Nuclear Museum Logo
Nuclear Museum Logo

National Museum of Nuclear Science & History

John Arnold’s Interview

Manhattan Project Locations:

John Arnold joined the Manhattan Project in 1943 when the MED tasked his employer, the Kellogg Corporation, with developing a special barrier for the gaseous diffusion plant in Oak Ridge. Arnold discusses his role as director of research and development and process engineering at the plant, where he supervised the assembly and testing of what would become the K-25 plant. In his interview, Arnold describes the challenges of creating a suitable barrier that could withstand the corrosive effects of uranium hexafluoride gas while remaining porous enough to allow smaller atoms of uranium-235 to pass through.

Date of Interview:
July 27, 1965
Location of the Interview:


Stephane Groueff: Now if you can give me briefly your background and where you came from and how you got connected with Kellex. Were you a Kellogg man?

John Arnold: Yes, I was a Kellogg man, and at the time I was working on an ammonia plant at Sterlington.

Groueff: Where is that?

Arnold: In Louisiana.

Groueff: Louisiana?

Arnold: Yeah, for Kellogg doing some experimental and development work to produce ammonia from a waste gas product at the Commercial Solvents Plant at Sterlington. They had a plant there where they made carbon black by taking natural gas and subjecting it to heat and cracking it into its elements of carbon and hydrogen.

Groueff: You’re a chemist?

Arnold: A chemical engineer by training.

Groueff: Most of the Kellogg people were chemical engineers?

Arnold: Chemical engineers mostly, but a large number of mechanical engineers as well, but it was a chemical engineering construction firm basically. A good part of their work was in the petroleum refining business, but they did some chemicals. At that time they weren’t called petrochemicals but ammonia was one, methanol was another—that type of thing.

At the completion of that project, which was, as I recall, about in November or December of 1942, I came back to take over the direction of Kellogg’s activity under the OSRD in the study of the feasibility of building a uranium isotope separation plant.

Groueff: Because Kellogg was already involved since the beginning of the year, I think. I was told by [Manson] Benedict and [Clarence] Johnson that they were already told by [Percival] Keith about this possibility, and they were briefed by [Eger] Murphree—

Arnold: Yes, that’s right.

Groueff: So, a secret meeting, and they started working at the Kellogg Laboratory in New Jersey. So all of this is in ’42 before the formation of Kellex?

Arnold: That was Kellogg, and it was called Project X, indicating that it was secret. Subsequently when the Manhattan Project was formed and Kellogg was seeking a contract to be the designers of this gas diffusion plant, because of the complications of accounting and a different type of business, this was to be a cost plus a fee type of business. They had fixed price contracts and so on. The allocation of costs for reimbursable purposes and so on was so complex that Kellogg decided they would form a separate corporation, and there would be no doubt that all the costs to that corporation would be costs against the project. So they took the “X” from Project X, “Kel” from Kellogg, and called it the Kellex Corporation.

Groueff: So you were called from Louisiana to head—

Arnold: I had completed my work in Louisiana, and I came back to the research laboratories at Jersey City where I had worked, and really I was working out of the Jersey City laboratories when I was in Louisiana.

Groueff: But who told you first about the secret thing? You didn’t know when you came back?

Arnold: Keith told me about it and asked me if I would take over the direction of this work. There was, at that time, some loose ends between the work that Benedict and Johnson and [Lombard] Squires were doing.

Groueff: They were the first men? The men that you found—Benedict, Johnson, Squires.

Arnold: It’s pretty hazy to me as to how many others there were there, but there were probably a group of six or eight, something like that.

Groueff: A small group?

Arnold: A small group. There was some lack of coordination between what they were doing among themselves, and also what they were doing in conjunction with Columbia University at that time. At that time, this was a project that was dominated by Columbia University.

Groueff: Did you have a contract for that?

Arnold: We had a contract with the Office of Scientific Research and Development.

Groueff: But not anything very big and important? 

Arnold: No.

Groueff: Just for one type of work?

Arnold: Well, to make really an engineering feasibility study of the fundamental concepts that—

Groueff: But not to make the plant?

Arnold: Not to make the plant, no.

Groueff: Just to see whether it is possible to?

Arnold: To see whether it is feasible.

Groueff: And to give estimates?

Arnold: And to point out the problems and the risk areas, and to a degree, as well as one could, accurately on the basis of the data available indicated what further development work would have to be done and how long it would take to do it. Get a master plan so to speak, although it was really a very rough, primitive plan of what it would take to wind up with enough U-235.

Groueff: The engineering side of Columbia University’s project. But they were the important people at that time of the project?

Arnold: That is right.

Groueff: And [John] Dunning was the leading man, and [Harold] Urey was supervising?

Arnold: Well, both Dunning and Urey had very important parts in it and I was never quite sure who was the head of it.

Groueff: Nominally it was Urey, but the real sort of work was done by Dunning, which actually created this friction.

Arnold: Yes, I think so.

Groueff: But in your organization, the man in charge of that was Keith since the beginning?

Arnold: Yes. Well, Keith was in charge of it because Keith was the executive officer who had charge of all of Kellogg’s engineering activities at that time.

Groueff: Oh, he was the executive for the whole thing?

Arnold: The whole company, yes.

Groueff: So he was a very important man there?

Arnold: Oh yes.

Groueff: And there was also Mr. [Morris Woodruff] Kellogg, no?

Arnold: There was Mr. Kellogg, who organized the company, and then there were other vice presidents. There was Mr. Harverson, Mr. Johnson, and they were both more in the finance and sales effort, and Keith was the engineering executive.

Groueff: He was relatively young, no? In his forties?

Arnold: Yes, he was born in the 1900s, so he was forty-two.

Groueff: So it was quite a position for him. Was the company very important at that time?

Arnold: Yes, the company at that time was a recognized world leader in the design of chemical engineering plants, and this was the basis of their selection. Murphree, who as I recall it was a member of one of the original OSRD committees, was the director of research and development for Standard Oil of New Jersey. It was he who recommended the Kellogg Company, and primarily because Keith was a recognized, dynamic—

Groueff: He was known in the business among you engineers? He was a well-known figure and known for his dynamism?

Arnold: Yes, I would say there were three outstanding, creative engineers in the petroleum processing business at that time. He was one of the three. There was DeFlores and there was—it escapes me for the moment. He worked for UOP [Universal Oil Production Company]—Gus[tav] Egloff. These three individuals really, working independently and developing independent processes, did account for a large amount of the independent thinking that went into the rapid development of petroleum refining technology in this country.

Groueff: So he had enough prestige? So you knew him very well.

Arnold: I worked for him so I knew him very well, but he was well known in the industry. Yes, indeed.

Groueff: And how was he at that time? The same dynamic and full of ideas? What would you say was his main quality?

Arnold: I think that he had, for one thing, the ability to attract and interest very highly intelligent, competent people and was not only inventive himself, but he was a catalyst for ideas from others. I think as I look upon Keith, really his greatest attribute was his tremendous capacity to relate things almost through intuition rather than arrive at a conclusion by a series of exact calculations. 

Groueff: He wasn’t a methodical type, sort of meticulous worker. He was more intuition and ideas.

Arnold: Right, but this is not to infer that he didn’t have a tremendous intellectual capacity for understanding ideas. He could talk on any subject, whether it be music or—

Groueff: Yeah, I was surprised. He knows about paintings, about history, about Italian renaissance and things like that.

Arnold: He is very widely read and reads very rapidly, and I guess if you visit him at his home—

Groueff: I have been several times to his office and he is quite a gourmet. He knows about French wines and cooking and things like that, and paintings. The contact with him was pleasant. He wasn’t one of those strong but rather bullish men that intimidate people, no?

Arnold: Well, he was impatient at times and he was capable of making a hard decision when the moment came to make a hard decision regardless of what effect it might have. If I were wrong and someone else was right and it came to decide who was right, he was ready to make the decision. If it was a technical decision as to whether we should make the barrier in the form of tubes or flat plates, no matter how many people were working on the flat plate diligently, he was capable of cutting it off and saying, “That’s not for us.”

Groueff: That’s what he has in common with people like Groves—the strength to make a decision without too much hesitation. When he wanted, he could turn his charm on.

Arnold: Oh indeed, yes. He also had the capacity to have individual groups doing if not the same thing, related things to get a competitive thrust to the development of ideas and competition.

Groueff: A good psychologist?

Arnold: Yes, he was a good psychologist.

Groueff: He knew people?

Arnold: He had the ability of commanding the respect of anyone no matter what their intellectual attainment. He could charm Hugh Taylor or the janitor.

Groueff: So when Urey came here, he was the dominant figure in this project?

Arnold: Right.

Groueff: And he asked you to become in charge of the coordination and organizing the group, but it still didn’t have the proportions that it took later? I mean, you didn’t know that it was going to be something colossal?

Arnold: Well we knew it was colossal if it ever went forward, and of course we were all enthused about it when we heard of it, but to have a full understanding of all of its implications and the prospects of bringing it to a successful conclusion gave everyone pause at least. I wouldn’t say at that time we all felt, “Well, this is the biggest thing that Kellogg Company has ever done,” which it later turned out to be. We did have a number of sessions with the Columbia University group and with the committee that Keith talks about of Warren Lewis and [Crawford] Greenewalt and Roger Williams reviewing progress—

Groueff: Ah yes, the reviewing committee. Tom Gary—DuPont people.

Arnold: Until finally there was a decision made – in which I have no knowledge of what went on and who pulled the strings and whatnot – but the decision was made to negotiate a contract with Kellogg to undertake to build the plant. At that time, Kellogg organized the Kellex, and we took some space in the Woolworth Building.

Groueff: It was only Kellex—not Kellogg?

Arnold: It was only Kellex. One of the contributions of Kellogg was to have Mr. Keith devote full time to this project where he had been involved in many other projects. In fact, all the technical projects of the company. We moved across the street from the transportation building where we had our main offices and the Jersey City laboratory where we had our Research and Development laboratories to the Woolworth Building just directly across the street.

Groueff: When was that? The end of ’42 or the beginning of ’43?

Arnold: I believe it was the beginning of ’43, but I am not very clear now as to just when it was. I think it was about that time.

Groueff: And the decision was made probably by the Washington people and the Manhattan District and all these committees—Dr. [Vannevar] Bush and Keith. Your side was with Keith, who was negotiating the high policy involvement of the company on this?

Arnold: Keith and the other Kellogg executives.

Groueff: I see. You weren’t in charge of the Kellogg side of the operation? You wouldn’t negotiate with the outsiders for the Manhattan District at all, that was Keith?

Arnold: Let me tell you how I came into it. I was in charge of all of the activities reporting directly to Keith when we were in the OSRD phase, so to speak. But then as we organized the company to take it over, then we split the responsibilities. I assumed administrative and technical responsibility for the more highly technical side, and Al Baker assumed responsibility for the engineering, drawings, procurement of equipment—that type of thing.

Groueff: By technically, you mean also the scientific development of components and the cascade and all this?

Arnold: I had the title, at that time, of director of research and development and process engineering at the startup of the plant. To give you an idea of what this means, it is simply this: that when you design a plant, you first try to determine what does the plant have to do chemically and engineering-wise. You make flow sheets, very much like you see flow sheets of battles or anything else. It is just a lot of blocks and so on. Then you start to say how big the vessels have to be, what the pressures and temperatures are, and what has to be inside the vessels, and how it flows from one vessel to another.

This is what is known as the process. You carry out unit operations. You pump. How much do you have to pump? Up to what pressure do you have to pump it? You transfer heat if something is hot and something is cool and you want to preserve the energy, you go through a heat transfer mechanism. How big does that have to be? How much surface does it have? If you have a cascade, how many stages do you have to have in the cascade, and what size do those stages have to be? What is the circulation around a stage? What is the net flow up the cascade, and what is the net flow down the cascade? Where does the feed come in? That type of thing.

Groueff: The job of the design of all this was the job of Manson Benedict?

Arnold: Manson Benedict, at that time, was head of the Process Department. The other sections that I had—we had actual research and development work done, and that was under Clarence Johnson.

Groueff: Mostly on barriers and things like that?

Arnold: Well yeah, the whole gamut really. Then we had development of pumps and seals, so mechanical equipment valves and the pumps and seals were under George Watts with [Jusdon] Schweringen working on the seals. The development of the valves and instruments was under Al Baker as part of his engineering assignment.

Now, once the plant was ready to operate, then the supervision of the operation—to be sure that we were turning the right valves and we had the right procedures written up and we had the right instruction classes for training the operators, guiding the operators on the start-up—

Groueff: Wasn’t that your job?

Arnold: Yes.

Groueff: Wasn’t that Union Carbide?

Arnold: Let me tell you how that works so you clearly understand the distinction. When a contractor like Kellogg sells a job to Union Carbide—now actually this was to be owned by the government, but in effect they were selling it to the operator—then Kellogg writes all the basic operating instructions, tells how the equipment is to be operated, and then demonstrates the guarantee, so to speak.

They will come in with a limited number of people, and we’ll supervise what valves are to be turned. We won’t actually turn the valves, but we’ll supervise what valves will be turned, what pumps will be started, when they’ll be started, how fast the compressors will be run and so on. And then show that when you do it the way that Kellogg says, that you will, in fact, produce the guaranteed quality and quantity of material. Union Carbide had the job of operating the plant, but the supervision of the startup was my responsibility.

Groueff: I see. They take it only once it is working perfectly. The startup is your responsibility if something doesn’t—

Arnold: Yeah. I could describe it like this: if you were to buy a boat and you own it, you are going to operate it, you are going to maintain it. But until you are satisfied that it can go through the paces that I represented it could go through, I would sit alongside you and say, “Well, push the throttle a little harder, throw the gear up into this corner and it will go backwards.”

Groueff: And you even will bring your crew for the first day or so?

Arnold: Well, I’d say it was a matter of months that we were there.

Groueff: At the end you moved to Oak Ridge?

Arnold: I didn’t move to Oak Ridge, although I spent a lot of time there at the guest house. I never did move my residence. Now, Clarence Johnson did, and Clarence Johnson was directly under me and he’d moved over from the development phase to the operating phase.

Groueff: So, to continue your story, in the end of ’42 you came here and you were given the responsibility for the Project X, and then you were informed that the company was going to form a new company and Kellex moved to Woolworth Building. Now, how many people did you employ at that time at Kellex?

Arnold: Well, we started out with about, I would say, a dozen or something of us. We had to recruit not only from Kellogg staff but from industry. A large number of people that we finally wound up with, and I forget the number now, but it was about 3000 or something like that.

Groueff: 3000 in New York?

Arnold: We had them all around the country. We had some in New Jersey, we had some up at the SAM [Special Alloy Materials] Laboratory here, we had some out at the barrier plant.

Groueff: 3000 people about?

Arnold: That’s my recollection of it, but it’s not 300, that’s for sure. It’s something of that order.

Groueff: So it became an enormous operation?

Arnold: Yes.

Groueff: How did you recruit the top people? That could be very interesting to explain to the public. Kellex was a very unusual firm in the sense that the members were all very well established and top people in their own companies or industries, no? Scott was already vice president of a big company, and George Watts was very well established.

Arnold: Yeah, I think he was vice president or at least chief engineer of Standard Oil of Indiana.

Groueff: And then [James C.] Hobbs was a big authority in his field. Don’t you think that is a rather unusual conglomeration? Sort of it was like an all-star team, you know?

Arnold: Yes, and the way it was done was primarily through personal knowledge that started within the original group of individuals that could contribute and then panned out as individuals coming into the project would indicate others. We also had the great advantage of being able to go to any company where an individual was working that we desired, and seldom were we turned down when we wanted them. We had the support of the Army and everybody else.

Groueff: Who would decide among yourselves if you had a problem? Keith would discuss it with you and Baker? You decide there is a guy in Chicago?

Arnold: Yes and what we did was—

Groueff: Any one of you would suggest the person?

Arnold: Yes, I think to a very large degree we all did our own recruiting. When Benedict needed a man, to a large degree Benedict went out and got a man.

Groueff: And if it didn’t work, then you helped him, and you go higher to the Manhattan Project to sort of recruit him.

Arnold: Well, we didn’t have any real problem that way because even Benedict or Johnson could tell the story very effectively, and we had a great many people that really wanted and were anxious to work with us.

Groueff: But you were not supposed to tell them the real whole story at the beginning, no?

Arnold: No, we had a simple principle that we were only supposed to divulge that part of the classified information that was essential to a reasonable performance of the duty of the individual.

Groueff: But would you tell them that the purpose was to build an atomic bomb?

Arnold: No, not necessarily. Now, there were many of us that knew this was the purpose but—

Groueff: Well, if you wanted to recruit some important engineer, you go and talk to him. If you ask me to go work for you my first question will be, “What are we going to produce?” What would you say?

Arnold: We would say, “This is a top secret project that has the highest priority, and we would like you, with your specialized knowledge, to join us in producing a valve,” if it was a valve. But on the other hand, if we wanted to ask someone to come into our process design group where we were going to fashion the fundamental separation of the uranium atoms or isotopes, then of course he would know very well what we were doing.

Groueff: And usually it worked? People didn’t want to know more, and you didn’t have problems?

Arnold: Well, this is another situation. We did have security problems constantly, as you would understand in a situation like this. We had security officers.

Groueff: And they had to approve when you bring a colleague or engineer, you had to tell the security people to get him clearance?

Arnold: Right.

Groueff: And if they don’t, you can’t employ them?

Arnold: We certainly have to take him off sensitive work. There were all kinds of problems. Sometimes we had just weak characters. You know the problems you have with security risks and that nature. We had some individuals suspected of other security problems, and we were never told any more details of the case than was necessary to take whatever action was required. 

Sometimes we were asked to take people off sensitive work and quietly squeeze down on the flow of confidential information, but there was never any abrupt firing. So that would indicate that we had a hold on a situation, and we were about to expose the activities of a group. 

I wasn’t in security at that time, but I assume the feeling was they would rather know where these people were and what they were doing and following what they were doing and finding out what leads they could get, rather than to just close the door. We had no problems in recruiting good people, and this is indicated by the very large number of extremely competent individuals.

Groueff: Did that cause a new problem of having subordinate people, who were by then used to being the top man in their company? Wouldn’t that create some problems, to be the boss of people who are such important people in their companies?

Arnold: Well, one of the ways that we accomplished this was a relatively loose organization. We didn’t have the hard lines that one would have and we didn’t have the titles spread around the way you might have in a normal organization. Added to that, Keith had a tremendous capacity for details, and himself was always probing—even in the Kellogg company—way down the line to fellows at the laboratory bench to see what was in the test tube type of things so that everyone had a sense, at least to a degree, of working for Keith.

Groueff: Which, I understand sometimes created some problems like with Al Baker, because I understand Keith was impatient, and he would go over his head and go down, as you said, to a worker. And I understand that Baker, by temperament, was a different man and wanted very orderly, methodical organization with everybody knowing his place and reporting through the line. The two men, I understand, were very opposite in temperament and way of working.

Arnold: Well, they were opposite in many ways. As you say, they both were dynamic, strong individuals, and they both had a knack of getting a lot of work out of their fellow workers. They were both leaders, there is no question about that. Now, Al never had the intellectual capacity nor the intellectual inquisitiveness to lead him to want to understand my side of the business, for example.

On the other hand, he had a tremendous capacity for taking ideas and converting them into an operating plant. Just tremendous. And he had a fantastic capacity for not only organizing a group to accomplish this in, but recruiting people for it. Probably the greatest contributions he made were to recruit people for the engineering phases of the work and to organize it into an effective team that was schedule conscious.

He was a great heckler of mine, for example, because many of the things we were still trying to develop he wanted answers right now so that he could put them into steel and concrete. I could readily understand his problems, but we have to understand our own problems as to, “Were we really ready?”

Groueff: Yeah, it’s very unusual also to work in a normal operation because the engineering people will wait until your research people finish the job, and then it would go to some pilot plant.

Arnold: Well, we had to work this simultaneously. We did have a pilot plant, of course, in Jersey City where we tried out parts of the cascade, but we were also building the plant before we got to run and get all the data from the cascade in Jersey. We had terrific responsibilities with respect to whether or not the designs that we were putting into being—if they didn’t work the first time—could be modified so that they would work.

One of our biggest problems was corrosion—both corrosion of the internals of the containers and also the corrosion of the barrier. Part of this was the in-leakage of air through the rotating seals; you have seen in a motorboat with the propeller sticking out the back and the little stuffing box that seals that shaft so that the water doesn’t come into the boat? Well, these seals were somewhat more sophisticated, but they were on centrifugal compressors or pumps, as they were called, to prevent the air from leaking into the process, which was run under vacuum.

Groueff: And there was no such thing invented at that time in existence?

Arnold: There were no such things in existence which would meet the requirements of the real load in leakage. We could only tolerate a very, very minor amount of particularly moisture getting in, otherwise the moisture in the hexafluoride would react in a way to cause solids, which would plug up the holes.

Groueff: And you couldn’t have any oiling for the bearings and things like that, right?

Arnold: Right, that would have the same problems. But compressive seals is an old, old art. This was just something much finer than had ever been done before – order of magnitude fine.

Groueff: What were the main sort of most dramatic difficulties and obstacles that you met in your part of the work? Were there moments when the whole thing seemed hopeless?

Arnold: I don’t think there were moments when it seemed hopeless. I think there were moments when the schedule seemed less than firm, when we ran into problems a solution to which was essential before we could complete the plan. Until we got those solutions, we really couldn’t tell when we were going to get the plan on stream.

Groueff: Would you say the problems were on the cascade, seals, the barrier, and the valves and the corrosion?

Arnold: The corrosion was a big problem. We had to condition all the surfaces that this hexafluoride was exposed to, otherwise we would get reaction. The unfortunate part of it was that almost invariably when we got a reaction to the hexafluoride, which was a gas under these conditions, we would get products which were solid. Then if it started to float around it would plug up the barrier.

Groueff: Then this process of nickel plating was developed?

Arnold: On the barrier? You mean for the internals?

Groueff: For all the pipes and everything actually was nickel-plated inside?

Arnold: Right, nickel plated. Then we had to learn how to take that nickel-plated surface, which was still attacked by the hexafluoride, and react it with fluorine, which is one of the components of uranium hexafluoride. And then get a layer—almost a monomolecular layer—a thin skin.

Groueff: Of the same gas?

Arnold: It had not the same gas, but it had the same component, the same active component basically. Uranium hexafluoride was used first because this was a gaseous diffusion plant, and uranium is something like lead, only a little bit more so. To convert that into a gas, you have to change it into a chemical compound. One of the few compounds known that will vaporize is uranium hexafluoride. Now, fluorine itself is a very light gas.

Groueff: And you gave it fluorine?

Arnold: We would prepare the surface with fluorine.

Groueff: Was that the contribution that you mentioned of Elsie?

Arnold: Elsie had a great deal to do with this, but this again was just as complex as the barrier. I believe the fundamental idea of treating with fluorine to stabilize the barrier, as we called it, was a contribution of the Columbia group. We had to clean the surfaces first.

Groueff: I thought nickel-plating was enough to prevent corrosion?

Arnold: I hope this isn’t classified. I don’t believe it is.

Groueff: No, we have a lot about all the nickel-plating and the corrosion. What would be classified is all the specifications—how exactly it is done and how much of it. But I was told about the nickel plating also by Mr. Keller from Chrysler, because I think they developed this matter. Then there was this Bart Company.

Arnold: Bart over here in Jersey.

Groueff: They worked for you?

Arnold: Chrysler built the largest diffusion units and they nickel-plated the insides of those. Bart plated the inside of the piping that connected one stage to the next.

Groueff: As I understand at that time in 1942 and 1943, nickel-plating the inside of curved surfaces was quite an unusual process.

Arnold: Yes, and a very expensive process. There had been no real economic need for it up until that time.

Groueff: So they found some good techniques?

Arnold: Yes, they developed some fine techniques, and it worked out very well.

Groueff: And the cascade, where was that work done? In New Jersey, or here in the Woolworth Building?

Arnold: The cascade itself—the process design of it really is the development of a mathematical model. You know that a barrier with certain sized holes will pass on the average a certain number of molecules of the 235 and a relatively lower percentage of the molecules of the 238 so you get a certain enrichment, but it is not complete. Then you have to calculate how many times you have to repeat this, and this is what is the cascade.

The reason it is called a cascade is that you start out with a very large amount of uranium, which has a low abundance of the U-235—as I recall seven parts per thousand of U-235—and the rest is U-238 and traces of other compounds. You have to handle a large amount of that material, and now as you go into the first stage of the feed point you get a separation, but there isn’t a great deal of difference between the concentration of the enriched material. It has more U-235, but just a little bit more. The depleted material has U-235, but just a little bit less, so you have to keep working both of them.

Groueff: Tons of times?

Arnold: If we just follow the rejected material, it goes into the feed stage, and the U-235 goes through the barrier more in proportion to its concentration than the U-238, so the depleted material passes by the barrier and it flows down into the next stage. In this stage now, some U-235 flows out of it, and that U-235 now has a concentration of about the same as the feed, so it flows back up into the first stage. So the heavy material starts wandering down.

Groueff: And the lighter up?

Arnold: And the lightest starts going up. Now, as you go toward the end of the feed section, you are still handling massive materials because most of the material is flowing down—so these are very large stages, but they do cascade. They get a little bit smaller. As you flow up to enrich the uranium, then it—being a smaller fraction of the total—goes up, and gets really much and much smaller until finally the top stages of the cascade are very small. So that is what it is called. It’s a cascade. It goes like this.

Groueff: It’s a mathematical thing to calculate?

Arnold: Yes, you can calculate what the separation is for each stage, and then you calculate how many stages.

Groueff: How can they even calculate if they don’t have the barrier or the pump? You have to work with something to know.

Arnold: You have to have in your mind a barrier, and then you have to go to the laboratory.

Groueff: A theoretical barrier?

Arnold: It’s a theoretical barrier in that it isn’t in existence, but it is also a fundamental barrier in that it is the separation that is reasonable to obtain and also gives you a reasonable sized cascade.

Groueff: It must be difficult to work without having the actual thing—the same thing that they didn’t know what kind of pumps or compressors.

Arnold: No, once you decide what kind of barrier you are going to get, then you decide what kind of a pump you need to go along with that barrier. I’m at a loss to know how much of the barrier is classified and how much isn’t. I just haven’t kept up with what has been declassified. 

Basically, on the basis of the progress at that time, in very small samples at Columbia we and the process design group decided what was a reasonable barrier with the concurrence of the Columbia group. Then we designed a cascade on the basis of that.

Groueff: So you worked very closely with them?

Arnold: We worked very, very closely with them all throughout the operation.

Groueff: And it was a friendly?

Arnold: It was a friendly rivalry, surely was. There was contention, but everyone was trying to really do the job. There was a tremendous thrust of human beings to get our boys home and this war over with, so I would say we worked very well with them. Now, when you start to say who gets credit for what and that sort of thing, that’s a different story.

Groueff: I imagine then probably this principle that you described of heavier gas going down and lighter going up and getting smaller and smaller, the principle probably was developed by Dunning and his group, but all the size figures of how much and how many stages and all this was done by Benedict for this particular plant and all the calculations and mathematical problems? But the principle was known already.

Arnold: Well the principle was known even before the Columbia group started it. It’s difficult to explain, but it’s a principle of separation that is used in distillation, for example. Murphree was a world authority on distillation. He was able to describe the cascade and to develop the mathematics of it himself from his knowledge of distillation. It wasn’t really a fundamental contribution—that is, the cascading. The individual stage of separation was.

Groueff: And the difficulty was to work with this terrible gas, because there was another gas that was done already and had been done.

Arnold: Yes.

Groueff: But not with this corrosive—

Arnold: The difficulties with the gas were twofold. One was that the two molecules we were trying to separate were almost identical. One had a molecular weight of 235 and the other 238, and this is a good indication as to how rapidly they move in a gas.

All molecules of a gas bounce back and forth at various energy levels, but the lighter ones bounce back with greater speed than the heavier one, so that if you wanted to separate light hydrogen from methane by this process, then the number of times a hydrogen atom would slip through a hole simply because it hits where the hole is so much faster than methane would be what we would call a very high separation factor. That would take just a few stages of the cascade.

Groueff: Cascades were known, for instance, probably to separate two different gases of a mixture or something like that. Which will be relatively easy.

Arnold: Yeah.

Groueff: That was for the first time what you did to separate isotopes of the same gas?

Arnold: I’m having a little trouble, I think, getting through this principle that the fundamental separation was not as unique as you might think. This wasn’t a great invention.

Groueff: I see.

Arnold: The great invention was to—

Groueff: The principle was known, applied. So actually, what was unique and completely without any precedent was the barrier and the seal for this fantastic precision, and the plating and all the fighting the corrosion. What other problems did you have? The leaking problem, I understand that was fantastic also, and you had to use this Alfred Nier’s spectrometer.

Arnold: Mass spectrometer and helium to test to see whether the joints and—

Groueff: I know more or less what a spectrometer is because they used the principle also in electromagnetic separation, but how do you use the spectrometer for leakage?

Arnold: It’s very simple. If you visualize a beer can, and I want to find out if it is real tight, then I could cut a little hole in the cover and put a pipe in there, and then I could put a vacuum pump and start pulling out all the air out of that can. Now, I put a plastic envelope around the can, and in that plastic envelope I put helium.

Now there was no helium in the air that I am pumping out of that can, so what I pump out I run through a mass spectrometer to see if that gas that is being pumped out contains helium. I am just using the mass spectrometer as an indication as to whether the mass four is present and if it is then it’s leaking. We had certain tolerances for each different size vessel and so on.

Groueff: But how did you do that? It was before the startup of the plant. You would put helium through every pipe and every diffuser? You’d fill it with helium and then you’d put it in a bag or something?

Arnold: Well actually, for the most part we put the helium on the outside between this envelope of plastic.

Groueff: You put envelopes everywhere? That’s a tremendous job. I saw the plant from outside, and it runs for miles. So every single pipe or pump or valve was tested this way? You put the envelope outside and filled between the walls and the bag, you put helium? And then how do you do it physically with the spectrometer? You go from spot to spot?

Arnold: No, well, you have to go along with a probe if there is a leak. Some of it we did massively, and if there are no leaks, we passed them. If there were leaks, then you have to go back sometimes with a little probe and find out where the leak is—just like a little water hose, for example. When you hit the right spot, then you get a deflection on the mass spectrometer.

Groueff: How big were the spectrometers? Portable?

Arnold: Oh yeah, about that big and that high. Whereas a normal mass spectrometer will scan a wide range of masses, this was just designed to detect helium and nothing else. It was a very simple gadget. We had hundreds of them. This testing was done. Some of the components were tested in the manufacturing shops. We would bring them to Oak Ridge and test only the closures.

Groueff: So you connect it with a tube, with a pipe?

Arnold: Everything was welded so that when we got through we had pipes conducing the light gas up to one stage to another. We had pipes conducting the depleted gas from one stage down to the next, so that gas that didn’t flow through the barrier really was coming down the cascade, and the gas that flowed through the barrier was going up the cascade. All of this was welded tight.

Groueff: And vacuum inside?

Arnold: Vacuum inside. No flanges, no air. Moisture was the worst thing that we could have.

Groueff: How do you fight against moisture?

Arnold: Keep the air out; it carries the moisture. In certain instances, we used dry nitrogen to surround some of this equipment so that if there was leakage it was dry.

Groueff: So you pumped the air out of every piece of equipment?

Arnold: Oh yeah, we had to do that first.

Groueff: Was this done at the K-25 building, or it was done first in this conditioning building?

Arnold: No, in the conditioning building. That was to take these surfaces and treat them with fluorine to condition them so that the hexafluoride wouldn’t attack the metal surfaces.

Groueff: I see. And then when they were already conditioned, you would bring them to K-25 and you would assemble them. Then you pull the air out and put all the plastic bags and stuff. That must have been a tremendous job.

Arnold: It was. It was a tremendous job. Now, let me go back. When we were conditioning the equipment, we had to be sure that we didn’t have a lot of moist air in them. We either flushed them out with dry nitrogen or got rid of the moisture before we put the fluorine in because the fluorine and moisture react very rapidly. Prior to that conditioning, we had to clean everything and get all the last bit of oil.

Groueff: How did you clean it?

Arnold: We had all kinds of solvents and Oakite—commercial cleaners.

Groueff: Liquids?

Arnold: Yeah, liquids.

Groueff: So you washed actually the interior of every pipe?

Arnold: To be sure it is free of all foreign matter.

Groueff: By hand?

Arnold: Some by hand and some by pumping. The men in that area generally wore white gloves so that they wouldn’t leave fingerprints on the inside.

Groueff: Where was that made? What building?

Arnold: It was all done in the conditioning building, but many of the components were treated in fabrication plants as well.

Groueff: In K-25?

Arnold: Yeah.

Groueff: That’s what Claude Center from Union Carbide was telling me, how in certain parts, as you said, they had to wear gloves, and everybody had to change clothes and not touch anything with fingers.

Arnold: I guess they still practice that.

Groueff: It’s amazing. He tells me that now they employ much fewer people. Everything is automation.

Arnold: I suppose they still have the bicycles to ride around.

Groueff: Yeah, they have bicycles. Unfortunately, they don’t let anybody in. No visitors, so I couldn’t be admitted. That applies to even high-ranking generals, senators—nobody. I understand they don’t let even allies like English scientists, French scientists; it is very top secret. He tells me that they use much, much fewer people now. When you started it was half automation, no?

Arnold: Yes, indeed it was.

Groueff: For instance, in your experience as an engineer, did you ever have anything similar? Have you seen anything similar?

Arnold: Similar with respect to the controls?

Groueff: Yeah.

Arnold: I would say that fundamentally our then-modern refineries had controls of a similar class, although they were much less elaborate.

Groueff: But the standards were not that high?

Arnold: No, nothing like that.

Groueff: So for an engineer, it was absolutely without precedent?

Arnold: Absolutely.

Groueff: And also the size?

Arnold: The size, of course, was tremendous.

Copyright 1965 Stephane Groueff. From the Stephane Groueff Collection, Howard Gotlieb Archival Research Center at Boston University. Exclusive rights granted to the Atomic Heritage Foundation.