While Kistiakowsky struggled with that dilemma the theoreticians began to glimpse how a successful implosion mechanism might be designed.
The previous spring the Polish mathematician Stanislaw Ulam, then thirty-four years old and a member of the faculty at the University of Wisconsin, had found himself unhappy merely teaching in the midst of war: “It seemed a waste of my time; I felt I could do more for the war effort.” He had noticed that letters from his old friend John von Neumann often bore Washington rather than Princeton postmarks and deduced that von Neumann was involved in war work; now he wrote asking for advice. Von Neumann proposed they meet between trains in Chicago to talk and turned up impressively chaperoned by two bodyguards. Eventually Hans Bethe sent along an official invitation. In the winter of 1943 Ulam and his wife Franchise, who was then two months pregnant, rode the Sante Fe Chief to New Mexico as so many others had done before them. “The sun shone brilliantly, the air was crisp and heady, and it was warm even though there was a lot of snow on the ground — a lovely contrast to the rigors of winter in Madison.”
The day of his arrival Ulam met Edward Teller for the first time — he was assigned to Teller's group — who “talked to me on that first day about a problem in mathematical physics that was part of the necessary theoretical work in preparation for developing the idea of a ‘super’ bomb.” Teller's preemption of Ulam's first days at Los Alamos for Super calculations was symptomatic of the discord that had been widening between him and Hans Bethe, who needed every available theoretical physicist and mathematician to concentrate on the difficult problem of implosion. Teller had contributed enthusiastically and crucially to the most interesting part of the work. “However,” Bethe complains, “he declined to take charge of the group which would perform the detailed calculations on the implosion. Since the theoretical division was very shorthanded, it was necessary to bring in new scientists to do the work that Teller declined to do.” That was one reason the British team had been invited to Los Alamos.
Teller recalls no specific refusal. “[Bethe] wanted me to work on cal-culational details at which I am not particularly good,” he counters, “while I wanted to continue not only on the hydrogen bomb, but on other novel subjects.”
The Los Alamos Governing Board reevaluated the Super once again in February 1944, learning that despite deuterium's more favorable cross section it would still be difficult to ignite. A Super would almost certainly require tritium. The small tritium samples studied so far had been transmuted in a cyclotron by bombarding lithium with neutrons. Large-scale tritium production, like large-scale plutonium production, would require production reactors, but the piles at Hanford were unfinished and previously committed. “Both because of the theoretical problems still to be solved and because of the posssibility that the Super would have to be made with tritium,” reports the Los Alamos technical history, “it appeared that the development would require much longer than originally anticipated.” Work could continue — the Super was too portentous a weapon to ignore — but only to the extent that it “did not interfere with the main program.”
Von Neumann soon drafted Ulam to help work out the hydrodynamics of implosion. The problem was to calculate the interactions of the several shock waves as they evolved through time, which meant trying to reduce the continuous motion of a number of moving, interacting surfaces to some workable mathematical model. “The hydrodynamical problem was simply stated,” Ulam comments, “but very difficult to calculate — not only in detail, but even in order of magnitude.”
He remembers in particular a long discussion early in 1944 when he questioned “all the ingenious shortcuts and theoretical simplifications which von Neumann and other… physicists suggested.” He had argued instead for “simpleminded brute force — that is, more realistic, massive numerical work.” Such work could not be done reliably by hand with desktop calculating machines. Fortunately the laboratory had already ordered IBM punchcard sorters to facilitate calculating the critical mass of odd-shaped bomb cores. The IBM equipment arrived early in April 1944 and the Theoretical Division immediately put it to good use running brute-force implosion numbers. Hydrodynamic problems, detailed and repetitious, were particularly adaptable to machine computation; the challenge apparently set von Neumann thinking about how such machines might be improved.
Then a member of the newly arrived British mission made a proposal that paid his mission's way. James L. Tuck was a tall, rumpled Cherwell proteg6 from Oxford who had worked in England developing shaped charges for armor-piercing shells. A shaped charge is a charge of high explosive arranged in such a way — usually hollowed out like an empty ice cream cone with the open end pointed forward — that its normally divergent, bubble-shaped shock wave converges into a high-speed jet. Such a ferocious jet can punch its way through the thick armor of a tank to spray death inside.
It had just become clear from theoretical work that the several diverging shock waves produced by multiple detonators in Neddermeyer's experiments reinforced each other where they collided and produced points of high pressure; such pressure nodes in turn caused the jets and irregularities that spoiled the implosion. Rather than continue trying to smooth out a colliding collection of divergent shock waves, Tuck sensibly proposed that the laboratory consider designing an arrangement of explosives that would produce a converging wave to begin with, fitting the shock wave to the shape it needed to squeeze. Such explosive arrangements were called lenses by analogy with optical lenses that similarly focus light.
No one wanted to tackle anything so complex so late in the war. Geoffrey Taylor, the British hydrodynamicist, arrived in May to offer further insight into the problem. He had developed an understanding of what came to be called Raleigh-Taylor instabilities, instabilities formed at the boundaries between materials. Accelerate heavy material against light material, he demonstrated mathematically, and the boundary between the two will be stable. But accelerate light material against heavy material and the boundary between the two will be unstable and turbulent, causing the two materials to mix in ways extremely difficult to predict. High explosive was light compared to tamper. All of the tamper materials under consideration except uranium were significantly lighter than plutonium. Raleigh-Taylor instabilities would constrain subsequent design. They would also make it difficult to predict bomb yield.
As the IBM results clarified shock-wave behavior the physicists began seriously to doubt if a uniform wrap of HE could ever be made to produce a symmetrical explosion. Complex though explosive lenses might be, they were apparently the only way to make implosion work. Von Neumann turned to their formulation. “You have to assume that you can control the velocity of the detonation wave in a chemical explosive very accurately,” Kistiakowsky explains, “so if you start the wave at certain points by means of detonators you can predict exactly where it will be at a given time. Then you can design the charge.” It was soon clear that the velocity of the converging shock waves from the several explosive lenses that would surround the bomb core could vary by no more than 5 percent. That was the demanding limit within which von Neumann designed and Kistiakowsky, Neddermeyer and their staffs began to work.
In the spring of 1944 the two difficult personal conflicts — between Teller and Bethe and between Kistiakowsky and Neddermeyer — forced Oppenheimer to intervene. First, Bethe writes, Teller withdrew from fission development:
With the pressure of work and lack of staff, the Theoretical Division could ill afford to dispense with the services of any of its members, let alone one of such brilliance and high standing as Teller. Only after two failures to accomplish the expected and necessary work, and only on Teller's own request, was he, together with his group, relieved of further responsibility for work on the wartime development of the atomic bomb.
A letter from Oppenheimer to Groves on May 1, 1944, seeking to replace Teller with Rudolf Peierls, corroborates Bethe's account: “These calculations,” it says in part, “were originally under the supervision of Teller who is, in my opinion and Bethe's, quite unsuited for this responsibility. Bethe feels that he needs a man under him to handle the implosion program.” It was, Oppenheimer notes, a question of the “greatest urgency.”
Ulam remembers that Teller threatened to leave. Oppenheimer stepped in then to save him for the project. He encouraged Teller to give himself over to the Super — encouragement, Teller wrote in 1955, perhaps disingenuously, that he needed to move him on from the immediate task at hand:
Oppenheimer… continued to urge me with detailed and helpful advice to keep exploring what lay beyond the immediate aims of the laboratory. This was not easy advice to give, nor was it easy to take. It is easier to participate in the work of the scientific community, particularly when a goal of the highest interest and urgency has been clearly defined. Every one of us considered the present war and the completion of the A-bomb as the problems
