Oops, I completely forgot that hydrogen would be solid at 3K...   
      
   At this temperature, heat pumps are too inefficient to be useful. Theoretical   
   max efficiency is Thot/(Thot-Tcold), which for liquid hydrogen is 14/(14-3) =   
   1.28, meaning you produce several more watts of waste heat per watt of heat   
   removed, and that's at    
   the most perfect theoretical performance.   
      
   helium will be neither good enough at absorbing heat per kg, per m^3 nor even   
   that good as a propellant, so it won't cut it by itself.   
   So you will need solid hydrogen (as cold as possible, let's arbitrarily say   
   1K), and a liquid helium loop to keep the surface of the ship at 3K. This   
   complicates the design.   
   On the other hand, wouldn't solid hydrogen be less prone to escape and   
   embrittle everything?   
      
   So as long as we have a (initially) solid heat sink and a liquid helium loop,   
   it is probably not that much more complicated to have one type or another, or   
   even several at the same time.   
   Still, I would stick with hydrogen, at least for the most part, for its   
   capacity to absorb more heat per kg at low temperature, and for its   
   performance as a propellant.   
      
   Hydrogen absorbing more energy per kg is a feature, not a liability: energy is   
   free at as great a quantity as needed with the Sun, thanks to solar-thermal   
   propulsion. If you need more energy for a given dV, simply increase the mirror   
   aperture.   
   This may not hold in the outer reaches of the system, though I am not sure how   
   the design should be adapted for those conditions, and it probably depends on   
   the required mission.   
   But in the inner Solar System, pretty much the only important factor is dV per   
   kg, regardless of how much energy it requires. (As long as the plume itself is   
   not visible, so light sail or photon drive is out)   
      
   Bulk is a drawback, but not such a big one I suspect. By far the most energy   
   received is from the Sun, and this is taken care of with the solar-thermal   
   engine. For the rest, you end up with a long, thin cone, but this craft   
   doesn't have to manoeuvre    
   anyway.   
   A numerical analysis would be necessary to see how this plays out, but with   
   the better dV of hydrogen, I suspect it still holds the advantage.   
      
   As for an insulated heat sink where heat is bottled instead of released, how   
   good would insulation need to be for long-term storage? how good are   
   existing/theoretical insulators? How well can you insulate the coolant loop   
   itself?   
      
   One case where such a heat sink could be useful is, for the brief moment the   
   mission becomes active and you have to start a high-power engine, you want to   
   keep things cold just a little bit longer (until right when the attack is   
   launched).   
   Here, compactness may be useful as it requires less insulation material, but   
   again, I can't tell what would be best.   
      
   > Not likely, unless if you are talking about pre-industrial colonisation. The   
   infrastructure required to make them viable in the first place will also   
   harden them against any point attack. Space habitats especially will be spread   
   over large volumes.   
      
   We are talking about a stealth impactor massing at least several tons   
   (possibly much more) closing in at interplanetary speeds, so probably dozens   
   of km/s. Kinetic energy is going to be counted in kT eq. TNT   
   Being stealthy, the defenders will probably have only a few seconds at best to   
   react, if they detect it at all before impact. So neither time to intercept it   
   nor to brace for impact or evacuate for shelters.   
      
   Unless this is dispersed shirt-sleeve environment (e.g. the countryside   
   surrounding cities), no civilian installation will survive. Space stations and   
   surface domes stand no chance. Even hardened habitats (like a hollowed   
   asteroid) won't be safe: not    
   only can it aim at a weak point, and being a long thin cone it can be built as   
   an armour-piercing projectile, but it can even pack a hydrogen bomb to   
   detonate once the armour is passed (similarly to present-day bunker-busters on   
   a larger scale).   
   I suspect even if Mars has started terraforming, a few well-placed impacts   
   could break those efforts.   
      
   > Again, this is only an issue for pre-industrial colonies. Once you have   
   colonies of sufficient size to build their own military forces, it would be   
   far easier for them to protect their "space" than it will be for Earth to   
   maintain its control.    
      
   The point is precisely that you can't intercept it, as you don't see it coming.   
   Even if other sides can build their own, this wouldn't work as a MAD system,   
   as MAD requires second-strike capability. With this, your opening salvo can   
   destroy the other side's launchers/attempts at building a launcher.   
   The only way to have second-strike is to already have stealth impactors flying   
   around, ready to change course and impact, but they have limited autonomy so   
   you have to keep launching them every few years. It requires several sides to   
   develop the    
   capability at about the same time (or the first one will be able to destroy   
   the other sites before completion), but building one in the first place is   
   probably going to cause a (non-stealthy) war before completion.   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)