A note on Argon, nuclear reactors in space and alternative power sources.
IIRC Xenon has been the preferred fuel for ion engines in space. It's *much* easier to store than Helium (the smallest atom) and being larger than Argon *should* be easier to ionize.
However while Argon is likely to be harder to ionize (higher voltage power supply needed) it is also the most *common* noble gas in the atmosphere and perhaps most important of all, the *cheapest* as it's used in welding quite a lot.
Being cheap and readily available (I'd expect large welding supply stores to have at least some in stock) can pay dividends when your research is on a budget.
The problem with nuclear reactors in space (at least those in the *open* literature) is that they just are not *big* or current enough. The last *actual* US orbiting *reactor* was SNAP-10A in 1965, which was good for about 500W.
The proposed SP100 (100Kw) for SDI in the early 80's never got out of design. Los Alamos has been designing something called the SAFE400, designed to deliver 100Kw of electrical power, as a sort of private project of the director, and that's about it in the US (of course the NRO, USAF and USN *could* have one on every big sat they've been orbiting since the 80s and simply not *told* anyone. However the presence of high temperature neutron emitting IR sources *anywhere* in Earth orbit would be a pretty big hint *someone* was using them and the list of countries that could is a short one).
The US bought 6 TOPAZ reactors in the early 90s (and AFAIK still has them) but I don't think they bought any *fuel*. However even if they were good to go they would deliver about 5Kw
Bottom line. No one is going to design, build and qualify a space nuclear reactor in the timescale needed (although a space reactor design in the 100Kw+ range, which gave more power than all but the *biggest* solar arrays, *would* be a good idea to have on the shelf for the future)
Ground based turbo generators in the 200Kw range are used for emergency power and typically fueled by gasoline or natural gas. Assuming sea level air has a density 1.22521 kg/m3. and 20% of that mass is O2 then LOX is 4562x more dense (per cubic metre). It *should* be possible to build a power pack (LOX/fuel tanks, turbo-generator, controls) into a package you could stow on the Shuttle.
While it would be enough to *start* the engine it's *very* doubtful to get the kind of run time that would be needed to show up any flaws that *only* shoe up in long term tests (936 Hrs, possibly with a re-start following flip over at the half way mark if you really want to go to Mars)
BTW the way to avoid the exhaust interfering with the thrust measurements is to fit the exhaust pipe with a T end piece that exhausts in equal and opposite directions, canceling any thrust.
Wiring this thing *directly* to the ISS power system and directly mounting it to the structure is *much* the simplest option.