faulty) underlying assumption about propagation of the fusion reaction.

So what did the Soviets do with this information? Other documents in the series give some indication of that, and I've included the full set here (warning: large PDF, 13.5 MB), although it is completely in Russian.

The 1948 intelligence data is identified as "Material No. 713." It includes a brief, near verbatim summary (Document No. 32) by the physicist Yakov Terletsky (the same one who interviewed Bohr at Beria's request), as well as a brief report by Terletsky explaining what this material gave them compared to previous information about the American H-bomb work (Document No. 33). The latter is interesting; they seem most interested in the new theoretical information about the conditions required for deuterium fusion than they are about the specifics of the designs given. The strongest phrase is one where Terletsky says that the intelligence information will help them get beyond general, theoretical calculations and move towards the actual design or construction of a 'deuterium superbomb, and thus reduce the time required for the practical implementation of the superbomb idea."5

Document No. 34 includes an order by Beria that Kurchatov and Vannikov be required to write analyses of the intelligence information, and that Khariton be consulted on the information. This was made just a few days after Terletsky's report. Vannikov and Kurchatov's analysis is included as Document 35. They seemed quite encouraged and interested in the intelligence, and claim it will help them a lot. Of note is that they in particular mention that, among the useful things in the document, they thought that "the ideas about the role of particles and photons in the transmission of the explosion to the deuterium are new."6 So they do seem to have picked up on that, though it is again mixed into a lot of other details. They then used this material to propose that the USSR start a full-fledged Super program, along the lines of the unanswered questions (and even some of the answered ones) reflected in the intelligence information.

The end of Beria's April 1948 memo written as a result of the Fuchs intelligence, instructing that Khariton's opinion should be sought, especially with respects to the future work of the KB-11 (Arzamas-16) laboratory.

The end of Beria's April 1948 memo written as a result of the Fuchs intelligence, instructing that Khariton's opinion should be sought, especially with respects to the future work of the KB-11 (Arzamas-16) laboratory.

One thing that comes out in this as well is that the Soviet scientists at this point only had one other significant intelligence source related to the Super work, from late 1946 (Material No. 462, which I've uploaded here.) This appears to be a summary of the Super lectures that Enrico Fermi gave at Los Alamos, and is focused entirely on the Classical Super approach to the bomb, with many uncertainties. If these two caches were the only significant espionage they had on the American Super program before starting their own Super program, that's pretty interesting in and of itself, and helps put some pretty strict limitations on what they would have gotten out of the data.

Looking at all this, even with the knowledge that there is probably a lot more to the story, I come away with the following conclusions. First, Bernstein is probably right when he says that the Fuchs-von Neumann approach wouldn't have helped the Soviets very much in terms of arriving at the Teller-Ulam design. As he puts it:

Part of the irony of this story is that the unlikely collaborators, John von Neumann and Klaus Fuchs, produced a brilliant invention in 1946 that could have changed the whole course of the development of the hydrogen bomb, but was not fully understood until after the bomb had been successfully made.

I think perhaps this might go a little too far in praising radiation implosion — it is brilliant of a sort, but it is only one piece in the overall puzzle. The bigger issue on the road to the Teller-Ulam design was not so much the idea that the radiation could be used to transmit the energy, or even to implode the secondary, but getting away from the Classical Super notion of starting a small reaction that would then propagate onward. Indeed, the real breakthrough in the end appears to have been getting out of that mindset altogether. Ulam's big idea was of total compression of the secondary by putting the whole thing in a "box," which Teller then realized could be done more efficiently with radiation implosion. Radiation implosion is just a part of the overall mechanism, one which Ulam later insisted was actually not even required.

But my second, perhaps deeper conclusion is that this intelligence appears to have been much more important than has been previously thought. It didn't give the Soviets the right idea of how to make an H-bomb. But it did seem to convince them that the Americans were taking this work very seriously, and making serious progress, and that they should set up their own dedicated H-bomb program as soon as possible. That's a big deal, from an organizational standpoint, arguably a much bigger deal than the idea that it gave them some hint at the final design.

The Soviets were talking about a serious H-bomb program in 1948, before they had a fission bomb, and before USA was really committing itself to making a hydrogen bomb. In this sense, while it isn't clear that this intelligence saved them any real time on the bomb, it did convince them it was worth spending time on. In the end, that was what produced their successful hydrogen bomb models, in the end. Not the intelligence itself, but the program spurred on by the intelligence. And so in that sense, Fuchs does have a very real role in the Soviet hydrogen bomb program, even if his specific ideas were not realized to be relevant until after the fact. Our focus on the importance of individual design secrets can lead us to underestimate the importance of programmatic and organizational decisions in weapons development.7 We tend to focus on the question of, "did this fact get transmitted, and was it appreciated?" But facts, by themselves, do not build bombs. What they can do, though, is inspire scientists to think that the bombs can and should be made, so that they start the laborious process of actually making them. If the Fuchs intelligence did have this result, then it was very important indeed.

Notes
  1. Note that it is, and probably will always be, an application. Secret patent applications cannot be granted until they are non-secret. And even then, the Atomic Energy Act of 1946 explicitly bans the patenting of atomic bombs. For the long, thrilling history of secret atomic patents, check out my page on them and my various articles on the history of the policy. []
  2. Jeremy Bernstein, "John von Neumann and Klaus Fuchs: an Unlikely Collaboration," Physics in Perspective 12 (2010), 36-50. []
  3. The "detonator" description is very strange. For one thing, using only 40% enriched uranium (I am sure that the U-233 is a typo, because it is not in the Russian version, but the 40% is repeated in both) seems strange for 1946, and there is a marked difference between the specificity of one part of the gun-type design (48.64 kg) and the other (22-24 kg). This may be some kind of strange transcription error; the original drawing that the above diagram is based on says 22.36 kg. 5% efficiency is ridiculously high for such a description, too — "Little Boy" had about a 1% efficiency with 80% enriched uranium. If 5% of the U-235 in the "detonator" underwent fission, it would be around 24 kilotons in yield — somethings quite achievable by less speculative means. []
  4. The 1946 Record of Invention describes the object of the device as follows: "To provide an improved method and means for initiating a self-sustaining thermo-nuclear reaction which minimizes the amounts of materials employed." (My emphasis.) When you compare this design with other Classical Super designs, it is clear, I think, that they are really trying to keep the amount of tritium down to a minimum, by starting the fusion with the heavy compression of a very small tritium-rich zone. Given that in 1946, the supply of tritium was minuscule, this would be a pretty appealing aspect of such a design. []
  5. "Материал #713а, в целом, позволяет перейти от общих теоретических расчетов к конструированию дейтериевой сверхбомбы и т[аким] о[бразом] сократить время, необходимое для практического осуществления идеи сверхбомбы." []
  6. "Приведенные в материале #713а принципиальные соображения о роли трития в процессе передачи взрыва от запала из урана-235 к дейтерию, соображения о необходимости тщательного подбора мощности уранового запала и соображения о роли частиц и квантов при передаче взрыва дейтерию являются новыми." []
  7. Michael Gordin makes this point excellently in his excellent Red Cloud at Dawn when discussing why the Smyth Report is actually a pretty important document for the Soviets: it didn't give them any details about how to build a bomb, but it did tell them how to start a bomb-building research program. []

Tags: , , , , , , ,


17 Responses to “The Spy, the Human Computer, and the H-bomb”

  1. Ben says:

    I know this isn’t the main point of your post, but if Ulam ultimately decided radiation was not necessary for implosion, what replaced it? Neutrons?

    • Ulam’s original idea just used mechanical shock. It’s not clear that he thought that fusion could be caused with just that (though he seems to think this is the case), but the original concept was probably using the mechanical shock of an atomic bomb to achieve incredible compressions in fissile material — which would make very large bombs indeed, even if they were not strictly hydrogen bombs. Ulam thought that Teller’s approach, with radiation implosion, represented an improvement on his idea, but insisted that his own original idea would have worked to some degree as well.

      • Wotan says:

        Alex, if i’m not mistaken i remember reading that the “mechinical shock” idea was tested at Castle Koon (and famously fizzled). This always puzzled me because i couldn’t understand why they decided to try that even after Mike (and Castle Bravo) gave a conclusive proof of radiation-implosion. Does this story rings any bell with you?

  2. Allen Thomson says:

    > (I am sure that the U-233 is a typo, because it is not in the Russian version, but the 40% is repeated in both)

    Undoubtedly a typo on the U233; the Russian version explicitly says it’s U235:

    Активным веществом детонатора служит U2З5 40%-ной чистоты в количестве 71 кг.

    I don’t understand the 40% either. Perhaps Carey Sublette or George Herbert could opine.

    • Right.

      I have two theories:

      1. It’s some kind of typo/mistake/whatever. Seems kind of unlikely but a possibility.

      2. It’s some kind of idea they were floating around in 1946 or so to extend the size of the stockpile by using less material.

      I do know that some of the speculative “primaries” they were proposing in the early days (when they were still focused on getting the biggest primaries possible to start the Classical Super) were massive gun-type devices (Teller had one he dubbed “Little Edward,” which is a little disturbing as a code-name), presumably with the idea that they could somehow combine a lot of U-235 very quickly that way (e.g. double guns or something). So I also wonder if this isn’t something related to that line of thinking.

      In any event, it’s one of the ways in which this design looks very foreign by later standards — it is clearly something developed in a very particular context, one quite removed from what eventually became the H-bomb!

      • Allen Thomson says:

        I’m tempted to think that 40% is a misreading of an original 90% as 4 and 9 are pretty similar in appearance, especially if hand-written. But it needs study — maybe there is a physical reason why 71 kg of 40% U235 in the indicated geometry makes sense.

        • Yeah, I was thinking about that as well. In the full file, you can see that some of this was written by hand, which would have made such an error easy. But still, you’d think the translator/interpreter (e.g. Terletskii) would have noticed it. Perhaps it is another reason for Beria to be paranoid!

  3. I should point out that I am the author of the analysis and interpretation of the FvN diagram which I provided to Jeremy Bernstein. It is entirely and exclusively my work.

    I was unaware that he had published it under his own name. Did he not give me complete credit for the analysis?

  4. Regarding the Little Edward trigger composition: 40% U-233 must be an error (possibly two errors), as this makes no sense from any perspective.

    If one had U-233, the rough equivalent of plutonium in performance, but harder to make (breeding material is not naturally present in uranium fuel) and thus more costly, one would not dilute it to 40% with anything.

    A 40%, or 70%, U-235 trigger makes some sense in 1946 when the Oak Ridge gaseous diffusion topping cascade was still under construction, and the Calutrons were slated for permanent shut-down and a lower enrichment product was readily available, but typical weapon-grade uranium was in shorter supply.

  5. Peter Solem says:

    I get the sense that the Von Neumann-Fuchs patent might have more to do with the development of boosted fission weapons (in which the solid plutonium core is replaced by a hollow plutonium sphere filled with a mixture of deuterium and tritium gas) than with the development of actual fusion weapons. This is still a very significant issue, since this vastly reduced the weight of the fission weapons such that delivery by submarine-hosted intercontinental ballistic missiles became possible.

    “To fission more of a given amount of fissile material, a small amount of material that can undergo fusion, deuterium and tritium (D-T) gas, can be placed inside the core of a fission device. Here, just as the fission chain reaction gets underway, the D-T gas undergoes fusion, releasing an intense burst of high-energy neutrons (along with a small amount of fusion energy as well) that fissions the surrounding material more completely. This approach, called boosting, is used in most modem nuclear weapons to maintain their yields while greatly decreasing their overall size and weight.”

    Source: Technologies Underlying Weapons of Mass Destruction December 1993
    available via http://www.fas.org/spp/starwars/ota/9344.html

    • Cthippo says:

      Building on this idea, this patent may be why the Soviets went down the road that led to the Sloika or “Layer cake” design.

      Something to understand about Soviet design principals in this era is that they were allergic to innovation. Stalin ordered, and enforced, that western designs be copied exactly, rather than spending precious resources trying to improve on them. This shows up in things like Soviet copies of British jet engines having the British manufacturers name cast into the soviet parts. The Soviet technicians no doubt knew that casting someone else’s name into a part was pointless and added difficulty, but they were told to copy it exactly, no exceptions.

      Obviously this approach wouldn’t work in the realm of nuclear weapons where there was no prototype to copy however they probably gave more credence to the information they got through espionage than perhaps it deserved (having Beria, the intelligence head, running the program probably also ensured they paid a lot of attention to the intel). Fuchs left Los Alamos in 1946 and so the Soviet data on the current thinking in American weapons design was largely cut off. I suspect that the soviet program continued working on boosted fission long after the Americans wrote it off simply because it was the last piece of hard information they had from the American program.

  6. kme says:

    Your conclusion is somewhat reminiscent of the idea that by the end of the war, the only secret really worth keeping was the fact that the fission bomb did actually work – once you were certain that the bomb could be built, there were hundreds of scientists around the world who could have figured out how to do it.

  7. The diagram in your blog, which is in Russian and the diagram in the xtract from the Atomnyi proekt SSSR are different. Anyone know why? The paper in Riabev’s book is almost certainly from Fuchs. I have found a hand written draft of the figures and tables in a file in the National Archives. Its in his hand writing. But there is no diagram.

    • I’m confused as to which diagrams you’re talking about — all of the diagrams here are scanned from Atomnyi proekt SSSR directly. There are technically two from Atomnyi proekt SSSR, one that is hard to make out and is probably an original sketch, and the other which is the more carefully done one in Russian.

  8. […] over by Groves’ deputy, Brig. General Thomas Farrell. Among the scientists in attendance were John von Neumann and William Penney (but not […]