Engineering
These minibots need designing as replicable units
If you make a factory unit that churns out minibots then you have to scale the factory unit when you need to scale the minibots. So it's the factory unit which is self-replicating. That seems okay. If the factory unit is made of minibots then that's how the minibots replicate themselves but both aspects need designing. One minibot can't make a new minibot but lots of minibots can combine to make a minibot factory.
Minimizing the number of specialist minibots seems sensible. On the other hand if some are construction parts and some are processing parts and some are bot team long-range communication parts, that could be three specialisms immediately.
The bot factory has to be able to construct logic circuitry. Silicon chips. Capacitor/batteries fed from the external energy feed. Every bot has at least this close-coupling communications chip so they can self-organize.
Every bot has to be mobile. The site bots are in a gravity well, the mirror bots aren't, that's two forms of mobility.
Mining
What the landing fab has to be able to build from scratch is a mining site able to replicate with local materials. The planning model for that will be exciting.
Getting factory units from the Mercury surface site into orbit
We have used the word railgun. It might not be the right word but it can change. We need a power source to put a mirror into Mercury orbit.
- The planet rotation is insignificant
- The package mass is 20 tonnes
- Orbit velocity is 4.1 km/s
KE = ½mv2
½ 2E4kg 1.6E7 m2/s2
is 1.6E11 Joules
Current best supercapacitor delivery is 1E5 J/kg (of capacitor mass)
so the mass of the required supercapacitor is 1.6E6kg to get one package into orbit.
1600 tonnes.
That's not bad at all. 1600 tonnes of energy store to get a 20 tonne factory into Mercury orbit. The rail itself will be a lot more than that. I assume we can pulse the rail so it can have several packages spaced along it at once, and that each zone has several supercapacitors, and that each supercapacitor has a charge-discharge cycle and we can determine how fast it is, which will show how many need to be switched in and out to keep the launch frequency at the rail's optimum. And perhaps the supercapacitor is a buffer and most of the energy can come from the receivers rather than the store.
For reference, scale up by ten to get from escape velocity to solar orbit velocity. I have no idea how much of that can be performed by using the mirror as a sail to get out of Mercury orbit into solar orbit.