> Has anybody thought through the engineering of those silica reaction vessels (I haven't searched)? I don't think it can work.
> If I understand the concept shown in the Wikipedia page, it depends on uranium hexafluoride operating at 25,000°C in fused silica containers which transmit hard UV that heats hydrogen external to the silica containers and is exhausted through nozzles, propelling the vehicle.
> Silica has very poor thermal conductivity, which means it can't be cooled readily by conduction. This means that the inner container wall will melt even if the outer wall coolant is liquid hydrogen because the wall material can't transport the heat flux. One might be able to keep a 10µm silica shell from melting, but nothing thicker. Too thin to be mechanically sound.
The wikipedia article is actually pretty terrible when it comes to the specifics, as if whoever wrote it just read the abstract of one paper (there were at least 50 IIRC). They used fused silica in the early proof of concept (without nuclear fuel) because they were still proving the basics like the irrotational vortex injectors and compression of the core. The technology for making extremely pure single crystal beryllium oxide has since been developed and has found commercial applications, especially in aerospace and semiconductors. Single crystal BeO has far better thermal conductivity and UV transparency, making it possible to cool the reactor walls fast enough.
The bigger problem is neutron bombardment which rapidly degrades the containment chamber's transparency and without replacing the neon with water (impossible), there's no way to counteract that. However, since the reactor is so easy to shut down (stop injecting neon and open the exhaust vents), even weekly or monthly replacements become economically feasible. All of the equipment needed for it are already part of the design (centrifuges for separation of the gas + fuel, modular reactor instead of solid piece of steel/concrete, etc.). The nuclear lightbulb design is unique because all of the other factors that make classic nuclear fission so expensive are eliminated (redundancies of active safety measures, meltdown-proof shielding, etc.) so there's a lot more money to spend on the actual operation and maintenance of the reactor.
Furthermore: the program was canceled before they started the research but they had plans to investigate ways of generating power from the reactor through magnetohydrodynamics instead of turbines/photovoltaics. The whole field was still in its infancy at that point but they had some ideas to test that would have replaced the transparent containment vessel with much stronger (and cheaper) reflective material, eliminating the hardest part of the design.
> Also, the uranium hex will be completely dissociated and ionized at 25K°C. This means the inner silica wall will be exposed to atomic fluorine, and will corrode away immediately. Uranium at 25k°C probably reacts with SiO2, but I don't know that. Uranium silicides and oxides are known compounds, so it's likely. It's a horrific corrosion environment. At temperatures like this, everything reacts with everything else.
The pressure applied on the core is (if I remember correctly) in the hundreds of atmospheres, if not thousands, with an absurd angular momentum. The vast, vast majority of particles that are shaved off of the core cool extremely quickly and are either sucked out or pushed back into the core. Any corrosion feeds into the above maintenance schedule and there was some discussion about seeding the neon with other elements to moderate chemical reactions.
If you can access them, I highly recommend reading at least the big summary paper. All of these problems were discussed, many were even tested, and solutions were found that require technology we have developed by now. All of the problems you brought up were at the top of their minds because they were designing the nuclear lightbulb primarily as a rocket engine where maintenance would have been impossible.