A better idea, from a poster here:
http://nextbigfuture.com/2014/08/elon-musk-says-that-sustainable-human.html
"I really think that if you have that 10 gigawatt-year (thermal) reactor, it would make more sense to "take the hit" and turn it to electricity, then using a high-amperage ion accelerator, accelerate
caesium or
barium or some other I / II column heavy ion down the track, and out the end. Using superconducting microwave resonators (Q > 10⁹), the exhaust Isp could be well over 1,000,000. Quite easily, in fact."
"Now this wouldn't result in much
quantitative thrust, but it would result in a
lot of kinetic energy and momentum transfer over years of service. Thousands of times more thrust than even
nuclear fired reactions such as you've been imagining."
"A sesame-seed's worth of radium, placed on a pinhead, mounted a few inches between a lead plate and an aluminum foil wire grid (quite airy), all three mounted on a block of teflon or glass, with the lead isolated, but the grill electrically attached to the pin, will do the following. The radium will emit a powerful flux of alpha particles."
"The lead plate will temporarily absorb the ones flying its way. They will charge the lead plate positive. With conservation of charge, the radium + grid becomes quite negative. Over time, the lead plate will become so positive that it will
deflect the alpha particles flying its way. "
"They'll generally be deflected toward the negative charged grill. Thus, viewed well outside the system, there is a
net flux of high-speed alphas heading almost entirely away from the lead plate. Most of the kinetic energy of the spontaneous fission of the radium is converted to kinetic energy without heat loss."
"The amount of thrust is very small, of course. But the efficiency is very high, and the stored fission energy of the radium is maximally utilized. This same reaction can occur for any radionuclide species that emits charged particles. However, ones that generate the heaviest particles will give the best energy-to-monentum transfer."
I think that's the "fission fragment" method, or something like it.
http://en.wikipedia.org/wiki/Fission-fragment_rocket
That's great over a long haul, but it is best to kick it out of Earth Moon with a big high energy cryogenic upper stage, to get the best use away from our gravity well.