A now-forgotten reactor design path was the bright idea of General Electric in the late 1940s to build what they called an “intermediate breeder,” cooled by liquid sodium. This is not the time or place to explore the intermediate breeder concept but suffice to say that it didn’t really work, so GE modified what they had to attempt a submarine reactor, at a time when there were no submarine reactors and most didn’t believe there ever would be. Hyman Rickover, in many ways the father of the Light Water Reactor, tried two designs for the very first nuclear subs: that Light Water Reactor and GE’s sodium-cooled SIR (for “Sodium Intermediate Reactor”). The LWR won and the SIR lost, and this paragraph by one of Rickover’s key lieutenants, engineer Ted Rockwell, written four decades later, sums up well why the SIR was fated to lose out:
The SIR system had large steam generators and superheaters, in which hot sodium from the reactor heated up water to make steam to run the turbines. Heat-exchanger tubes of any type sooner or later develop small leaks, which in a pressurized-water reactor result in water leaking into water. But in the SIR, a leak would result in high-pressure steam and water leaking into the sodium, which could cause a disastrous explosion. Even if the leak were a tiny one, the reaction of sodium and water at the leak would produce sodium hydroxide – lye – which would quickly corrode the tiny leak into a large one. To prevent such an occurrence, the SIR heat-exchanger tubes were made double-walled, with mercury in the space between the sodium and the water. In this design, sensitive mercury detectors in the steam and sodium systems could signal the start of a leak before sodium and water contacted each other. However, fabricating such a design was tricky, and the presence of toxic mercury on a submarine would always be a concern, no matter how carefully it was isolated from the atmosphere.
NOTE
Rockwell, Theodore. 1992. The Rickover Effect: How One Man Made a Difference. Naval Institute Press, Annapolis, pp. 210-211