Apart from the plutonium plant at Hanford, the heart of our effort to produce material for a fission bomb was Oak Ridge. Here were located all our uranium separation plants—the plants designed to separate the easily fissionable Uranium-235 from the more abundant but much less fissionable isotope, Uranium-238.
There were a number of ways we thought this could be done, but for practical reasons, to suit our immediate purposes, they were whittled down to two, the electromagnetic process and the gaseous diffusion process.1 The construction for these was authorized late in 1942, at about the same time that we gave the go-ahead signal for the plutonium process which was put at Hanford. In 1944 we decided to build a thermal diffusion process plant, which was also placed at Oak Ridge.
A full discussion of the Oak Ridge plants and the research and theory behind them would take volumes. There is room here to give only a sketchy idea of that enormous, security-hedged complex known as the Clinton Engineer Works and to mention briefly a few of the main troubles and headaches that plagued us in our attempt to turn out material for a weapon that might end the war.
We had decided at the start that the several uranium process plants at Oak Ridge should be well separated, so that in case a disaster struck one it would not spread to or contaminate the others. For that reason, the electromagnetic and gaseous diffusion plants were located in valleys some seventeen miles apart. Later, when the thermal diffusion plant was built, we had to disregard this policy and put it quite near the steam-generating plant for the gaseous diffusion process, in order to take advantage of its supply of extra steam.
The electromagnetic plant (Y-12, to give it its code name) was built in a restricted area of about 825 acres in the central-southeastern part of the reservation, approximately five miles from the commercial district of Oak Ridge, which was the town site of the over-all development.2 From the standpoint of employment, this plant was the largest in the Clinton works. It was the first on which construction was started (February, 1943) and the first to go into operation (the first units were ready in November, 1943). Indeed, for almost a year it was the only plant that was operating, and until December 31, 1946, the only plant that was turning out the final product—that is, the fully enriched uranium needed for an atomic bomb.
In every way, this process was one of the major efforts of the MED. Its purpose was to separate Uranium-235 from uranium as it occurs in nature and to do so in sufficient quantity, and of the concentration necessary, for use in atomic weapons. It is a physical rather than a chemical process, although a great deal of chemistry is involved in the handling of the material. Basically, electromagnetic separation of isotopes is based on the principle that an ion describes a curved path as it passes through a magnetic field. If the magnetic field is of constant strength, the heavier ions will describe curves of longer radii. Therefore, the various isotopes of an element, since they differ in mass, can be isolated and collected by such an arrangement.3
To apply this principle on a large scale, it was necessary to carry out a vast program of physical and chemical research. Research in many allied fields, such as biology, metallurgy and medicine, was also required. We then had to design, build and operate an extremely large plant with equipment of incredible complexity, without the benefit of any pilot plant or intermediate development: to save time we had early abandoned any idea of a pilot plant for this process. Always we were driven by the need to make haste. Consequently, research, development, construction and operation all had to be started and carried on simultaneously and without appreciable prior knowledge.
We would never have attempted it if it had not been for the great confidence that we, particularly Bush, Conant and I, had in the ability and drive of Dr. Ernest O. Lawrence of the University of California. Rather early in the American effort, Lawrence had proved to his own satisfaction that electromagnetic separation was feasible, but he stood almost alone in this optimism. The method called for a large number of extremely complicated, and as yet undesigned and undeveloped, devices involving high vacuums, high voltages and intense magnetic fields. As a large-scale method of separating Uranium-235, it seemed almost impossible. Dr. George T. Felbeck, who was in charge of the gaseous diffusion process for Union Carbide, once said it was like trying to find needles in a haystack while wearing boxing gloves. It seemed likely, though, that it could fill our need for more than microscopic samples of U-235 for experimental purposes at Los Alamos. The main reason we embarked upon this project, however, was that through it we hoped to achieve large-scale production.
As it turned out, our decision to go ahead with it was fully justified, for, as we had hoped, it enabled us to get the essential early samples of U-235 for Los Alamos and, later, the necessary U-235 for the Hiroshima bomb. Without it we would also have been seriously delayed in the design of the plutonium bomb.
Long before the essential research was well started and before the equipment could be designed, we had to start designing and constructing the building to house it. Stone and Webster was in charge of this operation. The research on which all design was based was carried out in the Radiation Laboratory of the University of California, under the direction of Dr. Lawrence; and to operate the plant we selected Eastman Kodak, whose subsidiary, Tennessee Eastman, was an extremely competent organization with much experience in chemical processes.
There were only three electrical suppliers in the country whom we considered suitable for the manufacture of the type of equipment in the quantities needed for the electromagnetic project. To avoid overloading them, we divided our requirements among them: General Electric produced power supply equipment, Allis-Chalmers the magnets, and Westinghouse the process bins and allied parts.
Building and operating such a plant as this presented many new industrial problems as laboratory experiments using raw materials measured in grams were expanded to handle tons. Proper liaison between research, design, manufacture, construction and operation people was absolutely essential to success, and fortunately it was excellent. All the companies involved had such representatives as were needed at the Berkeley laboratories; more than fifty key specialists were transferred from Berkeley to Tennessee Eastman; and another large group of engineers and physicists from the laboratory was maintained at Oak Ridge, where they gave invaluable assistance with the installation of the equipment and its operation.
From the beginning, we realized that the plant would have to be enormous, and also that it would be very costly. The first estimate for construction alone was for an unrealistic sum of between $12 and $17 million; soon afterward this was increased to $35 million. These figures were for a plant much smaller than the one we finally built. In its first report to President Roosevelt early in December, 1942, the Military Policy Committee estimated the cost of the entire project as of the order of $400 million. At that time we thought that over $100 million would be needed for this process as a whole. In fact, all these estimates were only guesses, made before anyone had a clear concept of what would be needed for an operating plant and what its productive capacity should be. Exclusive of the value of silver borrowed from the Treasury for electrical conductors, the construction costs, by December 31, 1946, totaled $304 million; research cost $20 million, the engineering $6 million and operation $204 million. The cost of operating power was almost $10 million.
We knew before we started building that two separate stages might be required to complete the uranium separation to the necessary degree of enrichment. We called them alpha and beta. Alpha would start with natural uranium and turn out a product much richer in U-235, though it might be short of the requirements for a weapon. If sufficiently enriched, it would be used in the bomb; but if not, we would achieve the necessary concentration by feeding it through a beta stage. Initially, the equipment for the alpha separation process was arranged in the shape of large ovals, each containing ninety-six magnets and ninety-six bins or tanks. Almost immediately the quite appropriate term “race track” was applied to the setup, by one of the Berkeley scientists, I believe.
Because beta would use feed material from alpha; because its design would be a modification of alpha, dependent on alpha experience, and because originally we were not sure it would be needed, emphasis was placed on completing the alpha installation first. Eventually, however, the Y-12 plant comprised five alpha buildings, of nine racetracks, and three beta buildings, of eight racetracks with thirty-six bins each, as well as numerous chemistry and other auxiliary buildings. All were large—two of the alpha buildings, for example, were 543 feet by 312 feet—and each contained a fantastic labyrinth of equipment and piping. Much of the equipment was considerably bigger, closer in tolerance and more demanding as to accuracy than any similar equipment ever designed. Much of it was of revolutionary design. Much of it was required in large quantities. Because of material and labor shortages, the always overriding necessity for speed and the badly overloaded manufacturing plants of the suppliers, all of it was built under trying conditions.
Setting up a plant of this size in the short time we had demanded an extremely well-organized and co-ordinated field force. Stone and Webster interviewed some 400,000 people for construction jobs and brought together a large force of experienced construction men from all parts of the country to fill the key positions.
Tennessee Eastman, for its part, immediately began training its key people, and sent a number of them to Berkeley to gain experience in the operation of the experimental units which were being set up in the University of California Radiation Laboratory.
At the same time, a drive was started to recruit the necessary labor for the operation of the plant. Originally we had thought we would need a work force of 2,500. This was a sad underestimate, resulting from our inability to anticipate how complex and difficult the job would be and how many units would be needed. Eventually we had over 24,000 on the payroll. The great difficulty was that our personnel procurement program had to ensure the operation of a plant for which the needs were unknown, but the planning had to be completed and the organization ready as the various units became operational.
In many cases during the training of workers, the announced aims of their operations were completely distorted in order to avoid any unnecessary disclosure of classified information. However, their jobs were always described in such a way as to impress the workers with the importance of being alert and careful.
To begin with, classes were held in Knoxville, at the University of Tennessee. Subsequently, they were conducted at a special school at Oak Ridge and later in the separation plant itself. To give training and develop operating techniques for the beta process without losing valuable alpha products, the beta race tracks were run for some time on the basic feed material before using the enriched product from the alpha tracks.
One of our chief difficulties was a shortage of electrical workers. This became so acute that we had to turn for help to Under Secretary of War Patterson. Out of this appeal came an agreement known as the Patterson-Brown plan (Edward J. Brown was president of the International Brotherhood of Electrical Workers). It provided for the payment to employees of round-trip transportation and subsistence, a guarantee of no loss of seniority rights and a job on return to their former employers after completing at least ninety days’ service at the project. Provision was also made for the official recognition of employers who released men in response to our appeal. This plan was a lifesaver, as was the co-operative attitude of Al Wegener, an official of the Brotherhood.
There was an almost complete absence of labor trouble, despite the fact that as many as four crafts were often involved in setting up a single piece of apparatus. The total time lost on the job from work stoppages, including jurisdictional disputes, was less than eight thousand man-hours as compared with the almost 67 million man-hours worked on the electromagnetic plant.
To me our excellent labor relations were a great satisfaction. The credit is due to the District organization, to the contractors, to Colonel C. D. Barker, in charge of labor relations in the office of the Chief of Engineers, to the patriotism and at times the forbearance of the employees, and to the co-operation of the union leaders of the trades involved.
I was particularly impressed by the attitude of one young union leader in connection with an effort to organize the powerhouse workers at Oak Ridge. We could not permit or even consider the unionization of the operating forces of any of the plants turning out U-235 because we simply could not allow anyone over whom we did not have complete control to gain the over-all, detailed knowledge that a union representative would necessarily gain. Neither could we permit the discussions between workers that would be bound to occur in union meetings. And obviously we could not have security officers present to monitor their meetings. Also, some information would inevitably filter back to the International Brotherhoods of the various unions.
When this organizing effort was begun at Oak Ridge, many were already union members but were necessarily inactive. If the powerhouse was organized there would surely be serious attempts and a great deal of agitation to unionize the entire gaseous diffusion process, which would soon enter into production. Naturally, strikes for any reason were out of the question.
To forestall any difficulty, I asked Patterson for his aid. We arranged a meeting at the White House with James F. Byrnes, who was then the Administrative Assistant to the President, and Fred Behler, the organizer of the union in question. He listened attentively to our reasons why unionization would be detrimental to the work we were doing. He then asked me about the importance of the work. I gave him my views and they were reinforced immediately by Mr. Patterson.
Behler then said, “If we don’t organize now, we’ll never be able to, because there will certainly be a plantwide election after the war, and as a craft union we’ll be snowed under.”
I replied that I expected that after the war the need for security would be less important and that there undoubtedly would be a plant-wide election, as he said.
I was not only pleased but extremely proud of him as a representative of American workers when he said, “General, in view of what you have told me about the importance of this work and your feelings that any attempt to unionize would be injurious to the country’s welfare, I want to assure you that we’ll make no effort to organize these men; we’ll discourage any effort that is made, and we will do this with the full realization that this means that ultimately these men will not belong to our union.”
Our one case of persistent labor trouble occurred at Hanford. There we were unable to get what we considered to be a proper output from our pipe fitters. Efforts to correct the situation through appeals to the local union officials were ineffective. The matter was so serious that G. M. Read of du Pont and I arranged to meet Mr. M. P. Durkin in Chicago. Durkin, later Secretary of Labor, was at that time head of the International Union. At this meeting we told him of Han-ford’s great importance to the war effort, emphasizing that every day’s delay in its completion could well mean hundreds of Americans killed and wounded. We appealed for his assistance in increasing the production of these workers and in securing more of them. He did not seem to be at all interested in our pleas, and our meeting was not a success.
Later, when our needs grew even more pressing, we were unable to find enough pipe fitters to maintain our schedule. Investigation showed that there simply were not enough in the United States to fill the demands. The solution we adopted was to locate a considerable number of pipe fitters, all union members, who had been inducted into the Army. These men were given the opportunity to be furloughed to the inactive reserve on condition that they would accept employment at Hanford as civilians at the going rates of pay.
When they arrived they were kept together as a group so that their output would not be held down by the pressure of any union officials or of the men already working there. In a direct comparison on identical work, they produced about 20 per cent more than the other men. Pressure was brought on them to slow down, but they refused. A typical comment was: “I’m not working as hard as I did in the Army, nobody’s shooting at me, I’m being paid a lot more and, what’s more important, I’ve a lot of friends in my old outfit that I hope to see come back alive.” As time went on, the other men were apparently shamed into greater effort, with the result that their output went up about 10 per cent.
Not long after he became general manager for Stone and Webster at Oak Ridge, in 1944 F. C. Creedon4 told me he wanted to hold a special meeting of all his supervisors, down through foremen and including even some straw bosses, and wanted me to talk to them. He felt this might stimulate morale and thus increase the construction output.
Creedon was not a man one would expect to favor appeals of this kind. To him, as to me, they would be embarrassing and normally seem a waste of time. Since he thought it would help, I felt the chances were that it might, and agreed to do it.
On my next visit to Oak Ridge I talked for five or ten minutes to some two thousand of these men. I was not introduced by name but merely as the general in charge of the work for the War Department. The reason for this was to avoid drawing attention to me personally; this was our policy throughout the project until security no longer required it. (My wife once commented that I was undoubtedly the most anonymous major general in the history of the United States Army.)
As simply as possible, I told the group that, as the officer in charge, I could state positively, both officially and personally, that their work was of extreme importance to the war effort, and that my views were a true reflection of those of the Chief of Staff, General Marshall, of Secretary of War Stimson and of President Roosevelt. I added that they could see for themselves how important it was from the terrific effort we were making, our obviously enormous expenditures in money and labor, and our evident ability to obtain materials that were in critically short supply. I said nothing about what we were working on or our hope that its success would quite possibly end the war. There was no flowery oratory; I would have been incapable of it, and it certainly would not have appealed to the audience.
Creedon estimated that after this meeting the efficiency of his construction operations improved by as much as 15 to 20 per cent. I never quite believed this, but the progress reports did indicate an increase of well over 10 per cent. This was far beyond anything I had anticipated; indeed, I would have been pleased with any improvement at all. In my opinion, whatever success the talk had was a result of Creedon’s understanding, as an experienced construction leader, of the mood of his men and how to improve it.
For much the same purpose, to wring out the utmost in the way of support for the electromagnetic project, we invited the presidents of Westinghouse, General Electric and Allis-Chalmers to visit Oak Ridge. The speed with which the plant could be completed depended largely on how quickly their companies could deliver the key parts, and we wanted these men to see for themselves just how their equipment was to be used and to gain a firsthand realization of the complexity and magnitude of the project. The results of the visits were quite noticeable. Because major items of equipment were not such a controlling factor at Hanford, we did not follow this procedure there.
From beginning to end, the problems encountered in building and running the Y-12 plant, in their variety and often plain unexpectedness, would have taxed the ingenuity and industry of Hercules. That they were solved was due to the magnificent leadership at all levels that we had at Oak Ridge. One of the greatest difficulties was the failure of the process equipment to arrive in the proper sequence for orderly installation. After all, there were fabulous amounts of this, including the enormous oval-shaped electromagnets, the process bins and the units enclosed in them, the control cubicles, motor generator sets, vacuum systems, chemical recovery equipment and thousands of smaller parts. An idea of the quantity of material used may be gained from the fact that 128 carloads of electrical equipment alone were received in a two-week period. Special warehouses had to be built to hold the equipment while we waited for the key items that had to be installed first. Highly secret process equipment was stored in a special area under armed guard and was not unpacked until it had been moved to the place where it was to be installed.
Actually, the work on the first race track was well under way before the structure for the opposite end of the building was finished. The moment the overhead cranes were set and the concrete roof poured, we started to unload and place the heavy magnets.
For security reasons, parts of the buildings were partitioned off when the construction reached an advanced stage, and special passes were issued to the workmen who had to enter these areas. At first the resulting confusion was extreme, but before long the men became used to the restrictions, and there did not seem to be any appreciable slowdown in the work.
One of the most serious snags we ran into was when the magnets in the first race track began to act up after a comparatively short period of operation. These magnets were very large (about 2(X x 20’ x 2’) and were much more powerful than any common magnet in use. They were encased in heavy, welded steel. After much theorizing about what could be causing the trouble, we broke one open in the hope that a visual examination would supply the answer, though this meant that the magnet would have to be shipped back to the manufacturer for rebuilding.
There were two possible sources of trouble. The first lay in the design, which placed the heavy current-carrying silver bands too close together. The other lay in the excessive amount of rust and other dirt particles in the circulating oil. These bridged the too narrow gap between the silver bands and resulted in shorting, often intermittent, which had made it more difficult to determine what, if anything, was really wrong.
To me, all this seemed entirely inexcusable, for we should have made certain that nothing like this would ever happen. The design should have provided for a much greater factor of safety. The manufacturing specifications should have made more adequate provision for rigid cleanliness. The design and erection specifications for the circulating system for the oil used to cool the magnets should have prevented the entrance of rust and dirt.
What made it even worse was the fact that so many of us failed to foresee the hazards that this design entailed. I know I should have. It was more than annoying, too, to learn too late that Lawrence had once had a similar difficulty with one of his cyclotrons. Fortunately, this error did not in the end set us back in our cumulative production of U-235, but only because it was discovered so early that the extraordinary precautions taken thereafter enabled us to eliminate many minor stoppages that otherwise would have occurred.
When the trouble was discovered, we had completed one race track, were well along with the second, and were starting on the third. Immediate and drastic measures were called for. The magnets were removed from the track and sent back to Milwaukee to be cleaned and rebuilt. The silver bands were unwound and then rewound to allow greater spacing. A special pickling plant was built and all installed piping was taken out and passed through it to eliminate every bit of rust. The equipment was then reassembled under conditions as free from contamination as we could devise. All new piping was similarly handled. The magnets were immediately redesigned. After that, we had no more trouble from that source.
Although we were certain sabotage was not involved, in our detailed review of the situation we found that it would be possible for a saboteur, who would have to be an employee on one particular assignment, to throw iron filings into a feed opening in the oil circulation system and thus put an entire section of track out of action. Steps were taken at once to station counterintelligence agents on and around these spots.
One difficulty, which was unforeseen, because we lacked experience with magnets of such enormous power, was that the magnetic forces moved the intervening tanks, which weighed some fourteen tons each, out of position by as much as three inches. This put a great strain on all the piping connected to them. The problem was solved by securely welding the tanks into place, using heavy steel tie straps. Once that was done, the tanks stayed where they belonged.
Another headache was ordering necessary spare parts in proper quantities. This was virtually impossible since, lacking operating experience, we had no intelligent basis on which to base our estimates. The estimates were guesses; some proved to be shrewd and some rather wild. What seemed to be a very generous order for some items turned out to be insufficient, and in a few cases temporary shutdowns of some equipment resulted. In other cases, the quantities ordered proved to be greatly in excess of what was needed. Extremely rare chemicals were procured in small quantities for laboratory uses. These included samarium, rhenium, yttrium and other rare earths. Other substances that had previously had very limited application were needed in staggering quantities. For example, each alpha track used four thousand gallons of liquid nitrogen every week.
One incident that delayed production on a bin in an alpha track for several days involved a mouse. In some unknown way, he got into the vacuum system, where his presence prevented the bin from reaching the necessary high vacuum. After several days of trouble-shooting failed to reveal the source of the trouble, the run was terminated and the bin opened. The remains of the mouse, a bit of fur and a tail, disclosed what had caused the trouble, but no one ever learned how he got into the system in the first place.
More serious in effect was the suicidal action of a bird which perched on an outside wire in such a way as to short the electrical system. We had to shut down an entire building, and, because of the nature of the process, it was several days before operations again became normal.
There were numerous other causes for shutdowns—ekctrical storms, accidental tripping of switches and accidents in serving the motor generators supplying currents in the magnets. Once, the stone used to grind the commutators of the generators broke and nicked one of them, delaying operations in that track for some time.
Because of the fabulous value of some of the materials, strict accountability was essential to avoid waste. Waste such as piping, scrap cloth, filter cloths, papers, rubber gloves, clothing and the like had to be carefully saved in order to recover the small concentrations of uranium, particularly of Uranium-235. Inventories of the alpha cycle were made every four weeks and of the beta cycle every two. Constant studies were made to find out where losses occurred.
Early in 1946 an additional safeguard was adopted—a lie detector. It was used chiefly on people who had access to the final product chemistry building, to make certain that no one had taken, or knew of anyone who had taken, material from the plant. The first tests were carried out under the supervision of the inventor of the instrument, and one of his assistants was retained at Y-12 to conduct tests whenever necessary.
The value of one of the materials we used in quantity necessitated what was virtually a separate operation in itself. Preliminary design calculations on the Y-12 electromagnetic plant in the summer of 1942 had indicated that enormous quantities of conductor material would be required. Because the demands for copper to be used in defense projects far exceeded the national supply, the Administration had decided that the need for copper should be reduced by substituting for it silver borrowed from the Treasury Department.
Colonel Marshall thereupon called on the Under Secretary of the Treasury, Daniel Bell. Mr. Bell said that he might be able to make available some 47,000 tons5 of free silver, together with 39,000 tons more which could be released from the backup of silver certificates, if Congress authorized its use through appropriate legislation. At one point early in the negotiations, Nichols, acting for Marshall, said that they would need between five and ten thousand tons of silver. This led to the icy reply: “Colonel, in the Treasury we do not speak of tons of silver; our unit is the Troy ounce.”
Under the terms of the final agreement, the silver required by the project was to be withdrawn from the West Point Depository. Six months after the end of the war an equal amount of silver would be returned to the Treasury. It was further agreed that no information would be given to the press on the removal of the silver, and that the Treasury would continue to carry it on their daily balance sheets. Our relations with the Treasury were most cordial, and Mr. Bell and the various officials of the Mint and the Assay Office were always very pleasant and helpful.
Because of the natural reluctance of any private company to accept the responsibilities for safeguarding and accounting for the large amounts of silver that were involved, the MED had to carry out this responsibility with its own forces. This meant organizing separate guard and accountability units, establishing special inspection procedures employing special consultants and arranging to convert the silver into the conductors that we so urgently needed.
We accepted the Treasury’s certification of the bar weights of the silver as we took it over at West Point. Then we delivered it to a processor, who cast the bullion bars into billets which could be extruded into forms more suitable for manufacture into bus bars, magnet coils and similar items. The casting was done by the Defense Plant Corporation and by the U.S. Metal Refinery Company. For the large magnets which used the bulk of the silver, Phelps Dodge Copper Products Company then extruded the billets into strips, which were rolled into coils about the size of a large automobile tire. These coils were shipped to Allis-Chalmers, where they were wound, suitably insulated, around the steel bobbin plate of the magnet casing.
Special MED guards watched the silver at all times while it was being processed, and accompanied every shipment except that of the final magnets from Allis-Chalmers to the Clinton works. We decided that at this point we could achieve adequate security by sending unguarded railway cars over different routes on varying time schedules. The silver coils were encased in large, heavy, steel shells which were completely welded together. Although silver is a valuable commodity, to have made away with any great amount of it during shipment would have been a major task, as our experience in opening one of these shells at Oak Ridge later confirmed. Moreover, the railroads always followed our shipments carefully, and we would have known immediately if any car had been waylaid.
Despite the great total value of the silver involved and the thousands of small, easily stolen pieces of silver that were being fabricated into equipment, we sought to strike a reasonable balance between security and economy. Useless restrictions were avoided. Many of the precautions that we took were aimed primarily at concealing our interest in the silver, and included the use of coded commercial bills of lading, the direction of all shipments to nonmilitary personnel and the requirement that our officers wear civilian clothing in many of the plants they inspected. Naturally all communications, oral, telephonic and written, were carried on in accordance with our established procedures for handling highly secret matters.
Every person involved in this work was thoroughly investigated before being employed and anyone who was not properly cleared was denied access to the areas where our work was being done, which were completely shut off from the rest of the plants. Within the process areas the silver was guarded twenty-four hours a day and seven days a week.
Routine security inspections were performed by District security and intelligence personnel. General inspections of the entire silver program were also made by other District personnel not directly connected with the program, and normally we used a different inspector on each occasion. This introduction of fresh viewpoints proved of great value in disclosing possible loopholes.
Our accounting system was very detailed, for it had to reflect the disposition, including all intermediate steps, of more than fourteen thousand tons of silver, to the nearest ounce. To assist us in this work, and as an added precaution, we employed the services of a well-known New York accounting and auditing firm to maintain a complete running audit of the silver account.
No recovery operation was undertaken unless the recoverable amounts were expected to be of more value than the cost of recovery. Nevertheless, throughout the entire operation we lost only .035 of one per cent of the more than $300 million worth of silver we had withdrawn from the Treasury.
The electromagnetic process entailed a number of special hazards: uranium is toxic as well as radioactive. Some of the raw materials were also extremely difficult to handle. High temperatures and pressures were involved and many irritants such as phosgene had to be used. Liquid nitrogen was used in large quantities at a temperature of –196° Centigrade. Huge amounts of electricity were used throughout the process. Each control cubicle, for example, of which there were ninety-six for each alpha track and thirty-six for each beta, consumed about as much electricity as a large radio station.
In order to ensure complete compliance with the established Corps of Engineers’ safety regulations and to impress everyone with my belief in their value, I set up specific requirements soon after the work was begun. These included an immediate telegraphic report to me personally whenever a fatal or near fatal accident occurred, to be followed with a complete written report the next day, giving the details of the accident, the cause and the steps that were being taken to prevent similar accidents. We had eight fatal accidents in all of our plant operations through December, 1946. Five people were electrocuted, one was gassed, one was burned and one was killed by a fall.
The task of whipping the bugs out of equipment, overcoming failures, low efficiencies and losses, with untrained personnel, while surrounded by the dirt and din of construction work was a prodigious one, but Tennessee Eastman proved more than capable. Despite all the difficulties that had to be overcome, the first shipment of enriched uranium was sent to Los Alamos in March, 1944, just a few days more than a year after the construction of the plant was begun. The concentration was not great, but it was sufficient for urgently needed experimental work at Los Alamos.
The most crucial point came just before the Hiroshima bomb. In a letter to Oppenheimer dated July 3, 1945, Nichols gave the predicted amount of material that would be shipped to Los Alamos by July 24, 1945. Activity at Oak Ridge became even more feverish than usual. By the time of the deadline, the production exceeded by a considerable amount the schedule Nichols had promised.
The chemical side of the electromagnetic process, in fact of the entire project, has often been treated as a simple auxiliary to its more eye-catching atomic physics aspects. Actually, chemistry was the beginning and the end of each of the separation processes. Production efficiency could be won or lost in chemistry, as well as in physics. Each was indispensable. The chemists and the chemical engineers had to keep constant pace with the possibilities unfolded in the new developments in physics. They, too, had to design processes without adequate information. And the chemical facilities had to be ready in time to meet the rest of the schedule. They always were.
The gaseous diffusion process, later termed the K-25 project, was a large scale multistage process for the separation of U-235 from U-238 by means of the principle of molecular effusion. The method was completely novel. It was based on the theory that if uranium gas was pumped against a porous barrier, the lighter molecules of the gas, containing U-235, would pass through more rapidly than the heavier U-238 molecules. The heart of the process was, therefore, the barrier, a porous thin metal sheet or membrane with millions of submicro-scopic openings per square inch. These sheets were formed into tubes which were enclosed in an airtight vessel, the diffuser. As the gas, uranium hexafluoride, was pumped through a long series, or cascade, of these tubes it tended to separate, the enriched gas moving up the cascade while the depleted moved down. However, there is so little difference in mass between the hexafluorides of U-238 and U-235 that it was impossible to gain much separation in a single diffusion step. This was why there had to be several thousand successive stages.
The basic scientific research on the gaseous diffusion process was done by Columbia University’s SAM6 Laboratory in New York City under the leadership of Dr. Harold C. Urey, with Dr. John R. Dunning as chief physicist. In November, 1942, the S-1 Committee of the OSRD gave it a third-place priority, behind the electromagnetic and plutonium plants. The following month a letter contract was signed with the M. W. Kellogg Company for the extensive research and development, design, procurement and related services necessary to build a plant to produce U-235 of the purity and the quantity found needed for atomic bomb production. For operational reasons, as well as for security, Kellogg set up a wholly owned subsidiary, Kellex, to handle this project. Close co-operation was promptly established between Columbia and Kellex. Kellex took fundamental data, developed by Columbia, developed them further and applied them to the design of the production plant.
Kellex was headed by Mr. P. C. Keith, a brilliant scientific engineer. At first some thought was given to having the Kellex organization handle the construction, too, but I decided against this on the grounds that they would be fully occupied in the development and preparation of the engineering plans. Instead, we gave the building contract to the J. A. Jones Company of Charlotte, North Carolina, with whose excellent work on several Army construction jobs I was already familiar. Union Carbide and Carbon Corporation was chosen as the operator. Here the work was under the general supervision of James A. Rafferty, a vice president of Union Carbide, with Lyman A. Bliss as his principal assistant. Felbeck was in direct charge of the operation.
Our power needs for K-25 we knew would be high. For many months before, as well as during, the construction of the plant, we labored under the belief that if the plant was shut down through power failure or for any other reason—for as much of a fraction of a second—it would take many days, some said seventy, to get it back into full operation. This proved to be quite erroneous, but nevertheless it was the basis on which the plant was designed and built. And it was the main reason we decided to build a special steam-generating plant, instead of depending wholly on TVA, whose power had to come over many miles of wire and was therefore subject to interruption from both natural causes and sabotage.
There were also a number of minor advantages to be derived from a steam-generating plant near the K-25 plant, including a better means of supplying the wide range of variable-frequency power for which the plant was designed. The use of power at a number of varying frequencies was rather typical of many features of MED. We could not afford to make our task easier by using standard equipment if we knew that special equipment would work but were not sure the standard would.
To protect our power supply from sabotage, we built an underground conduit from the steam-generating plant to the gas diffusion plant. In spite of our precautions, we did have one case of attempted sabotage. This was sabotage in its crudest form—the driving of a nail through a rubber-coated electric power cable. Fortunately, it was discovered long before we had to depend on this cable for the supply of electricity. We could never discover the culprit; it could have been done either by an enemy agent, an enemy sympathizer or by a disgruntled workman, most probably the last.
Without question the most serious problem that confronted us throughout was our inability to produce until late 1944 the barrier material which was the heart of the process. This prevented the orderly installation of the production equipment. It meant that before the first unit could be put in operation, some $200 million had been spent on construction and on the purchase of special equipment, and most of this had been done before we knew even that a satisfactory barrier could be made in the quantities we would require. Yet in spite of this major unknown factor, we had to press ahead with the construction of one of the largest industrial plants ever built, comprising over forty acres of floor space.
The first separation of uranium isotopes by the gas diffusion method had been accomplished at Columbia in January of 1942, but it was not until October of that year that the first laboratory unit was operated. The barrier in this unit was about the size of a silver dollar. While other pilot units were built to test various components, no complete pilot plant or semi-works was ever constructed. The Oak Ridge plant was a first in every sense, and its design, involving many acres of barrier, was based on this small piece less than two square inches in area. Even this practical foundation soon disappeared when it became known that the material used in the first filter could never be employed in the main plant. One of the great contributions made by Union Carbide during the design and construction period was in connection with the development of a satisfactory barrier, and with the co-operation of a large number of firms mass production was achieved in ten months after manufacture was started.
The design and manufacture of the thousands of pumps needed to force the gas through the diffusion units was solved by the Allis-Chalmers Company. Keith had canvassed the entire industry, and had come to the conclusion that our one hope for obtaining the pump we needed within the time limit lay in that firm. When he first approached them they were unwilling, primarily because they were already seriously overloaded with war work. He then asked several Carbide men and me to visit Milwaukee with him, with the aim of persuading Allis-Chalmers to undertake the job. They were already playing an important role in the project in the production of the enormous magnets used in the electromagnetic process and their president was most reluctant to consider our request. Finally, after warning us that they were so overloaded with war work that he did not see how they could possibly undertake it, he consented to our talking with his chief engineer. We were amazed when, after we had described in some detail the exacting performance specifications, he replied, “Yes, we can do that. We have already manufactured pumps of the same type, but of course of much smaller capacity.” The contract was accepted and perfectly performed.
On our way back from Milwaukee, we stayed overnight in Chicago. In our hotel rooms we talked at some length about another design problem: how to handle a breakdown within a particular unit. In the course of the discussion, I expressed surprise that it was thought to be a problem, since all that was necessary was to cut out the particular unit that had broken down. The difference between the makeup of the gas varied from diffuser to diffuser so slightly as to be un-noticeable and almost unmeasurable, and I asked how the diffusers could ever tell the difference. That casual question immediately suggested the answer. As so often occurs, it was a case of a simple solution occurring immediately to someone who had not been struggling for months with the problem.
To minimize the effects of gas corrosion, it was first proposed that we use solid nickel for the some hundred miles of process piping. K. T. Keller, the head of the Chrysler Corporation, which was to produce the diffusing units, pointed out that our demands in that case would exceed the entire nickel production of the world, and insisted that heavy nickel plating on the inside of the larger pipe, four inches and above, was feasible. To attempt to heavily nickel-plate the interior of the quantity of pipe we needed was an unprecedented undertaking, but it was solved by a small manufacturer in Belleville, New Jersey, the Bart Laboratories. They developed a novel method in which they used the pipe itself as an electroplating tank. The pipe was rotated during the operation in order to obtain a uniform thickness of deposit. Their success eliminated what otherwise would have been a most difficult situation.
Instruments capable of distinguishing between the isotopes of a particular element had to be developed. This requirement was also unprecedented. Then there was the problem of regulating the gas flow within the system, which was extremely long, and was ultrasensitive to pressure waves and variations. It was feared these would cause surges.
I was never so concerned about the problem of surges as were many of our people. This was because I approached it with the background of an engineer rather than of a theoretical scientist. Many of the British scientists were extremely pessimistic. They expressed the view that it would be practically impossible to regulate the flow so as to avoid fluctuations and internal disturbances, and that the resulting surges would seriously impair operating efficiency, and might even damage the barriers. Some went so far as to say that the plant would be inoperable.
We had to be absolutely sure that in the hundreds of miles of piping the total leakage of air into the system, particularly through the welds, would not exceed that which would enter through a single pinhole. This problem was solved by industrial engineers. By using helium gas with an improved mass spectrometer, we were able to detect all leaks before the individual piping assembly was installed, and because we could not permit any leakage, no matter how slight, we could not tolerate normal commercial shop welding of the pipe connections, so special welding techniques had to be developed. Once we realized the importance of complete tightness in the gas diffusion plant, we did everything we could think of to achieve it. One step was to operate a special school for over two hundred employees of the various firms involved in the manufacture of equipment or the operation of the K-25 plant.
A very serious question arose in connection with the gas diffusion program when our special scientific safety committee told us that there was a distinct chance of an accidental atomic explosion resulting from an accumulation of highly fissionable material at some point within the piping system. Because of the distinguished membership of this committee, the utmost consideration had to be given to their fears, even though I disagreed with them. In this I found considerable comfort and support in the advice of Nichols, Conant and Tolman. Each expressed his feeling that the committee was overly apprehensive, and that there was no serious danger, or in fact any appreciable possibility, that an explosion would occur. I discussed this very thoroughly with both Keith and Felbeck. They agreed that the chances were extremely remote, and the opinion of the safety committee was ignored.
The cleaning and conditioning of equipment prior to installation was vital and the closest practical approach was made to surgical conditions. This involved the complete removal of dirt, grease, oxide, scale, fluxes and other extraneous matter. Any such material, even in small amounts, could very well have caused a complete failure.
The cleaning methods were based on procedures developed by the Chrysler Corporation. The individual steps were not too unusual in industrial practice, but the combination of all of them, their rigorous-ness and their application to the thousands of pieces of equipment were unheard of. Depending upon the particular item, as many as ten steps were required. For some parts the cleaning was done at the factory under the supervision of Kellex inspectors. Many, however, had to be cleaned just before installation, which required the construction of a special conditioning area at K-25.
Aside from the cleaning of equipment, the general cleanliness control measures at the K-25 plant were so rigid as at first to interfere considerably with normal methods of construction. All workers changed into clean outer clothing from head to foot upon entering a restricted building. Initially everyone had to comply with this requirement. Special lockers were provided for all concerned, including Nichols and myself. We had to change like everyone else even if we were going to walk into the area and out again in a few minutes.
Everything possible was done to eliminate dirt and dust. Vacuum cleaners were used instead of brooms, and dust mops were used in order to avoid raising dust by dry sweeping. As experience was gained, it was found possible to lessen the severity of the rules to some degree.
In scheduling the construction we established five objectives:
First, the completion of one cell of the main cascade; second, the completion of one entire process building; and third, the completion of a sufficient portion of the plant to enable the production of a lightly enriched material. As early as January 1, 1944, we set the date for this as January 1, 1945. Fourth, the completion of additional portions so as to enable the production of a maximum amount of U-235 by the critical date, as yet unset, for the first bomb. And fifth, the completion of the entire plant so as to permit maximum production.
The goal of producing slightly enriched material by January 1, 1945, was not quite met, but the later goals were achieved, although with very little time to spare.7
The gaseous diffusion process as it was built was essentially an ail-American effort. However, the British were made aware of the general details of its design. It should be realized that the first work in this area was begun in England, even though the efforts were confined to theory and to the laboratory.
Several preliminary meetings were held with several of the British representatives (W. A. Akers, F. Simon and R. Peierls) during the spring of 1942 in the United States, at which the principles of diffusion separation and possible types of plant design were discussed. The views of the British group on plant design, however, were quite different from the American, and their methods and equipment were dissimilar to ours. For that reason the latter discussions did not prove to be of much practical value to the United States development.
In the fall of 1943 the British sent over a strong delegation of scientists and engineers to learn of our plans on the gaseous diffusion process. They spent the period from September, 1943, to January, 1944, going over our designs in detail.
They did not agree with us on the course we were following. To many of our people the principal reason for this appeared to be because it was not based on the design theories they had originally developed. The barrier material was discussed with them in considerable detail, but their views did not influence the selection of either our first barrier, which was later abandoned, or the second.
In December of 1943, they proposed that an investigation be made of a method by which the gas would be recycled back and forth. Theoretically this had considerable merit, since it would have permitted a considerable reduction in the number of stages required. But it would have necessitated a large increase in the total barrier requirements, and the production of the barrier was still an unsolved problem, after several years of investigation. It also would have greatly increased the complexity of design of the converter and of the stage operation.
At this late date, I was strongly opposed to any major change such as the British suggested, for any one of them would have seriously delayed the completion of the plant. To me this was completely unacceptable.
The British group also suggested designing the plant as a cascade of cascades. In such a design the plant is not a simple long cascade, but rather is compartmentalized and consists of a number of smaller groups of stages, each in itself set up as a complete cascade and each connected with its neighbors. The system is quite complicated, and thus has many disadvantages. The American plan appeared to be satisfactory and as its design was progressing on schedule, we could see no advantage in changing it. We did examine their proposal carefully to make certain that our decision was sound. The discussions on this point were quite helpful in suggesting means for controlling operating disturbances.
There were many other suggestions made by the British, none of which was adopted. Their principal value lay in the stimulation the Kellex engineers derived from discussing them. This is always of value in any scientific development.
From February to May, 1944, a group from the British delegation gave us some valuable aid in solving some of our theoretical problems. This group included Peierls, C. F. Kearton, Skyrne and Fuchs. They were stationed in New York, but they were free to travel and they did.
I have mentioned earlier that the first uranium separation process I looked into on being told of my appointment to the MED was that involving liquid thermal diffusion. The basic apparatus is a column. It consists of a long, vertical, externally cooled tube with a hot concentric cylinder inside. What makes this an effective separation method is the fact that one isotope tends to concentrate near the hotter of two surfaces, and then moves upward.
From a practical standpoint, thermal diffusion was not suitable as an independent process because of the incredibly large amount of steam required. The production costs would have been staggering. A minimum rough estimate was two billion dollars, and I would not have considered this a safe figure, but would have raised it to at least three billion if I had thought the work would have to be undertaken. Moreover, the research, though it had been carried on by Philip Abelson in a most competent manner, had been extremely limited.
He had started his investigations at the Carnegie Institute in Washington, and in the summer of 1941 had succeeded in actually separating a certain amount of U-235. The interest of the Navy was aroused by its hope for a better power source, and it was not long before it began to support the work, first at Carnegie and then moving it with Abelson to the Naval Research Laboratory in Washington. Here the research went forward for several years under Navy auspices and using Navy funds. Abelson’s progress was followed closely by the OSRD and later by the MED. The process and the results obtained were reviewed on several occasions by competent scientific groups. No one was particularly impressed by its possibilities.
However, in June of 1944, Oppenheimer suggested to me that it might be well to consider using the thermal diffusion process as a first step aimed at only a slight enrichment, and employing its product as a feed material for our other plants. As far as I ever knew, he was the first to realize the advantages of such a move, and I at once decided that the idea was well worth investigating.
Just why no one had thought of it at least a year earlier I cannot explain, but not one of us had. Probably it was because at the time the thermal diffusion process was studied by the MED we were thinking of a single process that would produce the final product. No one was considering combining processes. This step came much later when we decided to limit the initial enrichment goal of the gas diffusion plant and to use its product as feed material for the beta stage of the electromagnetic plant.
If I had appreciated the possibilities of thermal diffusion, we would have gone ahead with it much sooner, taken a bit more time on the design of the plant and made it much bigger and better. Its effect on our production of U-235 in June and July, 1945, would have been appreciable. Whether it would have ended the war sooner, I do not know. It would not have affected the date of the Alamogordo test, since for that we used plutonium in the implosion-type bomb. Even with an increased production rate, we could not have afforded to use Uranium-235 in a test.
A few days after Oppenheimer made his suggestion, we took steps to carry it out. By this time the Navy was building a pilot plant at the Philadelphia Navy Yard. It was nearly complete and most of the operating techniques had been planned.
In order to get a large-scale plant into operation just as quickly as possible, we decided to use a single contractor for the entire operation. The H. K. Ferguson Company,8 an outstanding engineering firm, was given the assignment on June 26, 1944, and on the same day Admiral King, at my request, directed that copies of the Navy plans be turned over to them. Ferguson engineers also visited the Naval Research Laboratory and the Philadelphia Navy Yard and made many engineering sketches.
To expedite the design and construction, I ordered that, insofar as possible, all process features of our plant, particularly the basic column assemblies, should be Chinese copies of those at the Philadelphia pilot plant. A great deal of time was also saved by frequently using field engineering sketches instead of the customary more formal drawings.
Various sites were hastily considered, including one in Detroit, but we finally settled on an area at Oak Ridge near the K-25 powerhouse, where steam was immediately available. There was also an advantage in having the new plant next to the K-25 plant, which was to use its product. Above all, we wanted to avoid building and operating classified plants in a new area. This process was given the security designation S-50.
When we decided on its location, Union Carbide expressed concern over the possibility of an explosion caused by the high steam pressures we would be using and its effect on their gaseous diffusion operation. They were also worried lest the powerhouse water supply might be contaminated by S-50 process material, which would cause trouble in the big steam plant. I did not think there was too great a danger of a damaging explosion, and we guarded against the possibility of contamination by installing a number of special recording devices.
The basic piece of equipment was the isotope separation column, 102 of which were arranged to form an operating unit which we termed a “rack.” The column was a vertical pipe, forty-eight feet long, of nickel pipe surrounded by a copper pipe. The copper pipe was encased in a water jacket contained in a four-inch galvanized-iron pipe. The copper pipe was cooled with water at a moderate temperature. The columns were arranged in three groups, each of seven racks, making a total of 2,142 columns.
The columns at the Naval Research Laboratory were few in number and had not been standardized, yet Ferguson had to arrange for their manufacture on a mass basis. Over twenty manufacturers were consulted but none of those considered qualified wanted to undertake the job. Finally Mehring and Hanson and the Grinnell Corporation agreed to attempt it. After considerable difficulty, both firms worked out production methods that permitted them to make as many as fifty columns a day apiece. On July 5, nine days after getting the assignment, Ferguson placed the first order for the manufacture of process columns, and on the ninth, the clearing of the site began.
Training of operating personnel at the Philadelphia Navy Yard got under way in August. In September our progress received a severe blow when there was a bad explosion in the pilot plant, which injured several persons and disrupted the pilot plant installations. While we thought that it was probably due to external causes (as it proved to be) rather than to the process or to faulty designs, we could not be certain without thorough investigation, which, of course, held up the work for a while.
The importance of early production had dictated severe construction schedules. While six months were considered optimistic for the construction of the main process building, I set a schedule of 120 days to begin operation. The highest priorities were used and everything was done to speed the work. Riggers, pipe fitters, welders, sheet metal workers, electricians, carpenters and operation personnel worked feverishly on immediately adjacent parts of the unit. As soon as the riggers put a column in place and the pipe fitters completed its installation, it was pressure-tested and conditioned by the operators. Sixty-nine days after the start of construction, one-third of the plant was complete, and preliminary operation began. On October 30, the first product was drawn off, and peak production was reached the following June.
At first the operation was not too satisfactory. We were badly handicapped by the complete lack of pilot plant experience, the lack of trained people and, as it turned out, by an insufficient supply of steam. We were also plagued by leaks of the process material and by high-pressure steam leaks. It was not until January, 1945, that these difficulties were entirely ironed out.
The application of high steam pressures (a thousand pounds per square inch in this case) is a risky undertaking, particularly where the equipment has been designed and installed so hurriedly. The resulting leaks could have been avoided if we had had more time for study in the design phases of the project.
On one of the racks, owing to the hurried construction, there were large quantities of high-pressure steam escaping. The resulting clouds of vapor and the noise made it difficult for the operators to function. Yet we continued to operate the rack in spite of conditions that under any normal procedures would have called for an immediate shutdown. When the plant reached full operation, this rack was taken out of production so that it could be put into satisfactory condition.
By October it was evident that too much steam was leaking out at the screwed unions at the bottom and top of the columns. We decided, therefore, to replace all of them with new welded connections. By January, all the twenty-one racks were again ready except for a few small piping changes found to be necessary on the basis of our operating experience.
As the various plants at Oak Ridge were gradually put into operation, the main problem we faced was how best to use them, in combination and separately, to get the greatest possible output of fissionable uranium in the shortest possible time.9
The over-all problem was extraordinarily complex, not only because of the constantly changing relationship between the quantity of U-235 produced and the degree of its enrichment, but also because the product of one process was the feed material for another, and this also was subject to change. Moreover, as additional equipment was put into use—and in the spring of 1945 this was expected to, and did, occur almost daily—the production would increase also, and it would increase still further as the enrichment of the material fed into the individual processes was augmented. We were certain, too, to have increased efficiencies in the operation of the various plants as we gained experience. But the effects of this could only be guessed at.
A sound solution was of such importance to our success that I asked Nichols to give the problem even closer personal attention than he might normally have devoted to it. Consequently, in December, 1944, he set up a special group of scientists and engineers, for the most part officers and enlisted men, under the charge of Major A. V. Peterson. Their job was to work out the best practical plan of operation, despite the changing characteristics, performances and capacities of the different processes.
When they started their study, the exact amount of U-235 that would be required for a bomb was still not known. Peterson discussed with Oppenheimer the relative importance of quantities and enrichments, and the enrichment to be aimed at was fixed. Under his close supervision, Peterson’s group was able to work out an optimum program of operations, based on the most efficient production interplay between the electromagnetic alpha, beta, gaseous diffusion and thermal diffusion plants. This schedule was constantly revised as new efficiencies were achieved and as changes occurred in the dates for placing equipment into service.
By May, 1945, we reached the conclusion that our pre-Yalta estimates of being ready early in August were reasonable and that we should have accumulated enough material for one bomb by late July. After Nichols, Peterson, Oppenheimer and I had carefully reviewed the situation, July 24 was finally set as the deadline date.
And by the end of that day, enough uranium—and a little bit more —had been shipped to Los Alamos for the manufacture of the first bomb to be dropped on Japan.
1 Scientific research into the centrifuge and liquid thermal diffusion processes had not progressed as rapidly as it had into the other processes adopted for production in the fall of 1942. To avoid further extending our already strained resources we decided to put these aside.
2 See Appendix V, page 424.
3 The basis for the process grew out of A. J. Dempster’s use in 1918 of a simplified mass spectograph, and E. O. Lawrence’s development of the cyclotron during the thirties.
4 Formerly with the Rubber Administration and before that with me on the construction of Ordnance facilities.
5 Converted from Troy ounces.
6 Code name originally based on Substitute (or Special) Alloy Materials.
7 At the peak of construction employment the working force at K-25 totaled some 25,000. The cost of the K-25 design, engineering and procurement to the end of the fiscal year 1946 was $253 million, and the estimated total costs for the completed contracts were $275 million.
8 As a nationwide engineering construction company, the Ferguson Company operated on a closed shop basis. To avoid the union problem after the plant began to operate, a subsidiary operating company, the Fercleve Corporation, was formed. The MED then entered into two separate contracts; one for construction with Ferguson and one for operation with Fercleve. This arrangement worked out to everyone’s satisfaction.
9 See Appendix VI, page 427.