Making of the Atomic Bomb
I was sitting in the tank when the first explosion went off. George Kistiakowsky was in one tank and I was in the other. We were looking through the periscopes and all that happened was that it blew a lot of dust in our eyes.2157 And then—we hadn’t thought about this possibility at all—the whole forest around us caught on fire. These pieces of white-hot metal went flying off into the wild blue yonder setting trees on fire. We were almost surrounded.
Implosion lens development had begun the previous winter, says Bethe, when John von Neumann “very quickly designed an arrangement which was obviously correct from the theoretical point of view—I had tried and failed.”2158 Now in the fall and winter of 1944–45 Kistiakowsky had to make the theoretical arrangement work.
An optical lens takes advantage of the fact that light travels at different velocities in different media. Light traveling through air slows when it encounters glass. If the glass curves convexly, as a magnifying glass is curved, the light that encounters the thicker center must follow a longer path than the light that encounters the thinner edges. The effect of these differing path lengths is to direct the light toward a focal point.
The implosion lens system von Neumann designed was made up of truncated pyramidal blocks about the size of car batteries. The assembled lenses formed a sphere with their smaller ends pointing inward. Each lens consisted of two different explosive materials fitted together—a thick, fast-burning outer layer and a shaped slow-burning solid inclusion that extended to the surface of the face of the block that pointed toward the bomb core:
The fast-burning outer layer functioned for the detonation wave as air around an optical lens functions for light. The slower-burning shaped inclusion functioned as a magnifying glass, directing and reshaping the wave. A detonator would ignite the fast-burning explosive. That material would develop a spherical detonation wave. When the apex of the wave advanced into the apex of the inclusion, however, it would begin burning more slowly. The delay would give the rest of the wave time to catch up. As the detonation wave encountered and burned through the inclusion it thus reshaped itself from convex to concave, from a spherical wave expanding from a point to a spherical wave converging on a point, emerging fitted to the convex curve of the spherical tamper. Before the reshaped wave reached the tamper it passed through a second layer of solid blocks of fast-burning explosive to add to its force. The heavy natural-uranium tamper then served to smooth out any minor irregularities as the spherical shock wave compressed it passing through to the plutonium core.
Kistiakowsky would apologize after the war for a research program “too frequently reduced to guesswork and empirical shortcuts” because the field had been grossly neglected.2159 “Prior to this war the subject of explosives attracted very little scientific interest,” he wrote in an introduction to a technical history of X Division’s work, “these materials being looked upon as blind destructive agents rather than precision instruments; the level of fundamental knowledge concerning detonation waves—and strong shock waves induced by them in the adjacent non-explosive media—was distressingly low.”2160 To support its experiments X Division expanded an explosives-casting site a few miles south of Anchor Ranch, constructing roughhewn earth-sheltered timber buildings because hauling in concrete would have delayed the work.
Not until mid-December 1944 did a lens test look promising; the eighteen 5-kilogram bombs Groves told George Marshall he hoped to have on hand by the second half of 1945 he also thought might explode so inefficiently that each would be equivalent to no more than 500 tons of TNT, down from the 1,000 tons Conant had heard estimated in October.
Kistiakowsky had to fight once more with Parsons before he won the field. “So much pessimism was developing about our ability to build satisfactory lenses,” he recalls, “that Captain Parsons began urging (and he was not alone in this) that we give up lenses completely and try somehow to patch up the non-lens type of implosion.”2161 Kistiakowsky thought that alternative hopeless. Early in 1945 Groves came out to monitor the debate. In the end Oppenheimer took Kistiakowsky’s side and decided for lenses. Parsons’ Ordnance Division then restricted its work to the uranium gun, Little Boy, and to engineering the weapons for the battlefield. X and G Divisions worried about implosion.
Finishing the high-explosive castings by machining them was the most dramatic innovation Kistiakowsky introduced. He wanted to shape the HE components entirely by machining from solid pre-cast blocks but lacked sufficient time to develop and build the elaborate remote-controlled machinery the innovative technology would have required. He settled instead for precision casting with machine finishing and used his limited supply of machinists primarily to turn out the necessary molds. Molds gave him “the greatest agony,” he remembers; the HE components of the bomb totaled “something in the nature of a hundred or so pieces, which had to fit together to within a precision of a few thousandths of an inch on a total size of five feet and make a sphere. So we had to have very precise molds.”2162 Eventually mold procurement paced Fat Man’s testing and delivery.
But even with the necessary molds on hand, casting HE was far from simple, another technology that had to be learned by trial and error. In February 1945 Kistiakowsky chose an explosive called Composition B to serve as the fast-burning component of Fat Man’s lenses and a mixture he had commissioned from a Navy research laboratory, Baratol, for the slow-burning component.2163 Composition B was poured as a hot slurry of wax, molten TNT and a non-melting crystalline powder, RDX, that was 40 percent more powerful than TNT alone. Baratol slurried barium nitrate and aluminum powder with TNT, stearoxyacetic acid and nitrocellulose:
We learned gradually that these large castings, fifty pounds and more each, had to be cooled in just certain ways, otherwise you get air bubbles in the middle or separations of solids and liquids, all of which screwed up the implosion completely. So it was a slow process. The explosive was poured in and then people sat over that damned thing watching it as if it was an egg being hatched, changing the temperature of the water running through the various cooling tubes built into the mold.2164
The wilderness reverberated that winter to the sounds of explosions, gradually increasing in intensity as the chemists and physicists applied small lessons at larger scale. “We were consuming daily,” says Kistiakowsky, “something like a ton of high performance explosives, made into dozens of experimental charges.”2165 The total number of castings, counting only those of quality sufficient to use, would come to more than 20,000. X Division managed more than 50,000 major machining operations on those castings in 1944 and 1945 without one explosive accident, vindication of Kistiakowsky’s precision approach. A RaLa test on February 7, 1945, showed definite improvement in implosion symmetry. On March 5, after a strained round of conferences, Oppenheimer froze lens design. However scarce plutonium might be, no one doubted that Fat Man would have to be tested at full scale before a military weapon could be trusted to work.
* * *
A problem small in scale but difficult of solution was the initiator, the minuscule innermost component of the bombs.2166 The chain reaction required a neutron or two to start it off. No one wanted to trust a billion dollars’ worth of uranium or several hundred million dollars’ worth of plutonium to spontaneous fission or a passing cosmic ray. Neutron sources had been familiar laboratory devices for more than a decade, ever since James Chadwick bombarded beryllium with alpha particles from polonium and broke the elusive neutral particle free in the first place. In his early lectures at Los Alamos Robert Serber had discussed using a radium-beryllium source in a gun bomb with the radium attached to one piece of core material and the beryllium to the other, arranged to smash together when the gun was fired and the two core components mated to complete a critical assembly. Radium released dangerous quantities of gamma radiation, however, and Edward Condon noted in the Los Alamos Primer that “some other source such as polonium . . . will probably prove more satisfactory.”2167 Polonium emitted copious quantities of alpha particles energeti
c enough to knock neutrons from beryllium but very little gamma radiation.
The challenge of initiator development was to design a source of sufficient neutron intensity that released those neutrons only at the precise moment they were needed to initiate the chain reaction. In the case of the uranium gun that requirement would be relatively easy to meet, since the alpha source and the beryllium could be separated with the bullet and the target core. But the implosion bomb offered no such convenient arrangement for separation and for mixing. Polonium and beryllium had to be intimately conjoined in Fat Man at the center of the plutonium core but inert as far as neutrons were concerned until the fraction of a microsecond when the imploding shock wave squeezed the plutonium to maximum density. Then the two materials needed instantaneously to mix.
Polonium, element 84 on the periodic table, was a strange metal. Marie and Pierre Curie had isolated it by hand from pitchblende residues (at backbreaking concentrations of a tenth of a milligram per ton of ore) in 1898 and named it in honor of Marie Curie’s native Poland. Physically and chemically it resembled bismuth, the next element down the periodic table, except that it was a softer metal and emitted five thousand times as much alpha radiation as an equivalent mass of radium, which caused the ionized, excited air around a pure sample to glow with an unearthly blue light.
Po210, the isotope of polonium that interested Los Alamos, decayed to lead 206 with the emission of an alpha particle and a half-life of 138.4 days. The range of Po210’s alphas was some 38 millimeters in air but only a few hundredths of a millimeter in solid metals; the alphas gave up their energies ionizing atoms along the way and finally came to a stop. That meant the polonium for an initiator could be safely confined within a sandwich of metal foils. Sandwiching the foils in turn might be concentric shells of light, silvery beryllium. The entire unit need be no larger than a hazelnut.
“I think I probably had the first idea [for an initiator design],” Bethe remembers, “and Fermi had a different idea, and I thought mine was better for once, and then I was the chairman of a committee of three to watch the development of the initiator.”2168 Segregating the Po210 from the beryllium was straightforward. Making sure the two elements mixed thoroughly at the right instant was not, and the primary difference between initiator designs—many were invented and tested during the winter of 1944–45—was their differing mixing mechanisms. A quantity of Po210 equivalent in alpha activity to 32 grams of radium, thoroughly mixed with beryllium, would produce some 95 million neutrons per second, but that would be no more than nine or ten neutrons in the brief ten-millionth of a second when they would be useful in an imploding Fat Man to start the chain reaction; therefore the mixing had to be certain and thorough. Initiator design has never been declassified, but irregularities machined into the beryllium outer surface that induced turbulence in the imploding shock wave probably did the job: the Fat Man initiator may have been dimpled like a golf ball.
To supply ten neutrons to initiate a chain reaction men labored for years. Bertrand Goldschmidt, a French chemist who had once been Marie Curie’s personal assistant and who came to the United States after the invasion of France to work with Glenn Seaborg at the Met Lab, extracted the first half-curie of initiator polonium from old radon capsules at a New York cancer hospital (polonium is a daughter product of radium decay). Quantity production required using scarce neutrons from the Oak Ridge air-cooled pile to transmute bismuth one step up the periodic table to Po. Charles A. Thomas, research director for the Monsanto Chemical Company, a consultant on chemistry and metallurgy, took responsibility for purifying the Po, for which purpose he borrowed the indoor tennis court on his mother-in-law’s large and securely isolated estate in Dayton, Ohio, and converted it to a laboratory.
Thomas shipped the Po on platinum foil in sealed containers, but another nasty characteristic of polonium caused shipping troubles: for reasons never satisfactorily explained by experiment, the metal migrates from place to place and can quickly contaminate large areas. “This isotope has been observed to migrate upstream against a current of air,” notes a postwar British report on polonium, “and to translocate under conditions where it would appear to be doing so of its own accord.”2169 Chemists at Los Alamos learned to look for it embedded in the walls of shipping containers when Thomas’ foils came up short.
Initiator studies proceeded in G Division at a test site established in Sandia Canyon, one mesa south of the Hill. The Initiator group drilled blind holes in large turbine ball bearings—screwballs, the experimenters called them—inserted test initiators and plugged the holes with bolts.2170 After imploding the screwballs they recovered the remains and examined them to see how well the Po and Be had mixed. Mixing, unfortunately, could not be a conclusive measure of effectiveness. Bethe’s committee selected the most promising design on May 1, 1945, but only a full-scale test culminating in a chain reaction could prove definitively that the design worked.
* * *
Progress toward a Japanese atomic bomb, never rapid, slowed to frustration and futility across the middle years of the Pacific war. After the Imperial Navy had bowed out of atomic energy research Yoshio Nishina had continued patriotically to pursue it even though he privately believed that Japan in challenging the United States had invited certain disaster.2171 On July 2, 1943, Nishina had met with his Army liaison, a Major General Nobuuji, to report that he had “great expectations” for success.2172 He noted that the Air Force had asked him to study uranium as a possible aircraft fuel, as an explosive and as a source of power, and he had recently received a request for assistance from another Army laboratory, which had contributed 2,000 yen to his expenses. Nobuuji promptly discouraged such consultations. “The main point,” Nishina agreed, “is to complete the project as rapidly as possible.” His calculations, he told Nobuuji, indicated that 10 kilograms of U235 of at least 50 percent purity should make a bomb, although cyclotron experiments would be necessary to determine “whether 10 kg. will be sufficient, or whether it will require 20 kg. or even 50 kg.” He wanted help finishing his 60-inch cyclotron:2173
The 250-ton, 1.5 meter accelerator is ready for operation except for certain components which are unavailable as they are being used in the construction of munitions. If this accelerator is completed we believe we can accomplish a great deal. At this moment the U.S. plans to construct an accelerator ten times as great but we are unsure as to whether they can accomplish this.
The previous March Nishina had discarded as impractical under wartime conditions in Japan all methods of isotope separation except gaseous thermal diffusion. Otto Frisch had tried gaseous thermal diffusion (differing from Philip Abelson’s liquid thermal diffusion) at Birmingham early in 1941 and proved it inadequate for separating uranium isotopes, but Nishina had no knowledge of that secret work. The Riken team had designed a thermal column much like the laboratory-scale column Abelson had built at the Naval Research Laboratory in Washington: of concentric 17-foot pipes, the inner pipe heated to 750°F—electrically heated in the Riken configuration—and the outer pipe cooled with water.
Nishina did not meet again with Nobuuji until seven months later, in February 1944, when he reported difficulty producing uranium hexafluoride. His team had managed to develop a method for generating elemental fluorine but had not yet been able to combine the gas with uranium using an old and inefficient process that Abelson in the United States had discarded before he began his thermal-diffusion studies. Nishima also had a problem with his diffusion column that Abelson would have appreciated: it leaked. “To achieve an airtight system,” Nishina told Nobuuji, “we used [sealing] wax and finally achieved our goal. Solder could not be used because of the corrosive properties of the fluorine.” He was “in the middle of developing this [hexafluoride-generating] process but can see the end in sight.” His 1.5-meter cyclotron was now in operation but only at low energy; his explanation for that compromise comments pointedly on the condition of the Japanese industrial economy by 1944:
We have been unab
le to obtain any superior, high-frequency-generating vacuum tubes . . . for the cyclotron. . . . As a result of this constraint, the low operating voltages limit the population of neutrons we can produce. . . . In order to liberate many high-energy neutrons, a high-voltage vacuum tube is required. But, unfortunately, they are difficult to acquire.
By summer Nishina’s group had manufactured some 170 grams of uranium hexafluoride—in the United States hex was now being produced by the ton—and in July attempted a first thermal separation.2174 Gauges at the top and bottom of the column, intended to measure a difference in pressure—showing that separation was taking place—indicated no difference at all. “Well, don’t worry,” Nishina told his team.2175 “Just keep on with it, just keep giving it more gas.”2176
He reconvened with Nobuuji on November 17, 1944, to report that “since February of this year there has not been a great deal of progress.” He was losing as much as half his hexafluoride to corrosion effects:
We thought the materials we had used to make this apparatus for working with the [hexafluoride] were made of impure metals. Therefore we next used the most highly-refined metals available for the system. However, they were still eaten away. It was therefore necessary to reduce the pressure of the system . . . to compensate for this erosion.
The cyclotron was operating at higher but not yet full power; Nishina was using it, he told Nobuuji, “to assay the concentrated, separated material.” Significantly missing from the November 17 conference report is any mention of measurable separation of U235 from U238. Nishina’s staff had understood for more than a year that he did not believe his country could build an atomic bomb in time to affect the outcome of the war.2177 Whether he continued research out of loyalty, or because he thought such knowledge would be valuable after the war, or to win support for his laboratory and deferment from military service for his young men, the bare record does not reveal. On the occasion of the November 17 conference he once again complained of the lack of sufficiently powerful vacuum tubes for his cyclotron and told Nobuuji, contrary to the evidence of experiment, that the Riken’s efforts at isotope separation were “now at a midpoint in their practical solution.” Nobuuji might have been more helpful if he had understood even the most basic facts of the work. An exchange between the two men late in the meeting indicates the military liaison was as innocent of nuclear physics as a stone: