For long-term sustainable interplanetary travel, I'd look first for the most widely-available, highly-abundant volatile to use for propellant. What we're seeing from all the planetary science probes is: that answer is "water". It may not be pure, but ice is everywhere, and solid contaminants are really easy to separate with gravity, even if it is artificial (centrifugal force).
So, if we use the water, we also find it is really easy (and safe) to ship anywhere, because we can ship it frozen, with minimal vapor containment to prevent sublimation. An iceberg is not lost if you take a meteor hit; a remarkable safety feature! Just patch the hole in your vapor containment, which is holding at most about 6 mbar pressure. That containment might just be a big plastic bag. We do have to learn how to patch holes in plastic bags in-vacuo and in zero-gee. But that can be done. One way is to use an aluminized thermoplastic, and just heat-bond it with a hand-held “iron”.
OK, so water makes the most practical propellant material we have. So, now how exactly do we use it? Solar or nuclear electrolysis is a known way to produce hydrogen and oxygen, which can be both rocket propellants and fuel cell reactants. We already know we're going to do this at some level, and how to do it. These are known tinkertoys in-hand. But the energy cost to make the hydrogen and oxygen, store them successfully, etc, will never be trivial, no matter how the space commerce economy finally takes shape. You'd really like to use the water directly, as water, and just use your propulsive waste heat to melt the stored ice as you go.
That's a nuclear thermal rocket, a water-variant of the old NERVA. “Everybody” thinks the Isp will be far lower due to the molecular weight effect; and at the same core temperatures, it is. But water-as-steam will transfer a whole lot more heat as a coolant than hydrogen. You can run the reactor core at a higher temperature, and a whopping lot more generated power. I'd bet the as-finished Isp isn't as low as everybody fears, once the development is done. How would you like Isp 600+ seconds at engine T/W > 5? I'd bet real money this could be done, and within about 5 years of having resurrected the original LH2 NERVA, an item that itself should take at most about 5 years, if done by the right team.
The follow-on is a gas-core nuclear thermal rocket, which takes away the core temperature limitation, and eliminates the safety concerns associated with a "live" core inside an engine shell, between burns. Operated up to a modest power level, regenerative cooling is possible. How would you like 2000+ seconds of Isp at engine T/W > 30? Beyond that power level, you need a place to put all the waste heat, so a heavy radiator is probably needed. How would you like Isp 5000+ seconds at engine T/W around 0.1? Even an engine T/W like that is still high enough for an impulsive burn, for low-loss interplanetary travel. I'd bet we could have one of these working within about a decade of having the water-NERVA working.
High-Isp / high thrust nuclear propulsion, with plain water as the propellant. Water, which you can get nearly everywhere you go. Water, which is easy to store and ship as minimally-enclosed icebergs. To me, this sounds like a series of very smart technology development programs we should already be undertaking. Too bad no one is doing this stuff.
To me, it also sounds also like the government monopolies on nuclear power need to be broken, so that somebody who is actually motivated, can go do this stuff. Political anathema to some, I know. But it must be done.
I’d suggest testing these things on the moon, so that a plume capture feature is not needed on your nuclear engine test facility. The resulting facility savings might well exceed (in the long term) the costs of transporting all the necessary stuff to the moon.
This nuclear thermal rocket stuff starts looking really attractive if you use a thorium reactor. Thorium is very likely available "everywhere", too. It's very plentiful here, and on the moon, much more so than uranium. The probes should be looking for it "out there", shouldn't they? Too bad they're not. Not yet, anyway.
Just how sustainable and low-cost do you want to be?