Thorium in the Manhattan Project

July 21, 2014April 21, 2026Kirk SorensenAlvin Weinberg, blog, History, Nuclear Physics, Thorium

This spring I completed a master’s thesis at the University of Tennessee on the history of thorium research during the Manhattan Project. The thesis ran 476 pages, most of which was primary-source appendices — Seaborg’s Met Lab history volumes, the MUC-LAO meeting notes from April and May 1944, the Wigner-Young paper of May 1945, and related documents. I want to condense the historical argument here, for readers who don’t want to work through the full thesis but who care about the question: what was actually known about the thorium fuel cycle during the Manhattan Project, and why did plutonium become the fuel cycle the country built itself around?

The short answer

Thorium and uranium-233 were not overlooked during the Manhattan Project. They were known, characterized, and seriously evaluated. The decisions that privileged plutonium over U-233 were specific and documented. They were not the result of oversight or ignorance. They were the result of wartime priorities, compressed timelines, and a chain of technical decisions — each reasonable in isolation — that together put the country on a plutonium-and-enriched-uranium infrastructure by the end of 1946 and kept it there for the next eighty years.

The key documents are almost all public. Most of them have been public for decades. What they show is that by the spring of 1945, Eugene Wigner and Alvin Weinberg had produced a detailed paper-design for a thermal-spectrum, thorium-uranium-233 breeder reactor with circulating fluid fuel, that Arthur Compton and Colonel Nichols had discussed making this design the focus of a major Met Lab program, and that the concept was set aside not because it was shown to be unworkable but because the war ended three months later and the Met Lab dissolved into the postwar laboratory structure before the program could be mounted.

The story of thorium during the Manhattan Project does not appear in the conventional histories because it was not part of the path to nuclear weapons. But it was there and it was part of the story. The implications of what was done then continue to this day.

Thorium before the war

Thorium had been known as a radioactive element since Marie Curie’s 1898 work. Its half-life — about 14 billion years, roughly the age of the universe — made it one of the two longest-lived naturally occurring radioactive elements, along with uranium. In the early twentieth century it was used commercially in the Welsbach gas-mantle industry, where thorium dioxide’s incandescence made it the standard for gas lighting. It was, by the 1930s, a well-understood industrial mineral, more abundant in the earth’s crust than uranium by a factor of three or four.

What thorium was not, before 1940, was fissile. Otto Hahn and Fritz Strassmann’s discovery of fission in December 1938 had established that uranium-235 — the rare isotope of natural uranium — could be split by slow neutrons. Thorium-232, the only naturally occurring isotope of thorium, could not. Early neutron-irradiation experiments on thorium at laboratories in Europe had indicated that thorium-232 absorbed neutrons to form thorium-233, and it appeared likely that this decayed through beta-emission into protactinium-233. But where further decay proceeded beyond this point was an open question in the fall of 1940, when Glenn Seaborg directed his new graduate student John Gofman to take up the problem.

Room 307, Gilman Hall

Gofman was twenty-two years old, had just started his PhD at Berkeley under Seaborg, and was assigned the problem of conclusively identifying protactinium-233 and measuring its radioactive properties. Seaborg’s journal entry of September 23, 1940, records that Gofman had elected to carry out his graduate work under Seaborg’s direction and had accepted Seaborg’s suggestion to undertake, as his thesis project, the search for the missing 4n+1 heavy nuclei, starting with Pa-233.

The work was done in Room 307 of Gilman Hall, on the Berkeley campus — the same small room where Seaborg had separated and identified plutonium with Arthur Wahl and Joseph Kennedy earlier in 1940. Over the following year, Gofman worked through the chemistry of neutron-irradiated thorium, progressively confirming that the activities Meitner, Hahn, and Strassmann had reported from thorium-plus-neutrons in 1938 were indeed due to Pa-233 and its daughter U-233 rather than to fission-product contamination.

In early 1941, Gofman had accumulated enough U-233 from cyclotron irradiation of thorium to attempt a fission measurement. The critical experiment took place in late February. The U-233 was bombarded with slow neutrons. The fission count was unambiguous. Uranium-233 was fissile.

On April 23, 1941, Seaborg wrote a memo to the Uranium Committee chairman Lyman Briggs recommending that this result be treated as consequential. The memo — reproduced in Seaborg’s Early History of Heavy Isotope Research at Berkeley, which is now public — stated that U-233 had fission properties comparable to or possibly superior to those of U-235 and plutonium-239, and that thorium, which was abundant and cheaply available, could therefore be considered “a third available source of nuclear energy.”

The phrase is exactly right. From April 1941 forward, everyone at the top of the American nuclear program knew that there were three potential paths to fissile material, not two. Uranium-235 had to be separated from natural uranium by some physical process — gaseous diffusion, electromagnetic separation, centrifugal enrichment. Plutonium-239 had to be bred from uranium-238 in a reactor. Uranium-233 had to be bred from thorium-232 in a reactor. The three paths had different physical properties, different engineering requirements, and different strategic implications.

Why plutonium won the weapons race

The eighteen-month head start plutonium had over U-233 was decisive. Plutonium had been discovered in February 1941. Its physical and chemical properties had been characterized through 1941 and early 1942. By the time U-233’s fissile properties were confirmed in the spring of 1941, plutonium was already the focus of a serious production-reactor effort.

That effort had a crucial advantage. Plutonium could be produced in a reactor fueled with natural uranium — with the mass-238 uranium both acting as the fertile material and providing the chain-reacting isotope (mass-235, at 0.7 percent abundance) that sustained the reaction. No fissile starter charge was needed. The pile could be built from natural uranium and a moderator — graphite, at Chicago — and it would produce plutonium while it operated.

Thorium offered no such shortcut. Thorium-232 is not fissile. A reactor running only on natural thorium cannot achieve criticality. To produce U-233 from thorium, you first need a reactor running on something else — U-235 or plutonium — and you use the neutrons from that reactor to convert thorium to U-233. In 1942, with no operating plutonium reactor yet and no enriched U-235 available in useful quantities, the U-233 path was a second step that had to wait for the first.

So the Metallurgical Laboratory’s early reactor work focused on plutonium. CP-1 went critical in December 1942. The X-10 graphite reactor at Oak Ridge, intended as a plutonium-production pilot, began operating in November 1943 at 500 kilowatts thermal. Within months it was producing plutonium in gram quantities and thorium in its outer zones could be irradiated to produce small quantities of U-233 for characterization. By the middle of 1944, Oak Ridge was producing enough U-233 from X-10 irradiation to begin serious work on its nuclear properties.

What that work found was that U-233 was in some respects a better fissile material than either U-235 or plutonium-239. Its thermal fission cross-section was large. Its eta value — the number of neutrons released per neutron absorbed — was the highest of the three in a thermal spectrum. It had more favorable neutron-economy properties for a thermal breeder than any other known fissile material.

It was, however, contaminated with uranium-232.

The U-232 problem and the end of U-233 as a weapons material

Uranium-232 is produced as a side reaction in any thorium irradiation. Thorium-232 can absorb a neutron to form thorium-233, which beta-decays to protactinium-233, which beta-decays to the desired U-233. But a small fraction of the time, a thorium-230 impurity in the thorium, or a two-neutron reaction in the irradiated thorium, produces Pa-232 or directly produces U-232. And U-233 itself, when irradiated further, can undergo an (n,2n) reaction to produce U-232.

U-232 has a half-life of about 69 years. Its decay chain includes thallium-208, which emits an intense 2.6-MeV gamma ray. Any U-233 that contains significant U-232 is intensely radioactive by virtue of the thallium-208 daughter. The radiation dose from bulk U-233 is high enough to make weapons assembly hazardous and high enough to be detected at considerable distance by gamma-ray detectors.

Seaborg recognized this problem at the Met Lab sometime in 1944. The documentary evidence is clearest in Weinberg’s autobiography, The First Nuclear Era, where Weinberg recounts Seaborg pointing out that U-233 manufactured in Wigner’s proposed plutonium-to-U-233 converter reactor would be contaminated with U-232, which “probably fissioned spontaneously and also emitted many alpha particles,” and that “a bomb made of U-233 would therefore have to be carefully cleansed of any tramp light elements such as beryllium, since alpha particles interacting with beryllium would produce neutrons; and if spontaneous fission of U-232 was sufficiently probable, the U-233 gun-type weapon was unworkable.” Weinberg adds: “I was greatly impressed by Glenn’s insight into this difficulty, since at the time little was known about either U-233 or U-232.”

By the spring of 1945, serious consideration of U-233 as a weapons material had essentially ended. The implosion weapon design that Oppenheimer’s team at Los Alamos had developed for plutonium rescued Pu-239 from its analogous contamination problem (Pu-240 predetonation), but the implosion rescue didn’t have to be extended to U-233 because U-233 was no longer a weapons candidate. The Teapot series test in 1955 would include a composite Pu/U-233 core, but that test had substandard yield and no further weapons work with U-233 was done. The details remain classified, but the record is consistent: U-233 was evaluated, found problematic as weapons material because of U-232 contamination, and dropped.

This was not a failure. It was a feature. The same U-232 contamination that made U-233 problematic as a weapons material made it resistant to weapons diversion from a civilian power program. The intrinsic proliferation resistance of the thorium fuel cycle that advocates argue for today is exactly the property that caused U-233 to be set aside for weapons work in 1945.

The April 1944 meetings

What happened next is the most underreported episode in the early history of American nuclear power. By the spring of 1944, with plutonium production at X-10 beginning to reach gram quantities, Wigner and his colleagues had begun seriously thinking about what came after the war. A series of meetings convened in April and May of 1944 — MUC-LAO-17 on April 26, MUC-LAO-18 on April 28, MUC-LAO-19 on May 5, MUC-LAO-30 on July 6 — discussed specifically what the postwar reactor program should look like.

The meeting notes, reproduced in the appendices of my thesis, show the direction the conversation was taking. Wigner proposed a “converter reactor” that would use plutonium as its fissile driver and thorium as its fertile blanket, producing U-233 as a product. The converter could in principle be used to generate a U-233 stockpile that could start a subsequent thorium-U-233 thermal breeder. The rationale was that plutonium — abundant, by the end of the war, in the form of Hanford-produced material with some Pu-240 contamination — could serve as the starter charge for a thorium economy that would thereafter run on U-233 and thorium alone.

This was the outline of what would later become the molten-salt reactor program at Oak Ridge: thorium-U-233 as the operational fuel cycle, plutonium or enriched uranium as the startup material only. The key insight — that U-233 could serve as the equilibrium fissile material in a thermal-spectrum reactor while plutonium served only to launch the program — was articulated in these meetings in April and May 1944, eighteen months before the war ended.

Seaborg’s journal entry for January 17, 1945 records that “with regard to research and development, Compton said that he was interested to find that three different laboratories have proposed breeder piles for U-233.” Compton also noted that Colonel Nichols of the Manhattan Engineer District had told him that Nichols preferred pursuing the thorium-U-233 breeder over the plutonium-U-233 converter as the Met Lab’s next major effort. The thorium breeder concept was emerging as the consensus direction.

The Wigner-Young paper

On May 17, 1945, Wigner and a Met Lab colleague named Gale Young issued an internal paper titled Preliminary Calculations on a Breeder with Circulating Fuel. It is reproduced as an appendix to my thesis. The paper is seven pages long. It is almost certainly the earliest engineering-grade analysis of a fluid-fuel thorium breeder reactor ever produced.

The design Wigner and Young sketched had several features that would become familiar to anyone who has looked at the 1970 Oak Ridge MSBR reference design. The fuel was a uranium compound dissolved in water — aqueous homogeneous, in later terminology — rather than a solid ceramic encapsulated in metal cladding. The reactor was a single core containing both the fissile U-233 and the fertile thorium in solution. The fuel circulated through the reactor at some velocity and could be continuously processed to remove fission products, which would otherwise poison the chain reaction. The reactor would operate in a thermal neutron spectrum, which gave the best eta for U-233 and thus the best breeding performance.

Wigner and Young calculated the reactor’s neutron economy in detail. They allocated the roughly 2.3 neutrons that U-233 produces per fission to the various sinks: one neutron to maintain the chain reaction, one neutron to convert thorium to U-233 for breeding, the remaining 0.3 neutrons to overcome losses to fission-product absorption, leakage from the core, absorption by structural materials, and absorption by the chemical processing streams that could not be made arbitrarily fast. They concluded that the losses could be held within the breeding margin. The reactor would, they calculated, breed.

This was May 17, 1945. Nazi Germany had surrendered on May 8. The Trinity test was two months away. The war in the Pacific would end four months later. The thorium breeder concept had, by the time Germany fell, already been worked out to a level of engineering detail sufficient to begin serious paper design.

Weinberg, who had collaborated with Wigner on the converter reactor work the previous year, was by then making plans to move to Oak Ridge. He would write later, in The First Nuclear Era, that “the aqueous homogeneous circulating-fuel power breeder based on the 232Th-233U cycle became a kind of obsession for me. I came to Oak Ridge convinced that the laboratory, with its many chemists and chemical engineers, was the ideal place to develop Wigner’s aqueous homogeneous power breeder.” His train left Chicago in late May 1945. The Weinbergs — Alvin, his wife Marge, and their two-year-old son David — arrived in Oak Ridge in early June.

The dissolution

What happened next was that the war ended. Hiroshima and Nagasaki were bombed in August 1945. Japan surrendered on September 2. The Manhattan Engineer District began winding down almost immediately. The Met Lab’s operational priority shifted from wartime weapons production to postwar disposition. The Atomic Energy Act was passed in August 1946. The Atomic Energy Commission took over civilian nuclear responsibility from the military Manhattan Engineer District on January 1, 1947.

The thorium breeder program that Wigner, Young, Weinberg, and their colleagues had been developing did not, in that transition, receive a programmatic priority. It was not cancelled — it was simply not scaled up. The Oak Ridge laboratory that Weinberg arrived at in 1945 was reorganized as Clinton Laboratories, then as Oak Ridge National Laboratory, with new directors and new missions. The thorium work continued at a small scale. Aqueous homogeneous reactor experiments were built and operated in the late 1940s and early 1950s. The aircraft-reactor work of the mid-1950s used fluoride-salt rather than aqueous chemistry, and it was from that aircraft-reactor base that the molten-salt reactor program emerged in the late 1950s and reached its technical maturity in the MSRE in the 1960s.

But the twenty-year gap between the Wigner-Young paper of May 1945 and the MSRE criticality of 1965 is not a gap in the physics. The physics was understood. The gap is institutional. The wartime compression that had forced a decision between plutonium and U-233 as weapons priorities had, by the time of the compression’s end, built a production infrastructure — Hanford for plutonium, Oak Ridge and later Paducah and Portsmouth for U-235 — that was oriented around uranium and plutonium. The civilian reactor program that grew out of that infrastructure naturally used that infrastructure’s outputs. Light-water reactors fueled on enriched uranium, with plutonium as a spent-fuel byproduct, became the civilian norm because they fit the material flows the weapons complex was already producing.

Thorium, by contrast, required an infrastructure that didn’t yet exist. A thorium fuel cycle would have needed thorium mining at scale (the Welsbach mantle industry was too small), thorium fuel fabrication (the thorium dioxide industry was not oriented toward nuclear purity), U-233 breeder reactors (none existed in commercial form), and a closed reprocessing infrastructure that could handle the U-232 radiation levels associated with U-233 operations. None of this was in place in 1946. None of it was committed to being built. The weapons complex’s expansion through 1950 to 1955 built a plutonium and enriched-uranium infrastructure at a scale that, once in place, made thorium a progressively harder sell on institutional grounds regardless of its technical merits.

What the thesis showed that I had not fully appreciated before

Going through the documents carefully, what became clear to me was how differently the thorium story looks from the primary record than from the secondary histories. The conventional narrative — that plutonium and uranium-235 were simply the available fuels for the Manhattan Project while thorium was a postwar afterthought — is wrong. Thorium and U-233 were known and evaluated from the very beginning of serious nuclear work. The fuel cycle was understood in essential outline by 1942 and in engineering detail by May 1945. The technology that would later become the molten-salt reactor program was not an innovation of the 1950s. It was an extension of a line of thinking that had been active throughout the Manhattan Project.

What the primary record shows, in short, is that the Manhattan Project generation did not overlook thorium. They considered it, understood it, evaluated it as weapons material and found it problematic for that purpose, evaluated it as reactor fuel and found it promising, and began the engineering work on a thorium breeder before the war ended. The reasons that work was not brought to commercial fruition are historical and institutional rather than scientific.

That distinction matters because the thorium case today is made in the same terms that Wigner and Weinberg would have recognized: U-233 has the best eta in a thermal spectrum of any fissile material; thorium is more abundant than uranium by a factor of three or more; the fuel cycle is intrinsically proliferation-resistant because of the U-232 contamination that rendered it unsuitable for weapons; continuous processing of fluid fuel eliminates the fission-product poisoning that limits solid-fuel reactors; and the chemistry is well-behaved enough that a commercial system is engineerable with materials and techniques that were already available in 1970, let alone now.

The argument did not have to wait for us to make it. It was made, in essentials, in May 1945.

A note on sources

Everything I’ve described above is drawn from documents that are now publicly available. The most important are:

Glenn Seaborg, Early History of Heavy Isotope Research at Berkeley, August 1940 to April 1942 (Lawrence Berkeley Laboratory, 1976). This is Seaborg’s reconstruction of the Berkeley work on thorium and U-233 in 1940-1942, including the Gofman thesis work and the April 1941 memo.
Glenn Seaborg, History of Met Lab Section C-I (Lawrence Berkeley Laboratory, four volumes, 1977-1980). This is the four-volume reconstruction of the chemistry section’s work at the Metallurgical Laboratory from April 1942 through May 1946. The April and May 1944 meeting notes (MUC-LAO-17 through MUC-LAO-30) are included in Volume 3.
Alvin Weinberg, The First Nuclear Era: The Life and Times of a Technological Fixer (AIP Press, 1994). Weinberg’s autobiography covers the Met Lab years and his subsequent career at Oak Ridge. The U-232 contamination observation by Seaborg is recounted on pages 37-38.
Richard Hewlett and Oscar Anderson, The New World, 1939-1946: A History of the United States Atomic Energy Commission, Volume 1 (Pennsylvania State University Press, 1962). The official institutional history of the Manhattan Project’s civilian aspects.

The full primary-source appendices of my thesis (the Gofman thesis extracts, the Met Lab history extracts, the MUC-LAO meeting notes, and the Wigner-Young May 1945 paper) are available through the University of Tennessee’s TRACE repository. The thesis itself is titled Thorium Research in the Manhattan Project Era and was submitted in early 2014.

One correction

The introduction to the thesis opens with the sentence: “On August 8, 1967, Dr. Glenn Seaborg, chairman of the United States Atomic Energy Commission, sat at the controls of an experimental reactor in Oak Ridge, Tennessee.” This date is wrong. The MSRE achieved first criticality on uranium-233 on October 2, 1968, with the operations team present. Glenn Seaborg came to Oak Ridge six days later and, on October 8, 1968, took the reactor to 100 kW. The photograph that became the unofficial symbol of the MSR program — Seaborg at the console with Weinberg behind him and Ray Stoughton seated at his right — was taken on that day. My apologies for the error.