It took me until page 7 of the article before I found anything really relevant, and even given that I was already so many pages into the article it was still fairly fluffy stuff:
They came up with an even more efficient system, in which the core of the rocket was not a huge solid mass of ceramic, but it was a cloud of Uranium HexaFluoride gas. Since the core started out as a cloud of gas, it couldn't melt! Therefore it could get much hotter than a solid core rocket, and would thus be much more efficient.
This idea was dubbed the Gas Core Nuclear Rocket, or GCNR for short.
They built test models of the GCNR many years ago, and discovered a little problem. Since the core was a hot gas, when you pumped the fuel gas through it to get it hot, the radioactive core gas would leak out through the exhaust. This is a real problem. Luckily, they were able to figure out a way to get around this issue.
On page 8, they continue:
In a GCNR, the core is run SO hot, it lights up like a lightbulb, and then gets much, much, much hotter. The energy being given off goes above red hot, even goes above white hot, until the core is blazing away in the deep ultraviolet. Yes, it gets so hot you can't see it any more.
At those huge temperatures, the normally small radiative heat transfer mechanism grows until it is easily big enough to get the energy from the core into the reaction gas all by itself. You no longer need to mix the two gases together, and you can keep them separate. But how can we do that, if the core is so super hot?
The answer is fused silica.
Silica is very transparent to ultraviolet light. If we treat the core like a real lightbulb and put a dome of fused silica glass around it, the glass lets basically all of the ultraviolet energy shine right through. Even though it seems impossible, the smart fellows back in the 70's actually built test models of this type of system and made it work. Given the technology we have today, we can make fused silica of such perfect transparency that this works great.
A GCNR with one of these bulbs in it is called a nuclear lightbulb. With today's technology we can build these pretty easily.
If there are any engineers out there, I'd like your thoughts.
The above strikes me as a bit far-fetched (or, perhaps more fairly, too idealistic and based on rough analogies), but they do get into some descriptions about launching something like this on page 10. It starts out like this:http://members.shaw.ca/bru_b/Liberty_ship_pg10.html
To recap, the efficiency and power of the thruster is based on the difference in temperature between the fissioning mass and the reaction mass. If you run a solid core NTR much above 3000 C, it melts. This provides a firm "ceiling" on how efficient a solid core reactor can be. A gas core design STARTS melted. In addition, since all of the structure of the fuel mass is dynamic, a gas cored reactor is inherently safer than a solid core device. If a "hot spot" develops in a solid core, disaster ensues. If a hot spot develops in a gas core, the hot spot superheats and "puffs" itself out of existence. A gas core reactor is expected to operate at temperatures of 25,000C. The much higher temperature gradient makes the thruster inherently more efficient.
Second, a solid core reactor has a "fixed" core, since it is solid. A gas core reactor does not, and the radioactive fuel is easily "sucked" out of the core and stored in a highly non-critical state completely out of the engine! The fuel storage system I propose is a mass of thick walled boron-aluminum alloy tubing. As I said above, the fuel proper is uranium hexafluoride gas. UF6 is mean stuff, but we have decades of experience handling it in gaseous diffusion plants, and common aluminum and standard seals are available which resist attack from it. It is stoichiometric, fluorine is low activation, and UF6 changes phase at moderate temperatures, allowing it to be converted from high pressure gas to a solid and back again using nothing fancier than gas cooling and electrical heaters.