An Engineering History of the Manhattan Project

The Manhattan Project, the US program to build an atomic bomb during WWII, is one of the most famous and widely known major government projects: a survey in 1999 ranked the dropping of the atomic bomb as the top news story of the 20th century. Virtually everyone knows that the project built the bombs that were dropped on Hiroshima and Nagasaki. And most of us probably know that the bomb was built by some of the world’s best physicists, working under Robert Oppenheimer at Los Alamos in New Mexico. But the Manhattan Project was far more than just a science project: building the bombs required an enormous industrial effort of unprecedented scale and complexity. Enormous factory complexes were built using hundreds of millions of dollars worth of never-before-constructed equipment. Scores of new machines, analytical techniques, and methods of working with completely novel substances had to be invented. Materials which had never been produced at all, or only produced in tiny amounts, suddenly had to be manufactured in vast quantities.

This massive effort was required in part because of the enormous difficulty in producing fissile material, and in part because of the enormous uncertainty facing the project: it wasn’t known what the best method for manufacturing the fissile material needed for the bomb would be, what the design of the bomb should be, or whether a workable bomb could even be built. Developing the bomb required resolving this uncertainty, and the project needed to rapidly push forward knowledge and capabilities in many fields: not merely in the realm of nuclear chain reactions and atomic physics, but also in areas like precision explosives, metallurgy, welding, chemical separation, and electronics.

Because of the exigencies of war, this work needed to be done extremely rapidly. There wasn’t time to investigate promising approaches sequentially, or wait for more information before picking a particular course. Thus, multiple possible routes to the bomb — different fuels (and different fuel production techniques), different bomb designs, different components like triggers and tampers — were pursued simultaneously. Major commitments, like factories that cost hundreds of millions of dollars, were made before it was known whether they would even be useful. Design work began on the bombs when the nuclear fuel they would use hadn’t been produced in more than microscopic amounts.

Normally when trying to create a new technology, funding constraints and the need for economic returns determine how much time and effort can be spent on development. Efforts to create some new technology will often be small-scale until the surrounding conditions are right — until knowledge has caught up, or the necessary supporting infrastructure exists, or the input materials are cheap enough — and risk can be minimized. But with the Manhattan Project, these constraints didn’t exist. Funding was virtually unlimited in service of ending the war sooner, and the biggest perceived risk was that Germany would beat the US to the bomb. As a result, an extremely robust development effort could be justified, which thoroughly explored virtually every promising path to an atomic weapon (no matter how expensive or uncertain).

Beginnings of the project

The Manhattan Project began in June of 1942, when Colonel James Marshall of the Army Corps of Engineers was directed to create a new engineering district to lead the army’s efforts to develop an atomic weapon. Shortly after, Colonel Leslie Groves (who would soon be promoted to brigadier general) was selected to lead the project. At the time, the official name of the project was “Laboratory for the Development of Substitute Materials” (DSM for short), but Groves felt that this name would attract curiosity, and so a new name was selected based on the location of Marshall’s New York office: the Manhattan Engineer District.

By the time the Manhattan Project officially formed, the US was already at work developing an atomic bomb. Following the discovery of fission in 1938 by Otto Hahn and Fritz Strassmann, physicists began to speculate that a nuclear chain reaction might be possible, and that such a reaction could be used to build a bomb of unprecedented magnitude. In August the following year, Albert Einstein and physicist Leo Szilard sent a letter to president Roosevelt, warning him that a nuclear chain reaction might be used to build an extremely powerful bomb, and that the US should research atomic energy. Two months later, Roosevelt ordered the creation of an advisory committee on uranium, and by early 1940 US researchers (most notably Enrico Fermi) were working to create a sustained nuclear chain reaction.

In July of 1941, a report from the British MAUD Committee concluded that it was likely feasible to build an atomic bomb. It reached the US, and in October Roosevelt authorized expediting atomic bomb work. Bomb efforts accelerated following Japan’s attack on Pearl Harbor in December of 1941, and in February of 1942 the Metallurgical Laboratory was formed at the University of Chicago to study nuclear chain reactions and the chemistry of newly-created element plutonium. There, a team working under Enrico Fermi continued their work to create nuclear chain reactions, ultimately resulting in Chicago Pile-1, the world’s first self-sustaining nuclear reaction, in December of that year.

Early test pile at University of Chicago, 1942, via Wikipedia.

The path to the bomb

When the Manhattan Engineering District was formed in 1942, there was still a great deal of uncertainty surrounding the construction of an atomic bomb. Based on what was known at the time, it was believed that a bomb was probably feasible, and that due to the risks of Germany developing one it should be pursued. But it was far from clear what the surest path to success was.

The first major challenge came in producing sufficient fissile material to build a bomb. Fissile material splits and releases neutrons when struck by slow, “thermal” neutrons, making a nuclear chain reaction possible. At the time there were two major candidate materials: a rare isotope of uranium known as uranium-235 (U235), and plutonium, an element first synthesized by Glenn Seaborg in late 1940.¹

Using either would be very challenging. U235 makes up less than 1% of naturally-occurring uranium, and using it as a bomb material required separating it from the far more common U238. But the two isotopes were only distinguished by a tiny difference in their weights (U235 weighs about 1.3% less than U238). Some sort of filtering mechanism was needed that could act on this difference and create concentrations of U235 high enough for a bomb (a process known as enrichment).

There were several potential methods considered for separating U235:

• In the electromagnetic method, a beam of charged uranium tetrachloride particles would be fired through a magnetic field, which would alter their paths. Because of the difference in weight, the paths of U235 and U238 would be slightly different, and the U235 could be gathered at an appropriately placed collector.

• In the gaseous diffusion method, gaseous uranium hexafluoride would diffuse through a barrier with microscopic pores. The lighter U235 would diffuse more readily through the barrier due to Graham’s Law.

• In the liquid thermal diffusion method, a thermal gradient created in a uranium solution would cause a slight migration of heavier U238 to the cold side and the lighter U235 to the warm side.

• In the centrifuge method, uranium spun in a high-speed centrifuge would cause the heavier U238 to concentrate on the outer edge.

Of these methods, the electromagnetic method was the most developed, thanks to the efforts of Ernest Lawrence at the University of California, but it wasn’t clear if any of them (either alone or in combination) could actually produce U235 at the scale and speed needed. And no matter the method selected, an enormous industrial facility would be required: in 1939 Danish physicist Niels Bohr insisted that an atomic bomb could never be built “unless you turned the US into one huge factory.”

The main alternative to U235 was plutonium. As with U235, the main challenge would be collecting enough of it to build a bomb. Plutonium only occurs in trace amounts in nature (1 atom per 100 billion in uranium ore): collecting enough to build a bomb requires synthesizing it. Plutonium could be produced in a nuclear reactor (then called a “pile,” as it was essentially chunks of uranium piled high enough to create a self-sustaining nuclear reaction), but only in microscopic amounts. Producing a pound of plutonium required around 4000 pounds of uranium fuel, and producing enough for a bomb would require an enormous industrial facility, as with U235. Exacerbating this difficulty was the fact that while U235 and U238 could be handled comparatively easily, plutonium and other nuclear reactor byproducts were highly radioactive, requiring special handling.

Once enough fissile material had been collected, it then needed to be turned into a bomb. When enough fissile material is brought together in a small enough volume (the so-called critical mass), it can start a nuclear chain reaction, releasing enormous amounts of energy as more and more fissions were triggered. Because a chain reaction in critical mass could be started by spontaneous fission (fissile elements randomly splitting and releasing neutrons) or by cosmic rays, a bomb would have to start with a sub-critical mass of fissile material, turning it into a critical mass at detonation.

The most straightforward way to do this, it was thought, was to use a gun that would fire a sub-critical “bullet” into another sub-critical “target,” the combination of which would exceed the critical mass. But there were other mechanisms considered, including using an explosion to compress a sphere of fissile material (the so-called implosion method), as well as “autocatalytic” mechanisms in which “the chain reaction itself, as it proceeded, increased the neutron number for a time.”

In mid-1942, a gun-based plutonium bomb was generally considered most promising, but due to the lack of information and the great urgency, many promising paths were investigated simultaneously. Early on in the project, when resolving a debate about pile cooling systems, Leslie Groves stated that “The War Department considers this project important. There is no objection to a wrong decision with quick results. If there is a choice between two methods, one of which is good and the other looks promising, then build both.”

Perhaps no phrase better summarizes the philosophy of Manhattan Project than “build both.” It was ultimately decided to pursue both U235 and plutonium-based bombs. To produce the necessary U235, a production facility would be built near Knoxville, Tennessee, employing both electromagnetic separation and gaseous diffusion (and, eventually, liquid thermal diffusion). This plant, initially referred to as the Clinton Engineer Works, would later be named Oak Ridge. To produce the plutonium, another facility, the Hanford Engineer Works, would be built in southeast Washington. And while these plants were being built and producing fissile material, the design of the bombs themselves would be done at Los Alamos, New Mexico.

Oak Ridge and Uranium 235

The acquisition of the site for Oak Ridge was authorized in September of 1942, and construction of the electromagnetic separation plant began a few months later in February of 1943 by the firm Stone and Webster. To produce U235, the plant would use modified versions of Ernest Lawrence’s Nobel prize-winning cyclotron particle accelerators. Lawrence had been working on the devices, which he referred to as “calutrons” (after the University of California) since the spring of 1942, and while he confident that calutrons could be used for large-scale production of U235, “he stood almost alone in his optimism”:

The method called for a large number of extremely complicated, and as yet undesigned and undeveloped devices involving high vacuums, high voltages and intense magnetic fields. As a large-scale method of separating Uranium-235, it seemed almost impossible. Dr. George T. Felbeck, who was in charge of the gaseous diffusion process for Union Carbide, once said it was like trying to find needles in a haystack while wearing boxing gloves. - Now It Can Be Told

(According to Richard Rhodes, the total volume of high-vacuum required by the calutrons would eventually exceed the amount of vacuum produced everywhere else on earth at the time.)

The calutrons, arranged around a series of several “racetracks” with several dozen collection tanks attached, were divided into two stages. Partly enriched material from “alpha” racetracks would be fed into “beta” racetracks to be further enriched, eventually (it was hoped) producing 90% enriched U235. Each calutron only produced a tiny amount of U235 — Lawrence estimated that 2000 calutrons could produce 100 grams of enriched U235 per day — so a huge number of them were needed: the alpha and beta calutron buildings eventually occupied an area greater than 20 football fields, and the entire electromagnetic separation facility grew to 268 buildings, requiring 20,000 workers to build:

…The calutron structures of steel and brick and tile, chemistry laboratories, a distilled water plant, sewage treatment plants, pump houses, a shop, a service station, warehouses, cafeterias, gatehouses, change houses and locker rooms, a paymaster’s office, a foundry, a generator building, eight electric substations, nineteen water-cooling towers - for an output measured in the best of times in grams per day. - The Making of the Atomic Bomb

Building this enormous facility was rife with challenges. Copper, traditionally used for winding electromagnets, was in short supply due to the war, and so was substituted with silver (also a good conductor) borrowed from the US Treasury. Altogether 13,540 tons, worth $300 million ($6 billion in 2025 dollars) was borrowed, 99.964% of which was eventually returned. Because construction started so early, constant changes to already manufactured and installed equipment were required, a process that continued “long after the first major units of the plant began production operations.” Little of what was required to build the plant was off the shelf or standard:

Discouragingly few items were commercially available. Tanks, magnets, vacuum pumps, cubicles, and most of the chemical equipment, for example, were either completely new in design or so much larger or so much greater in capacity that nothing of the kind previously had been manufactured. Many less obvious items also carried performance specifications that far exceeded anything ever attempted on a commercial scale. For instance, the calutrons required electrical cable that could carry a high-voltage load continuously. The only commercial product that came near meeting this specification was the heaviest X-ray cable, and it was designed to operate intermittently. Even when the commercial equipment could be used, suppliers often had to add to their productive capacity or build entire new plants to furnish the items required in the enormous quantities they were needed. Thus, in the first equipping of the racetracks some eighty-five thousand vacuum tubes were required. In the case of one type of tube, procurement officials ordered in advance the entire national output for 1943 as well as that from a plant still under construction. In the early months of plant operation, when tubes burned out faster than predicted, some feared the racetracks might prove inoperable simply through inability to maintain the tube supply.

New methods had to be developed for machining and shaping the graphite in those parts of the calutron subject to intense heat. No standard material would endure the high potentials, mechanical strain, and temperature changes to which bushings in the high-voltage elements in the sources were continuously subjected. After months of investigation, Stone and Webster found an insulator made of zirconium oxide, a new and still very expensive substance. Similarly, use of large quantities of liquid nitrogen to condense moisture created a demand for a substance hitherto not produced on a commercial scale anywhere in the country. - Manhattan: The Army and the Bomb

These difficulties didn’t stop after construction was completed. When the first calutrons were turned on for testing in November of 1943, the extremely powerful magnets caused the equipment to “walk” by several inches: this was eventually resolved by tying them down with heavy steel straps. Testing also showed intermittent electrical shorts and unexpectedly high variation in the strength of the magnetic fields, problems that was eventually traced to dirt and rust within the electromagnets bridging the gap between the closely spaced silver windings. To fix this required rebuilding and redesigning the magnets, delaying production by a month.

Alpha racetrack at Oak Ridge.

As production came online in early 1944, the electromagnetic separation plant continued to deal with numerous problems: equipment and mechanical failures, electrical short circuits, vacuum leaks and various breakdowns. The operation “skirted the edge of chaos for months.” As late as August of 1944, the electromagnetic plant had only produced a small fraction of the expected U235, and it was unclear if enough would be produced to build a wartime bomb. But eventually these problems were ironed out, and by September of 1945 the alpha and beta calutrons had produced 88 kilograms of 84.5% enriched U235.

In addition to the electromagnetic separation plant, Oak Ridge was also the site of the gaseous diffusion plant, which began construction in May 1943. As with the electromagnetic process, the gaseous diffusion process only produced U235 in tiny amounts, and an enormous facility was needed to manufacture it in sufficient quantities to build a bomb. Gaseous diffusion worked on the principle that lighter U235 would be more likely to diffuse through a porous barrier than heavier U238, but the difference in diffusion rates was miniscule. A single gaseous diffusion step would only increase the fraction of U235 by a factor of 1.0043: to produce 90% enriched U235, the initial design of the plant called for a series of 4,600 gaseous diffusion stages. Upon completion, the gaseous diffusion plant was one of the largest buildings in the world.

K-25 gaseous diffusion plant.

As with the electromagnetic plant, construction began before the design of the diffusion process had been finished, and it was unclear if the plant would work at scale. The greatest challenge was finding an appropriate barrier for the gas to diffuse through. The barrier needed to have numerous microscopic pores, be robust enough to withstand exposure to extremely corrosive uranium hexafluoride gas, and be mass-producible. Researchers had experimented with “a great many materials” between 1941 and 1942, but none were suitable. The only common material sufficiently corrosion-resistant was nickel, but no form of nickel seemed to do the trick. An electro-deposited nickel mesh, invented by Edward Norris (a “self-educated Anglo-American interior decorator”) and Edward Adler appeared most promising, but it was so brittle that manufacturing it was incredibly difficult. A modified version of the Norris-Adler barrier, produced by a team from Kellex, Bell Labs, and Bakelite, appeared to work even better, though it too had problems. Work proceeded to further develop both barriers simultaneously, but progress was slow and by the end of 1943 “morale had plummeted”. It wasn’t until early 1944 that satisfactory barriers were being produced in sufficient quantities.

While the diffusion barrier was the biggest challenge with the gaseous diffusion process, it wasn’t the only one. The plant required a level of vacuum-tightness that had previously only been achieved in labs, demanding the development of novel methods of pipe welding and leak detection. Upon completion, it took 406 workers eight months to test the plant for leaks. More than 130,000 measuring instruments, many of them novel, were installed in the plant. It was likely the greatest number of instruments ever used in any plant in the world till that date, and they required months of testing and calibration. The plant had to be incredibly reliable, as “even slight variations in such factors as temperature and pressure could produce adverse effects.” Initially, it was believed that a slight power interruption could bring the diffusion plant offline for months, so a dedicated power plant was built specifically at Oak Ridge for the process. And because any contaminants could prove disastrous, the cleanliness standards for the plant approached surgical:

…Construction workers had to cleanse all pipes, valves, pumps, converters, and other items of equipment thoroughly before installation. Workmen in a special unit performed this vast operation in the large conditioning building, using equipment for solvent degreasing, alkaline cleaning, acid pickling, scratch brushing, surface passivation, and a variety of other procedures. When they finished, they sealed all openings to interior surfaces and kept them sealed until installation teams put the equipment into place.

To make certain no dust or other foreign matter polluted the system during installation, J. A. Jones instituted a rigid schedule of surgical cleanliness in installation areas. Isolating these areas with temporary partitions, the workers installed pressure ventilation, using filtered air. Then they cleaned the areas thoroughly, and inspectors carefully checked all personnel and material that entered them. Maintenance crews with mops and vacuum cleaners continued to remove any foreign substances that seeped in. When trucks had to enter, workers hosed them down at the entrances.

Workers wore special clothes and lintless gloves. Because certain work on equipment to be used in plant installations could not be done in the dirt-free areas, such as welding pipes and other small jobs, J.A. Jones installed special inflatable canvas balloons and the work was done inside them. The cleanliness control measures required many additional guards, inspectors, and supervisors… - Manhattan: The Army and the Bomb

This level of cleanliness extended to the design of the equipment itself:

[Uranium hexafluoride] attacked organic materials ferociously: not a speck of grease could be allowed to ooze into the gas stream anywhere along the miles and miles of pipes and pumps and barriers. Pump seals therefore had to be devised that were both gastight and greaseless, a puzzle no one had ever solved before that required the development of new kinds of plastics. (The seal material that eventually served at Oak Ridge came into its own after the war under the brand name Teflon.) A single pinhole leak anywhere in the miles of pipes would confound the entire system; Alfred O. Nier developed portable mass spectrometers to serve as subtle leak detectors. Since pipes of solid nickel would exhaust the entire U.S. production of that valuable resource, Groves found a company willing to nickel-plate all the pipe interiors, a difficult new process accomplished by filling the pipes themselves with plating solution and rotating them as the plating current did its work. - The Making of the Atomic Bomb

Because of these difficulties, the gaseous diffusion plant didn’t begin operating until February of 1945. While the plant was originally planned to produce uranium to the roughly 90% U235 enrichment needed to build a bomb, it was discovered that beyond 36.6% enrichment, different types of barrier and different types of pumps being designed would be required. In the latter half of 1943 the plant was thus redesigned to produce 36.6% enriched uranium (using 2892 diffusion stages) that would then be fed into the electromagnetic process. By the end of the war, the gaseous diffusion plant had “contributed substantially to the manufacture of the fissionable material used in the fabrication of atomic weapons”, and would become the primary method of producing enriched uranium in the early post-war years.

In addition to the electromagnetic and gaseous diffusion separation processes, a plant to separate U235 by liquid thermal diffusion was also built. Thermal diffusion had been considered by the Manhattan Project early on, but it appeared insufficiently promising and there were no initial plans to build a thermal diffusion plant. However, work on the process continued by the Navy as a method of producing fissile material for nuclear reactors. By late 1942, it appeared much more promising as a feasible separation method, and Leslie Groves recommended that it continue to be developed by the Navy. Eventually, in June of 1944 it was decided to build a thermal diffusion plant at Oak Ridge to produce partially-enriched uranium as an input to the electromagnetic separation process, as doing so would speed up overall U235 production. The plant came online in late 1944.

Like the other separation plants, there were struggles getting the plant built and operational. Early on there were numerous steam leaks and other equipment failures, and the “results scarcely seemed to justify the risks.” But eventually the plant “served its wartime purpose,” providing enough slightly enriched U235 to the electromagnetic separation plant and (later) the gaseous diffusion plant to build a uranium bomb by July 1945. After the war, however, it was found that the thermal diffusion plant was less economical than gaseous diffusion in producing enriched uranium, and it was shut down in September 1945, less than a year after starting operation.

Hanford and plutonium

Hanford F reactor, via Oregon Department of Energy.

As with U235, when plutonium was being considered as a possible fissile material for an atomic bomb there was a great deal of uncertainty around its large-scale production. When plans began to be formulated for plutonium manufacture in late 1942, the element had only been produced in microscopic amounts in cyclotrons — as late as December 1943, only two milligrams of plutonium had been manufactured. Producing the pounds of plutonium needed for a bomb would require a self-sustaining nuclear chain reaction in a nuclear reactor, which would create plutonium as a fission byproduct. The first such chain reaction was created in Chicago Pile-1 in December 1942, shortly after Du Pont had been (reluctantly) brought on as the contractor to build and operate a plutonium production plant.

As with uranium separation, there were a variety of potential ways to build a plutonium-producing nuclear reactor. Every design considered used uranium as a fuel, but there were a variety of options for cooling (water, helium, diphenyl, bismuth) and for moderators to slow down the neutrons (heavy water, graphite).² The initial reactor design used helium-cooling with a graphite moderator: helium wouldn’t absorb neutrons, and was an inert gas that wouldn’t corrode any of the reactor materials. But it was later shown that the neutron multiplication factor in a reactor, k, was high enough that coolants which absorbed more neutrons (such as water) could be made to work. Because a water-cooled reactor appeared far simpler to build, the design was changed to use water cooling. Du Pont, worried that pursuing a single reactor design was too risky, also continued to develop other designs, chiefly one moderated and cooled by heavy water, and several heavy water plants were built around the country for these purposes.

Plutonium production began with a smaller-scale, air-cooled test reactor, named X-10, built at Oak Ridge between February 1943 and January 1944. Full-scale production would take place at a remote facility built in southeastern Washington known as the Hanford Engineer Works. The first plutonium production reactor, B Reactor, began construction at Hanford in August of 1943, and first achieved criticality in September of the following year.

Like with uranium, plutonium production demanded a massive industrial facility, and Hanford was the biggest facility that Du Pont had ever constructed. An oral history of the Hanford site notes that “the building effort at Hanford from 1943 to 1945 can only be measured in superlatives”:

Consider the following: Project building crews used 1800 vehicles, including sedans, pick-up trucks, jeeps, and ambulances; 900 buses that had a total seating capacity of over 30,000; 1900 dump trucks and flat bed trucks; 240 tractor trailers; 44 railway locomotives and 460 railway cars; 5 locomotive cranes and 4 stiff leg derricks. The various construction teams built 386 miles of highways, 158 miles of track, poured 780,000 cubic yards of concrete, and erected housing for 5,000 women and 24,000 men. Excavation crews moved 25 million cubic yards of earth in the process. The overall cost was $350 million. - Working on the Bomb

The three production reactors built at Hanford (B, D, and F Reactors) were far more powerful than anything that had come before. Fermi’s Chicago Pile-1 produced a maximum of 0.2 kilowatts of power. The X-10 reactor at Oak Ridge produced just 500 kilowatts when it was first turned on (though output would later be raised to 4,000 kilowatts). The Hanford reactors were designed to produce 250,000 kilowatts.

But the largest structures at Hanford were the separation facilities to extract plutonium out of the soup of radioactive waste products the reactors generated: these were three 800 foot long, 8-story tall buildings that resembled “an ancient mausoleum”; they were so large they were referred to as “Queen Marys.” Within these enormous structures, plutonium was extracted from radioactive “slugs” of spent uranium fuel from the reactors:

Irradiated slugs ejected from a production pile would be stored in pools of water 16.5 feet deep to remain until the most intense and therefore short-lived of their fission-product radioactivities decayed away, the water glowing blue around them with Cerenkov radiation, a sort of charged-particle sonic boom. The slugs would then move in shielded casks on special railroad cars to one of the Queen Marys, where they would first be dissolved in hot nitric acid. A standard equipment group occupied two cells: a centrifuge, a catch tank, a precipitator and a solution tank, all made of specially fabricated corrosion-resistant stainless steel. The liquid solution that the slugs had become would move through these units by steam-jet syphoning, a low-maintenance substitute for pumps. There were three necessary steps to the separation process: solution, precipitation and centrifugal removal of the precipitate. These would repeat from equipment group to equipment group down the canyon of the separation building. The end products would be radioactive wastes, stored on site in underground tanks, and small quantities of highly purified plutonium nitrate. - The Making of the Atomic Bomb

As at Oak Ridge, there were numerous challenges in building a mammoth industrial facility employing completely novel production processes. The reactors “presented construction problems never encountered before, even by Du Pont's highly competent field forces.” Graphite bars of exceptional purity had to be fabricated and then machined to remove any sort of surface imperfections. Supplying cooling water for the reactors required “installation of a complex system of river pumps; purification, aeration, and distillation units; and retention basins for holding radioactive water until natural decay permitted its return to the Columbia.” New machines for fabricating the uranium fuel rods had to be designed and built, and a method for shielding the unprecedentedly large nuclear reactors had to be developed:

Ten months of work went into this before we could even begin to build it, with three more months before the first unit was completed…In the course of this, a special high-density pressed-wood sheet was developed in collaboration with an outside supplier. Then special sharp tools and operating techniques were required to cut various shapes from the standard manufactured widths…At the same time very detailed specifications for assembly, prescribing the closest of tolerances, were written. Some sixty manufacturers were invited to bid and refused, presumably because of the complexity of construction and the close tolerances required…but after methods were developed and prototypes were fabricated at du Pont’s shops in Wilmington eventually satisfactory suppliers were found. - Now It Can Be Told

Even seemingly simple items often proved enormously complex. Finding a way to prevent the uranium fuel (which was packaged into aluminum-clad slugs) from corroding took an enormous amount of effort:

Two years of trial-and-error effort had not produced canning technology adequate to seal the uranium slugs, which quickly oxidized upon exposure to air or water, away from corrosion. Only in August had the crucial step been devised, by a young research chemist who had followed the problem from Du Pont in Wilmington to Chicago and then to Hanford: putting aside elaborate dips and baths he tried soaking the bare slugs in molten solder, lowering the aluminum cans into the solder with tongs and canning the slugs submerged. The melting point of the aluminum was not much higher than the melting point of the solder, but with careful temperature control the canning technique worked. - The Making of the Atomic Bomb

Similarly, Leslie Groves notes that “seven months of persistent effort” were required to produce simple aluminum tubes that met the required specifications.

After enough plutonium had collected in the canned uranium slugs, it needed to be separated. The chemical process for this was comparatively straightforward to develop compared to the novel methods of mass-based separation developed for U235 (after experimenting with several possibilities, a method based on using bismuth phosphate was employed), but actually implementing it was a challenge. The materials to be processed were radioactive enough that the entire facility needed to be operable and maintainable remotely:

…Periscopes and other special mechanisms were incorporated into the plant design; all operations could thus be carried out in complete safety from behind the heavy concrete walls. The need for shielding and the possibility of having to replace parts by indirect means required unusually close tolerances, both in fabrication and in installation. This was true even for such items as the special railroad cars that moved the irradiated uranium between the piles and the separation plants. The tracks over which these cars moved were built with extreme care so as to minimize the chances of an accident. Under no circumstances could we plan on human beings directly repairing highly radioactive equipment. - Now It Can Be Told

Among the technologies developed to make remote operation possible were pipe flanges that could be connected by a remotely operated wrench, and the world’s first use of closed circuit TV.

The biggest crisis at Hanford came shortly after the first production reactor came online: a few hours after beginning operation the reaction began to slow, and within a few days it had shut itself down completely. Investigation revealed that this was being caused by a fission byproduct, an isotope of xenon known as Xenon 135 that had a massively greater probability of absorbing neutrons (known as a “neutron cross section”) than any previously discovered material. (Its cross section was 70 times larger than the previous largest measured cross section). This “xenon poisoning” hadn’t been noticed in earlier reactors because they hadn’t been run for long enough at a high enough power output.

The problem was ultimately resolved thanks to the conservative reactor design of the Du Pont engineers. The original Hanford reactor, designed chiefly by physicist Eugene Wigner, consisted of 1,500 “channels” for uranium fuel arranged in a cylindrical shape. Du Pont had squared this cylinder, adding more channels to the edges and bringing the total to 2004. When fuel was loaded into these extra channels, it was sufficient to overcome the xenon poisoning effect.

The xenon poisoning problem was overcome in December 1944, and by March 1945 the Hanford site had achieved full-scale plutonium production of around a pound and a half of plutonium per day.

Los Alamos

As design and construction of the huge facilities at Hanford and Oak Ridge began, work also proceeded on designing the bomb itself. This would chiefly be done at Los Alamos, New Mexico, the third of the atomic cities built for the Manhattan Project. Scientists led by Robert Oppenheimer began to arrive at Los Alamos in March of 1943 (though most of the facility was still under construction).

When work began at Los Alamos, the most promising method for building a bomb appeared to be the gun method: firing a sub-critical uranium or plutonium bullet into another sub-critical mass. But there were a variety of ways this might be done. Different gun arrangements, ranging from somewhat conventional gun mechanisms of various sizes and shapes to more exotic layouts like double-guns, rockets, and spherical guns were considered. As with the rest of the program, every design choice was mired in uncertainty. Since plutonium and U235 hadn’t yet been produced in large quantities, it wasn’t known exactly how much fissile material would be needed to create a critical mass, and thus how big the bullet and target needed to be. The neutron-reflecting properties of steel gun barrels (which would affect critical mass requirements) hadn’t yet been measured.

More generally, while the designers were reasonably confident that a gun-type bomb would work, because various nuclear properties and constants had at best been measured very imprecisely (if they’d been measured at all), they couldn’t be sure. A history of Los Alamos notes that in 1942, “the main obstacle to a theoretical understanding of the fission bomb was the uncertainty surrounding existing experimental data, in part the result of inadequate instrumentation and a lack of experience in the new field.” Things were so uncertain that it was not even 100% clear at the beginning of the program that plutonium would produce neutrons. A great deal of effort at Los Alamos was thus devoted to more accurate measurements and better understanding of nuclear physics: the neutron cross sections of various materials, the number of neutrons produced per fission, rates of spontaneous fission, and so on. These measurements were “constantly in flux” for much of the project, and the Los Alamos scientists were “plagued by worry about some unpredicted or overlooked mechanism of nuclear physics which might render our program unsound.”

Thus in 1943, while most design efforts were directed to developing a gun-type assembly for both uranium and plutonium, there were also parallel efforts on other types of bomb. Chief among these was the implosion method, which would use an explosion to compress a sphere of material enough to create a critical mass, though there were also investigations into the autocatalytic methods. Work also proceeded on what was referred to as the Super (better known today as the hydrogen bomb): using a fission explosion to trigger an even more destructive fusion explosion.

Beyond the mechanism for creating the critical mass, there were many other aspects of the bomb that needed to be figured out. To minimize the amount of fissile material needed to create a critical mass, the core needed to be surrounded by some type of material that would reflect neutrons back into the core and prevent them from escaping, but what material would work best and how it should be arranged wasn’t yet known. To ensure that the bomb detonated at the right time, it would also need some type of initiator: a mechanism that, when triggered, would create a sudden burst of neutrons to start the nuclear chain reaction.

More generally, everything about the atomic bomb was new, and almost nothing about its various aspects was known or could be assumed. Every step involved from taking the fissile material from Oak Ridge and Hanford and turning it into a bomb had to be worked out for the first time.

Fissile material would arrive from Hanford and Oak Ridge not as pure plutonium or uranium but as compounds — plutonium nitrate and uranium tetrafluoride, respectively — and methods for turning these into metallic plutonium and uranium needed to be created. Because impurities might affect the functioning of the bomb, purification methods also needed to be developed. It was initially believed that the plutonium in particular would need to be exceptionally pure, with no more than one part per hundred billion of light elements. Preventing contamination required, among other things, reagents that were “unbelievably purified,” electronic air cleaners, and an extensive lab-cleaning procedure performed by a dedicated service team. Creating effective plutonium purification methods took roughly a year, in part because initially only microscopic quantities of plutonium were available for experiments. Because purification required the ability to detect extremely small quantities of impurities, novel methods of “sub-micro” chemical analysis had to be developed.

Once metallic plutonium and uranium had been produced, methods for shaping the material — casting, rolling, pressing — also needed to be created. The need for purity, combined with the extreme reactivity of molten uranium and plutonium, meant that new crucible materials were needed: MIT spent an enormous amount of “time, effort, and expense” to develop a cerium sulfide crucible that could withstand plutonium's extreme reactivity and high expected melting point without introducing impurities.

Shaping metallic plutonium — never before produced — required understanding its material properties, which were found to be extremely strange: plutonium has been dubbed “the most complicated metal known to man.” Researchers discovered that plutonium had six different allotropes (physical arrangement of atoms), more than any other metal. Plutonium’s complexity made basic facts such as its melting point surprisingly difficult to determine, and the exotic cerium sulfide crucibles proved to be unnecessary when an unexpectedly low melting point was found. Most of what’s known about plutonium metallurgy was initially worked out at Los Alamos during the Manhattan Project.

In addition to uranium and plutonium, novel methods and processes had to be developed for a variety of other materials. Polonium, used in the initiator, needed to be procured in large quantities, extensively purified, and deposited on metallic foils. The investigation into polonium’s material properties has been described as “novel as that of plutonium.” Fabrication techniques were also developed for other materials with useful nuclear properties, such as Boron 10 (an isotope of boron with a large neutron cross section) and beryllium.

While a great deal was known about the mechanics of guns and projectiles at the beginning of the project, nothing like a gun needed to fire a sub-critical nuclear bullet had ever been built. The bullet needed to be fired as fast as possible to avoid the problem of predetonation — spontaneous fission starting a chain reaction before the pieces had been joined, leading to a “fizzle” — but the target also needed to stay roughly intact after the projectile’s impact. The gun also needed to be light and small enough that the resulting bomb wouldn’t be too heavy to carry, a requirement that was greatly aided by the realization that the gun would only need to fire once, and could be much less robust than conventional guns. A great deal of calculation and testing on guns of various calibers, propellants, and geometries of targets and bullets was required.

But novel as it was, this work on gun development was far more straightforward than what was required for the implosion bomb. Virtually nothing was known about the behavior of materials when imploded, or how to create an explosion that would compress a sphere of material symmetrically. Resolving this lack of knowledge began simply — setting off explosives on the outside of pipes and seeing how they deformed — but quickly ramped up in complexity. A variety of methods and instruments had to be created to study the interior of materials as they were being imploded. Some of these, such as using X-rays and high-speed cameras, were adaptations of existing measurement techniques, pushed to much higher levels of performance: getting high-speed cameras to work, for instance, required months of experiments with multiple camera designs to create one that was fast enough.

Other implosion analysis methods, such as the “RaLa” method, were far more novel. The RaLa method involved placing an isotope of radioactive lanthanum (hence RaLa), an emitter of gamma rays, at the center of the material to be imploded. When the material began to compress, it would become denser, resulting in fewer gamma rays penetrating. By surrounding the implosion with gamma ray detectors, a detailed progression of the geometric changes in the imploded material could be recorded.

While the RaLa method was a valuable source of implosion information, implementing it was fiendishly complicated. Even gathering and manipulating the lanthanum was difficult due to its intense radioactivity. Lanthanum was shipped from Oak Ridge in special lead-lined trucks, which were driven 24 hours a day. The assembled material reached 100 curies of radioactivity at a time when most radioactive experiments didn't exceed a fraction of a curie: a Los Alamos chemist noted that “no one ever worked with radiation levels like these before, ever, anywhere in the world.” Using RaLa required things like a “mechanical chemist” to remotely manipulate the radioactive material, and, initially, a mobile laboratory built inside repurposed M4 tanks.

It was initially hoped that a symmetrical compression could be created simply by adding enough explosive detonation points around the spherical bomb core, but RaLa and other measurement methods revealed “jets” of core material shooting ahead of the rest of the collapsing mass. Dealing with the jets and other asymmetries, and creating a symmetrical collapse of the bomb core, would be one of the main difficulties of the implosion bomb program.

New analytic techniques and devices weren’t limited to measuring implosions. New devices were made for, among other things, counting neutrons, collecting electrons, discriminating between different electronic pulses, and measuring projectiles with microwaves. Over a thousand pieces of electronic equipment were built at Los Alamos, many of them novel or higher performance than any other equipment then available, including better amplifiers, oscilloscopes, and counting circuits.

As with Oak Ridge and Hanford, even seemingly straightforward project elements were often enormously complex development efforts due to the novel requirements of the bomb. The triggering mechanism, for instance, couldn’t rely on off the shelf components, as they were considered insufficiently reliable: even a 1% chance of failure was far too high for a bomb in which hundreds of millions of dollars had been invested (the acceptable failure rate was eventually decided to be 0.01%). And when dropped, the bomb needed to automatically trigger at a specific elevation, a capability which didn’t yet exist. After a great deal of testing and development, a trigger circuit using radar altimeters, barometric switches, and electronic clocks was eventually created.

All this development work took place in largely the same form as the rest of the project: for nearly every decision or device, multiple promising paths were investigated, often at great expense, in the hopes that one or more could be made to work. This often led to a cascade of branching investigations, where each path branched into several possible paths, each of which might branch into more possible paths, and so on. The implosion method was just one of multiple bomb designs investigated, and within the implosion investigation RaLa was just one of multiple analytic techniques created to study implosions (a history of Los Alamos described the implosion studies as a “seven-pronged experimental program”). Bombs using both plutonium and U235 fuels were pursued, and for each material multiple methods of processing them were studied.

Because so little was known, and progress needed to be made quickly, these investigations often relied on brute force empiricism: running dozens or hundreds of experiments while systematically varying different experimental parameters. Early implosion studies repeatedly tested pipes surrounded by explosives, systematically varying the size of pipes, explosive arrangement, and the type of explosive used. These systematic investigations continued when more advanced diagnostic methods became available: to determine implosion parameters like symmetry, collapse velocity, and amount of compression, an “exhaustive” test program was initiated, where “every possible parameter was varied”. Systematic trial-and-error testing was also used for the design of the gun, the projectiles, and the target.

The pivot to implosion at Los Alamos

The strategy of investigating multiple promising paths proved its worth when the first shipments of reactor-produced plutonium began to arrive at Los Alamos from Oak Ridge in the spring of 1944. Prior to this Los Alamos had worked only with plutonium produced in cyclotrons, which consisted chiefly of the isotope Plutonium-239. However, reactor-produced plutonium was found to also have significant amounts of a different isotope, Plutonium-240. This isotope was found to undergo spontaneous fission much more readily than Plutonium-239 or U235. Its rate of spontaneous fission — a million times higher that of U235 — was so high in fact that the presence of Plutonium-240 made a gun-type plutonium bomb infeasible: so many neutrons would be produced by spontaneous fission that a chain reaction would be triggered before the bullet met the target, blowing the bomb apart and creating a fizzle.

The situation seemed dire. There was no time to design and build a separation plant to remove Plutonium-239 from Plutonium-240. To make use of the Hanford plutonium, the only options appeared to be build a composite bomb that mixed plutonium and uranium together (which would be low efficiency and have a comparatively small yield), or to use a different bomb mechanism capable of creating critical mass much more quickly than a gun could. That meant the implosion method, but in early 1944 work on the implosion bomb was far behind behind the gun bomb: until then the implosion bomb had been of secondary importance, a backup in case the gun didn’t work. It was still far from certain whether a workable implosion bomb could be built.

In July 1944, Robert Oppenheimer ordered a halt on further work on the plutonium gun, and to step up efforts on a plutonium implosion bomb. Within two weeks Los Alamos had been completely reorganized to focus on solving the problems of the implosion bomb. Work on the various implosion analysis methods accelerated, and the first RaLa test was completed in late September. An extensive exploration of solutions to the problem of jets and asymmetrical compression was undertaken, and the development of plutonium purification and metal fabrication methods continued (made easier by the fact that an implosion bomb could tolerate a much higher level of plutonium impurities). By early 1945, aided by the discovery plutonium’s unexpectedly low melting point (which made finding a workable crucible much easier), plutonium was successfully being purified, reduced to a metal, and worked by various methods, shortly before the first “batch” (a mere 80 grams) of plutonium arrived from Hanford.

The asymmetrical compression problem was eventually solved by the use of explosive lenses, and by changing from a hollow to a solid core of fissile material. Explosive lenses — shaped explosives that would focus the explosion and create a converging pressure wave — were first suggested by James Tuck, who arrived at Los Alamos in May 1944, but using them to build a bomb was fiercely difficult. No theory yet existed for analyzing and predicting the behavior of explosive lenses, and no methods existing for fabricating the carefully shaped explosives to the level of precision required. To design the lenses, there was no choice but to take an iterative approach: designers made guesses about effective lens shapes, tested them, and used the feedback to refine their designs.

Explosive lenses around a bomb core, via Atomic Heritage Foundation.

At the same time, methods had to be developed for casting and machining the explosives. This was both enormously difficult (since the explosives had to be extremely uniform and precisely shaped) and dangerous (since machining risked explosion). A history of Los Alamos describes the challenges of explosive casting:

It was particularly difficult to cast the high explosives accurately and avoid cracks, bubbles, and other imperfections. Cooling cycles had to be long to minimize thermal stress cracks. Castings had to be wrapped in insulation before being transported between buildings…Casting technology developed slowly and painfully at Los Alamos, by a succession of reasonable steps, that consistently failed to give completely satisfactory results. Eventually, the problems were overcome… - Critical Assembly

Machining the explosives was similarly difficult. The jigs and fixtures required to hold the explosives required several months of development, and the explosive machining methods created were considered “revolutionary.”

The work of figuring out casting and machining methods, and of creating workable lens shapes, demanded enormous amounts of explosives: over the course of the project James Tuck noted that “well over twenty thousand castings were delivered to the firing sites, while the number of castings rejected because of poor quality or destroyed for other reasons is several times this figure.” At the peak of the implosion program, Los Alamos was using 100,000 pounds of high explosive a month.

Creating symmetrical compression in the bomb core also required very precise detonating of the explosives. Early tests were done using primacord detonators (a cord of high explosive surrounded by textiles), but these were found to be far too imprecise, and investigation into other types of detonators was pursued. After extensive experimentation, a spark gap switch (a sort of explosive spark plug) combined with an exploding bridgewire detonator was developed.

Another major design problem on the implosion device was the initiator. The implosion bomb would require a new, more precise type of initiator than used on the gun bomb, one that would be triggered at the moment of highest compression in the bomb core. As late as early 1945, it wasn’t clear whether such an initiator could be built. But continued experiments and testing eventually resulted in what appeared to be a workable design. Named “the urchin,” it consisted of a beryllium sphere and pellet, with polonium between the two. When the core was compressed, the sphere would be crushed, mixing the beryllium and polonium and emitting neutrons.

Until very late in the Manhattan Project, it remained unclear if a workable implosion bomb could be built. In the last weeks of 1944, James Conant, President of Harvard and chair of the National Research Defense Council which oversaw the Manhattan Project, stated that the “difficulties were still enormous” and “my own bets very much against it.” At that time, the problems of the modulated initiator and of sufficiently precise and accurate detonation still hadn’t been solved. But as the researchers continued to run down various problems over the following months, the outlook improved considerably. By April the head of the explosives division at Los Alamos could report that its major research and design gambles “had been won”, and there was growing confidence that a bomb of the design chosen — solid plutonium core, explosive lenses, with electric detonators and a modulated initiator — could be made to work. Development work was by no means complete, and there were many problems yet to be solved (design changes to the bomb continued to be made until a few days before the Trinity test on July 16 1945), but by spring of 1945 the “research” portion of research and development had largely concluded.

As Los Alamos scrambled to build an effective implosion bomb, work also continued on the uranium gun weapon. Because the gun device was considered much less risky and much more certain to work, its development didn’t have the same fervor as the plutonium bomb. (In fact, cancelling the plutonium gun made the uranium gun program easier: a uranium bullet could be fired at a much lower velocity, reducing the difficulties of building a working gun device). But there were still numerous development problems that needed to be overcome. As with plutonium, uranium metal reduction and working methods were finalized by late 1944, and by early 1945 the design of the gun was completed and assembly of it was being tested. Unanswered questions remained until surprisingly late in the program (as late as December 1944 the critical mass of U235 still hadn’t been precisely determined), but there was little uncertainty around whether the bomb would function (so little, in fact, that testing the uranium gun bomb was considered unnecessary). By May of 1945, the uranium gun weapon, code named “Little Boy,” was “ready for combat”.

Conclusion

The Manhattan Project has become synonymous with a difficult, expensive, and ambitious technological development project, and you often hear folks advocating for a “Manhattan Project for X.” So it's worth understanding why, specifically, the Manhattan Project was so difficult and expensive.

First, there were inherent physical difficulties in many of the tasks. The bombs required pounds of fissile material, and there was no easy way to produce it: any method chosen would require an enormous industrial-scale production facility. The Hanford Site for producing plutonium cost $350 million ($6.4 billion in 2025 dollars), and the Oak Ridge site cost $304 million ($5.5 billion in 2025 dollars), not including the cost of the borrowed silver for the electromagnets. Hundreds of millions more were spent on operating the facilities. Part of the expense and difficulty of the Manhattan Project came simply because manufacturing fissile material is expensive and difficult (and remains so today.)

The second difficulty with the Manhattan Project was that because of the great urgency, work had to proceed on the basis of very little information. Resolving the uncertainty often entailed expensive efforts that would have been greatly simplified (or eschewed altogether) had a slower pace been acceptable. Plants were built before the processes they would use had been completely defined, often requiring extensive rework after parts of them had been built. Time and effort was invested in creating a plutonium gun bomb that could have been avoided had the designers waited until reactor-produced plutonium (which due to the presence of Plutonium-240 wouldn’t work in a gun bomb) was available.

The third difficulty was that because so little knowledge existed around the nature of atomic physics and nuclear chain reactions, it was far from clear what the best route to an atomic weapon was. Because the field was so new, using only recently-discovered natural phenomena that were poorly understood, a great deal of effort was needed to resolve this uncertainty along numerous technological axes. Thus the Manhattan Project involved a large amount of trial and error experimentation, and of pursuing multiple paths of technological development — different bomb types, different fuels, different uranium separation methods, different tampers, different triggers, different implosion analysis methods — to create a workable bomb.

It’s this last difficulty that is most relevant for other technological development projects. Developing other technologies doesn’t necessarily require building enormous, industrial scale industrial facilities to even begin, and doesn’t necessarily require rapidly proceeding before the proper information and supporting technologies are available. But it will almost certainly require investigating various promising paths of development, partially-informed groping around until the right combination of methods and components is discovered. Indeed, this sort of exploration is the very essence of technological development.

Edison’s light bulb provides a useful comparison: inventing it didn’t require building an enormous, multi-million dollar factory to produce the components to experiment with. And Edison wasn’t forced to invent every single predecessor technology that a light bulb required. One of the reasons why an incandescent bulb wasn’t invented earlier is that, prior to the 1860s, vacuum pumps weren’t good enough. Edison’s bulb relied on the invention of the Sprengel mercury pump by Hermann Sprengel in 1865, which could create a high enough vacuum that incandescent lamps became feasible. But Edison was forced to explore a variety of different potential methods for creating a bulb until he created one that worked.

One thing that the Manhattan Project shows is that resolving this uncertainty, and figuring out what a technology should actually be, is hard. The Manhattan Project had some of the most brilliant scientific minds in the world working on it, but even with this collective brainpower it was far from clear what the best route to the bomb was. For almost every major design decision (including the successful ones), there was at least one genius or expert highly skeptical that it would work. Future Nobel Prize winner Hans Bethe initially resisted joining the program because he believed that building an atomic bomb wasn’t feasible at all. Early in the program, many (such as physicist Alfred Nier) felt that electromagnetic separation wouldn’t be a feasible way to isolate U235 (and indeed, until very late it looked like it wouldn’t be). British scientists were similarly pessimistic about the gaseous diffusion process, with some believing that the plant “would be inoperable” due to surges and fluctuations in the gas flow. Du Pont thought that a graphite moderated, water-cooled reactor might not work, and insisted on a heavy water reactor as a backup. Explosives expert William Parsons, head of the uranium gun program, was skeptical that explosive lenses would work, and argued that the only way to have an implosion bomb ready by summer of 1945 was with a non-lensed design. Enrico Fermi initially believed that a modulated initiator wouldn’t work, and would come up with a new reason why “every second day or so.”

It’s also notable that many of the options chosen didn’t turn out to be the best long-term. Post-war atomic weapons almost entirely eschewed gun-type mechanisms in favor of more efficient implosions. And of all the uranium separation methods explored, it was the one that was deemed least promising and not used at all by the Manhattan Project — gas centrifuges — that is primarily used for uranium enrichment today.

Not all technologies will require expensive physical facilities to produce, or require extremely rapid, expensive development. But resolving the uncertainty inherent in a new technology — figuring out what, exactly, the arrangement of phenomena needs to be to achieve some goal, and how that arrangement can be achieved — is part of the fundamental nature of creating a new technology. The Manhattan Project required an unusually large amount of this (advancing the state of the art in many different disciplines), but will always be required to some degree.

1. Consideration was also given to another isotope of uranium, U233.

2. Slowing down the neutrons made them more likely to be captured by the U235 and induce fission.