Atomic Energy for Military Purposes (The Smyth Report)
CHAPTER VI. THE METALLURGICAL PROJECT AT CHICAGO IN 1942
6.1. As has been made clear in Chapters IV and V, the information accumulated by the end of 1941 as to the possibility of producing an atomic bomb was such as to warrant expansion of the work, and this expansion called for an administrative reorganization. It was generally accepted that there was a very high probability that an atomic bomb of enormous destructive power could be made, either from concentrated U-235 or from the new element plutonium. It was proposed, therefore, to institute a intensive experimental and theoretical program including work both on isotope separation and on the chain-reaction problem; It was hoped that this program would establish definitely whether or not U-235 could be separated in significant quantities from U-238, either by electromagnetic or statistical methods; whether or not a chain reaction could be established with natural uranium or its compounds and could be made to yield relatively large quantities of plutonium; and whether or not the plutonium so produced could be separated from the parent material, uranium. It was hoped also that the program would provide the theoretical and experimental data required for the design of a fast neutron chain-reacting bomb.
6.2. As has been explained in Chapter V, the problems of isotope separation had been assigned to groups under Lawrence and Urey while the remaining problems were assigned to Compton's group, which was organized under the cryptically named "Metallurgical Laboratory" of the University of Chicago. In this chapter and the following two chapters we shall describe the work of the Metallurgical Laboratory and the associated laboratories up to June 1945. In later chapters we shall discuss isotope separation work and the work of the bomb development group, which was separated from the Metallurgical Laboratory early in 1943.
6.3. It would be futile to attempt an assessment of the relative importance of the contributions of the various laboratories to the overall success of the atomic-bomb project. This report makes no such attempt, and there is little correlation between the space devoted to the work of a given group and the ability or importance of that group. In deciding which subdivision of the atomic-bomb project should be discussed first and most fully, we have been governed by criteria of general interest and of military security. Some developments of great technical importance are of little general interest; others both interesting and important must still be kept secret. Such criteria, applied to the objectives and accomplishments of the various laboratories set up since large-scale work began, favor the Metallurgical Laboratory as the part of the project to be treated must completely.
6.4. In accordance with the general objective, just outlined, the initial objectives of the Metallurgical Laboratory were: first, to find a system using normal uranium in which a chain reaction would occur; second, to show that, if such a chain reaction did occur, it would be possible to separate plutonium chemically from the other material; and, finally, to obtain the theoretical and experimental data for effecting an explosive chain reaction with either U-235 or with plutonium. The ultimate objective of the laboratory was to prepare plans for the large-scale production of plutonium and for its use in bombs.
ORGANIZATION OF THE WORK
6.5. The laboratory had not only to concern itself with its immediate objectives but simultaneously to bear in mind the ultimate objectives and to work toward them on the assumption that the immediate objectives would be attained. It could not wait for a chain reaction to be achieved before studying the chemistry of plutonium. It had to assume that plutonium would be separated and to go ahead with the formulation of plans for its production and use. Consequently problems were continually; redefined as new information became available, and research programs were reassessed almost from week to week. In a general way the experimental nuclear physics group under E. Fermi was primarily concerned with getting a chain reaction going, the chemistry division organized by F. H. Spedding (later in turn under S. K. Allison, J. Franck, W. C. Johnson, and T. Hogness) with the chemistry of plutonium and with separation methods and the theoretical group under E. Wigner with designing production piles. However, the problems were intertwined and the various scientific and technical aspects of the fission process were studied in whatever group seemed best equipped for the particular task. In March 1942, Thomas Moore was brought in to head the engineering group. Other senior men in this group were M. C. Leverett, J. A. Wheeler and C. M. Cooper, who later succeeded Moore as head of the Technical Division. In the summer of 1942 the importance of health problems became apparent and a Health Division was organized under Dr. R. S. Stone. The difficult task of organizing and administering a research laboratory growing in size and complexity with almost explosive violence was carried out by R. L. Doan as Laboratory Director.
6.6. We have chosen to confine this chapter to the work of 1942 because a self-sustaining chain reaction was first achieved on December 2 of that year, at a time when the whole Chicago project was being appraised by a reviewing committee with the members particularly selected for their engineering background>1. That was a dramatic coincidence and also a convenient one for purposes of this report since either incident might be considered it to mark the end of an epoch at the Metallurgical Laboratory; Furthermore, in preparation for the reviewing committee's visit comprehensive report had been prepared, that report was generally known as the "Feasibility Report" and has been used extensively in preparing this chapter.
PLAN OF THIS CHAPTER
6.7. In this chapter we shall present the material in the order of the objectives given above. In Part I we shall discuss progress towards the initial objectives, including (a) procurement of materials, (b) the experimental proof of the chain reaction, (c) the chemistry of plutonium and some of the problems of separation, (d) some of the types of auxiliary experiments that were performed, and finally (e) the "fast neutron" work. Necessarily the work described in detail is only a sampling of the large amount of theoretical and experimental work actually performed, In Part II we shall discuss the possibilities that were considered for production piles and separation methods, and the specific proposals made in November 1942.
PART I: PROGRESS TOWARD THE INITIAL OBJECTIVES
PROCUREMENT OF MATERIALS
6.8. It has been made clear in earlier chapters of this report that the procurement of materials of sufficient purity was a major part of the problem, As far as uranium was concerned, it seemed likely that it would be needed in highly purified metallic form or at least as highly purified uranium oxide. The other materials which were going to be needed were either graphite, heavy water, or possibly beryllium, It was clear at this time that, however advantageous heavy water might be as a moderator, no large quantities of it would be available for months or years. Beryllium seemed less advantageous and almost as difficult to get Therefore the procurement efforts for a moderator were centered on graphite. As has been explained in Chapter V, procurement of uranium and graphite was not primarily the responsibility of, the Metallurgical Laboratory but was handled through E. V. Murphree and others on the "planning board." In fact, the obvious interest of the Metallurgical Laboratory in the problem fed to continual intervention by its representatives. A great deal of the credit for the eventual success in obtaining materials is due to N. Hillberry and later R. I. Doan, always supported by A. H. Compton.
6.9. Obviously there would be no point in undertaking this whole project if it were not going to be possible to find enough uranium for producing the bombs. Early indications were favorable, and a careful survey made in November 1942 showed that immediate delivery could be made of adequate tonnages of uranium ores.
URANIUM OXIDE AND URANIUM METAL
6.10. At the end of 1941 the only uranium metal in existence was a few grains of good material made on an experimental basis by the Westinghouse Electric and Manufacturing Company and others and a few pounds of highly impure pyrophoric powder made by Metal Hydrides Company. The only considerable amount of raw material then available in this country was in the form of a commercial grade of black uranium oxide, which could be obtained in limited quantities from the Canadian Radium and Uranium Co. It contained 2 to 5 percent of impurities and was the material which gave a neutron multiplication factor of only about 0.87 when used in an exponential pile.
6.11. By May 1942, deliveries averaging 15 tons a month of black oxide of higher purity and more uniform grade started coming in. Total impurities were less than 1 percent, boron comprised a few parts per million, and the neutron multiplication factor (k) was about 0.98. (It is to be remembered that the multiplication factor depends also on the purity of the graphite.) Deliveries of this material reached a ton a day in September 1942.
6.12. Experiments at the National Bureau of Standards by J. I. Hoffman demonstrated that, by the use of an ether extraction method, all the impurities are removed by a single extraction or uranyl nitrate. The use of this method removed the great hulk of the difficulties in securing pure oxide and pure materials for the production of metal. Early in May 1942, arrangements were completed with the Mallinckrodt Chemical Works in St. Louis to put the new grade of oxide through an ether extraction process on a production basis for a further reduction in impurity content and to deliver the final product as brown dioxide. Deliveries started in July 1942 at a rate of 30 tons a month. This oxide is now used as a starting point for all metal production, and no higher degree of purity can be expected on a commercial scale. In fact, it was a remarkable achievement to have developed and rut into production on a scale of the order of one ton per day a process for transforming grossly impure commercial oxide to oxide of a degree of purity seldom achieved even on a laboratory scale.
6.13. The process which Westinghouse had been using to produce the metal was the electrolysis of KUF5, at a cost of about 51,000 a pound. Since the KUF5, was produced photochemically under the action of sunlight this method constituted a political bottleneck in production. It was found that uranium tetrafluoridc could be used instead of KUF5 and steps were taken to have this salt produced at the Harshaw Chemical Company in Cleveland and at the du Punt plant in Penns Grove, New Jersey. Production started in August 1942 and by October 1942 was up to 700 pounds per day at Harshaw and 300 pound, per day at du Punt, the method of manufacture in both cases being the hydrofluorination of Mallinckrodt-purified dioxide.
6.14. As the result of this supply of raw materials to Westinghouse, and as a result of plant expansion, deliveries Westinghouse had accumulated to a total of more than 6,000 pounds by November 1942 and were expected to be at the rate of 500 pounds per day by January 1943. The purity of the metal was good, and the cost had dropped to $22 per pound.
6.15. Deliveries of acceptable metal from Metal Hydrides Co. were delayed for various reasons and were just beginning ill. November 1942. This company's production was supposed to reach a thousand pounds per week thereafter.
6.16. Neither the Westinghouse process nor the Metal Hydrides Process was entirely satisfactory. Intensive activity designed to accelerate metal production, and carried out independently by F. H. Spedding and his associates at Iowa State College at Ames, Iowa, and by C. J. Rodden at the National Bureau of Standards, resulted in the development of a satisfactory method. Production facilities were set up at Ames in the fall of 1942 and had already produced more than one ton by the end of November. The process was extremely simple, rapid and low cost.
6.17. Further research indicated additional changes that could be made to advantage, and by the middle of 1943 Spedding at Iowa and other producers who entered the picture were using the final production method adopted.
6.18. By the end of 1942 arrangements had been made by the Manhattan District to increase metal production by making greater use of the Mallinckrodt Chemical Works, the Union Carbide and Carbon Corporation, and the du Pont Company.
6.19. To summarize, almost no metal was available during most of 1942, a fact that seriously delayed progress as we shall see, but the production problems had been nearly solved by the end of 1942 and some 6 tons of metal were incorporated in the pile built in November 1942. The whole problem of procurement of metal was taken over by the Manhattan District at the end of the year, under the general direction of Colonel Ruhoff, formerly with the Mallinckrodt Chemical Works. From the point of view of the Metallurgical Project no further serious delays or difficulty have occurred because of metal shortages.
6.20. At the beginning of 1942 graphite production was still unsatisfactory but it was, of course, in quite a different condition from the metal production since the industrial production of graphite had already been very large. The problem was merely one of purity and priority. Largely through the efforts of N. Hilberry, the National Carbon Company and the Speer Carbon Company were both drawn into the picture. Following suggestions made by the experts of the National Bureau of Standards, these companies were able to produce highly purified graphite with a neutron absorption some 20 per cent less than the standard commercial materials previously used. Although efforts further to reduce the impurities have had some success, the purity problem was essentially solved by the middle of 1942 and large orders were placed with the cooperation of the War Production Board. As in the case of the metal, the graphite procurement problem was taken over by the Manhattan District.
THE CHAIN REACTION
FURTHER INTERMEDIATE EXPERIMENTS
6.21. At the time that the Metallurgical Project was organized, most of the physicists familiar with the problem believed that a chain-reacting pile probably could be built if sufficiently pure graphite and pure uranium metal could be obtained. Enough work had been done on resonance absorption, on the theory of absorption and diffusion of neutrons in a pile, and on intermediate experiments to make it possible to design a lattice structure that had a very good chance of maintaining a chain reaction. Nevertheless, there were uncertainties in the experimental data and in the approximations that had to be made in the theoretical calculations. There were two alternatives: (I) to build a pile according to the best possible design; (2) to make more accurate determinations of the pertinent nuclear constants, to perform intermediate experiments, and to improve the calculations. There is little doubt that the first alternative was the one likely to lead most rapidly to the production of plutonium. There were many important questions which could have been answered more rapidly by such an operating pile than by a series of small-scale experiments. Unfortunately, the necessary amounts of materials were not available and did not become available for nearly nine months. Consequently, it was necessary to choose the second alternative, that is, to accumulate all relevant or possibly relevant information by whatever means were available.
6.22. The major line of investigation was a series of intermediate experiments. The particular set-up for each intermediate experiment could be used to test calculations based on separate auxiliary experiments. For example, the proportion of uranium oxide to graphite was varied, oxides of different purities were used, oxide was used in lumps of various sizes and shapes and degrees of compression, the lattice spacing was varied, the effect of surrounding the uranium oxide units with beryllium and with paraffin was tried; and, finally, piles of identical lattice type but: of different total size were tried to see whether the values of the multiplication factor k (for infinite size) calculated from the different sets of results were identical. In general, E. Fermi had direct charge of investigations of effects of impurities, and S. K. Allison had charge of tests involving different lattice dimensions. All these experiments strengthened the confidence of the group in the calculated value of k and in the belief that a pile could be built with k greater than unity. In July enough purified uranium oxide from Mallinckrodt was available to permit building intermediate pile No.9. As in previous experiments, a radium-l beryllium neutron source was placed at the bottom of the lattice structure and the neutron density measured along the vertical axis of the pile. By this time it was known that the neutron density decreased exponentially with increasing distance from the neutron source (hence the name often used for experiments of this type, "exponential pile") and that, from such rates of decrease, the multiplication constant k for an infinitely large pile of the same lattice proportions could be calculated. For the first time the multiplication constant k so calculated from experimental results, came out greater than one. (The actual value was 1.007.) Even before this experiment Compton predicted in his report of July 1 that a k value somewhere between 1.04 and 1.05 could be obtained in a pile containing highly purified uranium oxide and graphite, provided that the air was removed from the pile avoid neutron absorption by nitrogen.
AN AUXILIARY EXPERIMENT; DELAYED NEUTRONS
6.23. We shall not mention a majority of the various auxiliary experiments done during this period. There was one, however, the study of delayed neutrons-that we shall discuss because it is a good example of the kind of experiment that had to be performed and because it concerned one effect, not heretofore mentioned, that is of great importance in controlling a chain-reacting pile.
6.24. From previous investigation, some of which were already published, it was known that about 1 per cent of the neutrons emitted in fission processes were not ejected immediately but were given off in decreasing quantity over a period of time, a fact reminiscent of the emission of beta rays from short-lived radioactive substances. Several half-lives had been observed, the longest being of the order of a minute.
6.25. It was realized early that this time delay gave a sort of inertia to the chain reaction that should greatly facilitate control. If the effective multiplication factor of a pile became slightly greater than 1, the neutron density would not rise to harmfully large values almost instantly but would rise gradually so that there would be a chance for controls to operate. (Other time intervals involved, such as those between collisions, arc too small to be useful.)
6.26. Because of the importance of this effect of delayed neutrons for control it was decided to repeat and improve the earlier measurements. (The fact that this was a repetition rather than a new measurement is also typical of much of the work in physics at this period.) A description of the experiment is given in Appendix 3. The results indicated that 1.0 per cent of the neutrons emitted in uranium fission are delayed by at least 0.01 second and that about 0.7 per cent are delayed by as much as a minute. By designing a pile such that the effective value of k, the multiplication factor, is only 1.01 the number of delayed neutrons is sufficient to allow easy control.
THE CHAIN-REACTING PILE
6.27. By the fall of 1942 enough graphite, uranium oxide, and uranium metal were available at Chicago to justify an attempts to build an actual self-sustaining chain-reacting pile. But the amount of metal available was small - only about 6 tons- and other materials were none too plentiful and of varying quality. These conditions rather than optimum efficiency controlled the design.
6.28. The pile was constructed on the lattice principle with graphite as a moderator and lumps of metal or oxide as the reacting units regularly spaced through the graphite to form the lattice. Instruments situated at various points in the pile or near it indicated the neutron intensity, and movable strips of absorbing material served as controls. (For a more complete description of the pile, see Appendix 4.) Since there were bound to be some neutrons present from spontaneous fission or other sources, it was anticipated that the reaction would start as soon as the structure had reached critical size if the control strips were not set in "retard" position. Consequently, the control strips were placed in a suitable "retard" position from the start and the neutron intensity was measured frequently. This was fortunate since the approach to critical condition was found to occur at an earlier stage of assembly than had been anticipated.
6.29. The pile was first operated as a self-sustaining system on December 2,1942. So far as we know, this was the first time that human beings ever initiated a self-maintaining nuclear chain reaction. Initially the pile was operated at a power level of ½ watt, but on December 12 the power level was raised to 200
ENERGY DEVELOPED BY THE PILE
6.30. In these experiments no direct measurements of energy release were made. The number of neutrons per second emitted by the pile was estimated in terms of the activity of standardized indium foils. Then, from a knowledge of the number of neutrons produced per fission, the resultant rate of energy release (wattage) was calculated.
6.31. Evidently this experiment, performed on December 2 just as a reviewing committee was appraising the Chicago project, answered beyond all shadow of doubt the first question before the Metallurgical Laboratory; a self-sustaining nuclear chain reaction had been produced in a system using normal uranium. This experiment had been performed under the general direction of E. Fermi, assisted principally by the groups headed by W. H. Zinn and H. L. Anderson. V. C. Wilson and his group had been largely responsible for developing the instruments and controls, and a great many others in the laboratory had contributed to the success of the enterprise.
RELATION BETWEEN POWER AND PRODUCTION OF PLUTONIUM
6.32. The immediate object of building a uranium-graphite pile was to prove that there were conditions under which a chain reaction would occur, but the ultimate objective of the laboratory was to produce plutonium by a chain reaction. Therefore we are interested in the relation between the power at which a pile operates and the rate at which it produces plutonium. The relation may be evaluated to a first approximation rather easily. A pile running stably must be producing as many neutrons as it is losing. For every thermal neutron absorbed in U-235 a certain number of neutrons, n, is emitted. One of these neutrons is required to maintain the chain. Therefore, assuming the extra neutrons all are absorbed by U-238 to form plutonium, there will be n -1 atoms of Pu239 formed for every fission. Every fission releases roughly 200 Mev of energy. Therefore the formation of n- 1 atoms of plutonium accompanies the release of about 200 Mev. Since n - 1 is a small number, we can guess that to produce a kilogram a day of plutonium a chain-reacting pile must be releasing energy at the rate of 500,000 to 1,500,000 kilowatts. The first chain-reacting pile that we have just described operated at a maximum of 200 watts. Assuming that a single bomb will require the order of one to 100 kilograms of plutonium the pile that has been described would have to be kept going at least 70,000 years to produce a single bomb. Evidently the problem of quantity production of plutonium was not yet solved.
THE CHEMISTRY Of PLUTONIUM
6.33. The second specific objective of the Metallurgical Laboratory was to show that, if a chain reaction did occur, it would be feasible to separate the plutonium chemically from the other material with which it is found. Progress toward this objective was necessarily slower than toward the attainment of a chain reaction. Initially little was done at the Metallurgical Laboratory on chemical problems although the extraction problem was discussed in a conference soon after the project was organized and the work of Seaborg's group at the University of California on plutonium was encouraged. On April 22-23, 1942, a general conference on chemistry was held at Chicago, attended by F. H. Spedding, E. W. Thiele, G. T. Seaborg, J. W. Kennedy, H. C. Urey, E. Wigner, N. Hilberry, G. E. Boyd, I. B. Johns, H. A. Wilhelm, I. Perlman, A. C. Wahl, and J. A. Wheeler. Spedding, in opening the meeting, pointed out that there were two main tasks for the chemists: first, to separate plutonium in the amount, and purity required for war purposes: second, to obtain a good understanding of the chemistry necessary for the construction and maintenance of the pile. The separation problem was to be studied by a new group at Chicago under the direction of Seaborg, by Johns and Wilhehn at Ames, and by Wahl and Kennedy continuing the work at California. Other closely related groups at Chicago were to be C. D. Coryell's, working on the fission products, and Boyd's on analytical problems. The chemistry group at Chicago has grown speedily since that time. A new building had to be constructed to house it late in 1942, and this building was enlarged subsequently. Altogether, the solving of many of the chemical problem, has been one of the most remarkable achievements of the Metallurgical Laboratory.
6.34. The first isotope of plutonium discovered and studied was not the 239 isotope but the 238 isotope, which is an alpha-ray emitter with a half-life of about 50 years. U-238 bombarded with deuterons gives 93Np238 which disintegrates to 94Pu238 by beta emission2.The first evidence of the actual existence of these new elements (ruling out the original erroneous interpretation of the splitting of uranium as evidence for their existence) was obtained by E. McMillan and P. H. Ahelson who isolated 93-238 from uranium bombarded with deuterons in the Berkeley cyclotron. This new element was identified as a beta emitter but the sample was too small for isolation of the daughter product 94-238. Later, enough Pu-238 was prepared to permit Seaborg, Kennedy and Wahl to begin the study of its chemical properties in the winter of 1940-1941 by using tracer chemistry with carriers according to practice usual in radiochemistry. By such studies many chemical properties of plutonium were determined, and several possible chemical processes were evolved by which Pu-239 might be removed from the chain-reacting pile. The success of experiments on a tracer scale led to plans to produce enough Pu-239 to be treated as an ordinary substance on the ultra-microchemical scale. Such quantities were produced by prolonged bombardment of several hundred pounds of uranyl nitrate with neutrons obtained with the aid of cyclotrons, first at Berkeley and later at Washington University in St. Louis. By the end of 1942, something over 500 micrograms had been obtained in the form of pure plutonium salts. Although this amount is less than would be needed to make the head of a pin, for the micro-chemists it was sufficient to yield considerable information; for one microgram is considered sufficient to carry out weighing experiments, titrations, solubility studies, etc.
6.35. From its position in the periodic table; plutonium might be expected to be similar to the rare earth or to uranium, thorium, or osmium. Which of these it will resemble most closely depends, of course, on the arrangement of the outermost groups of electrons and this arrangement could hardly have been predicted. On the whole, plutonium turned out to be more like uranium than like any of the other elements named and might even be regarded as the second member of a new rare-earth series beginning with uranium. It was discovered fairly early that there were at least two states of oxidation of plutonium. (It is now known that there are four, corresponding to positive valences of 3, 4, 5, and 6.) Successful microchemical preparation of some plutonium salts and a study of their properties led to the general conclusion that it was possible to separate plutonium chemically from the other materials in the pile, This conclusion represents the attainment of the second immediate objective of the Metallurgical Laboratory. Thus, by the end of 1942, plutonium, entirely unknown eighteen months earlier, was considered an clement whose chemical behavior was as well understood as that of several of the elements of the old periodic table
6.36. Besides the major problems we have mentioned, i.e., the chain reaction, the chemical separation, and the planning for a production plant, there were innumerable minor problems to be solved. Among the more important of these were the improvement of neutron counters, ionization chambers, and other instruments, the study of corrosion of uranium and aluminum by water and other possible coolants, the determination of the effects of temperature variation on neutron cross sections, the fabrication of uranium rods and tubes, the study of fission products, and the determination of the biological effects of radiation. As typical of this kind of work we can cite the development of methods of fabricating and coating uranium metal, under the direction of E. Creutz. Without the accomplishment of these secondary investigations the project could not have reached its goal. To give some further idea of the scope of the work, a list of twenty report titles is presented in Appendix 5, the 20 reports being selected from the 400 or so issued during 1942.
THE FAST-NEUTRON REACTION
6.37. The third initial objective of the Metallurgical Project was to obtain theoretical and experimental data on a "fast neutron" reaction, such as would be required in an atomic bomb. This aspect of the work was initially planned and coordinated by G. Breit of the University of Wisconsin and later continued by J. R. Oppenheimer of the University of California. Since the actual construction of the bomb was to be the final part of the program, the urgency of studying such reactions was not so great. Consequently, little attention was given to the theoretical problems until the summer of 1942, when a group was organized at Chicago under the leadership of Oppenheimer.
6.38. In the meantime experimental work initiated in most instances by G. Breit, had been in progress (under the general direction of the Metallurgical Project) at various institutions having equipment suitable for fast-neutron studies (Carnegie Institution of Washington, the National Bureau of Standards, Cornell University, Purdue University, University of Chicago, University of Minnesota, University of Wisconsin, University of California, Stanford University, University of Indiana, and Rice Institute). The problems under investigation involved scattering, absorption and fission cross section, the energy spectrum of fusion neutron." and the time delay in the emission of fission neutrons. For the most part this work represented an intermediate step in confirming and extending previous measurements but reached no new final conclusion. This type of work was subsequently concentrated at another site (see Chapter XII).
6.39. As indicated by the "Feasibility Report" (in a section Written by J. H. Manley, J. R. Oppenheimer, R. Serber, and E. Teller) the picture had changed significantly in only one respect since the appearance of the National Academy Report a year earlier. Theoretical studies now showed that the effectiveness of the atomic bomb in producing damage would be greater than had been indicated in the National Academy report. However, critical size of the bomb was still unknown. Methods of detonating the bomb had been investigated somewhat, but on the whole no certain answers had been reached.