University of California, Berkeley, Gilman Hall, Room 307
The room is not much to look at. Third floor of Gilman Hall, a buff-colored building on the Berkeley campus, with a view toward the bay if you crane around the corner. In the fall of 1940, room 307 held a handful of chemistry benches, some Geiger-Mueller counters numbered in sequence beginning with Z-1, a few Lauritsen quartz-fiber electroscopes, and two graduate students named Arthur Wahl and Gerhart Friedlander who were working under the direction of a twenty-eight-year-old instructor named Glenn Theodore Seaborg.
In that room, over the next two years, Seaborg’s group would discover two new isotopes — plutonium-239 and uranium-233 — and demonstrate that both could sustain nuclear fission. The plutonium discovery would become one of the most consequential acts of chemistry in human history, the key that unlocked the nuclear age. The uranium-233 discovery was equally profound, and almost no one would know about it for the rest of the twentieth century.
This is the story of why.
The Missing Chain
The world’s first nuclear reactor had not yet been built when John W. Gofman, a first-year graduate student from Cleveland, Ohio, accepted Glenn Seaborg’s suggestion for his doctoral thesis problem in the autumn of 1940. The suggestion was to look for something that might not exist: a radioactive element hiding in a gap that only mathematics predicted.
To understand the gap, you need to know a little about how heavy atoms fall apart.
The heaviest naturally occurring elements — uranium, thorium, radium — are unstable. Their nuclei shed particles over millions and billions of years, transforming themselves in a long cascade of decay from one element to the next until they reach a stable configuration. By the 1930s, physicists had traced three distinct pathways of heavy-element decay, each characterized by what remained when you divided an atom’s mass by four. One chain began with common uranium-238 and ended on stable lead-206. A second began with uranium-235 and ended on lead-207. A third began with thorium-232 and ended on lead-208.
Three chains. Four possible pathways. The fourth — the one whose decay products would leave a remainder of one when divided by four — seemed to have no representative in nature. Its members had apparently decayed away billions of years ago when the solar system was young. The only hint of their former existence was bismuth-209, a single lonely isotope whose mass divided by four left a remainder of one, suggesting that it had once been the stable terminus of a fourth and now-vanished chain.
The only way to recreate those missing isotopes, to fill in the ghost of a fourth decay chain, was to build them artificially with a machine powerful enough to rearrange the nucleus itself.
Ernest Orlando Lawrence had such a machine.
Lawrence was thirty-eight years old in 1940, a square-jawed Norwegian-American from the plains of South Dakota who had come to Berkeley in 1928 and proceeded to build the most powerful particle accelerators in the world. His instrument was the cyclotron: two hollow D-shaped electrodes facing each other across a gap, surrounded by an enormous electromagnet. Charged particles — protons, deuterons, helium nuclei — spiraled outward between the electrodes, gaining energy with each revolution, until they reached the outer edge of the magnetic field and flew off as a concentrated beam. The bigger the cyclotron, the more energy the beam carried. Lawrence’s 60-inch cyclotron, which came online in 1939, could accelerate deuterons to 16 million electron volts. Its magnet weighed 220 tons.
Glenn Seaborg had arrived in Berkeley two years before Lawrence’s 60-inch machine reached completion, as a young physical chemist freshly minted from UCLA. By 1940, he was directing a small group using the cyclotron’s output — a stream of high-energy neutrons produced as a byproduct of deuteron bombardments — to probe the nuclear properties of the heaviest known elements. What made Seaborg unusual among cyclotron researchers was that he was primarily a chemist. Where physicists saw nuclear events measured in bursts of ionizing radiation, Seaborg saw chemical separations: the careful, painstaking precipitation and reprecipitation of invisible quantities of new elements from solutions, isolating them from their parent materials, characterizing their behavior element by element.
That chemical precision was what made what followed possible.
On September 23, 1940, Seaborg recorded in his laboratory journal that a new graduate student, John W. Gofman, had accepted his suggestion to look for Pa-233 — protactinium-233 — the predicted intermediate step in the production of the missing fourth decay chain. The chain, if it existed, would begin with protactinium-233, proceed through various decay products, and eventually reach bismuth-209. Pa-233 would appear when thorium-232 absorbed a neutron, transforming into thorium-233, which decayed within twenty-six minutes into Pa-233.
Pa-233 was a ghost. No one had seen it. Hahn and Strassmann, who had discovered nuclear fission itself just two years earlier, had suggested that what Meitner and others had previously attributed to Pa-233 might actually be a zirconium fission product. Gofman’s job was to find out.
He bombarded thorium with neutrons at the cyclotron. He dissolved the irradiated material in water, precipitated thorium peroxyhydroxide, washed it with water and alcohol. He measured the decay. The twenty-six-minute half-life of Th-233 was there, exactly as predicted. Within weeks, Gofman had established that a twenty-five-day activity growing from the thorium was indeed Pa-233, not a zirconium fission product. He had located the first member of the missing chain.
The next member would be even more consequential.
If Pa-233 decayed by emitting a beta particle — transforming one of its neutrons into a proton — it would become a new isotope of uranium: uranium-233. No such isotope was known to occur in nature. Whether it existed, and if so what its properties were, was unknown. But Seaborg had a theory, and a graduate student willing to test it.
Two Discoveries in One Year
On the morning of February 25, 1941, Kennedy and Wahl and Seaborg — and Emilio Segre, who had joined the effort — completed a series of measurements that established beyond reasonable doubt that their sample contained plutonium-239, and that plutonium-239 underwent fission with slow neutrons at a rate 1.7 times greater than uranium-235.
They had discovered a new fissile isotope. It would not be the only new fissile isotope they would discover that year.
The plutonium measurements were done with painstaking care over weeks, using a magnetic field to deflect beta particles away from an ionization chamber so that the much rarer alpha particles could be counted. The numbers were clear. Plutonium-239, produced by the bombardment of uranium with neutrons, was fissile — it would split when struck by a slow neutron, releasing enormous energy and additional neutrons that could sustain a chain reaction. Within two months, Seaborg would summarize the results in a secret report to the Uranium Committee, the government body overseeing what would eventually become the Manhattan Project. Within eighteen months, Fermi would achieve the first self-sustaining chain reaction in a squash court under the stands of Stagg Field at the University of Chicago. Within four years, a bomb made from Seaborg’s plutonium would destroy Nagasaki.
The discovery of plutonium-239 is an extensively documented moment in the history of science. The discovery that happened sixty days later, in the same room, by the same team, is barely mentioned.
On April 10, 1941, John Gofman isolated purified Pa-233 from a thorium bombardment and mounted it in his ionization chamber to wait. He was waiting for alpha particles — the signature of a long-lived radioactive element growing in as Pa-233 decayed. If Pa-233 decayed to uranium-233, and if U-233 was radioactive with a measurable half-life, then over the following days and weeks, alpha counts would appear and grow.
On April 23, 1941, Seaborg measured the results and wrote in his journal:
A small but discernible quantity of alpha activity has grown into the beta-emitting Pa-233 samples isolated by Gofman on April 10. Measurements with our screen-windowed ionization chamber indicate that the 40 microcurie sample shows an alpha counting rate of 2 per minute, and our 27 microcurie sample shows an alpha counting rate of 1.2 per minute.
From these numbers — two alpha counts per minute in a 40-microcurie sample — Seaborg calculated a half-life. The math gave him approximately 100,000 years.
He understood immediately what this meant.
Of special importance is our demonstration through these results that U233 is sufficiently long-lived to be a practical source of nuclear energy should it be found to be fissionable with slow neutrons and should methods for its large scale production be developed.
In that sentence — carefully qualified, characteristically precise — Seaborg became the first person in history to recognize thorium as a potential energy source. The logic was straightforward but transformative. If U-233 had a half-life of roughly 100,000 years, it was stable enough to accumulate in useful quantities. If it proved fissile — if it split when struck by slow neutrons — it could sustain a chain reaction. And since U-233 could only be produced from thorium-232, any reactor running on U-233 would be, in effect, burning thorium. Thorium was three to four times more abundant in the Earth’s crust than uranium. The implications for the world’s energy supply were almost too large to contemplate.
The discovery had taken less than seven months. In the space of sixty days in the spring of 1941, Seaborg’s group had established the fissile properties of two new isotopes — plutonium-239 and uranium-233 — and had thereby identified both of the fuel cycles that would compete to power the nuclear age. The competition would last three decades. It would be decided not by physics but by war.
Plutonium-239 had a critical advantage over uranium-233: natural uranium contained both fissile material and the fertile precursor for plutonium. Thus, it could be produced from natural uranium in a reactor without requiring any enrichment. You loaded the reactor with natural uranium and ran it. Plutonium grew in the fuel as U-238 absorbed neutrons. You could build a reactor from materials you already had.
Thorium offered no such shortcut. Thorium was all fertile Th-232; there was no fissile thorium isotope found in nature. To produce U-233 from thorium in a reactor, you first needed a fissile material to start the chain reaction — either enriched uranium-235 or plutonium. And in 1941, the United States had essentially none of either. Producing them was the project. So there was no where to start with U-233 until you already already succeeded in enriching uranium or producing plutonium, and both accomplishments seemed far away at that point.
In the months that followed, as the government organized its response to the nuclear problem with increasing urgency, the logic of the situation was merciless. Plutonium-239 was needed now and could be produced now. U-233 could not be produced now but must wait for the success of others. The waiting would last longer than anyone imagined.
The Metallurgical Laboratory
In April 1942, Seaborg left Berkeley for Chicago. He was almost thirty years old, newly married, setting out to help produce plutonium for weapons on an industrial scale. He brought with him his notebooks, his separation techniques, and his undiminished interest in uranium-233 — an interest that no one in authority shared.
The organization he joined was called the Metallurgical Laboratory, a deliberate misnomer chosen to conceal its actual purpose from enemy intelligence. Housed in various buildings at the University of Chicago, the Met Lab was staffed by some of the most talented physicists and chemists in the country, many of them European refugees: Fermi and Segre from Italy, Szilard from Hungary, Wigner from Hungary, Franck from Germany. Their task was to design and build the reactors that would produce plutonium for the bombs that Oppenheimer’s group at Los Alamos was designing.
Seaborg’s chemistry division had a more specific assignment: to develop the chemical separation processes that would extract pure plutonium from the intensely radioactive irradiated uranium coming out of the reactors. This was formidably difficult work. The quantities of plutonium involved were measured in micrograms. The process had to be scaled from microscale laboratory chemistry to an industrial plant that would handle hundreds of tons of material. And it had to work reliably, because if it failed, the reactors would operate for nothing.
In the urgency of this work, uranium-233 was a distraction. It received a B/C priority from the Army, the rough equivalent of a polite footnote. Four men worked on it. Seaborg maintained a quiet personal interest but had no resources to devote to it.
What the B/C priority designation meant in practice was that U-233 research proceeded at the margins of the Manhattan Project — in quiet experiments at Oak Ridge, in Gofman’s continued work in Chicago, in the occasional meeting where someone raised the possibility that thorium might matter someday. It was always someday. The war demanded today.
There was one figure at the Met Lab who understood exactly what the thorium cycle offered and was already thinking, with characteristic mathematical precision, about how to use it. His name was Eugene Wigner, and he was perhaps the most rigorous mind in nuclear physics.
The November Cancellation
By the autumn of 1943, the Met Lab’s U-233 program had approximately four men working on it. The classified status of thorium’s energy potential meant that even this small effort was organizationally invisible. The contrast with the plutonium effort — hundreds of scientists, billions of dollars, the Hanford site taking shape in the Washington desert — was total.
Arthur Compton, the director of the Met Lab, had more pressing matters than uranium-233. He had Hanford to worry about. He had the chemical separation problem. He had the implosion calculations. He had Glenn Seaborg’s group making progress on plutonium chemistry at a rate that suggested the bomb would actually work. Against all of that, U-233 looked like exactly what the organizational charts said it was: a second-line insurance policy against plutonium failure, to be maintained at minimal cost until it was needed or clearly wasn’t.
On November 17, 1943, Compton delivered what he called his “State of the Nation” address to the Met Lab scientific staff. The U-233 program was cancelled.
The reasoning, as recorded in Seaborg’s diary, was circular in a way that had become characteristic of the Manhattan Project’s resource allocation decisions: “Without a P-9 [heavy water] power plant it is not possible to make enough U233 to be useful; therefore, this work should be discontinued for the present.”
The P-9 power plant was a proposed production reactor that would have irradiated thorium on a large enough scale to produce U-233 in useful quantities. The P-9 had been cancelled because resources were needed for plutonium production. Without the P-9, you couldn’t make enough U-233. Therefore, cancel U-233. The argument was airtight. It was also a perfect circle.
That evening, Seaborg made his own quiet protest. He asked that at least a small sample of U-233 be produced and its physical constants measured before the work was entirely abandoned. The request was granted. Four months later, Zinn’s measurements at Argonne would establish that eta for U-233 was 2.35 in a thermal neutron spectrum — higher than plutonium, higher than uranium-235, the best breeding ratio achievable in a thermal reactor.
The data arrived after the decision had been made. The decision stood.
Wigner’s Idea
Eugene Wigner was almost forty when he arrived at the Met Lab, a Hungarian theoretical physicist of extraordinary precision who had already contributed foundational results to quantum mechanics and would eventually win the Nobel Prize. At the Met Lab, his particular contribution was in reactor theory: understanding how neutrons moved through a reactor, how they were absorbed, how chain reactions were sustained. He was also, in the words of his student Alvin Weinberg, a man who could not see a problem without immediately searching for its general mathematical structure.
Wigner had been thinking about postwar energy since before the war began. He had seen the Nazi annexation of Austria and Hungary. He had watched the refugee scientists flood into American universities. He understood, perhaps better than anyone except Szilard, that the war would not be the end of the problem but only a particularly acute episode in a longer crisis of human civilization and energy. After the war, what would power the world?
The answer, Wigner believed, was reactors. Not just any reactors: breeder reactors. A breeder reactor was one that produced more fissile material than it consumed — that took a fertile material like thorium or U-238 and converted it, through neutron absorption, into more fuel than was burned. A breeder reactor, if it worked, was essentially a machine for creating fuel from a raw material that was not itself fuel. It was, in a precise thermodynamic sense, the alchemist’s dream realized: you could transmute common matter into energy.
What made breeding possible was a number that Wigner had been tracking carefully: the Greek letter eta, chosen to represent the number of neutrons produced per neutron absorbed in a fissile material. For a chain reaction to be self-sustaining, eta had to exceed one — one neutron to sustain the chain, with any excess available for other purposes. For breeding to be possible, the excess neutrons after sustaining the chain reaction and after accounting for all losses had to be sufficient to transmute at least one fertile atom into a new fissile atom. This required eta to be substantially greater than two.
Different fissile materials had different values of eta, and the values depended on the energy of the neutrons causing fission. In a thermal reactor — one in which the neutrons were slowed to thermal velocities by a moderating material — eta for plutonium-239 was about 2.05. Eta for uranium-235 was about 2.14. These values left a narrow margin for breeding after accounting for all the neutron losses in a real reactor.
But eta for uranium-233, in a thermal reactor, was about 2.4. This was not a small difference. It was the difference between breeding being barely possible and breeding being comfortably achievable. Wigner saw the implications clearly: the thorium cycle, in a thermal reactor, offered a better path to breeding than the plutonium cycle. And breeding was the key to using the Earth’s full inventory of nuclear fuel rather than just the tiny fraction represented by fissile U-235.
In the spring of 1944, as the Hanford production reactors were being built in the Washington desert and the Met Lab was thinking about what came after the war, Wigner convened a series of informal seminars on postwar reactor possibilities. He called the series “New Piles” — “pile” being the wartime term for a reactor. Fermi attended. Szilard attended. Wheeler and Franck and Morrison attended. Weinberg, Wigner’s young student and chief lieutenant, attended.
In these meetings, as Weinberg recalled decades later, the participants “allowed their imaginations to run riot.” All the different reactor concepts that had been sketched on blackboards and in classified memos could be laid out simultaneously and compared. Gas-cooled reactors. Heavy-water reactors. Fast reactors. Thermal reactors. And fluid-fueled reactors — reactors in which the fissile material was dissolved in a liquid rather than formed into solid fuel elements, so that fission products could be removed continuously, xenon poisoning could be avoided, and the fuel could circulate and be processed without shutting the reactor down.
At one of these meetings, Wigner told his group to look into molten fluoride salts as a fluid-fuel medium. The properties were promising: high melting point, low vapor pressure, stable chemistry under intense radiation, good heat transfer. One of his people investigated and reported back that molten fluorides looked very attractive indeed. Wigner filed the idea away.
The Xenon Lesson
On September 27, 1944, at a few minutes after midnight, the B-Reactor at Hanford, Washington, achieved criticality for the first time. The largest nuclear reactor ever built — 28 feet high, 36 feet long, containing 200 tons of uranium in 2,004 horizontal aluminum tubes — went critical on schedule, on budget, having been designed and constructed in less than a year. It was a remarkable industrial achievement.
Within two hours after it reached full power, it began to die.
The power level climbed as planned, then began to fall. The reactor’s operators, following procedures written for a machine that behaved as designed, pushed the control rods further and further out to compensate. The power fell anyway. By early morning on September 28, B-Reactor had shut itself down entirely. The operators were baffled. The design seemed correct. The calculations had been right. But the reactor was dead.
What had killed it was a ghost: a fission product for which no one had adequately accounted, xenon-135. Xenon appeared in the chain of decay products from certain fission events that were quite probable. It has the largest thermal neutron absorption cross-section of any known isotope — 3,085 times larger than that of U-235. A few parts per billion of xenon-135 in the reactor’s uranium was enough to absorb so many neutrons that the chain reaction could no longer sustain itself. The reactor was poisoning itself.
The solution was both elegant and crude: build the reactor with more excess reactivity than the design calculations required, so that even with xenon poisoning at its worst, there would still be enough free neutrons to sustain the chain reaction. The Hanford engineers added more uranium tubes. B-Reactor recovered. The crisis passed.
But the xenon lesson stayed. Everyone who had been thinking about reactor design now understood something new: fission products were not a minor inconvenience to be dealt with periodically when the fuel was replaced. They were an active, ongoing presence inside a running reactor, poisoning the neutron economy in real time. Solid fuel elements provided no mechanism for removing them during operation. The fuel had to be shut down, removed, cooled, dissolved in acid, chemically processed to remove the fission products, and reconstituted into new fuel elements — a massively expensive and complex operation that could only happen periodically.
The fluid-fueled reactor Wigner had been sketching — with its continuously circulating liquid fuel and continuous chemical processing — solved the xenon problem elegantly. If the fuel was liquid and mobile, the xenon could simply bubble out. This was not the only advantage of fluid fuel, but it was perhaps the clearest. Solid fuel accumulated poison. Liquid fuel could be continuously cleaned.
The young Alvin Weinberg, sitting in Wigner’s seminars and listening to Fermi’s analyses and absorbing the xenon lesson from Hanford, was beginning to see the outline of something extraordinary.
A Fifty-Quadrillion-Dollar Discovery
The xenon problem at Hanford had been, for all its terror, a problem of narrow scope. The B Reactor had to produce plutonium; it was poisoning itself; Wigner’s extra tubes allowed it to overcome the poisoning and begin production. One could sketch the sequence on a blackboard in an hour and a working physicist would understand it. The questions that occupied Glenn Seaborg through the autumn of 1944 were different in character. They were questions whose answers, once understood, would not fit on a blackboard and would not be understood in an hour. What Seaborg was beginning to work out, in the ordinary course of his laboratory’s measurement program on the properties of uranium-233, was the shape of an energy future that no one had yet imagined.
The work had begun two months earlier. On the evening of August 2, 1944, a small team of chemists under Leonard Katzin began working after hours at the Met Lab to extract the uranium-233 that had accumulated in thorium carbonate slugs irradiated for many months in the X-10 reactor at Oak Ridge. The quantities involved were absurdly small. Katzin expected to recover approximately five milligrams of uranium — five thousandths of a gram, a quantity that would be invisible to the unaided eye. But five milligrams of U-233 was a thousand-fold increase over what the Berkeley cyclotron had been able to produce. For the first time, there was enough uranium-233 in one place to measure its properties with real precision.
Katzin’s team worked through the first weeks of August. By August 12, they had completed the extraction. The recovered material weighed six milligrams — slightly more than expected. The small excess was the first sign of something the chemists had not fully anticipated.
When thorium is loaded into a reactor, the thorium is never perfectly pure. W. C. Johnson, one of the Met Lab chemists who reviewed Katzin’s work, pointed out that the thorium used in the X-10 irradiation contained a natural uranium impurity at a concentration of approximately 0.05 parts per million. By ordinary chemical standards, 0.05 parts per million is so small a contamination that the material would be considered pure. But when Katzin’s team chemically extracted all the uranium from the irradiated thorium, the natural uranium impurity — which had ridden along through the entire irradiation and extraction process — constituted about ten percent of the total uranium in the sample. The U-233 that Katzin had actually produced was mixed with a non-trivial quantity of ordinary U-235 and U-238.
This was a correction that had to be carried through every measurement. The half-life of U-233, which Katzin’s team had initially calculated assuming pure U-233, came out wrong — 120,000 years. With Johnson’s correction applied, the calculated half-life shifted to 153,000 years, much closer to the actual value (159,000 years) that later measurements would establish. The same correction had to be applied to all the other properties the team was measuring: fission cross sections, capture cross sections, neutron yields. Everything had to be recalibrated to account for the ninety percent of the material that was actually what it claimed to be, and the ten percent that was something else.
On September 9, 1944, a mass spectrographic analysis finally gave the team a precise measurement of the sample’s isotopic composition: U-233 was 85 ± 1.5 percent of the total uranium. The remaining 15 percent was natural uranium contamination, slightly higher than Johnson’s calculation had suggested. With the composition known, all the earlier measurements could be corrected definitively.
In late September, Seaborg made another trip to Montreal. The British-Canadian “Evergreen” group at the Montreal Laboratory had its own interest in uranium-233, pursued as part of the broader British nuclear program that operated in parallel with — and at times in tension with — the American Manhattan Project. Seaborg spent several days in Montreal discussing the chemistry of U-233 production, the chemical forms in which thorium could be irradiated, and the extraction techniques that could be used to separate uranium from irradiated thorium. One technique discussed was the use of fluorine gas to volatilize uranium away from thorium — fluoride volatility, a chemistry that would later become central to the molten-salt reactor program at Oak Ridge and that would return to prominence in the twenty-first century as a reprocessing method for used nuclear fuel.
Seaborg returned to Chicago. Through October, the measurements continued. By the end of the month, the Met Lab had enough data on U-233 to begin drawing conclusions.
On November 1, 1944, Seaborg sat down to review what the measurements showed. The figures on his desk, extracted from the laboratory’s corrected analysis of the Katzin material, were these. The thermal fission cross-section of U-233 was approximately 550 barns — similar to U-235. The total neutron absorption cross-section was 620 barns, of which the non-fission absorption was only 70 barns. The ratio of non-fission absorption to fission — the parameter the physicists called alpha — was 0.12. This was exceptionally low. For U-235, alpha was about 0.19; for plutonium-239, considerably higher still.
The low alpha meant that when a U-233 nucleus absorbed a thermal neutron, the overwhelming majority of the time it fissioned. Very few of the absorbed neutrons went into the non-productive formation of U-234. And when U-233 fissioned, it released on average slightly more neutrons per fission than U-235 did — 1.07 times more. Combining the two effects — the low alpha and the slightly higher neutron yield — meant that the number of neutrons released per neutron absorbed in U-233 was 1.13 times the value for U-235.
The number mattered because it determined whether breeding was possible. In a thermal-spectrum reactor, each fission must produce at least one neutron to sustain the chain reaction. For breeding, each fission must also produce at least one additional neutron to convert fertile material into new fissile material. And any real reactor will lose some neutrons to structural absorption, to leakage, to parasitic capture in fission products that accumulate as the fuel burns. The margin available for these losses is the neutrons-per-absorption value minus two.
For U-235 in a thermal reactor, the neutrons-per-absorption value was about 2.07. The margin for breeding was 0.07 — barely positive. Any real reactor would lose more than 0.07 neutrons per fission to leakage and structural absorption, and a U-235 thermal reactor could not, as a practical matter, breed.
For plutonium-239, the thermal value was worse — about 2.04. A plutonium thermal breeder was physically impossible. Plutonium could breed only in a fast neutron spectrum, which brought its own engineering difficulties.
For U-233 in a thermal reactor, the value Seaborg was now calculating was 1.13 times the U-235 figure — about 2.34. The margin for breeding was 0.34, an order of magnitude larger than for U-235 and more than ten times larger than for plutonium in a thermal spectrum. A U-233 thermal breeder was not merely possible. It had substantial engineering margin. A real reactor, with real neutron losses, could still breed.
Seaborg wrote in his diary entry for November 1, 1944:
I reviewed recent results on the physical constants of U-233. It appears that the fission cross section is about the same as for U-235 (550 barns); the total neutron absorption is only around 620 barns, giving a favorably small value for the parameter alpha (0.12). The number of neutrons per fission is 1.07 times the U-235 value. The number of neutrons given off per neutron absorbed is 1.13 times the U-235 value. These values are within the range to enable U-233 to be made from thorium by a chain reaction on the U-233 (i.e., to make breeding possible) — extremely important because it may make it possible to be independent of uranium once a supply of U-233 for starting purposes is on hand.
The entry is restrained in the way that laboratory notebooks are restrained. What it describes, in the terms its author was trained to use, is the discovery that natural thorium could be made into an essentially inexhaustible source of energy — that the world’s thorium reserves, which were three or four times larger than its uranium reserves and which had until then been regarded as mineralogically unremarkable, could, once a seed quantity of U-233 was in hand, be converted to fissile material at a rate that would sustain, and over time exceed, any rate of consumption.
What Seaborg was seeing, as he worked through the cross-section arithmetic on that first of November, was a fuel cycle that required no enrichment, no large-scale isotope separation, no rare material. Thorium was abundant. A breeder reactor fueled with U-233 and blanketed with thorium could, in principle, convert thorium to U-233 at a rate that kept pace with the rate at which U-233 was fissioned in the core. The reactor would be, from a resource standpoint, self-sustaining. The fuel it needed to continue operating was being generated by its own operation. The only input from outside was thorium.
The scale of what this meant can be rendered in any number of ways. Later analyses would estimate that the thorium contained in the top hundred meters of the Earth’s continental crust, if fully utilized in breeder reactors, could supply humanity’s entire energy demand for roughly a hundred thousand years. The thorium content of the United States alone, recoverable from existing mine tailings and monazite deposits, represented several millennia of American energy consumption. None of this was economically unavailable; none of it required elaborate extraction; none of it was geographically concentrated in unstable regions. Thorium was a commodity. It was available.
Seaborg understood this on November 1, 1944. He was, as far as we can establish from the documentary record, the first person to fully understand it. The fuel cycle had been sketched in outline by Gofman at Berkeley in 1941. The possibility of breeding had been proposed by Wigner. The superior eta of U-233 had been suspected for some months before Katzin’s extraction made precise measurement possible. But the full synthesis — the moment when the measurements, the fuel-cycle physics, and the resource implications all came together in one person’s mind — belongs to Seaborg, in the laboratory at Chicago, on the first of November 1944.
Many years later, in a remark preserved in John Gofman’s oral history, Seaborg would describe the November 1, 1944 insight as a “fifty-quadrillion-dollar discovery.” The number was a back-of-the-envelope estimate of the value of global thorium reserves if fully utilized in breeder reactors. By any reasonable accounting, it was correct. Nothing in human history — no agricultural revolution, no industrial revolution, no discovery of oil, no domestication of electricity — had represented a resource expansion of that order. Thorium in breeder reactors, if the technology could be built, was an energy endowment unlike any previous endowment.
What Seaborg did with this recognition in November 1944 was: very little. It was wartime. The Manhattan Project existed to build weapons, not to breed energy. The staff of Seaborg’s section was fully committed to plutonium chemistry. The Katzin extraction team would continue measuring U-233’s properties over the coming months, but the work was a measurement program, not a reactor program. And the U-233 program at the Met Lab had, as we have seen in the preceding chapter, been cancelled by Arthur Compton the previous November, a cancellation that had been made on explicit grounds of resource allocation and that had not been reversed.
So Seaborg made his diary entry. He continued his work on plutonium. The cross-section measurements of U-233 were published in the normal channels, classified but accessible to the relevant parts of the Manhattan Project. Wigner and Weinberg, who would use those measurements the following spring to design the first fluid-fuel thorium breeder reactor on paper, would read Seaborg’s numbers and draw their own conclusions.
The fifty-quadrillion-dollar discovery would wait. It would wait, at various levels of dormancy and revival, for the next eighty years — an interval during which the civilian nuclear-power industry that grew out of the Manhattan Project would commit itself almost entirely to the other fuel cycle, the one that fissioned U-235 and bred plutonium, the one that did not require thorium and did not produce U-233. The thorium cycle Seaborg had recognized in 1944 would be pursued, in its most developed form, at Oak Ridge in the 1950s and 1960s. It would be terminated in 1976. It would be revived as a topic of public discussion in the early years of the twenty-first century. And even now, it is beginning, finally, to be pursued as the commercial technology that its original discoverers recognized it could be. It has been a long path from November 1944 to the present day.
The April Meeting
On April 26, 1944, roughly a year before Germany’s surrender and sixteen months before Hiroshima, a group of the most brilliant nuclear physicists in the world gathered in Chicago to discuss the postwar future.
The meeting, recorded in the classified document MUC-LAO-17, was chaired by Robert Christy. The attendees included Fermi, Wigner, Szilard, Morrison, and a young theorist named Alvin Weinberg, among others. The nominal subject was the properties of heavy elements, but the real subject was the world after the bomb.
Philip Morrison, twenty-eight years old and already one of the most incisive minds in the group, presented something that no one had quite put together so starkly before: a global survey of fissile and fertile material resources. Uranium, Morrison reported, was not abundant. Known deposits would fuel the reactors the Manhattan Project was building for a generation at most. But thorium was different. Thorium was roughly ten times more abundant in the Earth’s crust than uranium. If the world’s energy economy was going to run on nuclear power in the long term — and Morrison and Wigner and Weinberg all believed it would — thorium was the key material. The world had enough thorium to power civilization for hundreds of thousands of years.
“There should be more work on the nuclear development of thorium,” Morrison said at the meeting.
Samuel Allison, one of the senior physicists present, added a refinement: if thorium were used as a blanket material in a fast breeder reactor — surrounding the fissile core and absorbing neutrons — it would be transmuted into U-233, which could then fuel thermal reactors. The fast breeder and the thermal thorium reactor could work together in a complementary system.
The meeting record shows both suggestions — more thorium work, thorium as blanket material — being raised explicitly. What it does not show is any decision to act on them. The war required plutonium. Thorium would wait.
Two days later, on April 28, the group met again (MUC-LAO-18). Fermi presented a more systematic analysis of breeding possibilities in various reactor systems. His conclusions were careful but clear: the thermal neutron spectrum offered better breeding ratios for U-233 than for plutonium. This was the physics confirming what Wigner had suspected. But Fermi also noted the practical difficulties of the thorium cycle, and the meeting moved on.
In July, the group met once more (MUC-LAO-30). This time, the discussion turned more explicitly to postwar power reactors. What would the world’s civilian nuclear power industry look like in twenty years? Several different reactor concepts were presented. Thorium breeders appeared on the list, alongside heavy-water reactors and fast plutonium breeders.
The meetings were a kind of intellectual inventory: everything that nuclear physics knew about future power reactors, laid out simultaneously by the people who knew it best. Reading these documents eighty years later, it is striking how clearly the participants saw the full landscape — every reactor type, every fuel cycle, every advantage and disadvantage. The thorium cycle’s merits were understood. The abundance of thorium was known. The superior eta of U-233 was being measured. Nothing that would be “discovered” by proponents of thorium energy decades later was unknown to these men.
And yet, within months, U-233 research at the Met Lab would be cancelled outright. The path not chosen was chosen against actively and consciously — not from ignorance but from the iron logic of military necessity.
The Fatal Flaw
Even as the cancellation order was being issued, Seaborg was thinking about a problem that the U-233 advocates had not fully reckoned with: uranium-232.
U-232 was an isotope that formed inevitably as a contaminant whenever U-233 was produced in a reactor. It formed through three distinct pathways: fast neutrons could knock a neutron out of U-233 directly, producing U-232; protactinium-231 — a natural contaminant in any thorium deposit that had been near uranium — would absorb a neutron to form Pa-232 and then decay to U-232; and thorium-232 itself, when struck by fast neutrons, could produce Pa-231, which could absorb another neutron to give U-232.
These three formation pathways made it essentially impossible to produce U-233 free of U-232 contamination. And U-232, with a half-life of only 69 years, decayed rapidly through a chain of daughter products that included several intense gamma emitters. The gamma radiation from U-232 daughters would make any weapon material containing significant U-232 extremely difficult to handle and would make even a partially assembled weapon dangerous to the assemblers.
In December 1944, Seaborg wrote a ten-page report — “Conversion of Pu-239 to U-233” — in which he laid out the U-232 problem with characteristic precision. He didn’t bury the bad news. “Of particular concern,” he wrote, “is the presence of U-232 that has four short-lived (hence long-range) alpha-particle decay products, which means that the problem of purifying the material from light element impurities assumes importance again as it originally did in the case of Pu-239.”
The parallel to plutonium was deliberate and damning. The Manhattan Project had already suffered one near-catastrophic discovery about plutonium: that reactor-produced plutonium was contaminated with Pu-240, which underwent spontaneous fission at a rate high enough to pre-detonate any gun-type weapon design. The entire implosion program at Los Alamos — with all its complexity and cost and uncertainty — existed because of this discovery. Seaborg was warning that U-233 might have an analogous problem: that U-232 contamination would make it as difficult to use in weapons as Pu-240 made reactor plutonium.
He was right. Weinberg, who was present for the discussions and later recalled them vividly, wrote: “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.”
The U-232 problem effectively ended serious consideration of U-233 as a weapons material. Plutonium, despite its own contamination problem, was further along and had a solution — implosion — already under development. U-233 had neither the head start nor the institutional momentum.
What the U-232 problem did not do was eliminate U-233 as a reactor fuel. Quite the contrary. The same gamma radiation that made U-232 dangerous in a weapon made U-233-bearing material self-safeguarding against theft or diversion — any significant quantity would announce itself to radiation detectors. More importantly, the U-232 problem was irrelevant to the reactor physics that made U-233 attractive: its superior eta, its excellent breeding potential in the thermal spectrum, its chemical compatibility with fluoride salt chemistry. For civilian power generation, U-232 was not a fatal flaw. It was a feature.
But the distinction between weapons and reactors was not one that the Manhattan Project had much interest in making in 1944. The program existed to build weapons. Materials that were useful for weapons received resources. Materials that were useful for reactors but not weapons did not.
The Document That Began Everything
On May 17, 1945, six weeks after Franklin Roosevelt died in Warm Springs, Georgia, three weeks before Germany surrendered, three months before the Trinity test, two young physicists at the Metallurgical Laboratory in Chicago sat down and wrote a document.
Eugene Wigner was fifty-two years old. Alvin Weinberg was twenty-five. Their colleague Gale Young was also twenty-five. Together, they had written what they called “Preliminary Calculations on a Breeder with Circulating Fuel” — document MUC-EPW-134 in the classified files of the Manhattan Project.
The document described a reactor in which uranium-233 was dissolved in heavy water and circulated through a core. Fission would take place as the fuel circulated. The heat would be extracted by a heat exchanger. Fission products — including xenon-135 — would either be removed by chemical processing or, in the case of gaseous products, would simply bubble out of the liquid fuel before they could poison the chain reaction. New U-233 could be produced by surrounding the core with thorium, which would absorb the excess neutrons and breed new fuel.
The document was careful and precise. Wigner and Weinberg had calculated the neutron losses to xenon and other fission products, the losses to structural materials, the losses to leakage from the core, and had concluded that with U-233’s favorable eta, the numbers worked. A thermal breeder operating on the thorium cycle was physically possible.
The key sentence was brief but exact. In explaining why the fuel had to be liquid and circulating rather than solid and stationary, Wigner wrote: “The most definite reason for using the 23 [uranium-233] in solution is the need for eliminating the Xe-135.”
Xenon-135. The same poison that had nearly killed B-Reactor at Hanford. The same poison that would have to be reckoned with in any solid-fueled reactor for the rest of nuclear history. In a fluid-fueled reactor, it was not a problem at all. The xenon bubbled out.
This document — six pages, hand-typed, classified — is the founding text of the molten salt reactor concept. Its core insight was that the problems of solid-fueled reactors were not universal problems of reactor physics but specific problems of a particular design choice. Change the design — make the fuel liquid, make it mobile, let it circulate — and the problems dissolved with it.
Weinberg understood this. In the weeks that followed, as Germany surrendered and Japan fought on and the Los Alamos team prepared its test shot for the New Mexico desert, he sat with this idea and turned it over. He recognized that a circulating-fuel breeder based on thorium and U-233 was not just a theoretical possibility but potentially the solution to a problem that no one had yet fully articulated: how to fuel an industrial civilization over centuries and millennia, not merely decades.
He had caught what he would later call, with characteristic wry understatement, “the homogeneous bug.” He never recovered.
The Summer of Reckoning
August 6, 1945. Hiroshima.
Glenn Seaborg learned the news on a train traveling west. He was returning to Berkeley after more than three years in Chicago, heading home to begin the next chapter of his career as a professor and researcher in the peacetime university. He kept his thoughts to himself as the train crossed the plains and mountains, sitting in the club car, watching other passengers react to what the radio and newspapers were saying.
He had spent three years making the plutonium that in three days would destroy a city. In his journals, he was characteristically measured about it. He had always believed that the weapon would shorten the war and save lives on balance — a judgment that still divides historians. What he understood with perfect clarity was that the world had changed irrevocably.
At Oak Ridge, Tennessee, Alvin Weinberg heard the news differently. He had been in this city — this secret city that still appeared on no map — for only three months, having boarded a train from Chicago with his wife Marge and their two-year-old son David on a hot May weekend just before Germany’s surrender. He had come to Oak Ridge with a specific mission: to help turn the wartime laboratory at Clinton into a peacetime research center, and to develop the fluid-fueled thorium breeder reactor that he and Wigner had sketched in document MUC-EPW-134.
The bomb that fell on Hiroshima confirmed, for Weinberg, the urgency of finding a better path. Nuclear weapons were now real. They would proliferate. The only answer to nuclear weapons was nuclear knowledge — understanding what the technology could do, building the reactors that would allow humanity to use nuclear energy for purposes other than destruction, creating the fuel cycles that would make war over uranium unnecessary.
The aqueous homogeneous circulating-fuel power breeder based on thorium and U-233, Weinberg wrote in his memoir, “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.”
He was thirty years old. He would spend the rest of his working life in that city, pursuing that obsession.
The Princeton Pamphlet
On August 12, 1945, three days after the Nagasaki bomb, the United States Army Corps of Engineers released a document that had been prepared for this moment since the previous autumn. Its official title was Atomic Energy for Military Purposes. It was a 227-page technical narrative written at Major General Leslie Groves’s request by a physics professor from Princeton named Henry DeWolf Smyth. Groves had considered it prudent to tell the world, within days of the weapons’ use, what the Manhattan Project had been and how it had built them — or at least to tell the world what Smyth and Groves thought the public should know. Princeton University Press published it immediately as a public service. It sold for $1.25. Copies reached bookstores within weeks. It was known, almost from the moment of its publication, as the Smyth Report.
The report did several things at once. It informed the American public that their government had built a new kind of weapon by industrial means of unprecedented scale. It provided a technical outline — carefully bounded by what Groves thought could be disclosed without aiding an adversary — of the physics and engineering that had made the weapons possible. It established, as no other document could, the narrative through which the Manhattan Project would be remembered: Bohr arriving in America in January 1939 with the news of fission from Meitner and Frisch, Fermi’s Stagg Field reactor in December 1942, the Hanford production piles producing plutonium for Trinity and Nagasaki. Almost every subsequent popular history of the bomb project has its structural bones taken from the Smyth Report.
In paragraph 2.21 of the report, deep in the second chapter, Smyth considered the question of alternative materials.
All our previous discussion has centered on the direct or indirect use of uranium, but it was known that both thorium and protoactinium also underwent fission when bombarded by high-speed neutrons. The great advantage of uranium, at least for preliminary work, was its susceptibility to slow neutrons. There was not very much consideration given to the other two substances. Protoactinium can be eliminated because of its scarcity in nature. Thorium is relatively plentiful but has no apparent advantage over uranium.
Three sentences. Every one of them was true, in a narrow sense, and the overall impression they created was entirely false.
By August 1945, Oak Ridge had been irradiating thorium for more than a year, converting it to uranium-233 in measurable quantities. Glenn Seaborg’s December 21, 1944 memorandum had analyzed the three distinct pathways by which uranium-232 contaminated U-233 produced from thorium and had proposed remedies. The April 1944 meetings at the Met Lab — documents MUC-LAO-17, MUC-LAO-18, and MUC-LAO-30 — had, with Philip Morrison leading the discussion, established thorium’s abundance in monazite sands and the thermal-spectrum breeding advantage conferred by U-233’s eta value of 2.4 versus plutonium’s 2.05. On May 17, 1945 — three months before the Smyth Report’s release — Eugene Wigner, Alvin Weinberg, and Gale Young had completed a memorandum, MUC-EPW-134, proposing a civilian power reactor using molten fluoride salts dissolving uranium and thorium, with continuous fission-product removal, operating on the thorium-U233 cycle. This document laid out, in 1945, essentially the reactor that Weinberg’s team at Oak Ridge would spend the next twenty-five years trying to build.
None of that was in the Smyth Report. A reader in Philadelphia or London or Moscow, reading what the United States government had chosen to publish about what it had discovered, came away with the impression that thorium had been considered during the war, found to offer no apparent advantage, and set aside. The fluid-fuel thorium thermal breeder that Wigner had proposed in May 1945 was not mentioned. The thorium-U233 cycle that Seaborg had recognized in April 1941 as “sufficiently long-lived to be a practical source of nuclear energy” was not mentioned. The public record of what the Manhattan Project had learned about alternative fuel cycles consisted, for the world’s reading public in 1945, of those three sentences.
Smyth’s choice was defensible on its own terms. The Manhattan Project had not, in fact, built a reactor to produce U-233 during the war. The bomb at Trinity and the bomb over Nagasaki had been plutonium weapons. The wartime technical achievement Smyth was tasked with describing was the plutonium weapon, and the wartime discussion of thorium had indeed concluded, correctly, that thorium offered no advantage for the immediate problem of making a uranium-235 or plutonium bomb in time to use against Japan. But a passage written to describe that immediate wartime choice became, by virtue of its placement in the foundational public document of the nuclear age, the accepted characterization of what thorium was good for.
The institutional forgetting of thorium that this essay traces began in August 1945, on a page in a cheap pamphlet published by Princeton University Press. Everyone who came later — Lilienthal, Oppenheimer, Rickover, Cisler, the engineers who designed Shippingport, the utility executives who ordered pressurized-water reactors by the dozen in the mid-1960s — inherited the baseline that the Smyth Report had set. When Walker Cisler’s engineers at Lake Angelus in November 1951 concluded that a fluid-fuel thorium breeder was the best long-term approach to civilian nuclear power, they were, in effect, contradicting the Smyth Report. So was Alvin Weinberg every time he argued for the molten-salt program from inside Oak Ridge. So was any engineer or policymaker through the long decades after who questioned why the United States had not built what Wigner had proposed.
The technical truth had always been available. It was buried in classified reports and in Met Lab seminar minutes that almost nobody read. The public truth, as the war ended, was the Smyth Report. And the public truth said thorium had no apparent advantage over uranium.
The End of the Beginning
On June 14, 1946, the day after the United States conducted its first postwar nuclear test at Bikini Atoll, a Navy captain named Hyman G. Rickover landed at the Oak Ridge airport aboard General Kenneth Nichols’s personal aircraft. He was forty-six years old, a 1922 Naval Academy graduate from Chicago, short and combative and possessed of a technical precision that was unusual in the officer corps and sometimes maddening to those who worked with or for him. He had been a captain since 1942 and was, by his own private assessment in 1946, near the end of a career that had proven his competence without winning him flag rank. He was at Oak Ridge, he believed, because Admiral Earle Mills at the Bureau of Ships had decided the Navy needed officers who knew something about nuclear reactors, and Rickover was the best candidate Mills could think of who was also available.
The assignment was to learn. The Navy had sent a group of six officers and three civilians, led by Rickover as the senior officer: Lieutenant Commanders Louis Roddis, James Dunford, and Miles Libbey, along with Lieutenant Raymond Dick and the civilians Amorosi, Emerson, and Blizard. They arrived over the following few days. Their role was to be embedded in the Clinton Laboratories — which was running, on a modest scale, a project under Farrington Daniels to design a high-temperature gas-cooled power reactor — and to pick up what they could about reactor physics and engineering. They had no defined assignments beyond this. In the Army’s tidy organization chart, Rickover was nominally deputy to Colonel Walter Williams, director of operations; in practice, within a week of his arrival, Rickover had found a private office at the laboratory and was ignoring administrative duties entirely.
The freedom was what he wanted. During the first months at Clinton, Rickover spent his time reading technical reports, questioning engineers, attending every informal lecture available, and signing up for every course the laboratory offered. By September 1946 the junior officers had all arrived, and Rickover began to operate them as a unit. He persuaded the Army authorities to let him write the officers’ fitness reports — the reports that would determine their future Navy promotions — even though the Bureau of Ships had not explicitly given him that authority. The officers understood from this what kind of commander they had. Through the fall and winter, the Navy group systematically worked through the technical literature, attended every Clinton laboratory lecture, wrote detailed written reports summarizing what each lecture or course contained, and began assembling a compendium of reactor technology more complete than anything available elsewhere in the Navy.
Rickover’s first written assessment, in a November 11, 1946 memorandum to Mills, was sobering. A nuclear propulsion plant for a submarine would take five to eight years to build with existing resources. The obstacles were not theoretical. The physics of fission was understood well enough. The obstacles were engineering: materials that could tolerate prolonged neutron bombardment without degrading, coolants that could remove reactor heat reliably and transport it to a useful steam cycle, heat exchangers and pumps and valves that would not leak or fail under service conditions, and shielding compact enough to fit inside a submarine hull. Nothing the wartime Manhattan Project had built was directly applicable. The Hanford production reactors had been designed to make plutonium, not to produce shaft horsepower; their shielding was thick because the reactors were the size of buildings, and nobody had needed to make a reactor small. The Navy reactor would require original research into every aspect of reactor engineering that an actual shipboard plant would face.
Through the spring and summer of 1947, as the Atomic Energy Commission was establishing itself in Washington after taking over from the Army on January 1, Rickover extended the Navy group’s reach beyond Clinton. He and the senior officers began a systematic tour of the Commission’s installations: Ames Laboratory at Iowa State College, Argonne outside Chicago, the Radiation Laboratory at Berkeley, Los Alamos in New Mexico. At each site Rickover sought out every scientist and engineer who had views on reactor technology and asked them the same questions. What kind of reactor should be built first? What were the coolant options? What materials would stand up to the neutron flux? How long would it take, with what budget, using what contractor? The replies varied widely, but three impressed him particularly. At Argonne, Walter Zinn — the nation’s foremost reactor expert, who had been Fermi’s principal assistant at the Stagg Field pile — favored a slow-neutron reactor using either water or helium as the heat-transfer medium, and argued for a land-based prototype to be built as soon as the design was solid. At Berkeley, Ernest Lawrence told Rickover the Navy should budget a hundred million dollars for a submarine reactor project, hire a major industrial contractor, and aim for a land-based prototype. “Real cash” was Lawrence’s phrase. At Los Alamos, Edward Teller said a power reactor could be built in two years if someone actually put the effort into it, and that the first one should be simple — uneconomical but workable — because building it would be the necessary step toward everything else. Teller also wrote to the Joint Research and Development Board that the Navy had an exceptional man in Rickover and should not lose him.
Rickover wrote his own summary of the tour in a long memorandum to Mills on August 20, 1947. His observation was pointed. “It is significant that during our entire tour,” he wrote, “of the many scientists contacted, not one was found who had a definite interest in and was working on the problem of furthering nuclear power.” The universities and the Commission’s laboratories were focused on research, the military on weapons. Civilian power reactors were not being pursued by anyone with real urgency. If the Navy wanted a submarine reactor, it would have to build the organization that pursued it. Rickover recommended that his own group be formally constituted as the Navy’s nuclear propulsion project within the Bureau of Ships, with authority over both the reactor development and the shipbuilding program, and that more young naval officers be assigned to the Commission’s laboratories to staff the work. When Mills did not respond, Rickover wrote again a week later proposing a less ambitious alternative: that his group at least be split between the Bureau of Ships and the Commission, representing the Navy’s interests inside the Commission’s own organization. That was the version Mills eventually accepted, though in a form diluted enough that by September 1947 Rickover’s group had been dispersed into different offices, the idea of an independent Navy nuclear project had been set aside, and Rickover himself was assigned as staff to Mills with no clear portfolio.
Sometime during that same autumn, back at Oak Ridge on one of his frequent visits, Rickover had the conversation that quietly determined everything that followed. The Daniels reactor project was in technical trouble. Farrington Daniels and his associates had been designing a helium-cooled high-temperature reactor for eighteen months; the difficulties were accumulating faster than they were being solved. Monsanto, which had been operating Clinton since the end of the war, was preparing to give up its contract. Harold Etherington, who led the Clinton reactor division, was looking for a project with a future. Rickover told Etherington bluntly that the Daniels reactor was going to be cancelled, that the Clinton engineering team should not waste itself on a dying project, and that they should instead begin work on a reactor that could actually propel a ship. What Rickover had in mind was specific. Eighteen months earlier, in April 1946, Alvin Weinberg and Forrest Murray had written a short classified report — AEC document Mon P-93, High Pressure Water as a Heat Transfer Medium in a Nuclear Power Reactor — proposing that ordinary water, held at high pressure to prevent boiling, could serve as both the moderator and the coolant in a power reactor. At the time the paper was written, Weinberg had regarded it as one of several interesting possibilities, not as his preferred approach. Weinberg’s own preference, then and throughout his career, was the fluid-fueled thorium breeder. But Weinberg’s April 1946 paper described, in rough outline, the technology Rickover now believed was the only one he could build on a Navy schedule with a Navy crew.
A pressurized-water reactor would use highly-enriched solid uranium fuel, water moderation, water cooling, and a conventional steam cycle. The physics could be calculated. The materials could be tested. The components could be ordered from existing industrial suppliers. Nothing was easy, but everything was known to be possible, which was more than could be said for any other reactor concept available in 1947. By late 1947, Rickover had persuaded Etherington and the Clinton reactor division to begin shifting their attention, informally at first, from the Daniels reactor toward Weinberg’s pressurized-water concept.
Rickover looked also at the fluid-fuel work that Weinberg himself was doing. He understood what Weinberg was building. He understood the advantages: no xenon problem, continuous processing, higher breeding ratios with U-233, better long-term fuel economy, fuel temperatures and power densities that a submarine could never achieve with solid fuel. He also understood the disadvantages: unknown materials behavior under radiation at operating temperatures, unknown corrosion chemistry between molten salts and structural alloys, unknown reliability of mechanical systems pumping radioactive liquids at high temperature. Everything about the fluid-fueled approach was harder to predict than everything about solid-fuel pressurized water. The gap was not permanent. With enough time and enough research funding, the fluid-fuel systems would become as engineered and predictable as the pressurized-water system would be. But in 1947, they were not. Rickover did not need the best long-term reactor. He needed a submarine reactor, built on a schedule that mattered for the Cold War, using materials and methods his engineers could specify with confidence.
He chose pressurized water.
He did not choose it because it was the best civilian power reactor. He chose it because it was the best submarine reactor that could be built on a five-year schedule. In that narrow judgment, at that narrow moment, he was correct. The USS Nautilus would go to sea in 1955, the first nuclear-powered submarine in history, and the pressurized-water reactor it carried would transform naval warfare. The civilian industry that followed would discover, over the subsequent thirty years, the less happy consequences of the fact that its entire technology base had been built for a submarine.
By the end of 1947, the pressurized-water reactor had the informal backing of the Clinton engineering division, of Rickover’s Navy group, and of key Commission scientists who had begun to believe that a Navy-led project might produce the first American power reactor. What it did not yet have was institutional form. That would require another year of bureaucratic construction in Washington — the formal establishment of the Naval Reactors Branch jointly within the Bureau of Ships and within the Commission’s Division of Reactor Development, a dual organization Rickover would eventually head. But the technological commitment had been made. The reactor that would power American civilian nuclear energy for the rest of the twentieth century had, in essence, been chosen in the fall of 1947 by a forty-six-year-old captain who had been passed over for flag rank.
Weinberg would continue his work on the fluid-fueled thorium breeder at Oak Ridge. The Clinton engineers who had shifted to pressurized water would follow the project Rickover built around it — first to Argonne, then in 1949 to the new Bettis Atomic Power Laboratory that Westinghouse would build near Pittsburgh specifically for Navy reactor work. The parting had taken place, though the men involved did not yet see it as a parting. Oak Ridge would pursue fluid fuel. Bettis would pursue pressurized water. The thorium line and the uranium line, which had diverged in the Metallurgical Laboratory four years earlier when Compton cancelled U-233 research, were now diverging again in a different form. This time the split would be engineering rather than physics, and it would last.
Epilogue: The Thorium in the Cans
On May 15, 1946, Seaborg convened the last meeting of the Heavy Isotopes Group at the Met Lab. He was about to leave for Berkeley permanently. Several of his colleagues were also departing, scattered back to universities and laboratories around the country. The wartime team was dissolving.
Someone at the meeting noted that 80 pounds of thorium carbonate were sitting in irradiated cans in a storage facility, awaiting chemical processing to extract the U-233 they contained. The cans had been there for months. No one was sure what would happen to them now.
Seaborg filed the note and departed for Berkeley.
The cans of thorium sat in storage. The knowledge of what they contained — and what that meant for the energy future of the world — was classified. The four-man team that had worked on U-233 dispersed. The documents reporting on the superior breeding potential of the thorium cycle went into the classified files of the AEC.
The world that emerged from the war was obsessed with plutonium: building more of it for weapons, eventually learning to use it for civilian power, spending billions on a fast breeder reactor that would breed even more plutonium from uranium-238. The thorium cycle, recognized by the most brilliant physicists of the Manhattan Project era as offering a superior path to long-term energy abundance, would wait in those classified files for decades.
Weinberg was in Oak Ridge with his obsession. Seaborg was in Berkeley with his periodic table. Wigner would leave Oak Ridge within two years, frustrated by administrative conflicts, and return to Princeton. Gofman would leave atomic energy entirely and become a prominent critic of radiation health standards.
And in Washington, the men who were building the new Atomic Energy Commission — who had inherited the weapons program, the production reactors, the classified files, and the institutional habits of the Manhattan Project — were deciding what civilian nuclear power would look like. They had an answer. It looked like the path they had already chosen. Within three years, the institutional weight behind that path would grow enormously, as a congressional staff director named William Borden, a senator named Brien McMahon, and a newly-appointed AEC chairman named Gordon Dean began building the largest weapons-production complex in American peacetime history — a complex that would shape every subsequent American decision about civilian nuclear power for the rest of the century.
The path not taken was already receding into the past.
Notes on Sources
This essay draws primarily on Glenn Seaborg’s wartime journals, compiled in five volumes published by Lawrence Berkeley National Laboratory between 1976 and 1980: Early History of Heavy Isotope Research at Berkeley (1976) and History of Met Lab Section C-I, Volumes 1 through 4 (1977–1980). These documents — Seaborg’s day-by-day account of the work in room 307 and at the Metallurgical Laboratory — are the only first-person narrative record of the thorium cycle’s discovery and early development. They had not been previously examined along this line of inquiry before Kirk Sorensen’s 2014 University of Tennessee thesis, Thorium Research in the Manhattan Project Era, from which this account draws extensively.
The meeting minutes MUC-LAO-17 (April 26, 1944), MUC-LAO-18 (April 28, 1944), MUC-LAO-30 (July 6, 1944), and MUC-EPW-134 (May 17, 1945) are primary source documents obtained from the Oak Ridge National Laboratory library and reproduced in Sorensen’s thesis. The founding sentence of the fluid-fueled reactor concept — “The most definite reason for using the 23 in solution is the need for eliminating the Xe-135” — appears in MUC-EPW-134.
Alvin Weinberg’s account of his arrival at Oak Ridge and his subsequent decades of work appears in his autobiography The First Nuclear Era: The Life and Times of a Technological Fixer (AIP Press, 1994). His 1997 speech to a Korean scientific delegation — “the MSR protohistory” — provides the clearest retrospective account of how and why the fluid-fueled reactor concept took shape at Oak Ridge.
The official history of the Manhattan Project era, The New World, 1939–1946 by Richard Hewlett and Oscar Anderson (Penn State University Press, 1962), provides institutional context but gives scant attention to the thorium cycle. Richard Rhodes’s The Making of the Atomic Bomb (Simon & Schuster, 1986) — the definitive narrative account of the same period — likewise gives thorium minimal treatment. The story told here, grounded in Seaborg’s primary sources, is a story those accounts did not tell.
The account of Rickover’s year at Oak Ridge (1946-1947) draws from Francis Duncan and Richard Hewlett, Nuclear Navy, 1946-1962 (University of Chicago Press, 1974), Chapter 2 (“The Idea and the Challenge”), particularly the sections “The Oak Ridge Assignment,” “The Navy Team,” “A Call for Action,” and “Reconstruction.” Duncan’s account, based on Navy reactor records and interviews with Rickover and surviving members of the original Navy group, provides the detailed chronology of the Navy group’s arrival in June 1946, the composition of the original team (Dunford, Libbey, Roddis, Dick, and the three civilians Amorosi, Emerson, and Blizard), Rickover’s systematic study method, and the summer 1947 tour of AEC installations that ended with Rickover’s August 20, 1947 memorandum to Admiral Mills. The conversation between Rickover and Harold Etherington in the fall of 1947, in which Rickover persuaded the Clinton reactor division to shift from the Daniels gas-cooled reactor toward pressurized water based on Alvin Weinberg’s April 1946 paper (AEC document Mon P-93), is also from Hewlett and Duncan. This detail is consequential: the pressurized-water reactor that Rickover eventually built for the Mark I prototype in Idaho and the USS Nautilus drew, for its core concept, on a paper Weinberg had written as one of several alternatives before he committed himself to the fluid-fueled thorium breeder.