<P> The thermal diffusion process was based on Sydney Chapman and David Enskog's theory, which explained that when a mixed gas passes through a temperature gradient, the heavier one tends to concentrate at the cold end and the lighter one at the warm end . Since hot gases tend to rise and cool ones tend to fall, this can be used as a means of isotope separation . This process was first demonstrated by Klaus Clusius and Gerhard Dickel in Germany in 1938 . It was developed by US Navy scientists, but was not one of the enrichment technologies initially selected for use in the Manhattan Project . This was primarily due to doubts about its technical feasibility, but the inter-service rivalry between the Army and Navy also played a part . </P> <P> The Naval Research Laboratory continued the research under Philip Abelson's direction, but there was little contact with the Manhattan Project until April 1944, when Captain William S. Parsons, the naval officer in charge of ordnance development at Los Alamos, brought Oppenheimer news of encouraging progress in the Navy's experiments on thermal diffusion . Oppenheimer wrote to Groves suggesting that the output of a thermal diffusion plant could be fed into Y - 12 . Groves set up a committee consisting of Warren K. Lewis, Eger Murphree and Richard Tolman to investigate the idea, and they estimated that a thermal diffusion plant costing $3.5 million could enrich 50 kilograms (110 lb) of uranium per week to nearly 0.9% uranium - 235 . Groves approved its construction on 24 June 1944 . </P> <P> Groves contracted with the H.K. Ferguson Company of Cleveland, Ohio, to build the thermal diffusion plant, which was designated S - 50 . Groves's advisers, Karl Cohen and W.I. Thompson from Standard Oil, estimated that it would take six months to build . Groves gave Ferguson just four . Plans called for the installation of 2,142 48 - foot - tall (15 m) diffusion columns arranged in 21 racks . Inside each column were three concentric tubes . Steam, obtained from the nearby K - 25 powerhouse at a pressure of 1,000 pounds per square inch (6,900 kPa) and temperature of 545 ° F (285 ° C), flowed downward through the innermost 1.25 - inch (32 mm) nickel pipe, while water at 155 ° F (68 ° C) flowed upward through the outermost iron pipe . Isotope separation occurred in the uranium hexafluoride gas between the nickel and copper pipes . </P> <P> Work commenced on 9 July 1944, and S - 50 began partial operation in September . Ferguson operated the plant through a subsidiary known as Fercleve . The plant produced just 10.5 pounds (4.8 kg) of 0.852% uranium - 235 in October . Leaks limited production and forced shutdowns over the next few months, but in June 1945 it produced 12,730 pounds (5,770 kg). By March 1945, all 21 production racks were operating . Initially the output of S - 50 was fed into Y - 12, but starting in March 1945 all three enrichment processes were run in series . S - 50 became the first stage, enriching from 0.71% to 0.89% . This material was fed into the gaseous diffusion process in the K - 25 plant, which produced a product enriched to about 23% . This was, in turn, fed into Y - 12, which boosted it to about 89%, sufficient for nuclear weapons . </P>

Where was much of the work that developed the atomic bomb done