The program to develop nuclear weapons was designed with two parallel paths, and only one had to succeed. Path 1 was to build a weapon using uranium purified to at least 80 percent fissile U-235. In path 1 were two parallel, independent sub-paths, designated Y-12 and K-25, to be imple­mented at Site X, or the Clinton Engineering Works at Oak Ridge. Y-12 was to separate U-235 from natural uranium using industrial-grade mag­netic mass spectrometers, or “calutrons,” developed by Ernest Lawrence at the University of California, Berkeley. In a calutron, uranium atoms are ionized in a vacuum chamber and accelerated into a negatively charged electrode. In the flight path is a powerful magnetetic field, turning the fly­ing ions through an angle of 180 degrees. The heavy U-238 ions have trou­ble making the turn, and the lighter U-235 ions are more likely to hit the electrode and be collected. The process was extremely slow, and managed to make enough U-235 concentrate in two years running 24 hours a day, seven days a week, to make exactly one bomb.

The priority of the bomb project is illustrated by Groves’s ability to commandeer materials. To build the alpha and beta calutrons, enormous magnet coils were needed. Sufficient copper of adequate purity was simply not available during the war. Copper supplies were so short that pennies were being minted from steel. Groves asked his scientists what material could be substituted for pure copper. They told him silver would do nicely. Groves arranged to have 14,700 tons (13,300 metric tons) of pure silver transferred to the Manhattan Project from the U. S. Treasury.

The second sub-path for U-235 concentration was K-25, which was an entirely different enrichment scheme. K-25 used gaseous diffusion, in which uranium hexafluoride gas is pressurized against a porous mem­brane. Gas molecules containing U-235 travel slightly faster than mole­cules containing U-238 and are more likely to diffuse and make it through to the other side of the membrane. By passing the gaseous uranium through hundreds of stages of diffusion, large concentrations of U-235 can be achieved. The K-25 plant was the largest building on earth. It cov­ered more than 2,000,000 square feet (609,600 square meters) of ground, cost $512 million dollars to build in 1944, and employed 12,000 full-time workers. For all its great potential as a uranium enrichment facility, the K-25 plant did not contribute any enriched uranium to the atomic bomb. By the time it was operating and producing enriched product, the war was over, but it continued operations until 1987 and made a great deal of enriched uranium for the nuclear power industry. After the end of the war, with the K-25 plant running at full speed, the Y-12 plant was shut

down and dismantled. The silver was given back to the Treasury Depart­ment, with expressed gratitude.

Path 2 of the Manhattan Project was to build a nuclear bomb using Pu-239, plutonium, as the fissile material. Using plutonium had one enor­mous advantage: It did not require enrichment. The disadvantage was that it could not be mined, as it was not known to exist in nature. It had to be produced, and the only way to produce it was by bombarding natural uranium with neutrons. The U-238 in natural uranium would activate into U-239 upon neutron capture, and this would beta decay into nep­tunium-239, which was unstable and would quickly beta decay into rela­tively stable Pu-239. In beta decay, a neutron transforms into a proton, and the decaying isotope retains its mass number while changing its atomic species.

Plutonium had been made at the Berkeley cyclotron in vanishingly small quantities, but for use in a bomb several kilograms would be nec­essary. The way to bombard U-238 with large neutron flux was to use a nuclear reactor, running at very high power. Any neutrons produced in the fission action that were not used to produce continuous fissions would have a good probability of being captured by U-238 atoms in the fuel, con­verting the otherwise useless U-238 to fissile Pu-239.

Enrico Fermi led a team at the new Argonne Laboratory to quickly design a reactor for the Hanford Works that could produce 250 megawatts of power. Each person in the United States uses an average of 1,400 watts of power, 24 hours a day, so the power output of this plutonium produc­tion reactor, named B-Reactor, could have produced enough power for 180,000 people. Power production, however, was not the goal of this reac­tor, and all the energy was exhausted into the Columbia River on the flat desert in central Washington State. The goal was simply to make neutrons and use them to convert U-238 into Pu-239.

B-Reactor for Site W was an upscale version of CP-1, still using the nat­ural uranium fuel and the highly efficient moderator made of pure graph­ite pressed into bricks. The uranium was formed into slugs, each about the size of a roll of quarters, sealed in aluminum casings. The fuel slugs were lined up in 1,500 aluminum tubes, running horizontally through a 1,200-ton (1,089-mt) graphite cylinder, 28 by 36 feet (8.5 by 11 m), lying on its side. River water was pumped through the aluminum tubing at 30,000 gallons (110,000 L) per minute to remove the energy from the reactor. The industrial complex supporting the plutonium production at Hanford, Washington, was about half the size of the state of Rhode Island, with 512

new buildings as well as temporary housing. After two years of construc­tion, involving 42,400 workers, on Tuesday evening, September 26, 1944, B-Reactor was ready to be powered up and start producing plutonium.

E. I. du Pont de Nemours and Company was responsible for the opera­tion and engineering of this enormous chemical plant. Enrico Fermi and a crowd of du Pont executives were on hand at the Hanford Works to watch the fuel being loaded and to observe the start-up of the new reactor. Everything checked out perfectly, and the operations staff withdrew the control rods in stages, just as Fermi had done in the first reactor experi­ment in Chicago. The pile went critical, with a self-sustaining chain reac­tion, at about midnight. Cautiously the power was increased to operating level, and by 2:00 a. m. the reactor was operating smoothly at high power, with cooling water flowing through the aluminum tubing.

All went well for about an hour, and then the operation took an unan­ticipated turn. For some reason, the operators had to begin stepping out control rods to maintain the critical condition, in which the number of neutrons lost to leakage, capture in U-238, or fission exactly equaled the number of neutrons produced in fission. Eventually, the operators ran out of control rods to pull, and B-Reactor died early Wednesday morning.

Fermi tried to remain calm and think through the failure, considering all possible reasons why the reactor had quit generating power. Groves, needless to say, was not happy with the unexpected behavior of B-Reactor and neither was du Pont. A theoretician from Princeton University, John A. Wheeler (1911-2008), stepped forward with an explanation. Using the time required for the effect to shut down the reactor as a clue, Wheeler predicted that iodine-135 was being produced as a fission product, with a radioactive half-life of 6.68 hours. There was nothing wrong with having iodine-135 in the reactor, but it decays into xenon-135, which is a vora­cious neutron absorber, 150 times more effective at neutron capture than the cadmium metal used in the control rods. The Xe-135 would decay with a half-life of 9.13 hours into something harmless, but meanwhile it was being made by the decay of I-135, and it was shutting the reactor down. Fermi and his team had not observed “xenon poisoning” in their low-power experiments because only at high power was enough I-135 pro­duced to lead indirectly to shutdown.

The solution to this problem of high-power shutdown was to build into the reactor enough excess ability to fission, or reactivity, to over­come this poisoning effect. Fortunately, Wheeler had insisted that du Pont build more fuel channels into the graphite pile than would be necessary to

sustain a reaction, adding millions of dollars to the cost of the project, just in case something unforeseen came about. The extra tubes would accom­modate another 504 lines of fuel slugs, and this was sufficient excess reac­tivity to overcome the xenon effect. Two additional graphite piles, the D-Reactor and the F-Reactor, were started up in December 1944 and Feb­ruary 1945, using the modified fuel loading of 2,004 tubes. By April 1945, plutonium was being chemically separated from spent fuel slugs from the three production reactors and shipped to Los Alamos every five days.

Nuclear physics lived at Site Y, or the Los Alamos Laboratory, on a mesa in northern New Mexico. The site was chosen by Oppenheimer and Groves as an out-of-the-way location, far from prying eyes and with plenty of room to experiment with explosives and radioactive materials. It was the location of a private boy’s academy named the Los Alamos Ranch School. On November 22, 1942, the Corps of Engineers submitted an appraisal of the site, which was found perfect, and the U. S. govern­ment bought it for $440,000. Construction of a large laboratory and living complex began immediately, and Oppenheimer crisscrossed the country recruiting nuclear scientists from every major university. Enrico Fermi, Hans Bethe, Edward Teller, Stanislaw Ulam, Seth Neddermeyer, George Kristiakowsky, and a host of other extremely capable physicists agreed to move to Los Alamos for the duration of the war. Leo Szilard, who was not acclimatized to remote, rustic conditions, balked. “Nobody could think straight in a place like that,” he complained. “Everybody who goes there will go crazy.” Isador I. Rabi (1898-1988), a very capable theorist from Hungary, also declined the invitation, thinking that he was more useful at the radar laboratory at MIT. “I’m very serious about this war,” he said. “We could lose it with insufficient radar.”

Oppenheimer needed scientists, support personnel, and equipment for neutron research. He managed to wrangle the cyclotron from Harvard University and two Van de Graaff linear particle accelerators from the University of Wisconsin to be used for artificially generating neutrons. In April 1943, the laboratory opened, with a school set up to indoctrinate the incoming scientists and reveal to them the purpose of the new facility. The lecturer was Robert Serber (1909-97), and the subject of his five-day course was the production of a practical military weapon. In the secret parlance of the time, it was referred to as the “gadget.”

The gadget was to bring together two subcritical pieces of fissile mate­rial quickly to form a hypercritical assembly. A hypercritical block of material would contain enough fissile U-235 or Pu-239 to form more than

an assembly that was critical, in which the neutron production and loss are balanced. It would be a runaway chain reaction in enough fissile iso­tope to be critical three or four times, and the uncontrolled nuclear fission reaction would proceed to increase at an explosive rate.

A nuclear power reactor and an atomic bomb both use the chain reac­tion of neutrons causing fission, but the goals and the methods are differ­ent. In a power reactor, neutrons are slowed down to thermal speed, or the speed of normal molecules and atoms, and are captured by uranium or plutonium nuclei, causing fission. It takes time to slow the neutrons down, and all the neutrons thrown out in the fission process, about 240 per 100 fissions, are not thrown out of the fission debris at once. Some take several seconds to pop out, and these factors slow down the response of a nuclear reactor to changes in criticality. This is an attractive feature, because it renders a nuclear reactor easy to control. The response time at the controls is slow and smooth.

There are therefore two modes of criticality. There is delayed critical­ity, in which all the neutrons, including the ones delayed from fission, are used to balance with the lost neutrons. There is also prompt critical­ity, in which only the instantly available neutrons are counted, and this requires a larger mass of fissile material, as there are fewer neutrons avail­able promptly to contribute to fission. Prompt fission is essentially instan­taneous, with no built-in delay to moderate the controls.

There are also two modes of fission. Thermal fission is used in power reactors, as it makes optimum use of available neutrons and it involves a slowing-down medium, or moderator, inserted between fissile parts, and this moderator can be used conveniently as a coolant. Fast fission uses only high-speed neurons, running at about 1 MeV, as they are thrown out in the fission process and before they hit anything to slow them down.

Power reactors use delayed, thermal fission. Atomic bombs use prompt, fast fission. A power reactor makes a slow climb from full shutdown mode to a critical, self-sustaining power level, and any change in power level is made slowly. It took two hours for B-Reactor to move from zero to full operating power, which gave the auxiliary equipment such as coolant pumps plenty of time to adjust to the high-power conditions. An atomic bomb, or gadget, goes from zero power to a massive power spike in less than a millisecond, or one one-thousandth of a second. The criticality is promptly achieved on the high end of the neutron energy spectrum so that no time is wasted slowing down or diffusing through intervening moderator, and the reaction is completely out of control, with no effort to

balance the losses and productions of neutrons. More neutrons are pro­duced and wasted than could ever be used for fission, and the mechanism is destroyed in a momentary burst of extreme energy.