The idea of an "atomic drive" was a science-fiction cliche by the 1930's, but it appears that Stanislaw Ulam and Frederick de Hoffman conducted the first serious investigation of atomic propulsion for space flight in 1944, while they were working on the Manhattan Project (2). During the quarter-century following World War II, the U.S. Atomic Energy Commission (replaced by the Department of Energy in 1974) worked with various federal agencies on a series of nuclear engine projects with names like Dumbo, Kiwi, and Pluto, culminating in NERVA (Nuclear Engine for Rocket Vehicle Application) (3). Close to producing a flight prototype, NERVA was cancelled in 1972 (4). The basic idea behind all these engines was to heat a working fluid by pumping it through a nuclear reactor, then allowing it to expand through a nozzle to develop thrust. Although this sounds simple the engineering problems were horrendous. How good were these designs? A useful figure for comparing rocket engines is specific impulse (Isp), defined as pounds of thrust produced per pound of propellant consumed per second. The units of Isp are thus seconds. The best chemical rocket in service, the cryogenic hydrogen-oxygen engine, has an Isp of about 450 seconds (5). NERVA had an Isp roughly twice as great (6), a surprisingly small figure considering that nuclear fission fuel contains more than a million times as much energy per unit mass as chemical fuel. A major problem is that the reactor operates at a constant temperature, and this temperature must be less than the melting point of its structural materials, about 3000 K (7).

A number of designs were proposed in the late 1940's and 1950's to get around the temperature limitation and to exploit the enormous power of the atomic bomb, estimated to be on the order of 10 billion horsepower for a moderate-sized device (8). The Martin Company designed a nuclear pulse rocket engine with a "combustion chamber" 130 feet in diameter. Small atomic bombs with yields under 0.1 kiloton (a kiloton is the energy equivalent of 1000 tons of the high explosive TNT) would have been dropped into this chamber at a rate of about one per second (9); water would have been injected to serve as propellant. This design produced the relatively small Isp of 1150 seconds, and could have yielded a maximum velocity change for the vehicle of 26,000 feet/second. The vehicle would have been boosted to an altitude of 150 miles by chemical rockets, and the extra 8000 ft/sec or so thus provided would have allowed it to escape the Earth's gravity (10). The Lawrence Livermore Laboratory produced a similar although much smaller design called Helios at about the same time (11).